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) · 6.19.14-200.fc43.x86_64 · x64
Node	v24.18.0 (v8 13.6.233.17-node.50)
Commit	44e55dc9 perf(65.10): the CPU takes two families back — flat hierarchical merges, sparse PageRank, and honest re-routing (+ the wide benchmark page)
Machine id	5cf3f79c

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

sweep: edges() @ 2k

294929×

sweep: nodes() @ 2k

172211×

core: edges() @ 2k

90329×

core: nodes() @ 2k

51711×

sweep: elements() @ 2k

33089×

core: elements() @ 2k

22329×

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

17105×

core: removeScratch() @ 2k

9872×

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

8472×

compare: same() @ 2k

7812×

trav: successors() @ 2k

782×

trav: predecessors() @ 2k

721×

mut-bulk: hide + show @ 2k

270×

flags: show + hide + visible @ 2k

239×

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

191×

degree: nodes.totalDegree() @ 2k

168×

154×

compound: commonAncestors @ 2k

130×

degree: nodes.maxDegree() @ 2k

127×

mut-bulk: positions(obj) @ 2k

116×

degree: node.degree() @ 2k

115×

mut-bulk: shift @ 2k

108×

viewport: reset() @ 2k

106×

flags: selectify + unselectify + selectable @ 2k

98×

mut-bulk: positions(fn) @ 2k

98×

flags: grabify + ungrabify + grabbable @ 2k

96×

flags: lock + unlock + locked @ 2k

96×

flags: panify + unpanify + pannable @ 2k

95×

core: options() @ 2k

85×

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

77×

algo: closenessCentrality (one root) @ 2k

71×

algo: pageRank (20 iters) @ 500

69×

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

68×

algo: closenessCentrality (one root) @ 500

68×

reparent round-trip @ 2k

67×

algo: dfs (whole graph) @ 500

63×

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

61×

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

61×

61×

algo: bfs (whole graph) @ 500

61×

algo: bfs (whole graph) @ 2k

60×

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

59×

algo: dijkstra + pathTo @ 2k

57×

algo: dfs (whole graph) @ 2k

57×

mut-bulk: lock + unlock @ 2k

57×

algo: closenessCentrality (weighted, one root) @ 500

56×

algo: dijkstra + pathTo @ 500

52×

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

51×

dimensions: positions( fn ) over 100 @ 2k

45×

collection: component() @ 2k

44×

traverse: components() @ 2k

39×

algo: tarjan SCC @ 2k

37×

algo: degreeCentralityNormalized @ 500

37×

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

36×

36×

algo: degreeCentralityNormalized @ 2k

36×

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

35×

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

34×

algo: tarjan SCC @ 500

33×

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

33×

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

33×

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

33×

collection: silentPositions( fn ) over 100 @ 2k

31×

algo: degreeCentrality (one root) @ 500

30×

algo: degreeCentrality (one root) @ 2k

30×

collection: difference @ 2k

29×

collection: intersection @ 2k

29×

algo: kruskal @ 2k

27×

set: difference() @ 2k

26×

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

26×

set: intersection() @ 2k

26×

26×

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

25×

emit: position write, no listeners @ 2k

25×

algo: hopcroft-tarjan biconnected @ 2k

25×

algo: hopcroft-tarjan biconnected @ 500

25×

algo: kruskal @ 500

24×

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

22×

trav: band roots() @ 2k

22×

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

21×

algo: bellmanFord @ 500

21×

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

21×

data: bulk write, numeric column @ 2k

20×

algo: bellmanFord @ 2k

20×

viewport: center( eles ) @ 2k

19×

17×

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

17×

core: add + remove @ 2k

17×

mut: position set @ 2k

17×

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

17×

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

17×

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

17×

core: forceRender() (headless no-op) @ 2k

16×

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

16×

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

16×

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

15×

dimensions: silentPosition @ 2k

15×

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

15×

style: renderedStyle (one prop) @ 2k

15×

mut-bulk: data set @ 2k

15×

collection: absoluteComplement @ 2k

14×

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

14×

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

14×

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

14×

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

14×

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

13×

13×

algo: betweennessCentrality (unweighted) @ 500

13×

query: two keys AND-ed @ 2k

12×

collection: toArray() @ 2k

12×

algo: betweennessCentrality (unweighted) @ 2k

11×

algo: fuzzyCMeans @ 500

11×

dimensions: relativePosition get @ 2k

11×

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

11×

11×

10×

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

9.2×

algo: betweennessCentrality (weighted) @ 500

9.1×

collection: removeScratch() @ 2k

9.1×

algo: kMedoids @ 500

9.0×

core: removeAllListeners() @ 2k

8.6×

emit: 1 core listener @ 2k

8.5×

core: batchData() @ 2k

8.5×

algo: hierarchicalClustering (threshold) @ 500

8.4×

core: collection() @ 2k

7.9×

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

7.9×

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

7.9×

data: removeData (whole collection) @ 2k

7.8×

element data: scratch get + set @ 2k

7.8×

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

7.6×

mut-bulk: select + unselect @ 2k

7.1×

query: predicate function (both sides) @ 2k

7.0×

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

7.0×

core: collection() @ 2k

6.7×

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

6.5×

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

6.3×

bounds: whole-graph boundingBox() @ 2k

6.3×

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

6.2×

emit: 10 core listeners @ 2k

6.1×

style: whole-object style() @ 2k

6.1×

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

6.1×

core: startBatch + endBatch + batching @ 2k

5.7×

collection: union @ 2k

5.7×

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

5.6×

set: union() @ 2k

5.6×

algo: affinityPropagation @ 500

5.3×

algo: hierholzer (directed) @ 500

5.2×

traversal: allAreNeighbors @ 2k

5.1×

trav: outgoers() @ 2k

5.1×

trav: band neighborhood() @ 2k

5.0×

trav: incomers() @ 2k

4.9×

compare: contains() @ 2k

4.9×

core: animated() @ 2k

4.9×

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

4.9×

viewport: zoom() get @ 2k

4.7×

algo: kMeans (fixed centroids) @ 500

4.6×

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

