Cytoscape.js GPU benchmarks

CPU	Intel(R) Core(TM) i9-9900K CPU @ 3.60GHz
Cores	8 physical / 16 logical · 3.6 GHz base · 5 GHz max
Memory	62.6 GiB
GPU	CoffeeLake-S GT2 [UHD Graphics 630] i915 Ellesmere [Radeon RX 470/480/570/570X/580/580X/590] (8.0 GiB) amdgpu
OS	Fedora Linux 43 (Workstation Edition) · 7.1.10-100.fc43.x86_64 · x64
Node	v24.18.0 (v8 13.6.233.17-node.50)
Commit	5107c36e bench(113.1): the machine fingerprint rounds RAM to the GiB; the archive re-stamped
Machine id	5cf3f79c

Speedup overview v3 p50 ÷ gpu p50, log scale · right of the 1× line = gpu faster

sweep: edges() @ 2k

284648×

sweep: nodes() @ 2k

171358×

core: edges() @ 2k

91736×

core: nodes() @ 2k

50610×

sweep: elements() @ 2k

29389×

core: elements() @ 2k

20386×

core: graph scratch() set + get (v3 restyles the graph) @ 2k

16479×

core: removeScratch() @ 2k

9509×

compare: same() @ 2k

8408×

core: graph data() set + get (v3 restyles the graph) @ 2k

8387×

trav: successors() @ 2k

740×

trav: predecessors() @ 2k

723×

mut-bulk: hide + show @ 2k

255×

flags: show + hide + visible @ 2k

231×

parent drag (shift subtree + bb settle) @ 2k

182×

compound: commonAncestors @ 2k

128×

degree: nodes.totalDegree() @ 2k

114×

viewport: reset() @ 2k

107×

mut-bulk: shift @ 2k

105×

degree: nodes.maxDegree() @ 2k

97×

flags: panify + unpanify + pannable @ 2k

97×

flags: grabify + ungrabify + grabbable @ 2k

96×

flags: selectify + unselectify + selectable @ 2k

95×

flags: lock + unlock + locked @ 2k

94×

89×

degree: node.degree() @ 2k

87×

core: options() @ 2k

85×

reparent round-trip @ 2k

79×

degree: indegree + outdegree (2 calls) @ 2k

76×

collection: same + anySame + contains (3 calls) @ 2k

73×

algo: closenessCentrality (one root) @ 2k

72×

algo: pageRank (20 iters) @ 500

70×

algo: closenessCentrality (one root) @ 500

65×

algo: dfs (whole graph) @ 2k

61×

dimensions: shift + silentShift (round-trip) @ 2k

60×

sweep: nodes(":selected") vs nodes({selected}) @ 2k

59×

mut-bulk: positions(obj) @ 2k

59×

viewport: viewport({ zoom, pan }) @ 2k

58×

algo: bfs (whole graph) @ 2k

58×

algo: dijkstra + pathTo @ 2k

57×

algo: bfs (whole graph) @ 500

56×

algo: dfs (whole graph) @ 500

56×

mut-bulk: lock + unlock @ 2k

54×

algo: closenessCentrality (weighted, one root) @ 500

52×

child drag (move + parent re-derive) @ 2k

51×

algo: dijkstra + pathTo @ 500

50×

dimensions: positions( fn ) over 100 @ 2k

50×

collection: component() @ 2k

45×

degree: minDegree + maxDegree (2 calls) @ 2k

43×

traverse: components() @ 2k

41×

algo: degreeCentralityNormalized @ 2k

38×

37×

algo: degreeCentralityNormalized @ 500

35×

algo: degreeCentrality (one root) @ 500

35×

algo: degreeCentrality (one root) @ 2k

35×

sweep: selected nodes ($("node:selected") vs filter({group, selected})) @ 2k

35×

algo: tarjan SCC @ 2k

35×

query: structural, parents (:parent vs { parent: true }) @ 2k

34×

query: structural, orphans (:orphan vs { child: false }) @ 2k

33×

bounds: getFitViewport() (compute without committing) @ 2k

33×

bounds: cy.fit() (whole-graph scan) @ 2k

33×

algo: tarjan SCC @ 500

31×

collection: silentPositions( fn ) over 100 @ 2k

31×

30×

collection: difference @ 2k

29×

collection: intersection @ 2k

29×

algo: kruskal @ 2k

28×

load: init from the definition form (construct + dispose) @ 2k

28×

set: difference() @ 2k

27×

set: intersection() @ 2k

27×

scn: select-all + fit (listeners on) @ 2k

26×

degree: minIndegree + maxOutdegree (2 calls) @ 2k

25×

emit: position write, no listeners @ 2k

25×

degree: maxIndegree + minOutdegree (2 calls) @ 2k

25×

algo: hopcroft-tarjan biconnected @ 500

25×

algo: hopcroft-tarjan biconnected @ 2k

25×

algo: kruskal @ 500

25×

mut: position set @ 2k

25×

trav: band roots() @ 2k

24×

core: id lookup ($("#id") vs $id()) @ 2k

24×

mut-bulk: positions(fn) @ 2k

23×

data: bulk write, object column (plain-array fallback) @ 2k

22×

data: bulk write, numeric column @ 2k

21×

algo: bellmanFord @ 500

20×

algo: bellmanFord @ 2k

20×

core: add + remove @ 2k

19×

19×

data: bulk write, one new dictionary entry per pass @ 2k

19×

data: bulk write, dictionary string column (4 values) @ 2k

18×

viewport: center( eles ) @ 2k

17×

scn: drag (select band of 100 + 8-step shift, position listener) @ 2k

17×

flags: active/activate/unactivate (3 calls) @ 2k

16×

query: membership ([bucket in ...] vs { in }) @ 2k

16×

flags: select + unselect + selected (band) @ 2k

16×

collection: absoluteComplement @ 2k

16×

15×

dimensions: silentPosition @ 2k

15×

viewport: minZoom/maxZoom get (2 calls) @ 2k

15×

query: data equality ([bucket = 3] vs { data: { bucket: 3 } }) @ 2k

14×

style: renderedStyle (one prop) @ 2k

14×

mut-bulk: data set @ 2k

14×

dimensions: padding + paddedWidth + paddedHeight (3 calls) @ 2k

13×

query: structural, childless (:childless vs { parent: false }) @ 2k

13×

core: selected ($(":selected") vs filter({selected})) @ 2k

13×

sweep: selected ($(":selected") vs filter({selected})) @ 2k

13×

algo: betweennessCentrality (unweighted) @ 500

13×

traversal: roots + leaves (2 calls) @ 2k

13×

collection: toArray() @ 2k

12×

query: two keys AND-ed @ 2k

12×

algo: fuzzyCMeans @ 500

11×

algo: betweennessCentrality (unweighted) @ 2k

11×

query: data comparison ([weight > 3] vs { gt }) @ 2k

11×

dimensions: relativePosition get @ 2k

10×

sweep: all nodes ($("node") vs filter({group})) @ 2k

9.5×

core: batch + per-id data() (was batchData, round 90) @ 2k

8.9×

collection: removeScratch() @ 2k

8.9×

8.6×

core: removeAllListeners() @ 2k

8.5×

algo: betweennessCentrality (weighted) @ 500

8.5×

algo: hierarchicalClustering (threshold) @ 500

8.3×

data: removeData (whole collection) @ 2k

8.2×

emit: 1 core listener @ 2k

8.2×

mut-bulk: select + unselect @ 2k

8.0×

emit: delegated (v3 selector vs v4 predicate) @ 2k

7.9×

scn: explore (expand 2-hop + select + fit) @ 2k

7.8×

emit: 1 element (ref-qualified) listener @ 2k

7.7×

algo: kMedoids @ 500

7.7×

element data: scratch get + set @ 2k

7.7×

query: predicate function (both sides) @ 2k

7.4×

core: collection() @ 2k

7.2×

7.1×

core: collection() @ 2k

7.0×

compound: orphans + nonorphans (2 calls) @ 2k

7.0×

6.7×

scn: edit (remove + re-add 256 nodes + cascade, add/remove listeners) @ 2k

6.7×

core: batch( fn ) (empty) @ 2k

6.6×

emit: 10 core listeners @ 2k

6.5×

mut-bulk: remove + re-add (256 nodes + cascade) @ 2k

6.4×

bounds: whole-graph boundingBox() @ 2k

6.2×

collection: union @ 2k

5.9×

style: bypass set + removeStyle round-trip @ 2k

5.9×

style: whole-object style() @ 2k

5.9×

core: startBatch + endBatch + batching @ 2k

5.9×

style: sheet swap (compile + applyAll) @ 2k

5.4×

traversal: allAreNeighbors @ 2k

5.3×

set: union() @ 2k

5.2×

algo: hierholzer (directed) @ 500

5.2×

trav: band neighborhood() @ 2k

5.1×

trav: outgoers() @ 2k

5.1×

traverse: incomers() @ 2k

5.0×

traverse: outgoers() @ 2k

5.0×

compare: contains() @ 2k

5.0×

algo: affinityPropagation @ 500

4.9×

core: animated() @ 2k

4.9×

trav: incomers() @ 2k

4.9×

core: nodes ($("node") vs filter({group})) @ 2k

4.8×

algo: kMeans (fixed centroids) @ 500

4.7×

compound: children + descendants + siblings (3 calls) @ 2k

4.7×

viewport: zoom() get @ 2k

4.7×

collection: length + empty + nonempty (3 calls) @ 2k

4.6×

mut: select + unselect @ 2k

4.2×

style: bypass set (ele.style name,value) @ 2k

4.1×

trav: neighborhood() @ 2k

4.0×

viewport: width() + height() + size() (3 calls) @ 2k

3.7×

collection: symmetricDifference @ 2k

3.7×

traverse: neighborhood() @ 2k

3.7×

collection: layoutDimensions() @ 2k

3.6×

compound: parent + parents + ancestors (3 calls) @ 2k

3.6×

algo: aStar @ 500

3.6×

collection: cy() @ 2k

3.5×

traversal: parallelEdges + codirectedEdges (2 calls) @ 2k

3.4×

algo: aStar @ 2k

3.4×

traversal: connectedNodes @ 2k

3.2×

traversal: sources + targets (2 calls) @ 2k

3.2×

dimensions: rendered dims (4 calls) @ 2k

3.1×

collection: byGroup @ 2k

3.1×

collection: diff @ 2k

3.1×

flags: takesUpSpace + interactive + transparent (3 calls) @ 2k

3.1×

anim: start + stop, 512-node collection @ 2k

3.0×

traversal: openNeighborhood + closedNeighborhood (2 calls) @ 2k

3.0×

algo: hierholzer (directed) @ 2k

2.9×

trav: band connectedNodes() @ 2k

2.9×

collection: animated() @ 2k

2.9×

collection: delayAnimation() @ 2k

2.9×

style: numericStyle @ 2k

2.9×

dimensions: width + height + outerWidth + outerHeight (4 calls) @ 2k

2.8×

load: cy.json() @ 2k

2.8×

core: window() @ 2k

2.8×

iter: map() @ 2k

2.8×

traverse: connectedEdges() @ 2k

2.7×

anim: viewport animate + stop @ 2k

2.7×

mut: data set @ 2k

2.7×

traversal: isLoop + isSimple (2 calls) @ 2k

2.7×

core: container() @ 2k

2.6×

trav: closedNeighborhood() @ 2k

2.6×

trav: connectedEdges() @ 2k

2.6×

dimensions: renderedPosition get @ 2k

2.5×

core: on('render') + off('render') (was onRender/offRender, round 90) @ 2k

2.5×

algo: closenessCentralityNormalized @ 500

2.5×

core: resize() / invalidateSize() @ 2k

2.4×

core: zoom set @ 2k

2.3×

flags: effectiveOpacity @ 2k

2.3×

flags: grabbed + hidden + inactive (3 calls) @ 2k

2.2×

collection: is (none match: the full walk) @ 2k

2.2×

core: destroyed() @ 2k

2.2×

core: isReady() @ 2k

2.2×

style: read one prop (width) @ 2k

2.2×

emit: on() + off() round-trip @ 2k

2.2×

collection: slice() @ 2k

2.2×

data: read the whole object @ 2k

2.2×

core: selectionType get @ 2k

2.0×

collection: min + max ( fn ) (2 calls) @ 2k

1.97×

element data: json() one element @ 2k

1.97×

element data: jsons() over 100 @ 2k

1.95×

iter: filter(fn) @ 2k

1.93×

algo: floydWarshall @ 500

1.93×

json: element @ 2k

1.88×

viewport: zoom( n ) set @ 2k

1.87×

scn: refresh (bulk data write + mapped labels + filter + fit, data listener) @ 2k

1.87×

core: gating setters round-trip (autolock/autoungrabify/autounselectify) @ 2k

1.87×

viewport: pan( p ) set @ 2k

1.86×

style: first apply on add (256-node band) @ 2k

1.86×

collection: eq/first/last (3 calls) @ 2k

1.83×

dimensions: renderedBoundingBox @ 2k

1.83×

style( 'text-max-width' ) (was #74 in the switch) @ 2k

1.77×

collection: indexOfId() @ 2k

1.76×

collection: element accessors (id/group/isNode/isEdge, 4 calls) @ 2k

1.74×

style( 'text-wrap' ) (was #73 in the switch) @ 2k

1.69×

viewport: panBy() @ 2k

1.68×

style: read one prop (color) @ 2k

1.67×

viewport: renderedExtent() @ 2k

1.64×

load: add a 256-node band (add + remove) @ 2k

1.59×

collection: some + every (2 calls) @ 2k

1.59×

anim: delay() (no channels touched) @ 2k

1.59×

collection: sort( fn ) @ 2k

1.58×

compound: isParent/isChild/isChildless/isOrphan (4 calls) @ 2k

1.58×

bounds: one node boundingBox() @ 2k

1.58×

position: get @ 2k

1.56×

sweep: filter(fn) @ 2k

1.55×

trav: band sources() @ 2k

1.53×

viewport: extent() @ 2k

1.52×

trav: band connectedEdges() @ 2k

1.52×

collection: allAre (all match: the full walk) @ 2k

1.41×

collection: reduce( fn ) @ 2k

1.40×

collection: indexOf @ 2k

1.38×

data: read one numeric key @ 2k

1.36×

anim: start + stop, one element @ 2k

1.35×

data: read one string key (dictionary decode) @ 2k

1.33×

traversal: source() + target() (2 calls) @ 2k

1.33×

traversal: edgesTo @ 2k

1.31×

style( 'target-distance-from-node' ) (was #150 in the switch) @ 2k

1.31×

core: pan() get (x32) @ 2k

1.30×

traversal: edgesWith @ 2k

1.29×

style( 'taxi-radius' ) (was #142 in the switch) @ 2k

1.28×

core: instanceString + headless + styleEnabled (3 calls) @ 2k

1.23×

style( 'border-width' ) (was #6 in the switch) @ 2k

1.21×

viewport: pan() get @ 2k

1.18×

collection: removed() + inside() (2 calls) @ 2k

1.18×

core: gating getters (5 calls) @ 2k

1.17×

core: hasElementWithId + hasCompoundNodes (2 calls) @ 2k

1.17×

algo: markovClustering @ 500

1.11×

style( 'background-color' ) (was #4 in the switch) @ 2k

1.09×

algo: kargerStein @ 500

1.06×

core: extent() @ 2k

1.05×

data: get @ 2k

1.05×

iter: forEach() @ 2k

1.05×

core: elements() memo hit (was mutableElements, round 90) @ 2k

1.04×

core: getElementById() @ 2k

1.04×

core: filter(fn) @ 2k

0.96×

0.1×1×10×100×1000×10000×100000×1000000×

Scaling speedup vs graph size

algorithms

group	× @ N=500	× @ N=2k
algo: bfs (whole graph)	56×	58×
algo: dfs (whole graph)	56×	61×
algo: dijkstra + pathTo	50×	57×
algo: aStar	3.6×	3.4×
algo: bellmanFord	20×	20×
algo: kruskal	25×	28×
algo: tarjan SCC	31×	35×
algo: hopcroft-tarjan biconnected	25×	25×
algo: hierholzer (directed)	5.2×	2.9×
algo: betweennessCentrality (unweighted)	13×	11×
algo: degreeCentralityNormalized	35×	38×
algo: closenessCentrality (one root)	65×	72×
algo: degreeCentrality (one root)	35×	35×
algo: pageRank (20 iters)	70×	–
algo: closenessCentrality (weighted, one root)	52×	–
algo: floydWarshall	1.93×	–
algo: closenessCentralityNormalized	2.5×	–
algo: markovClustering	1.11×	–
algo: hierarchicalClustering (threshold)	8.3×	–
algo: kMeans (fixed centroids)	4.7×	–
algo: kMedoids	7.7×	–
algo: fuzzyCMeans	11×	–
algo: affinityPropagation	4.9×	–
algo: kargerStein	1.06×	–
algo: betweennessCentrality (weighted)	8.5×	–

core+collection N=2,000 · 2.9 minv4 (gpu) v3

core: getElementById()