4.6×

traverse: outgoers() @ 2k

4.4×

traverse: incomers() @ 2k

4.3×

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

4.3×

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

3.9×

mut: data set @ 2k

3.9×

collection: layoutDimensions() @ 2k

3.9×

collection: symmetricDifference @ 2k

3.8×

traversal: connectedNodes @ 2k

3.7×

trav: neighborhood() @ 2k

3.7×

mut: select + unselect @ 2k

3.7×

collection: cy() @ 2k

3.6×

algo: aStar @ 500

3.5×

traverse: neighborhood() @ 2k

3.4×

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

3.4×

algo: aStar @ 2k

3.3×

dimensions: rendered dims (4 calls) @ 2k

3.3×

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

3.3×

collection: diff @ 2k

3.3×

anim: viewport animate + stop @ 2k

3.0×

collection: byGroup @ 2k

3.0×

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

3.0×

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

3.0×

collection: delayAnimation() @ 2k

2.9×

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

2.9×

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

2.9×

collection: animated() @ 2k

2.8×

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

2.8×

algo: hierholzer (directed) @ 2k

2.8×

style: numericStyle @ 2k

2.8×

trav: band connectedNodes() @ 2k

2.8×

core: window() @ 2k

2.8×

iter: map() @ 2k

2.8×

load: cy.json() @ 2k

2.8×

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

2.7×

trav: closedNeighborhood() @ 2k

2.7×

core: container() @ 2k

2.6×

trav: connectedEdges() @ 2k

2.6×

dimensions: renderedPosition get @ 2k

2.6×

core: onRender + offRender @ 2k

2.5×

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

2.5×

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

2.4×

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

2.4×

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

2.3×

flags: effectiveOpacity @ 2k

2.3×

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

2.3×

collection: slice() @ 2k

2.2×

core: destroyed() @ 2k

2.2×

data: read the whole object @ 2k

2.2×

core: isReady() @ 2k

2.2×

style: read one prop (width) @ 2k

2.2×

traverse: connectedEdges() @ 2k

2.2×

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

2.0×

algo: floydWarshall @ 500

1.98×

dimensions: renderedBoundingBox @ 2k

1.98×

core: selectionType get @ 2k

1.97×

viewport: pan( p ) set @ 2k

1.96×

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

1.95×

viewport: zoom( n ) set @ 2k

1.91×

element data: json() one element @ 2k

1.90×

iter: filter(fn) @ 2k

1.89×

json: element @ 2k

1.88×

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

1.87×

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

1.84×

viewport: panBy() @ 2k

1.83×

core: zoom set @ 2k

1.76×

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

1.75×

viewport: renderedExtent() @ 2k

1.68×

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

1.67×

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

1.66×

style: read one prop (color) @ 2k

1.65×

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

1.63×

bounds: one node boundingBox() @ 2k

1.63×

algo: closenessCentralityNormalized @ 500

1.63×

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

1.61×

collection: sort( fn ) @ 2k

1.61×

sweep: filter(fn) @ 2k

1.59×

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

1.57×

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

1.53×

collection: indexOfId() @ 2k

1.53×

viewport: extent() @ 2k

1.52×

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

1.50×

trav: band sources() @ 2k

1.49×

trav: band connectedEdges() @ 2k

1.47×

element data: jsons() over 100 @ 2k

1.42×

data: read one numeric key @ 2k

1.40×

position: get @ 2k

1.39×

core: mutableElements() (v4 materializes; v3 is O(1)) @ 2k

1.39×

collection: reduce( fn ) @ 2k

1.37×

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

1.37×

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

1.34×

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

1.33×

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

1.32×

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

1.30×

anim: start + stop, one element @ 2k

1.29×

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

1.26×

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

1.24×

collection: indexOf @ 2k

1.24×

core: filter(fn) @ 2k

1.22×

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

1.21×

traversal: edgesWith @ 2k

1.21×

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

1.21×

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

1.19×

data: get @ 2k

1.18×

traversal: edgesTo @ 2k

1.18×

core: gating getters (5 calls) @ 2k

1.11×

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

1.11×

algo: markovClustering @ 500

1.11×

algo: kargerStein @ 500

1.08×

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

1.07×

viewport: pan() get @ 2k

1.05×

core: extent() @ 2k

1.04×

iter: forEach() @ 2k

1.04×

core: getElementById() @ 2k

1.03×

1×10×100×1000×10000×100000×1000000×

Scaling speedup vs graph size

algorithms

group	× @ N=500	× @ N=2k
algo: bfs (whole graph)	61×	60×
algo: dfs (whole graph)	63×	57×
algo: dijkstra + pathTo	52×	57×
algo: aStar	3.5×	3.3×
algo: bellmanFord	21×	20×
algo: kruskal	24×	27×
algo: tarjan SCC	33×	37×
algo: hopcroft-tarjan biconnected	25×	25×
algo: hierholzer (directed)	5.2×	2.8×
algo: betweennessCentrality (unweighted)	13×	11×
algo: degreeCentralityNormalized	37×	36×
algo: closenessCentrality (one root)	68×	71×
algo: degreeCentrality (one root)	30×	30×
algo: pageRank (20 iters)	69×	–
algo: closenessCentrality (weighted, one root)	56×	–
algo: floydWarshall	1.98×	–
algo: closenessCentralityNormalized	1.63×	–
algo: markovClustering	1.11×	–
algo: hierarchicalClustering (threshold)	8.4×	–
algo: kMeans (fixed centroids)	4.6×	–
algo: kMedoids	9.0×	–
algo: fuzzyCMeans	11×	–
algo: affinityPropagation	5.3×	–
algo: kargerStein	1.08×	–
algo: betweennessCentrality (weighted)	9.1×	–

core+collection N=2,000 · 57.9 sv4 (gpu) v3

core: getElementById()

1.03×

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

26×

core: nodes()

51711×

core: edges()

90329×

core: elements()

22329×

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

5.6×

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

14×

core: filter(fn)

1.22×

core: collection()

6.7×

core: pan() get (x32)

1.33×

core: extent()

1.04×

core: add + remove

17×

core: zoom set

1.76×

traverse: neighborhood()

3.4×

traverse: connectedEdges()

2.2×

traverse: outgoers()

4.4×

traverse: incomers()

4.3×

traverse: components()

39×

degree: node.degree()

115×

degree: nodes.totalDegree()

168×

degree: nodes.maxDegree()

127×

data: get

1.18×

position: get

1.39×

json: element

1.88×

set: union()

5.6×

set: intersection()

26×

set: difference()

26×

iter: map()

2.8×

iter: forEach()

1.04×

iter: filter(fn)

1.89×

compare: same()

7812×

compare: contains()

4.9×

mut: data set

3.9×

mut: position set

17×

mut: select + unselect

3.7×

1 ns10 ns100 ns1 µs10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
core: getElementById()	v3	86.7 ns	90.4 ns	163 ns	83.3 ns	1897
core: getElementById()	gpu	84.4 ns	87.7 ns	127 ns	80.8 ns	1953
core: id lookup ($("#id") vs $id())	v3	2.00 µs	2.32 µs	3.54 µs	1.83 µs	69
core: id lookup ($("#id") vs $id())	gpu	76.7 ns	79.7 ns	153 ns	73.3 ns	2144
core: nodes()	v3	422 µs	435 µs	705 µs	412 µs	1607
core: nodes()	gpu	8.16 ns	8.54 ns	23.6 ns	7.73 ns	19874
core: edges()	v3	732 µs	743 µs	933 µs	701 µs	940
core: edges()	gpu	8.10 ns	9.23 ns	17.2 ns	7.72 ns	18398
core: elements()	v3	153 µs	163 µs	410 µs	148 µs	4274
core: elements()	gpu	6.84 ns	7.20 ns	24.4 ns	6.49 ns	23513
core: nodes ($("node") vs filter({group}))	v3	562 µs	591 µs	1.30 ms	543 µs	1179
core: nodes ($("node") vs filter({group}))	gpu	99.5 µs	106 µs	257 µs	68.1 µs	6419
core: selected ($(":selected") vs filter({selected}))	v3	357 µs	371 µs	665 µs	347 µs	1825
core: selected ($(":selected") vs filter({selected}))	gpu	24.8 µs	25.7 µs	26.6 µs	24.5 µs	12
core: filter(fn)	v3	423 µs	438 µs	660 µs	416 µs	1595
core: filter(fn)	gpu	345 µs	374 µs	771 µs	334 µs	1830
core: collection()	v3	459 ns	557 ns	1.61 µs	446 ns	296
core: collection()	gpu	68.0 ns	70.3 ns	98.4 ns	66.5 ns	2319
core: pan() get (x32)	v3	44.5 ns	46.6 ns	59.4 ns	42.5 ns	3651
core: pan() get (x32)	gpu	33.3 ns	36.5 ns	71.0 ns	32.7 ns	4694
core: extent()	v3	22.4 ns	38.6 ns	164 ns	21.4 ns	4328
core: extent()	gpu	21.5 ns	28.8 ns	118 ns	20.2 ns	5751
core: add + remove	v3	131 µs	141 µs	282 µs	119 µs	4907
core: add + remove	gpu	7.50 µs	8.78 µs	29.4 µs	7.13 µs	73573
core: zoom set	v3	1.23 µs	1.24 µs	1.34 µs	1.18 µs	132
core: zoom set	gpu	697 ns	717 ns	945 ns	679 ns	226
traverse: neighborhood()	v3	3.69 µs	3.97 µs	4.91 µs	3.49 µs	39
traverse: neighborhood()	gpu	1.08 µs	1.25 µs	2.67 µs	1.02 µs	129
traverse: connectedEdges()	v3	1.03 µs	1.15 µs	2.03 µs	983 ns	144
traverse: connectedEdges()	gpu	471 ns	521 ns	1.48 µs	436 ns	321
traverse: outgoers()	v3	2.31 µs	2.49 µs	3.33 µs	2.20 µs	64
traverse: outgoers()	gpu	531 ns	575 ns	1.31 µs	517 ns	278
traverse: incomers()	v3	2.32 µs	2.48 µs	3.24 µs	2.18 µs	65
traverse: incomers()	gpu	538 ns	579 ns	1.27 µs	519 ns	277
traverse: components()	v3	78.3 ms	80.6 ms	89.3 ms	74.0 ms	12
traverse: components()	gpu	2.03 ms	2.15 ms	3.22 ms	1.97 ms	310
degree: node.degree()	v3	3.33 µs	3.63 µs	4.59 µs	3.19 µs	42
degree: node.degree()	gpu	29.0 ns	30.2 ns	45.8 ns	27.6 ns	5404
degree: nodes.totalDegree()	v3	8.45 ms	9.25 ms	13.1 ms	7.43 ms	72
degree: nodes.totalDegree()	gpu	50.4 µs	51.9 µs	94.8 µs	47.5 µs	13448
degree: nodes.maxDegree()	v3	7.31 ms	8.15 ms	12.4 ms	6.81 ms	82
degree: nodes.maxDegree()	gpu	57.4 µs	58.6 µs	92.6 µs	54.0 µs	11961
data: get	v3	30.7 ns	31.7 ns	51.8 ns	29.4 ns	5393
data: get	gpu	25.9 ns	26.7 ns	39.2 ns	24.6 ns	6419
position: get	v3	33.8 ns	36.3 ns	131 ns	32.8 ns	4705
position: get	gpu	24.3 ns	27.2 ns	52.8 ns	22.4 ns	6156
json: element	v3	300 ns	389 ns	3.49 µs	292 ns	421
json: element	gpu	160 ns	166 ns	308 ns	152 ns	974
set: union()	v3	93.1 µs	106 µs	158 µs	90.9 µs	6477
set: union()	gpu	16.5 µs	17.4 µs	16.8 µs	16.4 µs	12
set: intersection()	v3	90.1 µs	104 µs	161 µs	86.8 µs	6648
set: intersection()	gpu	3.48 µs	3.53 µs	3.79 µs	3.40 µs	43
set: difference()	v3	90.5 µs	104 µs	149 µs	88.0 µs	6619
set: difference()	gpu	3.44 µs	3.49 µs	3.89 µs	3.42 µs	44
iter: map()	v3	28.6 µs	30.6 µs	52.9 µs	27.4 µs	22340
iter: map()	gpu	10.4 µs	11.1 µs	36.3 µs	9.94 µs	60926
iter: forEach()	v3	3.45 µs	3.70 µs	3.69 µs	3.39 µs	37
iter: forEach()	gpu	3.32 µs	3.36 µs	3.59 µs	3.26 µs	45
iter: filter(fn)	v3	245 µs	253 µs	457 µs	243 µs	2763
iter: filter(fn)	gpu	130 µs	134 µs	199 µs	125 µs	5207
compare: same()	v3	116 µs	120 µs	203 µs	111 µs	5835
compare: same()	gpu	14.9 ns	15.7 ns	29.7 ns	14.1 ns	10883
compare: contains()	v3	5.97 µs	6.00 µs	6.29 µs	5.90 µs	25
compare: contains()	gpu	1.21 µs	1.22 µs	1.28 µs	1.19 µs	131
mut: data set	v3	417 ns	439 ns	746 ns	401 ns	381
mut: data set	gpu	108 ns	113 ns	172 ns	103 ns	1493
mut: position set	v3	1.22 µs	1.25 µs	1.58 µs	1.18 µs	131
mut: position set	gpu	70.4 ns	75.1 ns	120 ns	68.0 ns	2259
mut: select + unselect	v3	2.60 µs	2.80 µs	3.49 µs	2.47 µs	56
mut: select + unselect	gpu	712 ns	787 ns	1.70 µs	658 ns	613177

materializers N=2,000 · 16.6 sv4 (gpu) v3

sweep: nodes()

172211×

sweep: edges()

294929×

sweep: elements()

33089×

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

9.2×

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

14×

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

35×

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

61×

sweep: filter(fn)

1.59×

1 ns10 ns100 ns1 µs10 µs100 µs1 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
sweep: nodes()	v3	415 µs	428 µs	533 µs	403 µs	1638
sweep: nodes()	gpu	2.41 ns	2.57 ns	5.71 ns	2.36 ns	64199
sweep: edges()	v3	741 µs	755 µs	1.06 ms	710 µs	925
sweep: edges()	gpu	2.51 ns	2.63 ns	5.14 ns	2.46 ns	63285
sweep: elements()	v3	159 µs	165 µs	254 µs	155 µs	4238
sweep: elements()	gpu	4.81 ns	5.05 ns	12.1 ns	4.64 ns	33459
sweep: all nodes ($("node") vs filter({group}))	v3	552 µs	582 µs	900 µs	531 µs	1199
sweep: all nodes ($("node") vs filter({group}))	gpu	60.2 µs	76.6 µs	197 µs	55.9 µs	9036
sweep: selected ($(":selected") vs filter({selected}))	v3	359 µs	376 µs	656 µs	347 µs	1858
sweep: selected ($(":selected") vs filter({selected}))	gpu	26.3 µs	26.7 µs	27.3 µs	26.0 µs	12
sweep: selected nodes ($("node:selected") vs filter({group, selected}))	v3	330 µs	344 µs	620 µs	320 µs	2029
sweep: selected nodes ($("node:selected") vs filter({group, selected}))	gpu	9.46 µs	9.50 µs	9.66 µs	9.37 µs	14
sweep: nodes(":selected") vs nodes({selected})	v3	552 µs	566 µs	867 µs	534 µs	1237
sweep: nodes(":selected") vs nodes({selected})	gpu	8.99 µs	9.04 µs	9.46 µs	8.86 µs	15
sweep: filter(fn)	v3	418 µs	429 µs	538 µs	413 µs	1630
sweep: filter(fn)	gpu	263 µs	280 µs	592 µs	257 µs	2484

mutators N=2,000 · 17.9 sv4 (gpu) v3

mut-bulk: select + unselect

7.1×

mut-bulk: hide + show

270×

mut-bulk: lock + unlock

57×

mut-bulk: positions(obj)

116×

mut-bulk: positions(fn)

98×

mut-bulk: shift

108×

mut-bulk: data set

15×

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

6.2×

10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
mut-bulk: select + unselect	v3	1.88 ms	1.94 ms	3.07 ms	1.82 ms	338
mut-bulk: select + unselect	gpu	264 µs	281 µs	498 µs	254 µs	2496
mut-bulk: hide + show	v3	60.7 ms	67.5 ms	80.6 ms	59.6 ms	12
mut-bulk: hide + show	gpu	225 µs	232 µs	424 µs	218 µs	2922
mut-bulk: lock + unlock	v3	2.20 ms	2.23 ms	2.63 ms	2.12 ms	300
mut-bulk: lock + unlock	gpu	38.6 µs	39.4 µs	42.7 µs	38.2 µs	12
mut-bulk: positions(obj)	v3	2.37 ms	2.43 ms	4.15 ms	2.31 ms	281
mut-bulk: positions(obj)	gpu	20.5 µs	22.0 µs	46.5 µs	19.3 µs	31382
mut-bulk: positions(fn)	v3	3.40 ms	3.60 ms	5.68 ms	3.24 ms	188
mut-bulk: positions(fn)	gpu	34.8 µs	36.9 µs	89.2 µs	32.9 µs	18410
mut-bulk: shift	v3	3.66 ms	3.88 ms	6.30 ms	3.57 ms	175
mut-bulk: shift	gpu	34.1 µs	35.3 µs	63.6 µs	32.4 µs	19158
mut-bulk: data set	v3	836 µs	861 µs	1.17 ms	829 µs	777
mut-bulk: data set	gpu	56.5 µs	58.6 µs	155 µs	55.6 µs	11834
mut-bulk: remove + re-add (256 nodes + cascade)	v3	41.3 ms	42.1 ms	46.3 ms	39.7 ms	12
mut-bulk: remove + re-add (256 nodes + cascade)	gpu	6.71 ms	6.87 ms	8.40 ms	6.07 ms	94

scenarios N=2,000 · 12.8 sv4 (gpu) v3

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

7.9×

scn: select-all + fit (listeners on)