1.04×

core: id lookup ($("#id") vs $id())

24×

core: nodes()

50610×

core: edges()

91736×

core: elements()

20386×

core: nodes ($("node") vs filter({group}))

4.8×

core: selected ($(":selected") vs filter({selected}))

13×

core: filter(fn)

0.96×

core: collection()

7.0×

core: pan() get (x32)

1.30×

core: extent()

1.05×

core: add + remove

19×

core: zoom set

2.3×

traverse: neighborhood()

3.7×

traverse: connectedEdges()

2.7×

traverse: outgoers()

5.0×

traverse: incomers()

5.0×

traverse: components()

41×

degree: node.degree()

87×

degree: nodes.totalDegree()

114×

degree: nodes.maxDegree()

97×

data: get

1.05×

position: get

1.56×

json: element

1.88×

set: union()

5.2×

set: intersection()

27×

set: difference()

27×

iter: map()

2.8×

iter: forEach()

1.05×

iter: filter(fn)

1.93×

compare: same()

8408×

compare: contains()

5.0×

mut: data set

2.7×

mut: position set

25×

mut: select + unselect

4.2×

1 ns10 ns100 ns1 µs10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
core: getElementById()	v3	94.8 ns	96.9 ns	120 ns	91.0 ns	1770
core: getElementById()	gpu	91.2 ns	94.3 ns	131 ns	87.7 ns	1816
core: id lookup ($("#id") vs $id())	v3	1.99 µs	2.33 µs	3.69 µs	1.83 µs	68
core: id lookup ($("#id") vs $id())	gpu	84.7 ns	88.0 ns	163 ns	80.7 ns	1943
core: nodes()	v3	415 µs	428 µs	683 µs	404 µs	1635
core: nodes()	gpu	8.21 ns	8.96 ns	19.6 ns	7.79 ns	19000
core: edges()	v3	724 µs	762 µs	1.45 ms	713 µs	917
core: edges()	gpu	7.89 ns	8.41 ns	17.0 ns	7.51 ns	20150
core: elements()	v3	142 µs	148 µs	247 µs	137 µs	4717
core: elements()	gpu	6.99 ns	7.36 ns	14.4 ns	6.54 ns	23103
core: nodes ($("node") vs filter({group}))	v3	550 µs	578 µs	1.27 ms	535 µs	1207
core: nodes ($("node") vs filter({group}))	gpu	114 µs	120 µs	275 µs	85.0 µs	5721
core: selected ($(":selected") vs filter({selected}))	v3	329 µs	343 µs	616 µs	319 µs	1999
core: selected ($(":selected") vs filter({selected}))	gpu	25.1 µs	25.9 µs	27.3 µs	24.8 µs	12
core: filter(fn)	v3	410 µs	422 µs	583 µs	399 µs	1656
core: filter(fn)	gpu	429 µs	458 µs	991 µs	415 µs	1511
core: collection()	v3	456 ns	549 ns	1.56 µs	447 ns	305
core: collection()	gpu	65.2 ns	69.9 ns	283 ns	63.1 ns	2414
core: pan() get (x32)	v3	44.7 ns	45.5 ns	59.0 ns	43.4 ns	3600
core: pan() get (x32)	gpu	34.3 ns	36.2 ns	63.0 ns	32.3 ns	4722
core: extent()	v3	22.7 ns	39.4 ns	134 ns	21.7 ns	4233
core: extent()	gpu	21.5 ns	28.3 ns	110 ns	20.5 ns	5852
core: add + remove	v3	129 µs	135 µs	272 µs	120 µs	5054
core: add + remove	gpu	6.70 µs	8.04 µs	33.3 µs	6.33 µs	80319
core: zoom set	v3	1.36 µs	1.37 µs	1.50 µs	1.24 µs	119
core: zoom set	gpu	586 ns	601 ns	874 ns	561 ns	271
traverse: neighborhood()	v3	3.55 µs	3.87 µs	4.92 µs	3.44 µs	40
traverse: neighborhood()	gpu	960 ns	1.12 µs	2.42 µs	930 ns	146
traverse: connectedEdges()	v3	1.06 µs	1.16 µs	1.96 µs	1.03 µs	140
traverse: connectedEdges()	gpu	386 ns	420 ns	1.36 µs	369 ns	399
traverse: outgoers()	v3	2.43 µs	2.62 µs	3.28 µs	2.36 µs	61
traverse: outgoers()	gpu	488 ns	575 ns	2.05 µs	476 ns	282
traverse: incomers()	v3	2.44 µs	2.62 µs	3.41 µs	2.34 µs	61
traverse: incomers()	gpu	488 ns	614 ns	2.05 µs	473 ns	261
traverse: components()	v3	80.9 ms	82.5 ms	86.0 ms	75.3 ms	12
traverse: components()	gpu	1.97 ms	2.11 ms	3.91 ms	1.91 ms	315
degree: node.degree()	v3	3.61 µs	3.94 µs	5.18 µs	3.40 µs	40
degree: node.degree()	gpu	41.5 ns	42.9 ns	76.9 ns	39.2 ns	3992
degree: nodes.totalDegree()	v3	7.24 ms	8.02 ms	12.9 ms	6.90 ms	82
degree: nodes.totalDegree()	gpu	63.7 µs	66.5 µs	127 µs	59.9 µs	10251
degree: nodes.maxDegree()	v3	7.16 ms	7.95 ms	11.5 ms	6.87 ms	84
degree: nodes.maxDegree()	gpu	73.6 µs	75.2 µs	116 µs	70.0 µs	9311
data: get	v3	30.4 ns	31.5 ns	59.7 ns	28.7 ns	5436
data: get	gpu	29.1 ns	30.6 ns	53.6 ns	27.6 ns	5588
position: get	v3	34.0 ns	36.0 ns	85.8 ns	32.9 ns	4757
position: get	gpu	21.8 ns	25.7 ns	56.4 ns	21.1 ns	6499
json: element	v3	300 ns	392 ns	3.54 µs	292 ns	423
json: element	gpu	160 ns	170 ns	355 ns	153 ns	957
set: union()	v3	94.9 µs	107 µs	170 µs	91.7 µs	6386
set: union()	gpu	18.1 µs	18.5 µs	18.7 µs	17.6 µs	12
set: intersection()	v3	94.0 µs	110 µs	169 µs	90.6 µs	6300
set: intersection()	gpu	3.48 µs	3.52 µs	3.85 µs	3.40 µs	43
set: difference()	v3	93.7 µs	103 µs	143 µs	90.5 µs	6548
set: difference()	gpu	3.44 µs	3.49 µs	4.12 µs	3.35 µs	44
iter: map()	v3	28.3 µs	30.0 µs	55.4 µs	27.2 µs	22973
iter: map()	gpu	10.3 µs	10.3 µs	10.4 µs	10.2 µs	13
iter: forEach()	v3	3.47 µs	3.77 µs	3.86 µs	3.41 µs	36
iter: forEach()	gpu	3.31 µs	3.35 µs	3.60 µs	3.26 µs	45
iter: filter(fn)	v3	246 µs	248 µs	312 µs	236 µs	2824
iter: filter(fn)	gpu	127 µs	132 µs	181 µs	123 µs	5314
compare: same()	v3	124 µs	128 µs	195 µs	116 µs	5484
compare: same()	gpu	14.7 ns	15.4 ns	24.7 ns	14.1 ns	11143
compare: contains()	v3	6.26 µs	6.29 µs	6.50 µs	6.15 µs	24
compare: contains()	gpu	1.26 µs	1.27 µs	1.37 µs	1.21 µs	131
mut: data set	v3	630 ns	657 ns	950 ns	602 ns	251
mut: data set	gpu	231 ns	240 ns	347 ns	225 ns	700
mut: position set	v3	1.21 µs	1.25 µs	1.45 µs	1.16 µs	132
mut: position set	gpu	49.2 ns	52.7 ns	100 ns	47.6 ns	3224
mut: select + unselect	v3	2.61 µs	2.81 µs	3.76 µs	2.47 µs	56
mut: select + unselect	gpu	620 ns	627 ns	730 ns	596 ns	258

materializers N=2,000 · 49.8 sv4 (gpu) v3

sweep: nodes()

171358×

sweep: edges()

284648×

sweep: elements()

29389×

sweep: all nodes ($("node") vs filter({group}))

9.5×

sweep: selected ($(":selected") vs filter({selected}))

13×

sweep: selected nodes ($("node:selected") vs filter({group, selected}))

35×

sweep: nodes(":selected") vs nodes({selected})

59×

sweep: filter(fn)

1.55×

1 ns10 ns100 ns1 µs10 µs100 µs1 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
sweep: nodes()	v3	413 µs	422 µs	496 µs	399 µs	1635
sweep: nodes()	gpu	2.41 ns	2.49 ns	5.60 ns	2.35 ns	66641
sweep: edges()	v3	715 µs	724 µs	862 µs	682 µs	966
sweep: edges()	gpu	2.51 ns	2.62 ns	4.86 ns	2.46 ns	63182
sweep: elements()	v3	141 µs	149 µs	230 µs	136 µs	4690
sweep: elements()	gpu	4.80 ns	4.91 ns	8.90 ns	4.55 ns	34453
sweep: all nodes ($("node") vs filter({group}))	v3	543 µs	567 µs	818 µs	526 µs	1232
sweep: all nodes ($("node") vs filter({group}))	gpu	57.4 µs	72.5 µs	175 µs	53.1 µs	9541
sweep: selected ($(":selected") vs filter({selected}))	v3	329 µs	341 µs	512 µs	322 µs	2051
sweep: selected ($(":selected") vs filter({selected}))	gpu	25.8 µs	26.2 µs	26.6 µs	25.4 µs	12
sweep: selected nodes ($("node:selected") vs filter({group, selected}))	v3	320 µs	328 µs	490 µs	309 µs	2131
sweep: selected nodes ($("node:selected") vs filter({group, selected}))	gpu	9.11 µs	9.16 µs	9.44 µs	8.94 µs	15
sweep: nodes(":selected") vs nodes({selected})	v3	531 µs	551 µs	791 µs	514 µs	1267
sweep: nodes(":selected") vs nodes({selected})	gpu	8.92 µs	8.94 µs	9.22 µs	8.79 µs	15
sweep: filter(fn)	v3	412 µs	419 µs	497 µs	399 µs	1673
sweep: filter(fn)	gpu	265 µs	273 µs	572 µs	235 µs	2550

mutators N=2,000 · 56.0 sv4 (gpu) v3

mut-bulk: select + unselect

8.0×

mut-bulk: hide + show

255×

mut-bulk: lock + unlock

54×

mut-bulk: positions(obj)

59×

mut-bulk: positions(fn)

23×

mut-bulk: shift

105×

mut-bulk: data set

14×

mut-bulk: remove + re-add (256 nodes + cascade)

6.4×

10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
mut-bulk: select + unselect	v3	1.97 ms	2.03 ms	3.23 ms	1.90 ms	328
mut-bulk: select + unselect	gpu	248 µs	257 µs	400 µs	236 µs	2724
mut-bulk: hide + show	v3	54.5 ms	59.3 ms	70.0 ms	52.2 ms	12
mut-bulk: hide + show	gpu	214 µs	224 µs	378 µs	206 µs	3070
mut-bulk: lock + unlock	v3	2.06 ms	2.10 ms	2.60 ms	2.00 ms	326
mut-bulk: lock + unlock	gpu	38.4 µs	38.8 µs	39.3 µs	37.8 µs	12
mut-bulk: positions(obj)	v3	2.16 ms	2.22 ms	3.29 ms	2.08 ms	309
mut-bulk: positions(obj)	gpu	36.8 µs	38.6 µs	74.4 µs	35.1 µs	17621
mut-bulk: positions(fn)	v3	3.40 ms	3.56 ms	5.03 ms	3.32 ms	192
mut-bulk: positions(fn)	gpu	146 µs	152 µs	240 µs	140 µs	4565
mut-bulk: shift	v3	3.41 ms	3.58 ms	5.07 ms	3.29 ms	190
mut-bulk: shift	gpu	32.3 µs	33.7 µs	61.3 µs	30.7 µs	20458
mut-bulk: data set	v3	876 µs	905 µs	1.40 ms	845 µs	766
mut-bulk: data set	gpu	64.7 µs	67.6 µs	132 µs	62.3 µs	9967
mut-bulk: remove + re-add (256 nodes + cascade)	v3	41.2 ms	42.7 ms	46.3 ms	40.0 ms	12
mut-bulk: remove + re-add (256 nodes + cascade)	gpu	6.45 ms	6.59 ms	8.54 ms	5.69 ms	100

scenarios N=2,000 · 37.9 sv4 (gpu) v3

scn: explore (expand 2-hop + select + fit)

7.8×

scn: select-all + fit (listeners on)

26×

scn: drag (select band of 100 + 8-step shift, position listener)

17×

scn: edit (remove + re-add 256 nodes + cascade, add/remove listeners)

6.7×

scn: refresh (bulk data write + mapped labels + filter + fit, data listener)

1.87×

10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
scn: explore (expand 2-hop + select + fit)	v3	288 µs	330 µs	910 µs	267 µs	2107
scn: explore (expand 2-hop + select + fit)	gpu	37.0 µs	42.1 µs	136 µs	33.7 µs	16126
scn: select-all + fit (listeners on)	v3	23.7 ms	23.7 ms	25.6 ms	22.2 ms	24
scn: select-all + fit (listeners on)	gpu	905 µs	953 µs	1.52 ms	856 µs	705
scn: drag (select band of 100 + 8-step shift, position listener)	v3	1.63 ms	1.67 ms	2.14 ms	1.56 ms	392
scn: drag (select band of 100 + 8-step shift, position listener)	gpu	97.7 µs	103 µs	205 µs	94.3 µs	6608
scn: edit (remove + re-add 256 nodes + cascade, add/remove listeners)	v3	45.2 ms	45.9 ms	48.4 ms	43.5 ms	11
scn: edit (remove + re-add 256 nodes + cascade, add/remove listeners)	gpu	6.79 ms	7.08 ms	9.81 ms	6.14 ms	92
scn: refresh (bulk data write + mapped labels + filter + fit, data listener)	v3	7.86 ms	8.43 ms	16.1 ms	7.30 ms	77
scn: refresh (bulk data write + mapped labels + filter + fit, data listener)	gpu	4.21 ms	4.37 ms	7.07 ms	3.80 ms	148

traversal N=2,000 · 2.0 minv4 (gpu) v3

trav: neighborhood()

4.0×

trav: closedNeighborhood()

2.6×

trav: connectedEdges()

2.6×

trav: outgoers()

5.1×

trav: incomers()

4.9×

trav: successors()

740×

trav: predecessors()

723×

trav: band neighborhood()

5.1×

trav: band connectedEdges()

1.52×

trav: band roots()

24×

trav: band sources()

1.53×

trav: band connectedNodes()

2.9×

100 ns1 µs10 µs100 µs1 ms10 ms100 ms1 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
trav: neighborhood()	v3	4.00 µs	4.03 µs	4.87 µs	3.34 µs	38
trav: neighborhood()	gpu	999 ns	993 ns	1.12 µs	873 ns	166
trav: closedNeighborhood()	v3	5.64 µs	5.52 µs	5.93 µs	5.10 µs	27
trav: closedNeighborhood()	gpu	2.15 µs	2.20 µs	2.67 µs	2.12 µs	72
trav: connectedEdges()	v3	995 ns	1.08 µs	1.84 µs	977 ns	155
trav: connectedEdges()	gpu	383 ns	399 ns	661 ns	370 ns	424
trav: outgoers()	v3	2.28 µs	2.40 µs	2.93 µs	2.21 µs	67
trav: outgoers()	gpu	449 ns	528 ns	1.96 µs	428 ns	315
trav: incomers()	v3	2.27 µs	2.40 µs	2.99 µs	2.17 µs	67
trav: incomers()	gpu	468 ns	556 ns	2.11 µs	454 ns	296
trav: successors()	v3	432 ms	435 ms	445 ms	428 ms	12
trav: successors()	gpu	584 µs	618 µs	1.80 ms	569 µs	1111
trav: predecessors()	v3	433 ms	434 ms	446 ms	424 ms	12
trav: predecessors()	gpu	600 µs	627 µs	1.82 ms	574 µs	1082
trav: band neighborhood()	v3	186 µs	198 µs	264 µs	179 µs	3521
trav: band neighborhood()	gpu	36.6 µs	37.3 µs	38.3 µs	36.2 µs	12
trav: band connectedEdges()	v3	24.9 µs	25.2 µs	25.4 µs	24.4 µs	12
trav: band connectedEdges()	gpu	16.4 µs	16.9 µs	18.0 µs	16.1 µs	12
trav: band roots()	v3	108 µs	118 µs	176 µs	103 µs	5869
trav: band roots()	gpu	4.52 µs	4.53 µs	4.69 µs	4.42 µs	34
trav: band sources()	v3	6.65 µs	6.65 µs	6.85 µs	6.42 µs	22
trav: band sources()	gpu	4.36 µs	4.41 µs	4.63 µs	4.26 µs	34
trav: band connectedNodes()	v3	14.3 µs	14.4 µs	14.8 µs	13.9 µs	12
trav: band connectedNodes()	gpu	4.92 µs	4.95 µs	5.13 µs	4.85 µs	30

algorithms N=2,000 · 4.6 minv4 (gpu) v3

algo: bfs (whole graph)

58×

algo: dfs (whole graph)

61×

algo: dijkstra + pathTo

57×

algo: aStar

3.4×

algo: bellmanFord

20×

algo: kruskal

28×

algo: tarjan SCC

35×

algo: hopcroft-tarjan biconnected

25×

algo: hierholzer (directed)

2.9×

algo: betweennessCentrality (unweighted)

11×

algo: degreeCentralityNormalized

38×

algo: closenessCentrality (one root)

72×

algo: degreeCentrality (one root)

35×

100 ns1 µs10 µs100 µs1 ms10 ms100 ms1 s10 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
algo: bfs (whole graph)	v3	51.3 ms	53.3 ms	59.4 ms	49.3 ms	9
algo: bfs (whole graph)	gpu	881 µs	913 µs	1.43 ms	863 µs	761
algo: dfs (whole graph)	v3	50.6 ms	51.5 ms	51.1 ms	50.1 ms	10
algo: dfs (whole graph)	gpu	833 µs	855 µs	1.23 ms	824 µs	816
algo: dijkstra + pathTo	v3	54.0 ms	54.9 ms	57.5 ms	52.9 ms	9
algo: dijkstra + pathTo	gpu	951 µs	970 µs	1.42 ms	918 µs	718
algo: aStar	v3	1.66 ms	1.70 ms	2.24 ms	1.61 ms	408
algo: aStar	gpu	493 µs	506 µs	617 µs	482 µs	1381
algo: bellmanFord	v3	230 ms	233 ms	282 ms	205 ms	12
algo: bellmanFord	gpu	11.7 ms	11.8 ms	12.4 ms	11.6 ms	56
algo: kruskal	v3	13.1 ms	14.2 ms	17.3 ms	12.8 ms	45
algo: kruskal	gpu	472 µs	490 µs	924 µs	465 µs	1426
algo: tarjan SCC	v3	22.8 ms	23.6 ms	26.6 ms	21.5 ms	26
algo: tarjan SCC	gpu	653 µs	794 µs	5.43 ms	602 µs	869
algo: hopcroft-tarjan biconnected	v3	89.5 ms	90.9 ms	93.9 ms	85.2 ms	12
algo: hopcroft-tarjan biconnected	gpu	3.58 ms	3.73 ms	4.84 ms	3.45 ms	182
algo: hierholzer (directed)	v3	23.2 ms	23.9 ms	26.3 ms	22.3 ms	25
algo: hierholzer (directed)	gpu	7.86 ms	8.05 ms	10.0 ms	7.66 ms	78
algo: betweennessCentrality (unweighted)	v3	4.02 s	4.04 s	4.10 s	3.98 s	12
algo: betweennessCentrality (unweighted)	gpu	356 ms	374 ms	431 ms	342 ms	12
algo: degreeCentralityNormalized	v3	14.8 ms	15.8 ms	19.3 ms	14.1 ms	40
algo: degreeCentralityNormalized	gpu	390 µs	402 µs	704 µs	385 µs	1738
algo: closenessCentrality (one root)	v3	60.2 ms	63.8 ms	74.2 ms	58.9 ms	12
algo: closenessCentrality (one root)	gpu	840 µs	872 µs	1.51 ms	827 µs	803
algo: degreeCentrality (one root)	v3	7.61 µs	7.40 µs	7.81 µs	6.79 µs	19
algo: degreeCentrality (one root)	gpu	216 ns	222 ns	312 ns	213 ns	763

algorithms N=500 · 10.2 minv4 (gpu) v3

algo: bfs (whole graph)

56×

algo: dfs (whole graph)

56×

algo: dijkstra + pathTo

50×

algo: aStar

3.6×

algo: bellmanFord

20×

algo: kruskal

25×

algo: tarjan SCC

31×

algo: hopcroft-tarjan biconnected

25×

algo: hierholzer (directed)

5.2×

algo: betweennessCentrality (unweighted)

13×

algo: degreeCentralityNormalized

35×

algo: closenessCentrality (one root)

65×

algo: degreeCentrality (one root)

35×

algo: pageRank (20 iters)

70×

algo: closenessCentrality (weighted, one root)

52×

algo: floydWarshall

1.93×

algo: closenessCentralityNormalized

2.5×

algo: markovClustering

1.11×

algo: hierarchicalClustering (threshold)

8.3×

algo: kMeans (fixed centroids)

4.7×

algo: kMedoids

7.7×

algo: fuzzyCMeans

11×

algo: affinityPropagation

4.9×

algo: kargerStein

1.06×

algo: betweennessCentrality (weighted)

8.5×

100 ns1 µs10 µs100 µs1 ms10 ms100 ms1 s10 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
algo: bfs (whole graph)	v3	12.5 ms	12.9 ms	17.6 ms	10.7 ms	49
algo: bfs (whole graph)	gpu	222 µs	226 µs	297 µs	206 µs	3094
algo: dfs (whole graph)	v3	11.3 ms	12.0 ms	14.6 ms	10.8 ms	54
algo: dfs (whole graph)	gpu	201 µs	207 µs	281 µs	194 µs	3376
algo: dijkstra + pathTo	v3	11.7 ms	12.5 ms	15.7 ms	11.1 ms	52
algo: dijkstra + pathTo	gpu	232 µs	241 µs	346 µs	222 µs	2875
algo: aStar	v3	442 µs	456 µs	744 µs	424 µs	1536
algo: aStar	gpu	124 µs	129 µs	179 µs	119 µs	5416
algo: bellmanFord	v3	16.8 ms	16.9 ms	18.1 ms	16.4 ms	38
algo: bellmanFord	gpu	823 µs	845 µs	1.92 ms	816 µs	823
algo: kruskal	v3	3.18 ms	3.46 ms	6.88 ms	3.15 ms	198
algo: kruskal	gpu	128 µs	136 µs	258 µs	124 µs	5122
algo: tarjan SCC	v3	5.29 ms	5.73 ms	9.07 ms	5.24 ms	117
algo: tarjan SCC	gpu	169 µs	176 µs	379 µs	149 µs	3893
algo: hopcroft-tarjan biconnected	v3	20.0 ms	20.9 ms	23.4 ms	19.0 ms	30
algo: hopcroft-tarjan biconnected	gpu	801 µs	827 µs	1.13 ms	768 µs	842
algo: hierholzer (directed)	v3	4.26 ms	4.49 ms	7.44 ms	4.09 ms	149
algo: hierholzer (directed)	gpu	819 µs	836 µs	1.12 ms	786 µs	813
algo: betweennessCentrality (unweighted)	v3	254 ms	256 ms	257 ms	251 ms	12
algo: betweennessCentrality (unweighted)	gpu	20.3 ms	20.3 ms	21.0 ms	20.0 ms	30
algo: degreeCentralityNormalized	v3	3.45 ms	3.79 ms	7.44 ms	3.33 ms	181
algo: degreeCentralityNormalized	gpu	97.3 µs	99.1 µs	137 µs	95.5 µs	7040
algo: closenessCentrality (one root)	v3	13.3 ms	14.2 ms	18.0 ms	12.8 ms	45
algo: closenessCentrality (one root)	gpu	203 µs	208 µs	267 µs	197 µs	3362
algo: degreeCentrality (one root)	v3	7.50 µs	7.32 µs	7.80 µs	6.67 µs	20
algo: degreeCentrality (one root)	gpu	213 ns	218 ns	326 ns	202 ns	781
algo: pageRank (20 iters)	v3	2.74 ms	2.85 ms	6.09 ms	2.68 ms	241
algo: pageRank (20 iters)	gpu	39.4 µs	44.9 µs	155 µs	38.5 µs	15405
algo: closenessCentrality (weighted, one root)	v3	13.4 ms	14.1 ms	17.2 ms	12.9 ms	45
algo: closenessCentrality (weighted, one root)	gpu	256 µs	262 µs	341 µs	247 µs	2672
algo: floydWarshall	v3	245 ms	270 ms	377 ms	243 ms	12
algo: floydWarshall	gpu	127 ms	127 ms	130 ms	124 ms	12
algo: closenessCentralityNormalized	v3	274 ms	273 ms	276 ms	268 ms	12
algo: closenessCentralityNormalized	gpu	112 ms	115 ms	130 ms	110 ms	12
algo: markovClustering	v3	4.14 s	4.36 s	4.63 s	4.07 s	12
algo: markovClustering	gpu	3.74 s	3.74 s	3.76 s	3.70 s	12
algo: hierarchicalClustering (threshold)	v3	34.7 ms	34.9 ms	35.9 ms	34.2 ms	16
algo: hierarchicalClustering (threshold)	gpu	4.19 ms	4.40 ms	6.52 ms	4.04 ms	154
algo: kMeans (fixed centroids)	v3	9.06 ms	9.23 ms	10.9 ms	8.83 ms	72
algo: kMeans (fixed centroids)	gpu	1.92 ms	1.96 ms	3.06 ms	1.86 ms	354
algo: kMedoids	v3	312 ms	276 ms	403 ms	108 ms	12
algo: kMedoids	gpu	40.5 ms	36.5 ms	43.3 ms	20.0 ms	16
algo: fuzzyCMeans	v3	77.9 ms	79.4 ms	82.0 ms	77.0 ms	12
algo: fuzzyCMeans	gpu	6.79 ms	6.83 ms	7.36 ms	6.53 ms	98
algo: affinityPropagation	v3	208 ms	211 ms	222 ms	198 ms	12
algo: affinityPropagation	gpu	42.2 ms	43.8 ms	45.6 ms	39.5 ms	9
algo: kargerStein	v3	321 ms	324 ms	338 ms	313 ms	12
algo: kargerStein	gpu	303 ms	305 ms	324 ms	293 ms	12
algo: betweennessCentrality (weighted)	v3	705 ms	710 ms	719 ms	698 ms	12
algo: betweennessCentrality (weighted)	gpu	82.9 ms	84.4 ms	85.6 ms	81.6 ms	12

mappers N=2,000 · 24.0 s

mapper: bulk write, color — cpu eval

1.31 ms

mapper: bulk write, color — gpu path

175 µs

mapper: bulk write, color + size — cpu eval

1.56 ms

mapper: bulk write, color + size — gpu color / cpu size

1.47 ms

mapper: bulk write, mapped labels (round-5 path) — label refresh

3.95 ms

100 µs1 ms10 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
mapper: bulk write, color	cpu eval	1.31 ms	1.36 ms	2.10 ms	1.28 ms	509
mapper: bulk write, color	gpu path	175 µs	185 µs	350 µs	172 µs	3793
mapper: bulk write, color + size	cpu eval	1.56 ms	1.60 ms	2.27 ms	1.52 ms	434
mapper: bulk write, color + size	gpu color / cpu size	1.47 ms	1.50 ms	1.94 ms	1.43 ms	456
mapper: bulk write, mapped labels (round-5 path)	label refresh	3.95 ms	4.20 ms	6.84 ms	3.41 ms	161

layouts N=2,000 · 2.1 minv4 (gpu) v3

89×

19×

37×

6.7×

15×

30×

8.6×

2.63 ms

8.09 ms

168 ms

10.3 ms

117 ms

115 ms

110 ms

layout: force seed — spectral vs scatter (20 iterations) — spectral (landmark MDS, the default)

115 ms

56.5 ms

454 µs

31.9 µs

535 ns

layout: contract fixed cost (empty impl vs bulk placement) — wrapper + bulk write

40.1 µs

7.1×

100 ns1 µs10 µs100 µs1 ms10 ms100 ms1 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
layout: grid	v3	11.3 ms	12.0 ms	15.2 ms	10.2 ms	53
layout: grid	gpu	127 µs	136 µs	512 µs	78.6 µs	4833
layout: preset	v3	5.62 ms	5.92 ms	8.99 ms	5.34 ms	108
layout: preset	gpu	295 µs	324 µs	691 µs	284 µs	2139
layout: preset (fn form)	v3	5.55 ms	5.83 ms	9.15 ms	5.28 ms	114
layout: preset (fn form)	gpu	150 µs	157 µs	289 µs	144 µs	4229
layout: circle	v3	9.52 ms	10.1 ms	14.9 ms	8.99 ms	64
layout: circle	gpu	1.42 ms	1.57 ms	4.06 ms	1.35 ms	435
layout: concentric	v3	31.7 ms	31.6 ms	36.1 ms	27.4 ms	18
layout: concentric	gpu	2.05 ms	2.31 ms	5.47 ms	2.00 ms	291
layout: breadthfirst	v3	387 ms	396 ms	433 ms	378 ms	12
layout: breadthfirst	gpu	13.0 ms	13.7 ms	19.1 ms	12.3 ms	45
layout: random	v3	5.57 ms	5.86 ms	9.32 ms	5.40 ms	114
layout: random	gpu	646 µs	702 µs	1.54 ms	628 µs	981
layout: radial vs breadthfirst-circle (unbalanced tree)	radial	2.63 ms	2.92 ms	5.88 ms	2.50 ms	233
layout: radial vs breadthfirst-circle (unbalanced tree)	breadthfirst circle	8.09 ms	8.51 ms	17.5 ms	7.72 ms	76
layout: flow vs breadthfirst (staged DAG)	flow	168 ms	181 ms	210 ms	165 ms	12
layout: flow vs breadthfirst (staged DAG)	breadthfirst	10.3 ms	10.5 ms	12.2 ms	9.96 ms	58
layout: force (CPU executor, 20 iterations)	gpu	117 ms	120 ms	123 ms	111 ms	12
layout: force constrained vs plain (CPU, 20 iterations)	constrained	115 ms	116 ms	127 ms	110 ms	12
layout: force constrained vs plain (CPU, 20 iterations)	plain	110 ms	115 ms	134 ms	101 ms	12
layout: force seed — spectral vs scatter (20 iterations)	spectral (landmark MDS, the default)	115 ms	115 ms	122 ms	111 ms	12
layout: force seed — spectral vs scatter (20 iterations)	scatter	56.5 ms	56.6 ms	57.0 ms	56.1 ms	9
layout: plumbing (layoutPositions vs setPositions)	finisher	454 µs	497 µs	1.06 ms	438 µs	1395
layout: plumbing (layoutPositions vs setPositions)	bulk	31.9 µs	40.8 µs	188 µs	29.9 µs	15982
layout: contract fixed cost (empty impl vs bulk placement)	wrapper only	535 ns	981 ns	3.58 µs	516 ns	159
layout: contract fixed cost (empty impl vs bulk placement)	wrapper + bulk write	40.1 µs	41.6 µs	45.1 µs	39.1 µs	12
layout: scope (eles.layout, 10% of nodes)	v3	987 µs	1.06 ms	2.39 ms	923 µs	574
layout: scope (eles.layout, 10% of nodes)	gpu	139 µs	155 µs	327 µs	130 µs	4412

style N=2,000 · 1.7 minv4 (gpu) v3

style: sheet swap (compile + applyAll)

5.4×

style: compile vs compile + apply — compile only (in batch)

24.4 µs

style: compile vs compile + apply — compile + apply

2.30 ms

style: first apply on add (256-node band)

1.86×

style: applyAll, flat vs compound (100 parents — the 14.6 partition) — flat

2.29 ms

style: applyAll, flat vs compound (100 parents — the 14.6 partition) — compound

2.79 ms

style: applyAll — constant vs state-case (the 57.1d partition) — constant sheet

2.32 ms

style: applyAll — constant vs state-case (the 57.1d partition) — state-case sheet