25×

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

16×

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

6.5×

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

1.84×

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	302 µs	346 µs	948 µs	275 µs	2002
scn: explore (expand 2-hop + select + fit)	gpu	38.2 µs	43.5 µs	188 µs	34.4 µs	15706
scn: select-all + fit (listeners on)	v3	24.2 ms	24.7 ms	26.3 ms	23.3 ms	24
scn: select-all + fit (listeners on)	gpu	955 µs	1.03 ms	2.07 ms	899 µs	656
scn: drag (select band of 100 + 8-step shift, position listener)	v3	1.60 ms	1.65 ms	2.15 ms	1.56 ms	404
scn: drag (select band of 100 + 8-step shift, position listener)	gpu	102 µs	115 µs	349 µs	97.0 µs	6044
scn: edit (remove + re-add 256 nodes + cascade, add/remove listeners)	v3	44.6 ms	45.0 ms	45.8 ms	43.6 ms	11
scn: edit (remove + re-add 256 nodes + cascade, add/remove listeners)	gpu	6.88 ms	7.15 ms	9.47 ms	6.32 ms	92
scn: refresh (bulk data write + mapped labels + filter + fit, data listener)	v3	7.61 ms	7.87 ms	9.99 ms	7.06 ms	82
scn: refresh (bulk data write + mapped labels + filter + fit, data listener)	gpu	4.14 ms	4.29 ms	6.96 ms	3.70 ms	152

traversal N=2,000 · 40.6 sv4 (gpu) v3

trav: neighborhood()

3.7×

trav: closedNeighborhood()

2.7×

trav: connectedEdges()

2.6×

trav: outgoers()

5.1×

trav: incomers()

4.9×

trav: successors()

782×

trav: predecessors()

721×

trav: band neighborhood()

5.0×

trav: band connectedEdges()

1.47×

trav: band roots()

22×

trav: band sources()

1.49×

trav: band connectedNodes()

2.8×

100 ns1 µs10 µs100 µs1 ms10 ms100 ms1 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
trav: neighborhood()	v3	3.78 µs	3.96 µs	5.20 µs	3.38 µs	39
trav: neighborhood()	gpu	1.02 µs	1.07 µs	2.00 µs	842 ns	154
trav: closedNeighborhood()	v3	5.76 µs	5.62 µs	6.10 µs	5.13 µs	26
trav: closedNeighborhood()	gpu	2.14 µs	2.30 µs	2.93 µs	2.07 µs	69
trav: connectedEdges()	v3	990 ns	1.07 µs	1.79 µs	967 ns	156
trav: connectedEdges()	gpu	388 ns	407 ns	742 ns	370 ns	415
trav: outgoers()	v3	2.28 µs	2.42 µs	3.02 µs	2.18 µs	66
trav: outgoers()	gpu	451 ns	528 ns	2.01 µs	442 ns	314
trav: incomers()	v3	2.24 µs	2.41 µs	3.21 µs	2.18 µs	66
trav: incomers()	gpu	454 ns	547 ns	2.07 µs	443 ns	301
trav: successors()	v3	438 ms	441 ms	453 ms	429 ms	12
trav: successors()	gpu	560 µs	591 µs	1.88 ms	545 µs	1153
trav: predecessors()	v3	431 ms	436 ms	455 ms	426 ms	12
trav: predecessors()	gpu	598 µs	635 µs	1.98 ms	568 µs	1071
trav: band neighborhood()	v3	186 µs	197 µs	276 µs	179 µs	3531
trav: band neighborhood()	gpu	37.3 µs	38.2 µs	40.5 µs	36.9 µs	12
trav: band connectedEdges()	v3	24.5 µs	24.8 µs	25.2 µs	24.1 µs	12
trav: band connectedEdges()	gpu	16.6 µs	17.5 µs	36.1 µs	15.7 µs	38848
trav: band roots()	v3	107 µs	116 µs	164 µs	104 µs	5970
trav: band roots()	gpu	4.84 µs	4.85 µs	5.09 µs	4.74 µs	31
trav: band sources()	v3	6.50 µs	6.57 µs	6.93 µs	6.25 µs	22
trav: band sources()	gpu	4.38 µs	4.43 µs	4.74 µs	4.29 µs	34
trav: band connectedNodes()	v3	13.9 µs	13.9 µs	14.0 µs	13.8 µs	9
trav: band connectedNodes()	gpu	4.99 µs	5.02 µs	5.51 µs	4.88 µs	30

algorithms N=2,000 · 1.5 minv4 (gpu) v3

algo: bfs (whole graph)

60×

algo: dfs (whole graph)

57×

algo: dijkstra + pathTo

57×

algo: aStar

3.3×

algo: bellmanFord

20×

algo: kruskal

27×

algo: tarjan SCC

37×

algo: hopcroft-tarjan biconnected

25×

algo: hierholzer (directed)

2.8×

algo: betweennessCentrality (unweighted)

11×

algo: degreeCentralityNormalized

36×

algo: closenessCentrality (one root)

71×

algo: degreeCentrality (one root)

30×

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	56.3 ms	55.7 ms	61.6 ms	49.6 ms	9
algo: bfs (whole graph)	gpu	931 µs	943 µs	1.42 ms	867 µs	735
algo: dfs (whole graph)	v3	48.4 ms	48.9 ms	49.2 ms	48.0 ms	10
algo: dfs (whole graph)	gpu	849 µs	865 µs	1.15 ms	827 µs	787
algo: dijkstra + pathTo	v3	54.2 ms	54.4 ms	56.5 ms	52.0 ms	9
algo: dijkstra + pathTo	gpu	945 µs	969 µs	1.26 ms	926 µs	717
algo: aStar	v3	1.69 ms	1.73 ms	2.44 ms	1.65 ms	399
algo: aStar	gpu	509 µs	525 µs	642 µs	492 µs	1331
algo: bellmanFord	v3	237 ms	238 ms	262 ms	217 ms	12
algo: bellmanFord	gpu	11.9 ms	12.2 ms	15.8 ms	11.9 ms	51
algo: kruskal	v3	13.5 ms	14.6 ms	18.5 ms	12.9 ms	43
algo: kruskal	gpu	495 µs	524 µs	788 µs	469 µs	1310
algo: tarjan SCC	v3	23.1 ms	23.9 ms	26.9 ms	21.5 ms	25
algo: tarjan SCC	gpu	625 µs	674 µs	1.19 ms	612 µs	1009
algo: hopcroft-tarjan biconnected	v3	86.0 ms	87.9 ms	96.3 ms	82.0 ms	12
algo: hopcroft-tarjan biconnected	gpu	3.48 ms	3.56 ms	4.51 ms	3.39 ms	189
algo: hierholzer (directed)	v3	22.5 ms	23.6 ms	26.3 ms	21.7 ms	25
algo: hierholzer (directed)	gpu	7.95 ms	8.07 ms	9.62 ms	7.70 ms	75
algo: betweennessCentrality (unweighted)	v3	4.06 s	4.06 s	4.11 s	4.00 s	12
algo: betweennessCentrality (unweighted)	gpu	359 ms	377 ms	437 ms	347 ms	12
algo: degreeCentralityNormalized	v3	14.4 ms	15.5 ms	18.6 ms	13.9 ms	41
algo: degreeCentralityNormalized	gpu	402 µs	414 µs	616 µs	396 µs	1685
algo: closenessCentrality (one root)	v3	60.4 ms	62.4 ms	72.0 ms	58.5 ms	12
algo: closenessCentrality (one root)	gpu	847 µs	880 µs	1.31 ms	817 µs	787
algo: degreeCentrality (one root)	v3	7.51 µs	7.41 µs	7.99 µs	6.63 µs	19
algo: degreeCentrality (one root)	gpu	248 ns	254 ns	338 ns	242 ns	668

algorithms N=500 · 6.9 minv4 (gpu) v3

algo: bfs (whole graph)

61×

algo: dfs (whole graph)

63×

algo: dijkstra + pathTo

52×

algo: aStar

3.5×

algo: bellmanFord

21×

algo: kruskal

24×

algo: tarjan SCC

33×

algo: hopcroft-tarjan biconnected

25×

algo: hierholzer (directed)

5.2×

algo: betweennessCentrality (unweighted)

13×

algo: degreeCentralityNormalized

37×

algo: closenessCentrality (one root)

68×

algo: degreeCentrality (one root)

30×

algo: pageRank (20 iters)

69×

algo: closenessCentrality (weighted, one root)

56×

algo: floydWarshall

1.98×

algo: closenessCentralityNormalized

1.63×

algo: markovClustering

1.11×

algo: hierarchicalClustering (threshold)

8.4×

algo: kMeans (fixed centroids)

4.6×

algo: kMedoids

9.0×

algo: fuzzyCMeans

11×

algo: affinityPropagation

5.3×

algo: kargerStein

1.08×

algo: betweennessCentrality (weighted)

9.1×

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	13.6 ms	14.4 ms	18.5 ms	11.6 ms	44
algo: bfs (whole graph)	gpu	223 µs	234 µs	403 µs	204 µs	2980
algo: dfs (whole graph)	v3	11.5 ms	12.3 ms	15.0 ms	11.1 ms	53
algo: dfs (whole graph)	gpu	183 µs	189 µs	245 µs	179 µs	3697
algo: dijkstra + pathTo	v3	12.4 ms	12.8 ms	15.7 ms	11.4 ms	50
algo: dijkstra + pathTo	gpu	240 µs	245 µs	303 µs	233 µs	2796
algo: aStar	v3	437 µs	447 µs	562 µs	424 µs	1563
algo: aStar	gpu	124 µs	129 µs	225 µs	121 µs	5414
algo: bellmanFord	v3	17.0 ms	17.2 ms	18.5 ms	16.6 ms	36
algo: bellmanFord	gpu	824 µs	846 µs	1.95 ms	789 µs	821
algo: kruskal	v3	3.18 ms	3.48 ms	6.88 ms	3.13 ms	197
algo: kruskal	gpu	133 µs	139 µs	257 µs	130 µs	4994
algo: tarjan SCC	v3	5.28 ms	5.65 ms	9.07 ms	5.12 ms	119
algo: tarjan SCC	gpu	159 µs	203 µs	421 µs	153 µs	3404
algo: hopcroft-tarjan biconnected	v3	19.9 ms	20.9 ms	23.8 ms	18.9 ms	30
algo: hopcroft-tarjan biconnected	gpu	805 µs	863 µs	1.46 ms	777 µs	807
algo: hierholzer (directed)	v3	4.36 ms	4.66 ms	7.62 ms	4.20 ms	143
algo: hierholzer (directed)	gpu	831 µs	859 µs	1.28 ms	803 µs	793
algo: betweennessCentrality (unweighted)	v3	260 ms	263 ms	267 ms	257 ms	12
algo: betweennessCentrality (unweighted)	gpu	20.6 ms	21.1 ms	22.2 ms	20.1 ms	29
algo: degreeCentralityNormalized	v3	3.43 ms	3.70 ms	6.97 ms	3.39 ms	182
algo: degreeCentralityNormalized	gpu	93.9 µs	97.1 µs	149 µs	92.2 µs	7194
algo: closenessCentrality (one root)	v3	13.4 ms	14.3 ms	17.4 ms	13.0 ms	45
algo: closenessCentrality (one root)	gpu	197 µs	205 µs	360 µs	188 µs	3407
algo: degreeCentrality (one root)	v3	7.39 µs	7.25 µs	8.05 µs	6.59 µs	20
algo: degreeCentrality (one root)	gpu	243 ns	254 ns	475 ns	237 ns	669
algo: pageRank (20 iters)	v3	2.76 ms	2.90 ms	6.43 ms	2.67 ms	236
algo: pageRank (20 iters)	gpu	40.1 µs	45.8 µs	175 µs	38.6 µs	15137
algo: closenessCentrality (weighted, one root)	v3	14.4 ms	15.2 ms	18.6 ms	13.6 ms	42
algo: closenessCentrality (weighted, one root)	gpu	260 µs	269 µs	461 µs	251 µs	2597
algo: floydWarshall	v3	248 ms	274 ms	384 ms	245 ms	12
algo: floydWarshall	gpu	125 ms	128 ms	128 ms	124 ms	12
algo: closenessCentralityNormalized	v3	274 ms	274 ms	277 ms	270 ms	12
algo: closenessCentralityNormalized	gpu	168 ms	169 ms	173 ms	165 ms	12
algo: markovClustering	v3	4.14 s	4.38 s	4.66 s	4.10 s	12
algo: markovClustering	gpu	3.75 s	3.76 s	3.78 s	3.74 s	12
algo: hierarchicalClustering (threshold)	v3	34.8 ms	35.1 ms	35.7 ms	34.4 ms	16
algo: hierarchicalClustering (threshold)	gpu	4.12 ms	4.35 ms	7.94 ms	4.03 ms	154
algo: kMeans (fixed centroids)	v3	9.05 ms	9.10 ms	9.77 ms	8.80 ms	73
algo: kMeans (fixed centroids)	gpu	1.96 ms	2.01 ms	3.24 ms	1.92 ms	345
algo: kMedoids	v3	377 ms	315 ms	452 ms	69.8 ms	12
algo: kMedoids	gpu	41.7 ms	38.8 ms	44.9 ms	20.6 ms	15
algo: fuzzyCMeans	v3	78.5 ms	79.7 ms	80.8 ms	77.9 ms	12
algo: fuzzyCMeans	gpu	6.95 ms	7.03 ms	7.94 ms	6.71 ms	95
algo: affinityPropagation	v3	208 ms	221 ms	282 ms	203 ms	12
algo: affinityPropagation	gpu	39.0 ms	39.5 ms	41.0 ms	38.4 ms	11
algo: kargerStein	v3	327 ms	332 ms	332 ms	316 ms	12
algo: kargerStein	gpu	303 ms	307 ms	315 ms	297 ms	12
algo: betweennessCentrality (weighted)	v3	730 ms	729 ms	740 ms	706 ms	12
algo: betweennessCentrality (weighted)	gpu	79.8 ms	81.7 ms	81.8 ms	79.4 ms	12

mappers N=2,000 · 8.2 s

mapper: bulk write, color — cpu eval

2.58 ms

mapper: bulk write, color — gpu path

176 µs

mapper: bulk write, color + size — cpu eval

3.19 ms

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

2.70 ms

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

4.08 ms

100 µs1 ms10 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
mapper: bulk write, color	cpu eval	2.58 ms	2.66 ms	4.36 ms	2.48 ms	255
mapper: bulk write, color	gpu path	176 µs	185 µs	328 µs	172 µs	3779
mapper: bulk write, color + size	cpu eval	3.19 ms	3.24 ms	3.87 ms	3.00 ms	212
mapper: bulk write, color + size	gpu color / cpu size	2.70 ms	2.77 ms	3.31 ms	2.64 ms	250
mapper: bulk write, mapped labels (round-5 path)	label refresh	4.08 ms	4.34 ms	7.15 ms	3.53 ms	156