2.30 ms

style: applyAll — state-case vs data-case (what the partition avoids) — state-case sheet

2.29 ms

style: applyAll — state-case vs data-case (what the partition avoids) — data-case sheet

10.2 ms

style: select+unselect 256-band — constant vs state-case sheet — constant sheet (skips restyle)

6.75 µs

style: select+unselect 256-band — constant vs state-case sheet — paint-only state sheet (round-61 diff path)

33.8 µs

style: select+unselect 256-band — constant vs state-case sheet — state-case sheet (all channels; full write)

383 µs

style: read one prop (color)

1.67×

style: read one prop (width)

2.2×

style: numericStyle

2.9×

style: renderedStyle (one prop)

14×

style: whole-object style()

5.9×

style: bypass set (ele.style name,value)

4.1×

style: bypass set + removeStyle round-trip

5.9×

style: select+unselect 256-band under the default sheet — bypass punch-out — no bypasses (diff path; the O(0) gate row)

34.4 µs

style: select+unselect 256-band under the default sheet — bypass punch-out — 128 of 256 bypassed (punched out to merged writes)

226 µs

style: sheet swap (applyAll) — 0 vs 128 bypasses — no bypasses

2.45 ms

style: sheet swap (applyAll) — 0 vs 128 bypasses — 128 bypassed

8.83 ms

10 ns100 ns1 µs10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
style: sheet swap (compile + applyAll)	v3	12.5 ms	12.8 ms	15.8 ms	11.8 ms	49
style: sheet swap (compile + applyAll)	gpu	2.30 ms	2.34 ms	2.92 ms	2.21 ms	285
style: compile vs compile + apply	compile only (in batch)	24.4 µs	26.7 µs	26.1 µs	23.8 µs	12
style: compile vs compile + apply	compile + apply	2.30 ms	2.32 ms	2.82 ms	2.21 ms	286
style: first apply on add (256-node band)	v3	1.34 ms	1.66 ms	6.85 ms	1.13 ms	404
style: first apply on add (256-node band)	gpu	721 µs	830 µs	2.44 ms	675 µs	825
style: applyAll, flat vs compound (100 parents — the 14.6 partition)	flat	2.29 ms	2.35 ms	3.08 ms	2.24 ms	294
style: applyAll, flat vs compound (100 parents — the 14.6 partition)	compound	2.79 ms	2.92 ms	4.63 ms	2.73 ms	236
style: applyAll — constant vs state-case (the 57.1d partition)	constant sheet	2.32 ms	2.40 ms	3.18 ms	2.20 ms	289
style: applyAll — constant vs state-case (the 57.1d partition)	state-case sheet	2.30 ms	2.32 ms	2.57 ms	2.23 ms	283
style: applyAll — state-case vs data-case (what the partition avoids)	state-case sheet	2.29 ms	2.33 ms	2.77 ms	2.21 ms	298
style: applyAll — state-case vs data-case (what the partition avoids)	data-case sheet	10.2 ms	10.4 ms	11.4 ms	9.89 ms	60
style: select+unselect 256-band — constant vs state-case sheet	constant sheet (skips restyle)	6.75 µs	6.79 µs	6.98 µs	6.66 µs	20
style: select+unselect 256-band — constant vs state-case sheet	paint-only state sheet (round-61 diff path)	33.8 µs	35.0 µs	63.7 µs	32.5 µs	18830
style: select+unselect 256-band — constant vs state-case sheet	state-case sheet (all channels; full write)	383 µs	396 µs	564 µs	371 µs	1724
style: read one prop (color)	v3	95.0 ns	108 ns	641 ns	82.0 ns	1992887
style: read one prop (color)	gpu	56.7 ns	62.3 ns	125 ns	54.9 ns	2740
style: read one prop (width)	v3	114 ns	123 ns	431 ns	104 ns	1940202
style: read one prop (width)	gpu	51.9 ns	55.9 ns	305 ns	50.0 ns	3052
style: numericStyle	v3	141 ns	135 ns	386 ns	86.0 ns	1768034
style: numericStyle	gpu	49.3 ns	51.6 ns	110 ns	47.5 ns	3332
style: renderedStyle (one prop)	v3	879 ns	941 ns	1.94 µs	842 ns	561104
style: renderedStyle (one prop)	gpu	64.0 ns	68.0 ns	179 ns	61.4 ns	2513
style: whole-object style()	v3	84.5 µs	89.5 µs	181 µs	81.8 µs	7785
style: whole-object style()	gpu	14.3 µs	15.9 µs	52.4 µs	13.5 µs	42749
style: bypass set (ele.style name,value)	v3	10.1 µs	10.1 µs	10.4 µs	9.81 µs	11
style: bypass set (ele.style name,value)	gpu	2.45 µs	2.70 µs	4.00 µs	2.32 µs	58
style: bypass set + removeStyle round-trip	v3	20.5 µs	21.8 µs	21.8 µs	20.0 µs	12
style: bypass set + removeStyle round-trip	gpu	3.46 µs	3.64 µs	4.44 µs	3.31 µs	42
style: select+unselect 256-band under the default sheet — bypass punch-out	no bypasses (diff path; the O(0) gate row)	34.4 µs	36.1 µs	68.9 µs	33.3 µs	19159
style: select+unselect 256-band under the default sheet — bypass punch-out	128 of 256 bypassed (punched out to merged writes)	226 µs	236 µs	388 µs	218 µs	2948
style: sheet swap (applyAll) — 0 vs 128 bypasses	no bypasses	2.45 ms	2.47 ms	2.74 ms	2.37 ms	264
style: sheet swap (applyAll) — 0 vs 128 bypasses	128 bypassed	8.83 ms	8.98 ms	10.8 ms	8.64 ms	74

style-bundle N=2,000 · 58.5 sv4 (gpu) v3

style( 'background-color' ) (was #4 in the switch)

1.09×

style( 'border-width' ) (was #6 in the switch)

1.21×

style( 'text-wrap' ) (was #73 in the switch)

1.69×

style( 'text-max-width' ) (was #74 in the switch)

1.77×

style( 'taxi-radius' ) (was #142 in the switch)

1.28×

style( 'target-distance-from-node' ) (was #150 in the switch)

1.31×

whole-object style() (gpu-only: v3 returns a different shape) — gpu node

13.3 µs

whole-object style() (gpu-only: v3 returns a different shape) — gpu edge

14.1 µs

colour-valued vs numeric reads (the string build) — gpu background-color (colour)

76.0 ns

colour-valued vs numeric reads (the string build) — gpu border-color (colour)

67.0 ns

colour-valued vs numeric reads (the string build) — gpu border-width (number)

60.9 ns

colour-valued vs numeric reads (the string build) — gpu width (number)

67.3 ns

the getters that ride readProp — gpu numericStyle( width )

67.9 ns

the getters that ride readProp — gpu effectiveOpacity()

20.7 ns

the getters that ride readProp — gpu renderedStyle( width )

86.8 ns

10 ns100 ns1 µs10 µs100 µs

All stats (table view)

group	bench	p50	avg	p99	min	samples
style( 'background-color' ) (was #4 in the switch)	v3	60.8 ns	62.4 ns	88.6 ns	57.0 ns	2747
style( 'background-color' ) (was #4 in the switch)	gpu	55.9 ns	58.2 ns	94.8 ns	53.3 ns	2940
style( 'border-width' ) (was #6 in the switch)	v3	60.4 ns	61.1 ns	72.8 ns	57.8 ns	2810
style( 'border-width' ) (was #6 in the switch)	gpu	49.8 ns	51.0 ns	65.2 ns	47.6 ns	3352
style( 'text-wrap' ) (was #73 in the switch)	v3	60.8 ns	62.1 ns	76.1 ns	59.3 ns	2755
style( 'text-wrap' ) (was #73 in the switch)	gpu	36.0 ns	37.3 ns	61.7 ns	34.6 ns	4603
style( 'text-max-width' ) (was #74 in the switch)	v3	60.5 ns	63.1 ns	113 ns	59.5 ns	2720
style( 'text-max-width' ) (was #74 in the switch)	gpu	34.2 ns	34.9 ns	47.4 ns	33.2 ns	4917
style( 'taxi-radius' ) (was #142 in the switch)	v3	62.7 ns	63.5 ns	74.3 ns	60.4 ns	2707
style( 'taxi-radius' ) (was #142 in the switch)	gpu	48.9 ns	52.5 ns	80.7 ns	47.4 ns	3244
style( 'target-distance-from-node' ) (was #150 in the switch)	v3	62.7 ns	63.8 ns	78.4 ns	60.5 ns	2691
style( 'target-distance-from-node' ) (was #150 in the switch)	gpu	48.0 ns	49.6 ns	81.0 ns	45.9 ns	3448
whole-object style() (gpu-only: v3 returns a different shape)	gpu node	13.3 µs	13.8 µs	26.2 µs	12.8 µs	48745
whole-object style() (gpu-only: v3 returns a different shape)	gpu edge	14.1 µs	15.2 µs	33.5 µs	13.5 µs	44287
colour-valued vs numeric reads (the string build)	gpu background-color (colour)	76.0 ns	78.3 ns	106 ns	73.7 ns	2184
colour-valued vs numeric reads (the string build)	gpu border-color (colour)	67.0 ns	69.3 ns	101 ns	65.3 ns	2471
colour-valued vs numeric reads (the string build)	gpu border-width (number)	60.9 ns	62.9 ns	91.0 ns	59.1 ns	2718
colour-valued vs numeric reads (the string build)	gpu width (number)	67.3 ns	69.6 ns	130 ns	65.5 ns	2460
the getters that ride readProp	gpu numericStyle( width )	67.9 ns	69.7 ns	96.5 ns	66.2 ns	2457
the getters that ride readProp	gpu effectiveOpacity()	20.7 ns	21.0 ns	30.3 ns	19.8 ns	8130
the getters that ride readProp	gpu renderedStyle( width )	86.8 ns	90.0 ns	124 ns	83.9 ns	1899

load N=2,000 · 58.6 sv4 (gpu) v3

load: init from the definition form (construct + dispose)

28×

load: v4 ingest forms (construct + dispose) — definition form

4.35 ms

load: v4 ingest forms (construct + dispose) — columnar form

3.28 ms

load: v4 ingest forms (construct + dispose) — wire buffer

2.88 ms

load: v4 ingest forms (construct + dispose) — def clone alone (control)

301 µs

load: conversion — toColumnarElements

692 µs

load: conversion — serializeElements (from columnar)

179 µs

load: conversion — serializeElements (from defs)

875 µs

load: conversion — deserializeElements

4.45 µs

load: export from a live graph — cy.serialize() (wire)

500 µs

load: export from a live graph — cy.json() (v4)

728 µs

load: cy.json()

2.8×

load: add a 256-node band (add + remove)

1.59×

load: add a 256-node band, v4 forms — definition form

2.22 ms

load: add a 256-node band, v4 forms — columnar form

1.89 ms

load: add a 256-node band, v4 forms — wire buffer

1.91 ms

1 µs10 µs100 µs1 ms10 ms100 ms1 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
load: init from the definition form (construct + dispose)	v3	135 ms	147 ms	173 ms	129 ms	12
load: init from the definition form (construct + dispose)	gpu	4.89 ms	6.39 ms	18.9 ms	4.24 ms	100
load: v4 ingest forms (construct + dispose)	definition form	4.35 ms	6.50 ms	16.6 ms	4.11 ms	102
load: v4 ingest forms (construct + dispose)	columnar form	3.28 ms	4.44 ms	13.2 ms	3.17 ms	151
load: v4 ingest forms (construct + dispose)	wire buffer	2.88 ms	4.24 ms	13.5 ms	2.75 ms	160
load: v4 ingest forms (construct + dispose)	def clone alone (control)	301 µs	627 µs	16.9 ms	209 µs	1063
load: conversion	toColumnarElements	692 µs	730 µs	1.14 ms	661 µs	949
load: conversion	serializeElements (from columnar)	179 µs	362 µs	957 µs	172 µs	1924
load: conversion	serializeElements (from defs)	875 µs	892 µs	1.32 ms	836 µs	779
load: conversion	deserializeElements	4.45 µs	4.30 µs	5.85 µs	3.25 µs	31
load: export from a live graph	cy.serialize() (wire)	500 µs	576 µs	2.04 ms	485 µs	1193
load: export from a live graph	cy.json() (v4)	728 µs	1.04 ms	6.66 ms	685 µs	653
load: cy.json()	v3	2.02 ms	2.40 ms	6.22 ms	1.93 ms	281
load: cy.json()	gpu	722 µs	812 µs	3.14 ms	696 µs	854
load: add a 256-node band (add + remove)	v3	3.60 ms	4.32 ms	10.6 ms	3.17 ms	152
load: add a 256-node band (add + remove)	gpu	2.26 ms	2.59 ms	7.77 ms	1.53 ms	260
load: add a 256-node band, v4 forms	definition form	2.22 ms	2.23 ms	4.85 ms	1.49 ms	306
load: add a 256-node band, v4 forms	columnar form	1.89 ms	1.93 ms	3.55 ms	1.30 ms	349
load: add a 256-node band, v4 forms	wire buffer	1.91 ms	1.96 ms	3.52 ms	1.29 ms	349

spatial N=2,000 · 2.0 minv4 (gpu) v3

pick: hit (the walk; one inside-test) — zoom 1

21.0 µs

pick: hit (the walk; one inside-test) — zoom 2

21.0 µs

pick: miss walk — exact vs 24 px touch halo (57.9) — exact

46.4 µs

pick: miss walk — exact vs 24 px touch halo (57.9) — padPx 24

44.4 µs

pick: shape inside-test, 2000 coincident misses — ellipse

87.2 µs

pick: shape inside-test, 2000 coincident misses — star

234 µs

pick: shape inside-test, 2000 coincident misses — round-hexagon

796 µs

pick: shape inside-test, 2000 coincident misses — barrel

696 µs

pick: shape inside-test, 2000 coincident misses — cut-rectangle

99.3 µs

pick: shape inside-test, 2000 coincident misses — polygon

211 µs

pick: round-hexagon inside-test by zoom (not affine-invariant) — zoom 1

796 µs

pick: round-hexagon inside-test by zoom (not affine-invariant) — zoom 2

802 µs

pick: background miss (full highWater walk) — gpu

44.9 µs

pick: label box in the scan (text-events) — text-events: no

44.9 µs

pick: label box in the scan (text-events) — text-events: yes

125 µs

box: elementsInBox by covered fraction — 10%

150 µs

box: elementsInBox by covered fraction — 50%

196 µs

box: elementsInBox by covered fraction — 100%

325 µs

box: label containment (boxSelectionIncludesLabels) — off

229 µs

box: label containment (boxSelectionIncludesLabels) — on

264 µs

box: overlap vs contain (straight edges) — contain

239 µs

box: overlap vs contain (straight edges) — overlap

481 µs

box: overlap vs contain (curved edges — the segment walk) — contain

1.25 ms

box: overlap vs contain (curved edges — the segment walk) — overlap

1.21 ms

bounds: whole-graph boundingBox()

6.2×

bounds: boundingBox, labels in vs out (v4) — includeLabels: true (default)

1.25 ms

bounds: boundingBox, labels in vs out (v4) — includeLabels: false

657 µs

bounds: cy.fit() (whole-graph scan)

33×

bounds: getFitViewport() (compute without committing)

33×

bounds: boundingBoxAt (v4 only) — gpu

1.49 ms

bounds: one node boundingBox()

1.58×

bounds: labelBoundingBox (v4 only) — gpu

450 ns

bounds: compound + taxi exact scan (rounds 54/92, v4 only) — warm (curve bbs memoized)

19.3 µs

bounds: compound + taxi exact scan (rounds 54/92, v4 only) — cold (geometry write per call)

207 µs

bounds: straight-only scan (ndex-shaped, round 92, v4 only) — warm

34.2 µs

bounds: straight-only scan (ndex-shaped, round 92, v4 only) — cold (geometry write per call)