layouts N=2,000 · 30.8 sv4 (gpu) v3

154×

17×

61×

13×

26×

36×

11×

120 ms

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

109 ms

57.0 ms

366 µs

31.6 µs

523 ns

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

39.3 µs

10×

100 ns1 µs10 µs100 µs1 ms10 ms100 ms1 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
layout: grid	v3	11.5 ms	12.4 ms	16.0 ms	10.6 ms	51
layout: grid	gpu	74.6 µs	80.5 µs	365 µs	72.7 µs	7961
layout: preset	v3	5.78 ms	6.46 ms	12.4 ms	5.52 ms	101
layout: preset	gpu	331 µs	350 µs	712 µs	294 µs	1837
layout: preset (fn form)	v3	6.07 ms	6.59 ms	11.6 ms	5.39 ms	102
layout: preset (fn form)	gpu	99.0 µs	111 µs	389 µs	97.1 µs	6181
layout: circle	v3	10.0 ms	10.4 ms	14.1 ms	9.11 ms	58
layout: circle	gpu	770 µs	879 µs	2.82 ms	757 µs	773
layout: concentric	v3	30.0 ms	29.5 ms	31.5 ms	27.0 ms	18
layout: concentric	gpu	1.18 ms	1.34 ms	4.65 ms	1.15 ms	507
layout: breadthfirst	v3	400 ms	413 ms	455 ms	395 ms	12
layout: breadthfirst	gpu	11.1 ms	11.7 ms	16.1 ms	10.1 ms	52
layout: random	v3	5.56 ms	5.87 ms	9.27 ms	5.30 ms	114
layout: random	gpu	520 µs	579 µs	1.55 ms	499 µs	1184
layout: force (CPU executor, 20 iterations)	gpu	120 ms	132 ms	143 ms	104 ms	12
layout: force seed — spectral vs scatter (20 iterations)	spectral (landmark MDS, the default)	109 ms	111 ms	118 ms	102 ms	12
layout: force seed — spectral vs scatter (20 iterations)	scatter	57.0 ms	57.1 ms	57.6 ms	56.1 ms	9
layout: plumbing (layoutPositions vs setPositions)	finisher	366 µs	406 µs	924 µs	357 µs	1707
layout: plumbing (layoutPositions vs setPositions)	bulk	31.6 µs	40.7 µs	255 µs	29.5 µs	16081
layout: contract fixed cost (empty impl vs bulk placement)	wrapper only	523 ns	969 ns	3.86 µs	502 ns	166
layout: contract fixed cost (empty impl vs bulk placement)	wrapper + bulk write	39.3 µs	41.1 µs	49.2 µs	38.1 µs	12
layout: scope (eles.layout, 10% of nodes)	v3	999 µs	1.10 ms	2.01 ms	946 µs	614
layout: scope (eles.layout, 10% of nodes)	gpu	97.8 µs	114 µs	271 µs	95.7 µs	5980

style N=2,000 · 33.8 sv4 (gpu) v3

style: sheet swap (compile + applyAll)

1.32×

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

29.1 µs

style: compile vs compile + apply — compile + apply

10.1 ms

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

1.63×

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

10.1 ms

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

10.9 ms

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

10.3 ms

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

10.4 ms

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

10.4 ms

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

14.5 ms

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

6.77 µs

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

32.7 µs

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

595 µs

style: read one prop (color)

1.65×

style: read one prop (width)

2.2×

style: numericStyle

2.8×

style: renderedStyle (one prop)

15×

style: whole-object style()

6.1×

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

3.3×

style: bypass set + removeStyle round-trip

4.3×

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

33.6 µs

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

333 µs

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

11.1 ms

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

11.4 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	13.6 ms	13.9 ms	18.0 ms	12.5 ms	45
style: sheet swap (compile + applyAll)	gpu	10.3 ms	10.5 ms	12.9 ms	9.90 ms	61
style: compile vs compile + apply	compile only (in batch)	29.1 µs	38.7 µs	75.1 µs	23.8 µs	16724
style: compile vs compile + apply	compile + apply	10.1 ms	10.5 ms	15.8 ms	9.89 ms	61
style: first apply on add (256-node band)	v3	1.33 ms	1.60 ms	6.80 ms	1.11 ms	426
style: first apply on add (256-node band)	gpu	814 µs	916 µs	2.72 ms	781 µs	754
style: applyAll, flat vs compound (100 parents — the 14.6 partition)	flat	10.1 ms	10.3 ms	12.5 ms	9.91 ms	64
style: applyAll, flat vs compound (100 parents — the 14.6 partition)	compound	10.9 ms	11.0 ms	13.5 ms	10.5 ms	59
style: applyAll — constant vs state-case (the 57.1d partition)	constant sheet	10.3 ms	10.6 ms	12.7 ms	10.1 ms	62
style: applyAll — constant vs state-case (the 57.1d partition)	state-case sheet	10.4 ms	10.6 ms	12.4 ms	10.1 ms	61
style: applyAll — state-case vs data-case (what the partition avoids)	state-case sheet	10.4 ms	10.5 ms	12.5 ms	10.0 ms	60
style: applyAll — state-case vs data-case (what the partition avoids)	data-case sheet	14.5 ms	15.0 ms	17.6 ms	14.0 ms	41
style: select+unselect 256-band — constant vs state-case sheet	constant sheet (skips restyle)	6.77 µs	6.80 µs	7.00 µs	6.66 µs	20
style: select+unselect 256-band — constant vs state-case sheet	paint-only state sheet (round-61 diff path)	32.7 µs	33.8 µs	57.4 µs	31.4 µs	19421
style: select+unselect 256-band — constant vs state-case sheet	state-case sheet (all channels; full write)	595 µs	611 µs	884 µs	564 µs	1095
style: read one prop (color)	v3	95.0 ns	107 ns	622 ns	86.0 ns	1981731
style: read one prop (color)	gpu	57.6 ns	63.0 ns	134 ns	55.7 ns	2708
style: read one prop (width)	v3	115 ns	122 ns	433 ns	95.0 ns	1876987
style: read one prop (width)	gpu	52.6 ns	57.3 ns	307 ns	50.9 ns	2974
style: numericStyle	v3	140 ns	140 ns	393 ns	103 ns	1743803
style: numericStyle	gpu	49.9 ns	54.0 ns	112 ns	48.4 ns	3160
style: renderedStyle (one prop)	v3	931 ns	1.01 µs	2.07 µs	891 ns	514253
style: renderedStyle (one prop)	gpu	62.8 ns	67.3 ns	175 ns	60.2 ns	2541
style: whole-object style()	v3	83.5 µs	88.9 µs	184 µs	82.2 µs	7825
style: whole-object style()	gpu	13.7 µs	15.4 µs	51.9 µs	13.1 µs	43954
style: bypass set (ele.style name,value)	v3	10.2 µs	10.2 µs	10.5 µs	9.88 µs	11
style: bypass set (ele.style name,value)	gpu	3.10 µs	3.36 µs	4.22 µs	2.97 µs	45
style: bypass set + removeStyle round-trip	v3	20.3 µs	21.9 µs	22.3 µs	19.7 µs	12
style: bypass set + removeStyle round-trip	gpu	4.74 µs	4.95 µs	5.55 µs	4.60 µs	29
style: select+unselect 256-band under the default sheet — bypass punch-out	no bypasses (diff path; the O(0) gate row)	33.6 µs	35.3 µs	72.4 µs	32.2 µs	19548
style: select+unselect 256-band under the default sheet — bypass punch-out	128 of 256 bypassed (punched out to merged writes)	333 µs	349 µs	536 µs	327 µs	1999
style: sheet swap (applyAll) — 0 vs 128 bypasses	no bypasses	11.1 ms	11.4 ms	14.3 ms	10.8 ms	57
style: sheet swap (applyAll) — 0 vs 128 bypasses	128 bypassed	11.4 ms	11.5 ms	12.5 ms	11.1 ms	55

style-bundle N=2,000 · 19.6 sv4 (gpu) v3

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

1.07×

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

1.26×

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

1.66×

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

1.67×

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

1.21×

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

1.21×

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

14.7 µs

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

76.1 ns

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

73.6 ns

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

66.4 ns

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

62.0 ns

the getters that ride readProp — gpu numericStyle( width )