34.4 µs

100 ns1 µs10 µs100 µs1 ms10 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
pick: hit (the walk; one inside-test)	zoom 1	21.0 µs	21.5 µs	39.2 µs	20.1 µs	31904
pick: hit (the walk; one inside-test)	zoom 2	21.0 µs	21.6 µs	22.9 µs	20.8 µs	12
pick: miss walk — exact vs 24 px touch halo (57.9)	exact	46.4 µs	46.6 µs	47.3 µs	45.1 µs	12
pick: miss walk — exact vs 24 px touch halo (57.9)	padPx 24	44.4 µs	45.6 µs	83.6 µs	43.0 µs	15258
pick: shape inside-test, 2000 coincident misses	ellipse	87.2 µs	89.5 µs	134 µs	83.2 µs	7802
pick: shape inside-test, 2000 coincident misses	star	234 µs	247 µs	391 µs	224 µs	2832
pick: shape inside-test, 2000 coincident misses	round-hexagon	796 µs	808 µs	949 µs	776 µs	866
pick: shape inside-test, 2000 coincident misses	barrel	696 µs	713 µs	987 µs	675 µs	979
pick: shape inside-test, 2000 coincident misses	cut-rectangle	99.3 µs	102 µs	154 µs	95.6 µs	6888
pick: shape inside-test, 2000 coincident misses	polygon	211 µs	215 µs	282 µs	203 µs	3253
pick: round-hexagon inside-test by zoom (not affine-invariant)	zoom 1	796 µs	814 µs	1.02 ms	783 µs	859
pick: round-hexagon inside-test by zoom (not affine-invariant)	zoom 2	802 µs	821 µs	1.16 ms	777 µs	852
pick: background miss (full highWater walk)	gpu	44.9 µs	45.4 µs	46.0 µs	44.3 µs	12
pick: label box in the scan (text-events)	text-events: no	44.9 µs	45.1 µs	45.8 µs	44.3 µs	12
pick: label box in the scan (text-events)	text-events: yes	125 µs	128 µs	202 µs	120 µs	5424
box: elementsInBox by covered fraction	10%	150 µs	157 µs	294 µs	145 µs	4470
box: elementsInBox by covered fraction	50%	196 µs	201 µs	269 µs	191 µs	3462
box: elementsInBox by covered fraction	100%	325 µs	338 µs	791 µs	308 µs	2065
box: label containment (boxSelectionIncludesLabels)	off	229 µs	239 µs	423 µs	221 µs	2930
box: label containment (boxSelectionIncludesLabels)	on	264 µs	273 µs	580 µs	252 µs	2557
box: overlap vs contain (straight edges)	contain	239 µs	247 µs	462 µs	228 µs	2826
box: overlap vs contain (straight edges)	overlap	481 µs	488 µs	802 µs	425 µs	1430
box: overlap vs contain (curved edges — the segment walk)	contain	1.25 ms	1.28 ms	1.66 ms	1.24 ms	545
box: overlap vs contain (curved edges — the segment walk)	overlap	1.21 ms	1.26 ms	1.94 ms	1.17 ms	554
bounds: whole-graph boundingBox()	v3	7.66 ms	7.92 ms	10.1 ms	7.44 ms	82
bounds: whole-graph boundingBox()	gpu	1.24 ms	1.26 ms	1.63 ms	1.21 ms	549
bounds: boundingBox, labels in vs out (v4)	includeLabels: true (default)	1.25 ms	1.28 ms	1.96 ms	1.21 ms	543
bounds: boundingBox, labels in vs out (v4)	includeLabels: false	657 µs	672 µs	971 µs	639 µs	1041
bounds: cy.fit() (whole-graph scan)	v3	7.55 ms	7.64 ms	8.56 ms	7.35 ms	88
bounds: cy.fit() (whole-graph scan)	gpu	231 µs	241 µs	319 µs	226 µs	2896
bounds: getFitViewport() (compute without committing)	v3	7.54 ms	7.71 ms	9.00 ms	7.23 ms	87
bounds: getFitViewport() (compute without committing)	gpu	231 µs	239 µs	303 µs	225 µs	2915
bounds: boundingBoxAt (v4 only)	gpu	1.49 ms	1.57 ms	2.71 ms	1.42 ms	434
bounds: one node boundingBox()	v3	1.63 µs	1.67 µs	2.31 µs	1.58 µs	98
bounds: one node boundingBox()	gpu	1.03 µs	1.04 µs	1.18 µs	994 ns	157
bounds: labelBoundingBox (v4 only)	gpu	450 ns	473 ns	891 ns	422 ns	354
bounds: compound + taxi exact scan (rounds 54/92, v4 only)	warm (curve bbs memoized)	19.3 µs	20.1 µs	43.0 µs	18.5 µs	34084
bounds: compound + taxi exact scan (rounds 54/92, v4 only)	cold (geometry write per call)	207 µs	214 µs	326 µs	200 µs	3260
bounds: straight-only scan (ndex-shaped, round 92, v4 only)	warm	34.2 µs	34.7 µs	36.0 µs	33.8 µs	12
bounds: straight-only scan (ndex-shaped, round 92, v4 only)	cold (geometry write per call)	34.4 µs	34.8 µs	35.4 µs	33.9 µs	12

data N=2,000 · 1.4 minv4 (gpu) v3

data: bulk write, numeric column

21×

data: bulk write, dictionary string column (4 values)

18×

data: bulk write, one new dictionary entry per pass

19×

data: bulk write, object column (plain-array fallback)

22×

data: removeData (whole collection)

8.2×

data: read one numeric key

1.36×

data: read one string key (dictionary decode)

1.33×

data: read the whole object

2.2×

query: data equality ([bucket = 3] vs { data: { bucket: 3 } })

14×

query: data comparison ([weight > 3] vs { gt })

11×

query: two keys AND-ed

12×

query: membership ([bucket in ...] vs { in })

16×

query: predicate function (both sides)

7.4×

query: structural, parents (:parent vs { parent: true })

34×

query: structural, childless (:childless vs { parent: false })

13×

query: structural, orphans (:orphan vs { child: false })

33×

10 ns100 ns1 µs10 µs100 µs1 ms10 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
data: bulk write, numeric column	v3	1.81 ms	1.86 ms	3.03 ms	1.69 ms	359
data: bulk write, numeric column	gpu	84.9 µs	89.7 µs	158 µs	81.3 µs	7797
data: bulk write, dictionary string column (4 values)	v3	1.80 ms	1.84 ms	2.43 ms	1.73 ms	368
data: bulk write, dictionary string column (4 values)	gpu	102 µs	107 µs	309 µs	93.3 µs	6534
data: bulk write, one new dictionary entry per pass	v3	1.84 ms	1.91 ms	3.24 ms	1.72 ms	358
data: bulk write, one new dictionary entry per pass	gpu	98.8 µs	104 µs	186 µs	94.5 µs	6747
data: bulk write, object column (plain-array fallback)	v3	1.81 ms	1.90 ms	3.05 ms	1.75 ms	363
data: bulk write, object column (plain-array fallback)	gpu	83.3 µs	85.8 µs	130 µs	79.0 µs	8146
data: removeData (whole collection)	v3	2.76 ms	2.95 ms	5.22 ms	2.64 ms	232
data: removeData (whole collection)	gpu	335 µs	348 µs	601 µs	326 µs	1965
data: read one numeric key	v3	38.3 ns	40.0 ns	53.6 ns	36.6 ns	4291
data: read one numeric key	gpu	28.2 ns	29.7 ns	41.7 ns	27.5 ns	5745
data: read one string key (dictionary decode)	v3	41.4 ns	43.0 ns	82.5 ns	39.7 ns	3991
data: read one string key (dictionary decode)	gpu	31.1 ns	31.8 ns	71.8 ns	26.7 ns	5381
data: read the whole object	v3	45.6 ns	46.8 ns	67.8 ns	43.5 ns	3676
data: read the whole object	gpu	21.0 ns	22.5 ns	43.4 ns	19.9 ns	7575
query: data equality ([bucket = 3] vs { data: { bucket: 3 } })	v3	664 µs	687 µs	1.04 ms	630 µs	1016
query: data equality ([bucket = 3] vs { data: { bucket: 3 } })	gpu	47.6 µs	51.6 µs	127 µs	41.3 µs	13381
query: data comparison ([weight > 3] vs { gt })	v3	762 µs	786 µs	1.17 ms	731 µs	888
query: data comparison ([weight > 3] vs { gt })	gpu	70.1 µs	73.6 µs	174 µs	57.0 µs	9439
query: two keys AND-ed	v3	801 µs	823 µs	1.07 ms	790 µs	847
query: two keys AND-ed	gpu	67.1 µs	70.4 µs	129 µs	64.4 µs	9892
query: membership ([bucket in ...] vs { in })	v3	1.02 ms	1.04 ms	1.38 ms	979 µs	673
query: membership ([bucket in ...] vs { in })	gpu	64.8 µs	71.5 µs	189 µs	61.6 µs	9479
query: predicate function (both sides)	v3	672 µs	687 µs	988 µs	655 µs	1013
query: predicate function (both sides)	gpu	90.3 µs	97.6 µs	168 µs	86.1 µs	7146
query: structural, parents (:parent vs { parent: true })	v3	526 µs	535 µs	634 µs	514 µs	1306
query: structural, parents (:parent vs { parent: true })	gpu	15.6 µs	17.7 µs	23.3 µs	15.1 µs	12
query: structural, childless (:childless vs { parent: false })	v3	793 µs	809 µs	988 µs	774 µs	862
query: structural, childless (:childless vs { parent: false })	gpu	60.4 µs	80.0 µs	284 µs	56.6 µs	8604
query: structural, orphans (:orphan vs { child: false })	v3	504 µs	524 µs	984 µs	489 µs	1335
query: structural, orphans (:orphan vs { child: false })	gpu	15.3 µs	15.8 µs	15.5 µs	15.1 µs	12

events N=2,000 · 1.3 minv4 (gpu) v3

emit: position write, no listeners

25×

emit: 1 core listener

8.2×

emit: 1 element (ref-qualified) listener

7.7×

emit: delegated (v3 selector vs v4 predicate)

7.9×

emit: 10 core listeners

6.5×

emit: on() + off() round-trip

2.2×

emit: flat vs phased (compound bubbling, 14.5) — orphan (flat emit)

300 ns

emit: flat vs phased (compound bubbling, 14.5) — child of 2 ancestors (phased)

877 ns

emit: flat vs phased, no listeners — orphan (flat)

87.5 ns

emit: flat vs phased, no listeners — child (phased)

513 ns

anim: start + stop, one element

1.35×

anim: start against a running animation (round-21 eviction) — disjoint channels (composes)

1.31 µs

anim: start against a running animation (round-21 eviction) — overlapping channels (evicts)

831 ns

anim: start + stop, 512-node collection

3.0×

anim: delay() (no channels touched)

1.59×

anim: controls (v4 only) — pause + resume

466 ns

anim: controls (v4 only) — reverse

877 ns

anim: viewport animate + stop

2.7×

10 ns100 ns1 µs10 µs100 µs1 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
emit: position write, no listeners	v3	1.31 µs	1.33 µs	1.60 µs	1.27 µs	114
emit: position write, no listeners	gpu	51.6 ns	54.5 ns	117 ns	49.6 ns	3132
emit: 1 core listener	v3	1.40 µs	1.41 µs	1.57 µs	1.35 µs	109
emit: 1 core listener	gpu	170 ns	191 ns	429 ns	165 ns	835
emit: 1 element (ref-qualified) listener	v3	1.40 µs	1.42 µs	1.58 µs	1.36 µs	111
emit: 1 element (ref-qualified) listener	gpu	182 ns	186 ns	260 ns	177 ns	848
emit: delegated (v3 selector vs v4 predicate)	v3	1.46 µs	1.55 µs	2.56 µs	1.40 µs	102
emit: delegated (v3 selector vs v4 predicate)	gpu	185 ns	195 ns	357 ns	178 ns	819
emit: 10 core listeners	v3	1.97 µs	1.99 µs	2.15 µs	1.88 µs	77
emit: 10 core listeners	gpu	304 ns	338 ns	681 ns	302 ns	487
emit: on() + off() round-trip	v3	462 ns	565 ns	1.44 µs	423 ns	289
emit: on() + off() round-trip	gpu	210 ns	256 ns	921 ns	180 ns	635
emit: flat vs phased (compound bubbling, 14.5)	orphan (flat emit)	300 ns	349 ns	2.08 µs	279 ns	965040
emit: flat vs phased (compound bubbling, 14.5)	child of 2 ancestors (phased)	877 ns	932 ns	1.26 µs	852 ns	169
emit: flat vs phased, no listeners	orphan (flat)	87.5 ns	97.3 ns	182 ns	86.0 ns	1698
emit: flat vs phased, no listeners	child (phased)	513 ns	545 ns	1.68 µs	494 ns	305
anim: start + stop, one element	v3	848 ns	1.25 µs	3.23 µs	804 ns	410765
anim: start + stop, one element	gpu	630 ns	1.06 µs	4.66 µs	592 ns	457809
anim: start against a running animation (round-21 eviction)	disjoint channels (composes)	1.31 µs	2.02 µs	4.05 µs	1.10 µs	75
anim: start against a running animation (round-21 eviction)	overlapping channels (evicts)	831 ns	1.43 µs	3.57 µs	805 ns	106
anim: start + stop, 512-node collection	v3	297 µs	419 µs	1.07 ms	286 µs	1527
anim: start + stop, 512-node collection	gpu	99.5 µs	111 µs	159 µs	94.8 µs	5782
anim: delay() (no channels touched)	v3	653 ns	955 ns	3.21 µs	638 ns	168
anim: delay() (no channels touched)	gpu	412 ns	719 ns	2.68 µs	399 ns	226
anim: controls (v4 only)	pause + resume	466 ns	740 ns	2.59 µs	444 ns	207
anim: controls (v4 only)	reverse	877 ns	1.52 µs	3.58 µs	847 ns	100
anim: viewport animate + stop	v3	553 ns	843 ns	3.04 µs	540 ns	192
anim: viewport animate + stop	gpu	203 ns	233 ns	1.38 µs	196 ns	716

store N=2,000 · 1.0 min

id index: set x 2000 (build) — gpu

262 µs

id index: single-key operations — get (probe)

66.6 ns

id index: single-key operations — has

51.0 ns

id index: single-key operations — idAt (memoized hit)

8.99 ns

id index: single-key operations — hashAt

7.02 ns

id index: remove + re-set (churn, drives the round-11 reclaim) — gpu

232 ns

adjacency: rebuild (4000 edges, two counting passes) — gpu

39.0 µs

adjacency: reads — outDegree (O(1))

6.11 ns

adjacency: reads — outEdges

45.1 ns

adjacency: reads — connectedEdges

119 ns

adjacency: addEdge + removeEdge (the overlay path) — gpu

140 ns

blob: write x 2000 (build) — gpu

103 µs

blob: single-record operations — rewrite in place (same length)

29.8 ns

blob: single-record operations — offsetOf

4.24 ns

blob: single-record operations — free + rewrite

151 ns

dirty: mark (per-write span coalescing) — contiguous marks

12.8 ns

dirty: mark (per-write span coalescing) — scattered marks

17.7 ns

dirty: take (a frame's drain) — gpu

590 ns

images: registry bookkeeping — acquire (existing url)

206 ns

images: registry bookkeeping — acquire (new url) + release

533 ns

charts: sheet apply, chart vs no chart — no chart

2.26 ms

charts: sheet apply, chart vs no chart — pie from data (blob record per node)

2.69 ms

charts: data write refreshing chart-values — gpu

1.48 ms

1 ns10 ns100 ns1 µs10 µs100 µs1 ms10 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
id index: set x 2000 (build)	gpu	262 µs	272 µs	409 µs	245 µs	2527
id index: single-key operations	get (probe)	66.6 ns	66.5 ns	126 ns	54.6 ns	2566
id index: single-key operations	has	51.0 ns	52.8 ns	81.0 ns	49.4 ns	3240
id index: single-key operations	idAt (memoized hit)	8.99 ns	9.22 ns	17.5 ns	8.53 ns	18401
id index: single-key operations	hashAt	7.02 ns	7.17 ns	12.3 ns	6.80 ns	23643
id index: remove + re-set (churn, drives the round-11 reclaim)	gpu	232 ns	235 ns	294 ns	222 ns	714
adjacency: rebuild (4000 edges, two counting passes)	gpu	39.0 µs	45.8 µs	193 µs	36.7 µs	15151
adjacency: reads	outDegree (O(1))	6.11 ns	6.27 ns	11.3 ns	5.78 ns	26946
adjacency: reads	outEdges	45.1 ns	46.3 ns	69.0 ns	44.1 ns	3691
adjacency: reads	connectedEdges	119 ns	179 ns	2.08 µs	111 ns	941
adjacency: addEdge + removeEdge (the overlay path)	gpu	140 ns	144 ns	203 ns	135 ns	1188
blob: write x 2000 (build)	gpu	103 µs	124 µs	638 µs	99.6 µs	5646
blob: single-record operations	rewrite in place (same length)	29.8 ns	30.3 ns	40.9 ns	28.8 ns	5628
blob: single-record operations	offsetOf	4.24 ns	4.39 ns	9.14 ns	4.10 ns	38343
blob: single-record operations	free + rewrite	151 ns	171 ns	501 ns	145 ns	999
dirty: mark (per-write span coalescing)	contiguous marks	12.8 ns	13.2 ns	22.5 ns	12.2 ns	12893
dirty: mark (per-write span coalescing)	scattered marks	17.7 ns	18.3 ns	30.0 ns	17.1 ns	9299
dirty: take (a frame's drain)	gpu	590 ns	690 ns	1.24 µs	582 ns	239
images: registry bookkeeping	acquire (existing url)	206 ns	211 ns	265 ns	200 ns	808
images: registry bookkeeping	acquire (new url) + release	533 ns	606 ns	2.83 µs	521 ns	272
charts: sheet apply, chart vs no chart	no chart	2.26 ms	2.30 ms	3.11 ms	2.20 ms	296
charts: sheet apply, chart vs no chart	pie from data (blob record per node)	2.69 ms	2.73 ms	3.18 ms	2.61 ms	251
charts: data write refreshing chart-values	gpu	1.48 ms	1.52 ms	1.98 ms	1.42 ms	458

surface N=2,000 · 8.8 minv4 (gpu) v3

viewport: zoom() get

4.7×

viewport: zoom( n ) set

1.87×

viewport: pan() get

1.18×

viewport: pan( p ) set

1.86×

viewport: panBy()

1.68×

viewport: extent()

1.52×

viewport: renderedExtent()

1.64×

viewport: width() + height() + size() (3 calls)

3.7×

viewport: minZoom/maxZoom get (2 calls)

15×

viewport: center( eles )

17×

viewport: reset()

107×

viewport: viewport({ zoom, pan })

58×

viewport: getCenterPan( eles ) — gpu

6.60 µs

core: instanceString + headless + styleEnabled (3 calls)

1.23×

core: hasElementWithId + hasCompoundNodes (2 calls)

1.17×

core: options()

85×

core: elements() memo hit (was mutableElements, round 90)

1.04×

core: gating getters (5 calls)

1.17×

core: gating setters round-trip (autolock/autoungrabify/autounselectify)

1.87×

core: interaction tuning getters (5 calls) — gpu

7.30 ns

core: graph data() set + get (v3 restyles the graph)

8387×

core: graph scratch() set + get (v3 restyles the graph)

16479×

core: batch( fn ) (empty)

6.6×

core: startBatch + endBatch + batching

5.9×

core: collection()

7.2×

core: compact() (no dead slots — the early-out) — gpu

16.7 µs

collection: element accessors (id/group/isNode/isEdge, 4 calls)

1.74×

collection: length + empty + nonempty (3 calls)

4.6×

collection: eq/first/last (3 calls)

1.83×

collection: slice()

2.2×

collection: toArray()

12×

collection: sort( fn )

1.58×

collection: reduce( fn )

1.40×

collection: min + max ( fn ) (2 calls)

1.97×

collection: some + every (2 calls)

1.59×

collection: same + anySame + contains (3 calls)

73×

collection: allAre (all match: the full walk)

1.41×

collection: is (none match: the full walk)

2.2×

collection: union

5.9×

collection: intersection

29×

collection: difference

29×

collection: symmetricDifference

3.7×

collection: absoluteComplement

16×

collection: byGroup

3.1×

collection: diff

3.1×

collection: indexOf

1.38×

traversal: openNeighborhood + closedNeighborhood (2 calls)

3.0×

traversal: connectedNodes

3.2×

traversal: sources + targets (2 calls)

3.2×

traversal: source() + target() (2 calls)

1.33×

traversal: edgesWith

1.29×

traversal: edgesTo

1.31×

traversal: parallelEdges + codirectedEdges (2 calls)

3.4×

traversal: roots + leaves (2 calls)

13×

traversal: allAreNeighbors

5.3×

traversal: isLoop + isSimple (2 calls)

2.7×

degree: indegree + outdegree (2 calls)

76×

degree: minDegree + maxDegree (2 calls)

43×

degree: minIndegree + maxOutdegree (2 calls)

25×

dimensions: renderedPosition get

2.5×

dimensions: relativePosition get

10×

dimensions: width + height + outerWidth + outerHeight (4 calls)

2.8×

dimensions: rendered dims (4 calls)

3.1×

dimensions: renderedBoundingBox

1.83×

dimensions: midpoint + renderedMidpoint (2 calls) — gpu

1.17 µs

dimensions: source/targetEndpoint (2 calls) — gpu

380 ns

dimensions: rendered source/targetEndpoint (2 calls) — gpu

389 ns

dimensions: shift + silentShift (round-trip)

60×

dimensions: silentPosition

15×

dimensions: positions( fn ) over 100

50×

dimensions: padding + paddedWidth + paddedHeight (3 calls)

13×

curves: isBundledBezier — gpu

20.6 ns

curves: controlPoints + renderedControlPoints (2 calls) — gpu

527 ns

curves: segmentPoints (undefined for bezier) — gpu

33.4 ns

compound: parent + parents + ancestors (3 calls)

3.6×

compound: children + descendants + siblings (3 calls)

4.7×

compound: orphans + nonorphans (2 calls)

7.0×

compound: commonAncestors

128×

compound: isParent/isChild/isChildless/isOrphan (4 calls)

1.58×

flags: select + unselect + selected (band)

16×

flags: selectify + unselectify + selectable

95×

flags: lock + unlock + locked

94×

flags: grabify + ungrabify + grabbable

96×

flags: show + hide + visible

231×

flags: active/activate/unactivate (3 calls)

16×

flags: panify + unpanify + pannable

97×

flags: takesUpSpace + interactive + transparent (3 calls)

3.1×

flags: effectiveOpacity

2.3×

element data: scratch get + set

7.7×

element data: json() one element

1.97×

element data: jsons() over 100

1.95×

element data: label() — gpu

15.2 ns

core: batch + per-id data() (was batchData, round 90)

8.9×

core: animated()

4.9×

core: zoomRange get — gpu

2.16 ns

core: resize() / invalidateSize()

2.4×

core: on('render') + off('render') (was onRender/offRender, round 90)

2.5×

core: removeScratch()

9509×

core: selectionType get

2.0×

core: isReady()

2.2×

core: window()

2.8×

core: container()

2.6×

core: destroyed()

2.2×

core: removeAllListeners()

8.5×

collection: cy()

3.5×

collection: indexOfId()

1.76×

collection: removed() + inside() (2 calls)

1.18×

collection: silentPositions( fn ) over 100

31×

collection: animated()

2.9×

collection: delayAnimation()

2.9×

collection: removeScratch()

8.9×

collection: component()

45×

collection: layoutDimensions()

3.6×

degree: maxIndegree + minOutdegree (2 calls)

25×

flags: grabbed + hidden + inactive (3 calls)