61.6 ns

the getters that ride readProp — gpu effectiveOpacity()

21.7 ns

the getters that ride readProp — gpu renderedStyle( width )

79.4 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	58.9 ns	61.8 ns	121 ns	56.9 ns	2772
style( 'background-color' ) (was #4 in the switch)	gpu	54.8 ns	57.5 ns	100 ns	52.6 ns	2978
style( 'border-width' ) (was #6 in the switch)	v3	63.6 ns	65.6 ns	107 ns	61.0 ns	2617
style( 'border-width' ) (was #6 in the switch)	gpu	50.5 ns	52.2 ns	80.9 ns	48.5 ns	3275
style( 'text-wrap' ) (was #73 in the switch)	v3	59.6 ns	61.3 ns	86.7 ns	57.3 ns	2789
style( 'text-wrap' ) (was #73 in the switch)	gpu	35.8 ns	37.2 ns	69.5 ns	35.0 ns	4607
style( 'text-max-width' ) (was #74 in the switch)	v3	58.1 ns	60.6 ns	98.7 ns	56.3 ns	2832
style( 'text-max-width' ) (was #74 in the switch)	gpu	34.8 ns	36.1 ns	66.1 ns	33.8 ns	4756
style( 'taxi-radius' ) (was #142 in the switch)	v3	59.9 ns	61.5 ns	102 ns	59.1 ns	2794
style( 'taxi-radius' ) (was #142 in the switch)	gpu	49.6 ns	54.7 ns	113 ns	48.4 ns	3117
style( 'target-distance-from-node' ) (was #150 in the switch)	v3	60.2 ns	62.4 ns	90.7 ns	58.5 ns	2751
style( 'target-distance-from-node' ) (was #150 in the switch)	gpu	49.6 ns	50.6 ns	86.5 ns	48.8 ns	3381
whole-object style() (gpu-only: v3 returns a different shape)	gpu node	13.3 µs	13.9 µs	24.6 µs	12.6 µs	48541
whole-object style() (gpu-only: v3 returns a different shape)	gpu edge	14.7 µs	15.5 µs	34.1 µs	14.1 µs	43147
colour-valued vs numeric reads (the string build)	gpu background-color (colour)	76.1 ns	80.2 ns	163 ns	73.6 ns	2129
colour-valued vs numeric reads (the string build)	gpu border-color (colour)	73.6 ns	76.0 ns	110 ns	71.8 ns	2253
colour-valued vs numeric reads (the string build)	gpu border-width (number)	66.4 ns	69.5 ns	107 ns	65.6 ns	2461
colour-valued vs numeric reads (the string build)	gpu width (number)	62.0 ns	64.0 ns	95.3 ns	59.4 ns	2672
the getters that ride readProp	gpu numericStyle( width )	61.6 ns	62.9 ns	78.0 ns	60.0 ns	2647
the getters that ride readProp	gpu effectiveOpacity()	21.7 ns	22.2 ns	32.5 ns	20.9 ns	7701
the getters that ride readProp	gpu renderedStyle( width )	79.4 ns	81.7 ns	111 ns	78.0 ns	2091

load N=2,000 · 19.7 sv4 (gpu) v3

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

6.1×

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

21.6 ms

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

12.6 ms

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

11.8 ms

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

300 µs

load: conversion — toColumnarElements

706 µs

load: conversion — serializeElements (from columnar)

190 µs

load: conversion — serializeElements (from defs)

869 µs

load: conversion — deserializeElements

4.60 µs

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

501 µs

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

717 µs

load: cy.json()

2.8×

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

1.37×

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

2.73 ms

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

2.37 ms

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

2.31 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	136 ms	153 ms	177 ms	134 ms	12
load: init from the definition form (construct + dispose)	gpu	22.4 ms	23.2 ms	26.9 ms	19.9 ms	25
load: v4 ingest forms (construct + dispose)	definition form	21.6 ms	22.1 ms	25.6 ms	19.6 ms	27
load: v4 ingest forms (construct + dispose)	columnar form	12.6 ms	13.4 ms	23.3 ms	12.0 ms	46
load: v4 ingest forms (construct + dispose)	wire buffer	11.8 ms	13.1 ms	23.6 ms	11.4 ms	48
load: v4 ingest forms (construct + dispose)	def clone alone (control)	300 µs	612 µs	12.7 ms	206 µs	1076
load: conversion	toColumnarElements	706 µs	797 µs	2.23 ms	673 µs	850
load: conversion	serializeElements (from columnar)	190 µs	364 µs	955 µs	173 µs	1908
load: conversion	serializeElements (from defs)	869 µs	902 µs	1.73 ms	858 µs	771
load: conversion	deserializeElements	4.60 µs	4.49 µs	5.78 µs	3.29 µs	29
load: export from a live graph	cy.serialize() (wire)	501 µs	578 µs	2.04 ms	485 µs	1189
load: export from a live graph	cy.json() (v4)	717 µs	1.04 ms	6.77 ms	683 µs	640
load: cy.json()	v3	2.02 ms	2.35 ms	5.55 ms	1.97 ms	289
load: cy.json()	gpu	735 µs	831 µs	2.77 ms	726 µs	837
load: add a 256-node band (add + remove)	v3	3.66 ms	4.38 ms	11.5 ms	3.27 ms	150
load: add a 256-node band (add + remove)	gpu	2.67 ms	2.89 ms	7.20 ms	2.14 ms	233
load: add a 256-node band, v4 forms	definition form	2.73 ms	2.78 ms	4.58 ms	2.16 ms	246
load: add a 256-node band, v4 forms	columnar form	2.37 ms	2.43 ms	3.35 ms	1.95 ms	279
load: add a 256-node band, v4 forms	wire buffer	2.31 ms	2.42 ms	3.33 ms	1.88 ms	284

spatial N=2,000 · 37.6 sv4 (gpu) v3

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

21.0 µs

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

21.3 µs

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

47.6 µs

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

45.2 µs

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

87.0 µs

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

234 µs

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

808 µs

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

683 µs

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

100 µs

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

211 µs

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

807 µs

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

793 µs

pick: background miss (full highWater walk) — gpu

45.6 µs

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

45.8 µs

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

125 µs

box: elementsInBox by covered fraction — 10%

149 µs

box: elementsInBox by covered fraction — 50%

200 µs

box: elementsInBox by covered fraction — 100%

344 µs

box: label containment (boxSelectionIncludesLabels) — off

238 µs

box: label containment (boxSelectionIncludesLabels) — on

263 µs

box: overlap vs contain (straight edges) — contain

245 µs

box: overlap vs contain (straight edges) — overlap

487 µs

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

1.29 ms

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

2.24 ms

bounds: whole-graph boundingBox()

6.3×

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

1.30 ms

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

747 µs

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

33×

bounds: getFitViewport() (compute without committing)

33×

bounds: boundingBoxAt (v4 only) — gpu

2.52 ms

bounds: one node boundingBox()

1.63×

bounds: labelBoundingBox (v4 only) — gpu

426 ns

bounds: compound + taxi conservative scan (round 54, v4 only) — warm (curve bbs memoized)

74.4 µs

bounds: compound + taxi conservative scan (round 54, v4 only) — cold (geometry write per call)