2.2×

curves: renderedSegmentPoints — gpu

35.9 ns

layout contract: edgeSlots + endpoints + degreeOf — gpu

34.7 µs

formats: isColumnarElements + isSerializedElements + isPackedIds — gpu

4.61 ns

0.100 ns1 ns10 ns100 ns1 µs10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
viewport: zoom() get	v3	5.61 ns	5.72 ns	10.2 ns	5.45 ns	29426
viewport: zoom() get	gpu	1.21 ns	1.24 ns	1.77 ns	1.18 ns	127419
viewport: zoom( n ) set	v3	615 ns	622 ns	746 ns	588 ns	267
viewport: zoom( n ) set	gpu	329 ns	338 ns	518 ns	310 ns	498
viewport: pan() get	v3	5.75 ns	6.63 ns	14.2 ns	5.38 ns	25344
viewport: pan() get	gpu	4.86 ns	5.06 ns	11.3 ns	4.76 ns	33457
viewport: pan( p ) set	v3	607 ns	626 ns	1.00 µs	582 ns	269
viewport: pan( p ) set	gpu	326 ns	333 ns	397 ns	316 ns	507
viewport: panBy()	v3	587 ns	593 ns	688 ns	568 ns	284
viewport: panBy()	gpu	348 ns	353 ns	537 ns	319 ns	471
viewport: extent()	v3	29.4 ns	45.1 ns	219 ns	28.0 ns	3736
viewport: extent()	gpu	19.3 ns	26.4 ns	108 ns	18.1 ns	6352
viewport: renderedExtent()	v3	27.1 ns	28.2 ns	50.1 ns	25.8 ns	6001
viewport: renderedExtent()	gpu	16.5 ns	23.1 ns	97.8 ns	15.1 ns	7325
viewport: width() + height() + size() (3 calls)	v3	20.8 ns	21.7 ns	43.0 ns	20.0 ns	7819
viewport: width() + height() + size() (3 calls)	gpu	5.58 ns	5.92 ns	12.4 ns	5.33 ns	28407
viewport: minZoom/maxZoom get (2 calls)	v3	14.7 ns	15.3 ns	26.6 ns	14.0 ns	11122
viewport: minZoom/maxZoom get (2 calls)	gpu	0.998 ns	1.03 ns	1.45 ns	0.977 ns	152604
viewport: center( eles )	v3	117 µs	122 µs	240 µs	110 µs	5684
viewport: center( eles )	gpu	6.68 µs	6.67 µs	6.75 µs	6.57 µs	21
viewport: reset()	v3	935 ns	1.02 µs	1.58 µs	910 ns	163
viewport: reset()	gpu	8.76 ns	9.07 ns	15.8 ns	8.43 ns	18636
viewport: viewport({ zoom, pan })	v3	920 ns	1.00 µs	1.54 µs	880 ns	164
viewport: viewport({ zoom, pan })	gpu	15.8 ns	16.6 ns	37.1 ns	15.0 ns	10241
viewport: getCenterPan( eles )	gpu	6.60 µs	6.60 µs	6.70 µs	6.48 µs	22
core: instanceString + headless + styleEnabled (3 calls)	v3	32.2 ns	33.2 ns	48.2 ns	30.9 ns	5133
core: instanceString + headless + styleEnabled (3 calls)	gpu	26.3 ns	27.0 ns	37.8 ns	25.4 ns	6322
core: hasElementWithId + hasCompoundNodes (2 calls)	v3	38.4 ns	41.6 ns	66.1 ns	37.5 ns	4117
core: hasElementWithId + hasCompoundNodes (2 calls)	gpu	32.8 ns	34.0 ns	60.4 ns	32.0 ns	5020
core: options()	v3	90.9 ns	93.8 ns	155 ns	88.1 ns	1821
core: options()	gpu	1.07 ns	1.18 ns	2.20 ns	1.05 ns	135670
core: elements() memo hit (was mutableElements, round 90)	v3	2.40 ns	2.48 ns	4.62 ns	2.35 ns	66654
core: elements() memo hit (was mutableElements, round 90)	gpu	2.30 ns	2.38 ns	4.07 ns	2.20 ns	69479
core: gating getters (5 calls)	v3	15.6 ns	16.3 ns	31.3 ns	15.0 ns	10505
core: gating getters (5 calls)	gpu	13.3 ns	13.7 ns	23.3 ns	12.3 ns	12456
core: gating setters round-trip (autolock/autoungrabify/autounselectify)	v3	11.2 ns	11.7 ns	22.8 ns	10.6 ns	14632
core: gating setters round-trip (autolock/autoungrabify/autounselectify)	gpu	6.00 ns	6.20 ns	13.2 ns	5.73 ns	27343
core: interaction tuning getters (5 calls)	gpu	7.30 ns	7.47 ns	11.5 ns	7.14 ns	22719
core: graph data() set + get (v3 restyles the graph)	v3	2.76 ms	2.80 ms	3.14 ms	2.67 ms	246
core: graph data() set + get (v3 restyles the graph)	gpu	329 ns	336 ns	491 ns	315 ns	497
core: graph scratch() set + get (v3 restyles the graph)	v3	2.79 ms	2.83 ms	3.22 ms	2.71 ms	243
core: graph scratch() set + get (v3 restyles the graph)	gpu	169 ns	172 ns	277 ns	162 ns	995
core: batch( fn ) (empty)	v3	545 ns	651 ns	1.60 µs	517 ns	254
core: batch( fn ) (empty)	gpu	82.6 ns	84.6 ns	111 ns	81.0 ns	2000
core: startBatch + endBatch + batching	v3	537 ns	657 ns	1.63 µs	516 ns	255
core: startBatch + endBatch + batching	gpu	90.7 ns	164 ns	2.86 µs	87.2 ns	1017
core: collection()	v3	447 ns	514 ns	1.50 µs	428 ns	326
core: collection()	gpu	62.1 ns	64.4 ns	96.5 ns	60.2 ns	2623
core: compact() (no dead slots — the early-out)	gpu	16.7 µs	21.5 µs	105 µs	16.4 µs	31937
collection: element accessors (id/group/isNode/isEdge, 4 calls)	v3	60.1 ns	62.7 ns	92.1 ns	55.7 ns	2719
collection: element accessors (id/group/isNode/isEdge, 4 calls)	gpu	34.5 ns	36.3 ns	64.4 ns	33.5 ns	4693
collection: length + empty + nonempty (3 calls)	v3	6.70 ns	6.92 ns	12.7 ns	6.34 ns	24506
collection: length + empty + nonempty (3 calls)	gpu	1.47 ns	1.51 ns	1.92 ns	1.43 ns	107651
collection: eq/first/last (3 calls)	v3	15.1 ns	15.6 ns	24.7 ns	14.6 ns	10959
collection: eq/first/last (3 calls)	gpu	8.24 ns	8.48 ns	15.3 ns	7.77 ns	20017
collection: slice()	v3	3.73 µs	3.83 µs	4.27 µs	3.63 µs	41
collection: slice()	gpu	1.71 µs	1.72 µs	1.87 µs	1.64 µs	95
collection: toArray()	v3	8.94 µs	9.08 µs	9.42 µs	8.77 µs	15
collection: toArray()	gpu	739 ns	760 ns	944 ns	717 ns	213
collection: sort( fn )	v3	14.2 µs	14.7 µs	14.5 µs	13.8 µs	12
collection: sort( fn )	gpu	8.97 µs	9.03 µs	9.26 µs	8.86 µs	15
collection: reduce( fn )	v3	156 ns	157 ns	171 ns	116 ns	1083
collection: reduce( fn )	gpu	111 ns	113 ns	129 ns	108 ns	1520
collection: min + max ( fn ) (2 calls)	v3	10.9 µs	11.0 µs	11.0 µs	10.8 µs	12
collection: min + max ( fn ) (2 calls)	gpu	5.54 µs	5.61 µs	5.91 µs	5.48 µs	26
collection: some + every (2 calls)	v3	887 ns	900 ns	1.15 µs	860 ns	181
collection: some + every (2 calls)	gpu	559 ns	560 ns	598 ns	539 ns	303
collection: same + anySame + contains (3 calls)	v3	7.49 µs	7.50 µs	7.64 µs	7.39 µs	19
collection: same + anySame + contains (3 calls)	gpu	102 ns	104 ns	120 ns	97.4 ns	1646
collection: allAre (all match: the full walk)	v3	3.61 µs	3.60 µs	3.96 µs	3.36 µs	43
collection: allAre (all match: the full walk)	gpu	2.56 µs	2.61 µs	3.40 µs	2.48 µs	60
collection: is (none match: the full walk)	v3	5.59 µs	5.60 µs	5.91 µs	5.37 µs	27
collection: is (none match: the full walk)	gpu	2.50 µs	2.56 µs	3.28 µs	2.50 µs	61
collection: union	v3	96.6 µs	103 µs	132 µs	93.5 µs	6704
collection: union	gpu	16.3 µs	16.6 µs	16.7 µs	15.7 µs	12
collection: intersection	v3	92.2 µs	101 µs	138 µs	90.3 µs	6835
collection: intersection	gpu	3.21 µs	3.22 µs	3.37 µs	3.12 µs	48
collection: difference	v3	93.6 µs	105 µs	209 µs	92.2 µs	6590
collection: difference	gpu	3.20 µs	3.23 µs	3.51 µs	3.13 µs	48
collection: symmetricDifference	v3	24.9 µs	25.4 µs	25.4 µs	24.4 µs	12
collection: symmetricDifference	gpu	6.69 µs	6.72 µs	6.91 µs	6.63 µs	21
collection: absoluteComplement	v3	6.34 ms	6.90 ms	10.6 ms	6.10 ms	98
collection: absoluteComplement	gpu	407 µs	413 µs	545 µs	393 µs	1696
collection: byGroup	v3	1.01 ms	1.03 ms	1.49 ms	972 µs	681
collection: byGroup	gpu	322 µs	331 µs	551 µs	238 µs	2118
collection: diff	v3	36.4 µs	36.9 µs	39.0 µs	34.5 µs	12
collection: diff	gpu	11.6 µs	11.6 µs	11.8 µs	11.4 µs	11
collection: indexOf	v3	32.2 ns	33.6 ns	66.9 ns	31.1 ns	5112
collection: indexOf	gpu	23.4 ns	24.2 ns	35.9 ns	22.6 ns	7063
traversal: openNeighborhood + closedNeighborhood (2 calls)	v3	9.33 µs	9.51 µs	10.2 µs	9.28 µs	14
traversal: openNeighborhood + closedNeighborhood (2 calls)	gpu	3.16 µs	3.37 µs	3.86 µs	3.05 µs	46
traversal: connectedNodes	v3	970 ns	1.05 µs	1.86 µs	941 ns	158
traversal: connectedNodes	gpu	303 ns	309 ns	373 ns	295 ns	549
traversal: sources + targets (2 calls)	v3	1.72 µs	1.89 µs	3.12 µs	1.68 µs	84
traversal: sources + targets (2 calls)	gpu	539 ns	744 ns	2.49 µs	511 ns	223
traversal: source() + target() (2 calls)	v3	74.7 ns	75.9 ns	96.6 ns	61.8 ns	2250
traversal: source() + target() (2 calls)	gpu	56.2 ns	57.3 ns	68.4 ns	53.9 ns	2989
traversal: edgesWith	v3	68.8 µs	71.6 µs	125 µs	66.4 µs	9739
traversal: edgesWith	gpu	53.3 µs	53.7 µs	54.9 µs	51.9 µs	12
traversal: edgesTo	v3	69.7 µs	72.2 µs	143 µs	67.5 µs	9626
traversal: edgesTo	gpu	53.1 µs	53.5 µs	54.3 µs	52.1 µs	12
traversal: parallelEdges + codirectedEdges (2 calls)	v3	3.56 µs	3.86 µs	4.62 µs	3.45 µs	40
traversal: parallelEdges + codirectedEdges (2 calls)	gpu	1.04 µs	1.24 µs	2.06 µs	1.03 µs	130
traversal: roots + leaves (2 calls)	v3	4.88 ms	5.29 ms	8.62 ms	4.69 ms	128
traversal: roots + leaves (2 calls)	gpu	389 µs	397 µs	575 µs	372 µs	1764
traversal: allAreNeighbors	v3	198 µs	211 µs	275 µs	193 µs	3303
traversal: allAreNeighbors	gpu	37.2 µs	38.1 µs	40.5 µs	36.9 µs	12
traversal: isLoop + isSimple (2 calls)	v3	131 ns	133 ns	149 ns	125 ns	1292
traversal: isLoop + isSimple (2 calls)	gpu	49.3 ns	50.7 ns	65.9 ns	47.5 ns	3372
degree: indegree + outdegree (2 calls)	v3	3.53 µs	3.91 µs	4.96 µs	3.48 µs	40
degree: indegree + outdegree (2 calls)	gpu	46.4 ns	47.7 ns	69.8 ns	44.9 ns	3578
degree: minDegree + maxDegree (2 calls)	v3	15.9 ms	17.0 ms	20.4 ms	15.3 ms	37
degree: minDegree + maxDegree (2 calls)	gpu	368 µs	377 µs	530 µs	356 µs	1861
degree: minIndegree + maxOutdegree (2 calls)	v3	9.07 ms	9.77 ms	13.9 ms	8.81 ms	67
degree: minIndegree + maxOutdegree (2 calls)	gpu	357 µs	363 µs	471 µs	347 µs	1933
dimensions: renderedPosition get	v3	73.8 ns	79.3 ns	181 ns	71.2 ns	2139
dimensions: renderedPosition get	gpu	29.2 ns	33.4 ns	117 ns	28.4 ns	5068
dimensions: relativePosition get	v3	832 ns	838 ns	997 ns	799 ns	200
dimensions: relativePosition get	gpu	79.4 ns	84.2 ns	157 ns	76.5 ns	2009
dimensions: width + height + outerWidth + outerHeight (4 calls)	v3	615 ns	619 ns	745 ns	590 ns	274
dimensions: width + height + outerWidth + outerHeight (4 calls)	gpu	219 ns	224 ns	343 ns	214 ns	759
dimensions: rendered dims (4 calls)	v3	706 ns	712 ns	856 ns	683 ns	235
dimensions: rendered dims (4 calls)	gpu	224 ns	230 ns	301 ns	218 ns	741
dimensions: renderedBoundingBox	v3	1.69 µs	1.73 µs	2.23 µs	1.64 µs	95
dimensions: renderedBoundingBox	gpu	926 ns	943 ns	1.26 µs	889 ns	177
dimensions: midpoint + renderedMidpoint (2 calls)	gpu	1.17 µs	1.18 µs	1.24 µs	1.13 µs	138
dimensions: source/targetEndpoint (2 calls)	gpu	380 ns	385 ns	432 ns	371 ns	440
dimensions: rendered source/targetEndpoint (2 calls)	gpu	389 ns	398 ns	545 ns	378 ns	426
dimensions: shift + silentShift (round-trip)	v3	289 µs	301 µs	632 µs	280 µs	2328
dimensions: shift + silentShift (round-trip)	gpu	4.82 µs	4.82 µs	4.90 µs	4.71 µs	31
dimensions: silentPosition	v3	870 ns	875 ns	972 ns	843 ns	191
dimensions: silentPosition	gpu	56.6 ns	58.9 ns	86.1 ns	54.9 ns	2880
dimensions: positions( fn ) over 100	v3	175 µs	181 µs	262 µs	169 µs	3861
dimensions: positions( fn ) over 100	gpu	3.48 µs	3.50 µs	3.59 µs	3.44 µs	45
dimensions: padding + paddedWidth + paddedHeight (3 calls)	v3	2.02 µs	2.04 µs	2.13 µs	1.98 µs	80
dimensions: padding + paddedWidth + paddedHeight (3 calls)	gpu	150 ns	153 ns	182 ns	149 ns	1118
curves: isBundledBezier	gpu	20.6 ns	21.2 ns	31.1 ns	20.0 ns	8042
curves: controlPoints + renderedControlPoints (2 calls)	gpu	527 ns	543 ns	750 ns	525 ns	302
curves: segmentPoints (undefined for bezier)	gpu	33.4 ns	34.1 ns	43.8 ns	32.0 ns	5014
compound: parent + parents + ancestors (3 calls)	v3	2.99 µs	3.27 µs	4.39 µs	2.80 µs	48
compound: parent + parents + ancestors (3 calls)	gpu	832 ns	1.18 µs	2.62 µs	791 ns	142
compound: children + descendants + siblings (3 calls)	v3	8.53 µs	8.58 µs	9.42 µs	7.73 µs	16
compound: children + descendants + siblings (3 calls)	gpu	1.82 µs	2.35 µs	3.76 µs	1.76 µs	67
compound: orphans + nonorphans (2 calls)	v3	3.72 µs	3.99 µs	4.77 µs	3.58 µs	39
compound: orphans + nonorphans (2 calls)	gpu	534 ns	763 ns	3.06 µs	519 ns	216
compound: commonAncestors	v3	9.60 µs	9.89 µs	11.0 µs	9.38 µs	14
compound: commonAncestors	gpu	75.3 ns	77.3 ns	134 ns	71.9 ns	2205
compound: isParent/isChild/isChildless/isOrphan (4 calls)	v3	71.7 ns	75.4 ns	169 ns	59.7 ns	2269
compound: isParent/isChild/isChildless/isOrphan (4 calls)	gpu	45.3 ns	46.7 ns	66.2 ns	43.3 ns	3659
flags: select + unselect + selected (band)	v3	193 µs	200 µs	414 µs	187 µs	3501
flags: select + unselect + selected (band)	gpu	12.4 µs	12.6 µs	13.1 µs	12.3 µs	10
flags: selectify + unselectify + selectable	v3	194 µs	201 µs	350 µs	188 µs	3482
flags: selectify + unselectify + selectable	gpu	2.03 µs	2.04 µs	2.15 µs	1.98 µs	81
flags: lock + unlock + locked	v3	191 µs	197 µs	321 µs	186 µs	3542
flags: lock + unlock + locked	gpu	2.04 µs	2.07 µs	2.71 µs	1.99 µs	78
flags: grabify + ungrabify + grabbable	v3	195 µs	199 µs	307 µs	189 µs	3509
flags: grabify + ungrabify + grabbable	gpu	2.04 µs	2.04 µs	2.20 µs	1.97 µs	81
flags: show + hide + visible	v3	2.61 ms	2.74 ms	4.07 ms	2.53 ms	252
flags: show + hide + visible	gpu	11.3 µs	11.3 µs	11.4 µs	11.2 µs	11
flags: active/activate/unactivate (3 calls)	v3	194 µs	200 µs	294 µs	188 µs	3500
flags: active/activate/unactivate (3 calls)	gpu	12.2 µs	12.3 µs	12.3 µs	12.2 µs	10
flags: panify + unpanify + pannable	v3	195 µs	199 µs	288 µs	188 µs	3502
flags: panify + unpanify + pannable	gpu	2.01 µs	2.02 µs	2.14 µs	1.96 µs	81
flags: takesUpSpace + interactive + transparent (3 calls)	v3	196 ns	200 ns	229 ns	191 ns	857
flags: takesUpSpace + interactive + transparent (3 calls)	gpu	63.9 ns	67.3 ns	83.5 ns	61.6 ns	2534
flags: effectiveOpacity	v3	49.1 ns	50.2 ns	60.9 ns	47.0 ns	3418
flags: effectiveOpacity	gpu	21.3 ns	22.9 ns	38.3 ns	20.5 ns	7480
element data: scratch get + set	v3	1.31 µs	1.32 µs	1.41 µs	1.28 µs	126
element data: scratch get + set	gpu	171 ns	176 ns	264 ns	166 ns	969
element data: json() one element	v3	297 ns	382 ns	3.53 µs	290 ns	437
element data: json() one element	gpu	151 ns	193 ns	550 ns	149 ns	830
element data: jsons() over 100	v3	39.3 µs	39.5 µs	40.4 µs	38.0 µs	12
element data: jsons() over 100	gpu	20.1 µs	21.5 µs	24.2 µs	15.4 µs	12
element data: label()	gpu	15.2 ns	15.9 ns	29.5 ns	14.4 ns	10710
core: batch + per-id data() (was batchData, round 90)	v3	2.37 µs	2.48 µs	3.02 µs	2.28 µs	64
core: batch + per-id data() (was batchData, round 90)	gpu	266 ns	351 ns	2.15 µs	259 ns	471
core: animated()	v3	7.52 ns	7.77 ns	13.9 ns	7.21 ns	21833
core: animated()	gpu	1.53 ns	1.57 ns	2.02 ns	1.50 ns	105160
core: zoomRange get	gpu	2.16 ns	2.30 ns	6.01 ns	2.01 ns	72452
core: resize() / invalidateSize()	v3	315 ns	320 ns	380 ns	302 ns	531
core: resize() / invalidateSize()	gpu	129 ns	131 ns	182 ns	124 ns	1291
core: on('render') + off('render') (was onRender/offRender, round 90)	v3	486 ns	541 ns	1.04 µs	469 ns	311
core: on('render') + off('render') (was onRender/offRender, round 90)	gpu	196 ns	203 ns	270 ns	192 ns	821
core: removeScratch()	v3	2.65 ms	2.70 ms	3.19 ms	2.58 ms	254
core: removeScratch()	gpu	279 ns	283 ns	365 ns	270 ns	600
core: selectionType get	v3	4.55 ns	4.93 ns	11.3 ns	4.35 ns	34200
core: selectionType get	gpu	2.26 ns	2.38 ns	6.17 ns	2.11 ns	70083
core: isReady()	v3	3.25 ns	3.37 ns	6.58 ns	3.18 ns	49765
core: isReady()	gpu	1.48 ns	1.53 ns	1.91 ns	1.45 ns	107963
core: window()	v3	4.10 ns	4.24 ns	7.39 ns	4.01 ns	39816
core: window()	gpu	1.48 ns	1.54 ns	2.20 ns	1.45 ns	106830
core: container()	v3	3.90 ns	4.05 ns	7.80 ns	3.75 ns	41602
core: container()	gpu	1.48 ns	1.53 ns	2.08 ns	1.45 ns	107330
core: destroyed()	v3	3.26 ns	3.38 ns	6.56 ns	3.25 ns	49772
core: destroyed()	gpu	1.48 ns	1.55 ns	2.45 ns	1.45 ns	106224
core: removeAllListeners()	v3	13.5 ns	14.2 ns	27.5 ns	13.0 ns	11951
core: removeAllListeners()	gpu	1.58 ns	1.63 ns	2.32 ns	1.55 ns	100650
collection: cy()	v3	7.58 ns	8.13 ns	16.6 ns	7.17 ns	20906
collection: cy()	gpu	2.16 ns	2.41 ns	6.24 ns	2.06 ns	69387
collection: indexOfId()	v3	66.6 ns	69.0 ns	132 ns	62.1 ns	2482
collection: indexOfId()	gpu	37.8 ns	38.4 ns	54.1 ns	36.0 ns	4456
collection: removed() + inside() (2 calls)	v3	17.7 ns	18.1 ns	26.9 ns	17.3 ns	9448
collection: removed() + inside() (2 calls)	gpu	15.1 ns	15.5 ns	25.0 ns	14.6 ns	11075
collection: silentPositions( fn ) over 100	v3	107 µs	113 µs	186 µs	104 µs	6193
collection: silentPositions( fn ) over 100	gpu	3.50 µs	3.51 µs	3.59 µs	3.46 µs	45
collection: animated()	v3	12.1 ns	12.4 ns	21.5 ns	11.9 ns	13778
collection: animated()	gpu	4.19 ns	4.30 ns	7.85 ns	4.00 ns	39293
collection: delayAnimation()	v3	467 ns	594 ns	1.97 µs	456 ns	283
collection: delayAnimation()	gpu	162 ns	171 ns	320 ns	156 ns	991
collection: removeScratch()	v3	1.79 µs	1.81 µs	2.22 µs	1.72 µs	90
collection: removeScratch()	gpu	202 ns	205 ns	242 ns	196 ns	833
collection: component()	v3	82.2 ms	83.1 ms	93.1 ms	75.1 ms	12
collection: component()	gpu	1.84 ms	2.06 ms	5.47 ms	1.78 ms	334
collection: layoutDimensions()	v3	611 ns	618 ns	805 ns	587 ns	274
collection: layoutDimensions()	gpu	169 ns	175 ns	294 ns	163 ns	973
degree: maxIndegree + minOutdegree (2 calls)	v3	9.05 ms	9.79 ms	13.7 ms	8.82 ms	68
degree: maxIndegree + minOutdegree (2 calls)	gpu	359 µs	371 µs	652 µs	348 µs	1889
flags: grabbed + hidden + inactive (3 calls)	v3	86.3 ns	90.2 ns	142 ns	80.4 ns	1902
flags: grabbed + hidden + inactive (3 calls)	gpu	38.4 ns	39.5 ns	56.8 ns	36.2 ns	4341
curves: renderedSegmentPoints	gpu	35.9 ns	37.0 ns	51.9 ns	34.5 ns	4627
layout contract: edgeSlots + endpoints + degreeOf	gpu	34.7 µs	36.3 µs	55.9 µs	29.7 µs	18962
formats: isColumnarElements + isSerializedElements + isPackedIds	gpu	4.61 ns	4.71 ns	9.66 ns	4.51 ns	34936

compaction N=20,000 · 53.9 s

cpu pick: background miss (full highWater walk) — peak

207 µs

cpu pick: background miss (full highWater walk) — compacted

48.7 µs

cpu pick: node hit — peak

223 µs

cpu pick: node hit — compacted

49.8 µs

isCurrent: current ref (forwards present vs absent) — no forwards (peak)

5.87 ns

isCurrent: current ref (forwards present vs absent) — forwards present (compacted)

5.85 ns

isCurrent: stale-ref repair (chase + rewrite, alloc included) — current (baseline alloc)

8.89 ns

isCurrent: stale-ref repair (chase + rewrite, alloc included) — stale → repair

48.3 ns

control: nodes({ selected }) scan (order-list, expected parity) — peak

22.4 µs

control: nodes({ selected }) scan (order-list, expected parity) — compacted

22.4 µs

control: whole-graph boundingBox (expected parity) — peak

100 µs

control: whole-graph boundingBox (expected parity) — compacted

101 µs

1 ns10 ns100 ns1 µs10 µs100 µs1 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
cpu pick: background miss (full highWater walk)	peak	207 µs	212 µs	318 µs	203 µs	3301
cpu pick: background miss (full highWater walk)	compacted	48.7 µs	49.3 µs	50.1 µs	48.2 µs	12
cpu pick: node hit	peak	223 µs	225 µs	275 µs	217 µs	3124
cpu pick: node hit	compacted	49.8 µs	50.3 µs	51.6 µs	48.8 µs	12
isCurrent: current ref (forwards present vs absent)	no forwards (peak)	5.87 ns	6.05 ns	9.51 ns	5.62 ns	28047
isCurrent: current ref (forwards present vs absent)	forwards present (compacted)	5.85 ns	6.04 ns	10.5 ns	5.63 ns	28129
isCurrent: stale-ref repair (chase + rewrite, alloc included)	current (baseline alloc)	8.89 ns	9.48 ns	21.4 ns	8.25 ns	17793
isCurrent: stale-ref repair (chase + rewrite, alloc included)	stale → repair	48.3 ns	49.9 ns	69.9 ns	46.8 ns	3421
control: nodes({ selected }) scan (order-list, expected parity)	peak	22.4 µs	27.1 µs	34.3 µs	21.5 µs	12
control: nodes({ selected }) scan (order-list, expected parity)	compacted	22.4 µs	23.4 µs	24.8 µs	21.8 µs	12
control: whole-graph boundingBox (expected parity)	peak	100 µs	104 µs	165 µs	96.4 µs	6684
control: whole-graph boundingBox (expected parity)	compacted	101 µs	103 µs	146 µs	97.5 µs	6756

compound N=2,000 · 27.7 sv4 (gpu) v3

parent drag (shift subtree + bb settle)

182×

child drag (move + parent re-derive)

51×

reparent round-trip

79×

1 µs10 µs100 µs1 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
parent drag (shift subtree + bb settle)	v3	251 µs	277 µs	593 µs	238 µs	2509
parent drag (shift subtree + bb settle)	gpu	1.38 µs	1.40 µs	1.61 µs	1.34 µs	114
child drag (move + parent re-derive)	v3	79.0 µs	86.0 µs	153 µs	74.8 µs	7982
child drag (move + parent re-derive)	gpu	1.56 µs	1.60 µs	1.90 µs	1.53 µs	101
reparent round-trip	v3	83.8 µs	90.6 µs	153 µs	80.6 µs	7602
reparent round-trip	gpu	1.06 µs	1.07 µs	1.21 µs	1.03 µs	153

curves N=2,000 · 1.6 s

curve premium: build (style apply + derivation) — curved

11.5 ms

curve premium: build (style apply + derivation) — straight

6.42 ms

curve premium: node.position x 1000 (drag) — curved

374 µs

curve premium: node.position x 1000 (drag) — straight

302 µs

curve premium: nodes().positions() (bulk move) — curved

319 µs

curve premium: nodes().positions() (bulk move) — straight

305 µs

curve premium: first read after a bulk move (cold) — curved

5.24 ms

curve premium: first read after a bulk move (cold) — straight

2.60 ms

curve premium: same read again (warm) — curved

1.42 ms

curve premium: same read again (warm) — straight

1.42 ms

curve premium: edge.midpoint x 4000 — curved

2.35 ms

curve premium: edge.midpoint x 4000 — straight

4.19 ms

curve premium: edge.controlPoints x 4000 — curved

2.34 ms

curve premium: edge.controlPoints x 4000 — straight

2.70 ms

curve premium: cy.elements().boundingBox() x 20 — curved

18.1 ms

curve premium: cy.elements().boundingBox() x 20 — straight

19.6 ms

curve premium: cy.fit() x 20 (store scan) — curved

3.71 ms

curve premium: cy.fit() x 20 (store scan) — straight

3.40 ms

curve premium: elementsInBox x 20 (half the graph) — curved

23.4 ms

curve premium: elementsInBox x 20 (half the graph) — straight

5.71 ms

curve premium: edge.hide/show x 1000 (bundle re-fan) — curved

4.50 ms

curve premium: edge.hide/show x 1000 (bundle re-fan) — straight

2.21 ms

100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
curve premium: build (style apply + derivation)	curved	11.5 ms	11.5 ms	11.5 ms	11.5 ms	1
curve premium: build (style apply + derivation)	straight	6.42 ms	6.42 ms	6.42 ms	6.42 ms	1
curve premium: node.position x 1000 (drag)	curved	374 µs	374 µs	374 µs	374 µs	1
curve premium: node.position x 1000 (drag)	straight	302 µs	302 µs	302 µs	302 µs	1
curve premium: nodes().positions() (bulk move)	curved	319 µs	319 µs	319 µs	319 µs	1
curve premium: nodes().positions() (bulk move)	straight	305 µs	305 µs	305 µs	305 µs	1
curve premium: first read after a bulk move (cold)	curved	5.24 ms	5.24 ms	5.24 ms	5.24 ms	1
curve premium: first read after a bulk move (cold)	straight	2.60 ms	2.60 ms	2.60 ms	2.60 ms	1
curve premium: same read again (warm)	curved	1.42 ms	1.42 ms	1.42 ms	1.42 ms	1
curve premium: same read again (warm)	straight	1.42 ms	1.42 ms	1.42 ms	1.42 ms	1
curve premium: edge.midpoint x 4000	curved	2.35 ms	2.35 ms	2.35 ms	2.35 ms	1
curve premium: edge.midpoint x 4000	straight	4.19 ms	4.19 ms	4.19 ms	4.19 ms	1
curve premium: edge.controlPoints x 4000	curved	2.34 ms	2.34 ms	2.34 ms	2.34 ms	1
curve premium: edge.controlPoints x 4000	straight	2.70 ms	2.70 ms	2.70 ms	2.70 ms	1
curve premium: cy.elements().boundingBox() x 20	curved	18.1 ms	18.1 ms	18.1 ms	18.1 ms	1
curve premium: cy.elements().boundingBox() x 20	straight	19.6 ms	19.6 ms	19.6 ms	19.6 ms	1
curve premium: cy.fit() x 20 (store scan)	curved	3.71 ms	3.71 ms	3.71 ms	3.71 ms	1
curve premium: cy.fit() x 20 (store scan)	straight	3.40 ms	3.40 ms	3.40 ms	3.40 ms	1
curve premium: elementsInBox x 20 (half the graph)	curved	23.4 ms	23.4 ms	23.4 ms	23.4 ms	1
curve premium: elementsInBox x 20 (half the graph)	straight	5.71 ms	5.71 ms	5.71 ms	5.71 ms	1
curve premium: edge.hide/show x 1000 (bundle re-fan)	curved	4.50 ms	4.50 ms	4.50 ms	4.50 ms	1
curve premium: edge.hide/show x 1000 (bundle re-fan)	straight	2.21 ms	2.21 ms	2.21 ms	2.21 ms	1

arrows N=2,000 · 8.6 s

arrow premium: build (style apply + arrow fold) — arrowed

6.94 ms

arrow premium: build (style apply + arrow fold) — none

6.33 ms

arrow premium: cy.style() re-apply x 4 — arrowed

7.68 ms

arrow premium: cy.style() re-apply x 4 — none

7.93 ms

arrow premium: node.position x 200000 (drag) — arrowed

21.6 ms

arrow premium: node.position x 200000 (drag) — none

21.6 ms

arrow premium: edge.sourceEndpoint x 200000 — arrowed

59.6 ms

arrow premium: edge.sourceEndpoint x 200000 — none

60.2 ms

arrow premium: edge.targetEndpoint x 200000 — arrowed

59.0 ms

arrow premium: edge.targetEndpoint x 200000 — none

58.7 ms

arrow premium: edge.midpoint x 200000 — arrowed

139 ms

arrow premium: edge.midpoint x 200000 — none

137 ms

arrow premium: edge.style('target-arrow-shape') x 200000 — arrowed

31.5 ms

arrow premium: edge.style('target-arrow-shape') x 200000 — none

27.5 ms

arrow premium: cy.elements().boundingBox() x 20 — arrowed

18.3 ms

arrow premium: cy.elements().boundingBox() x 20 — none

18.2 ms

arrow premium: elementsInBox x 20 (half the graph) — arrowed

5.47 ms

arrow premium: elementsInBox x 20 (half the graph) — none

5.53 ms

arrow premium: control: sourceEndpoint, bezier vs straight — arrowed

75.4 ms

arrow premium: control: sourceEndpoint, bezier vs straight — none

60.4 ms

arrow premium: edge column bytes — B/edge

156 ms

1 ms10 ms100 ms1 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
arrow premium: build (style apply + arrow fold)	arrowed	6.94 ms	6.94 ms	6.94 ms	6.94 ms	1
arrow premium: build (style apply + arrow fold)	none	6.33 ms	6.33 ms	6.33 ms	6.33 ms	1
arrow premium: cy.style() re-apply x 4	arrowed	7.68 ms	7.68 ms	7.68 ms	7.68 ms	1
arrow premium: cy.style() re-apply x 4	none	7.93 ms	7.93 ms	7.93 ms	7.93 ms	1
arrow premium: node.position x 200000 (drag)	arrowed	21.6 ms	21.6 ms	21.6 ms	21.6 ms	1
arrow premium: node.position x 200000 (drag)	none	21.6 ms	21.6 ms	21.6 ms	21.6 ms	1
arrow premium: edge.sourceEndpoint x 200000	arrowed	59.6 ms	59.6 ms	59.6 ms	59.6 ms	1
arrow premium: edge.sourceEndpoint x 200000	none	60.2 ms	60.2 ms	60.2 ms	60.2 ms	1
arrow premium: edge.targetEndpoint x 200000	arrowed	59.0 ms	59.0 ms	59.0 ms	59.0 ms	1
arrow premium: edge.targetEndpoint x 200000	none	58.7 ms	58.7 ms	58.7 ms	58.7 ms	1
arrow premium: edge.midpoint x 200000	arrowed	139 ms	139 ms	139 ms	139 ms	1
arrow premium: edge.midpoint x 200000	none	137 ms	137 ms	137 ms	137 ms	1
arrow premium: edge.style('target-arrow-shape') x 200000	arrowed	31.5 ms	31.5 ms	31.5 ms	31.5 ms	1
arrow premium: edge.style('target-arrow-shape') x 200000	none	27.5 ms	27.5 ms	27.5 ms	27.5 ms	1
arrow premium: cy.elements().boundingBox() x 20	arrowed	18.3 ms	18.3 ms	18.3 ms	18.3 ms	1
arrow premium: cy.elements().boundingBox() x 20	none	18.2 ms	18.2 ms	18.2 ms	18.2 ms	1
arrow premium: elementsInBox x 20 (half the graph)	arrowed	5.47 ms	5.47 ms	5.47 ms	5.47 ms	1
arrow premium: elementsInBox x 20 (half the graph)	none	5.53 ms	5.53 ms	5.53 ms	5.53 ms	1
arrow premium: control: sourceEndpoint, bezier vs straight	arrowed	75.4 ms	75.4 ms	75.4 ms	75.4 ms	1
arrow premium: control: sourceEndpoint, bezier vs straight	none	60.4 ms	60.4 ms	60.4 ms	60.4 ms	1
arrow premium: edge column bytes	B/edge	156 ms	156 ms	156 ms	156 ms	1

labels N=20,000 · 1.8 s

labels: breakLines x 20000 — gpu

84.0 ms

labels: estimateBlock x 20000 — gpu

91.3 ms

labels: addNode x 20000 — gpu

30.4 ms

labels: setLabel (build) x 20000 — gpu

97.7 ms

labels: setLabel (rewrite) x 20000 — gpu

101 ms

labels: boundingBox with 20000 label terms — gpu

3.04 ms

labels: computeSdf tier 1 (30x42) x 96 glyphs — gpu

14.7 ms

labels: computeSdf tier 2 (59x85) x 96 glyphs — gpu

37.8 ms

1 ms10 ms100 ms1 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
labels: breakLines x 20000	gpu	84.0 ms	84.0 ms	84.0 ms	84.0 ms	1
labels: estimateBlock x 20000	gpu	91.3 ms	91.3 ms	91.3 ms	91.3 ms	1
labels: addNode x 20000	gpu	30.4 ms	30.4 ms	30.4 ms	30.4 ms	1
labels: setLabel (build) x 20000	gpu	97.7 ms	97.7 ms	97.7 ms	97.7 ms	1
labels: setLabel (rewrite) x 20000	gpu	101 ms	101 ms	101 ms	101 ms	1
labels: boundingBox with 20000 label terms	gpu	3.04 ms	3.04 ms	3.04 ms	3.04 ms	1
labels: computeSdf tier 1 (30x42) x 96 glyphs	gpu	14.7 ms	14.7 ms	14.7 ms	14.7 ms	1
labels: computeSdf tier 2 (59x85) x 96 glyphs	gpu	37.8 ms	37.8 ms	37.8 ms	37.8 ms	1

transitions N=2,000 · 28.8 s

transition: auto-extent shift, whole-channel re-derive (2000 slots) — transitions off

1.20 ms

transition: auto-extent shift, whole-channel re-derive (2000 slots) — transitions on

3.78 ms

transition: bulk tween tick (2000 slots, one manager tick) — tick

129 µs

transition: explicit-domain write (O(changed) — one element) — transitions off

14.7 µs

transition: explicit-domain write (O(changed) — one element) — transitions on

18.3 µs

transition: sheet swap, whole-graph apply (2000 slots diff + tween) — transitions off

2.09 ms

transition: sheet swap, whole-graph apply (2000 slots diff + tween) — transitions on

4.88 ms

10 µs100 µs1 ms10 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
transition: auto-extent shift, whole-channel re-derive (2000 slots)	transitions off	1.20 ms	1.25 ms	2.11 ms	1.16 ms	550
transition: auto-extent shift, whole-channel re-derive (2000 slots)	transitions on	3.78 ms	3.97 ms	6.10 ms	3.49 ms	171
transition: bulk tween tick (2000 slots, one manager tick)	tick	129 µs	136 µs	245 µs	124 µs	5146
transition: explicit-domain write (O(changed) — one element)	transitions off	14.7 µs	15.1 µs	15.9 µs	14.2 µs	12
transition: explicit-domain write (O(changed) — one element)	transitions on	18.3 µs	18.6 µs	19.1 µs	17.8 µs	12
transition: sheet swap, whole-graph apply (2000 slots diff + tween)	transitions off	2.09 ms	2.14 ms	2.87 ms	2.03 ms	319
transition: sheet swap, whole-graph apply (2000 slots diff + tween)	transitions on	4.88 ms	4.46 ms	6.08 ms	3.61 ms	152

geometry-tween N=2,000 · 32.6 s

size tick: unlabelled (2000 slots) — tick

527 µs

size tick: labelled, center/center (re-anchor early-out) (2000 slots) — tick

562 µs

size tick: labelled, hanging (sidecar rewrite per tick) (2000 slots) — tick

1.22 ms

edge width tick (4000 slots) — bare

285 µs

edge width tick (4000 slots) — with rides (casing + match-line arrows)

722 µs

padding tick + auto-bounds flush (250 parents × 8 children) — tick + flush

173 µs

font-size tick (2000 labels) — wrap none (dims scale-patch)

350 µs

font-size tick (2000 labels) — wrapped (re-break per tick)

5.82 ms

baseline: paint (opacity) tick (2000 slots) — tick

151 µs

100 µs1 ms10 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
size tick: unlabelled (2000 slots)	tick	527 µs	553 µs	868 µs	509 µs	1244
size tick: labelled, center/center (re-anchor early-out) (2000 slots)	tick	562 µs	583 µs	898 µs	537 µs	1182
size tick: labelled, hanging (sidecar rewrite per tick) (2000 slots)	tick	1.22 ms	1.29 ms	2.08 ms	1.20 ms	520
edge width tick (4000 slots)	bare	285 µs	295 µs	537 µs	275 µs	2362
edge width tick (4000 slots)	with rides (casing + match-line arrows)	722 µs	744 µs	1.05 ms	685 µs	938
padding tick + auto-bounds flush (250 parents × 8 children)	tick + flush	173 µs	180 µs	359 µs	168 µs	3888
font-size tick (2000 labels)	wrap none (dims scale-patch)	350 µs	380 µs	879 µs	323 µs	1820
font-size tick (2000 labels)	wrapped (re-break per tick)	5.82 ms	5.91 ms	6.60 ms	5.55 ms	113
baseline: paint (opacity) tick (2000 slots)	tick	151 µs	162 µs	284 µs	144 µs	4305