146 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	21.6 µs	37.0 µs	20.2 µs	31813
pick: hit (the walk; one inside-test)	zoom 2	21.3 µs	21.8 µs	22.1 µs	21.0 µs	12
pick: miss walk — exact vs 24 px touch halo (57.9)	exact	47.6 µs	47.8 µs	48.4 µs	46.5 µs	12
pick: miss walk — exact vs 24 px touch halo (57.9)	padPx 24	45.2 µs	46.7 µs	88.9 µs	43.0 µs	14852
pick: shape inside-test, 2000 coincident misses	ellipse	87.0 µs	89.9 µs	142 µs	83.5 µs	7760
pick: shape inside-test, 2000 coincident misses	star	234 µs	249 µs	373 µs	229 µs	2773
pick: shape inside-test, 2000 coincident misses	round-hexagon	808 µs	823 µs	1.09 ms	775 µs	850
pick: shape inside-test, 2000 coincident misses	barrel	683 µs	697 µs	843 µs	663 µs	983
pick: shape inside-test, 2000 coincident misses	cut-rectangle	100 µs	102 µs	150 µs	95.4 µs	6830
pick: shape inside-test, 2000 coincident misses	polygon	211 µs	218 µs	351 µs	204 µs	3215
pick: round-hexagon inside-test by zoom (not affine-invariant)	zoom 1	807 µs	820 µs	1.07 ms	776 µs	853
pick: round-hexagon inside-test by zoom (not affine-invariant)	zoom 2	793 µs	807 µs	965 µs	771 µs	868
pick: background miss (full highWater walk)	gpu	45.6 µs	46.1 µs	46.7 µs	44.9 µs	12
pick: label box in the scan (text-events)	text-events: no	45.8 µs	46.3 µs	46.7 µs	45.4 µs	12
pick: label box in the scan (text-events)	text-events: yes	125 µs	130 µs	240 µs	121 µs	5348
box: elementsInBox by covered fraction	10%	149 µs	155 µs	310 µs	145 µs	4511
box: elementsInBox by covered fraction	50%	200 µs	205 µs	294 µs	192 µs	3403
box: elementsInBox by covered fraction	100%	344 µs	361 µs	846 µs	332 µs	1928
box: label containment (boxSelectionIncludesLabels)	off	238 µs	247 µs	382 µs	235 µs	2832
box: label containment (boxSelectionIncludesLabels)	on	263 µs	276 µs	624 µs	254 µs	2528
box: overlap vs contain (straight edges)	contain	245 µs	256 µs	493 µs	238 µs	2697
box: overlap vs contain (straight edges)	overlap	487 µs	495 µs	923 µs	420 µs	1409
box: overlap vs contain (curved edges — the segment walk)	contain	1.29 ms	1.33 ms	2.43 ms	1.23 ms	522
box: overlap vs contain (curved edges — the segment walk)	overlap	2.24 ms	2.29 ms	2.99 ms	2.16 ms	303
bounds: whole-graph boundingBox()	v3	8.11 ms	8.27 ms	10.8 ms	7.64 ms	79
bounds: whole-graph boundingBox()	gpu	1.28 ms	1.31 ms	1.57 ms	1.27 ms	523
bounds: boundingBox, labels in vs out (v4)	includeLabels: true (default)	1.30 ms	1.32 ms	1.92 ms	1.27 ms	526
bounds: boundingBox, labels in vs out (v4)	includeLabels: false	747 µs	760 µs	1.01 ms	740 µs	920
bounds: cy.fit() (whole-graph scan)	v3	7.77 ms	7.96 ms	9.37 ms	7.40 ms	83
bounds: cy.fit() (whole-graph scan)	gpu	238 µs	247 µs	318 µs	235 µs	2819
bounds: getFitViewport() (compute without committing)	v3	7.73 ms	7.87 ms	8.93 ms	7.48 ms	85
bounds: getFitViewport() (compute without committing)	gpu	236 µs	247 µs	381 µs	233 µs	2825
bounds: boundingBoxAt (v4 only)	gpu	2.52 ms	2.64 ms	3.51 ms	2.45 ms	257
bounds: one node boundingBox()	v3	1.65 µs	1.67 µs	1.99 µs	1.61 µs	96
bounds: one node boundingBox()	gpu	1.01 µs	1.05 µs	1.62 µs	975 ns	157
bounds: labelBoundingBox (v4 only)	gpu	426 ns	475 ns	937 ns	399 ns	352
bounds: compound + taxi conservative scan (round 54, v4 only)	warm (curve bbs memoized)	74.4 µs	78.1 µs	175 µs	72.2 µs	8934
bounds: compound + taxi conservative scan (round 54, v4 only)	cold (geometry write per call)	146 µs	151 µs	217 µs	145 µs	4614

data N=2,000 · 28.2 sv4 (gpu) v3

data: bulk write, numeric column

20×

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

17×

data: bulk write, one new dictionary entry per pass

17×

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

21×

data: removeData (whole collection)

7.8×

data: read one numeric key

1.40×

data: read one string key (dictionary decode)

1.30×

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)

7.0×

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	1.71 ms	1.76 ms	3.00 ms	1.65 ms	378
data: bulk write, numeric column	gpu	84.7 µs	88.6 µs	140 µs	82.6 µs	7896
data: bulk write, dictionary string column (4 values)	v3	1.69 ms	1.76 ms	2.12 ms	1.66 ms	380
data: bulk write, dictionary string column (4 values)	gpu	101 µs	107 µs	314 µs	96.7 µs	6513
data: bulk write, one new dictionary entry per pass	v3	1.75 ms	1.84 ms	3.05 ms	1.69 ms	372
data: bulk write, one new dictionary entry per pass	gpu	100 µs	106 µs	237 µs	94.9 µs	6604
data: bulk write, object column (plain-array fallback)	v3	1.72 ms	1.81 ms	3.00 ms	1.68 ms	381
data: bulk write, object column (plain-array fallback)	gpu	83.9 µs	86.1 µs	126 µs	81.7 µs	8122
data: removeData (whole collection)	v3	2.57 ms	2.62 ms	3.88 ms	2.49 ms	261
data: removeData (whole collection)	gpu	329 µs	345 µs	901 µs	325 µs	2002
data: read one numeric key	v3	38.4 ns	39.9 ns	118 ns	37.2 ns	4304
data: read one numeric key	gpu	27.4 ns	28.9 ns	44.4 ns	26.6 ns	5918
data: read one string key (dictionary decode)	v3	41.5 ns	43.3 ns	86.9 ns	39.5 ns	3968
data: read one string key (dictionary decode)	gpu	32.0 ns	33.5 ns	71.7 ns	27.9 ns	5104
data: read the whole object	v3	46.8 ns	48.6 ns	81.4 ns	44.5 ns	3538
data: read the whole object	gpu	21.1 ns	22.7 ns	48.3 ns	20.5 ns	7502
query: data equality ([bucket = 3] vs { data: { bucket: 3 } })	v3	661 µs	694 µs	1.22 ms	645 µs	1006
query: data equality ([bucket = 3] vs { data: { bucket: 3 } })	gpu	47.5 µs	50.4 µs	112 µs	40.9 µs	13280
query: data comparison ([weight > 3] vs { gt })	v3	784 µs	814 µs	1.34 ms	740 µs	856
query: data comparison ([weight > 3] vs { gt })	gpu	70.1 µs	73.2 µs	213 µs	57.5 µs	9488
query: two keys AND-ed	v3	786 µs	817 µs	1.27 ms	777 µs	852
query: two keys AND-ed	gpu	66.2 µs	69.7 µs	140 µs	65.1 µs	9981
query: membership ([bucket in ...] vs { in })	v3	1.01 ms	1.02 ms	1.36 ms	987 µs	680
query: membership ([bucket in ...] vs { in })	gpu	63.8 µs	67.8 µs	132 µs	61.8 µs	10188
query: predicate function (both sides)	v3	648 µs	665 µs	873 µs	631 µs	1049
query: predicate function (both sides)	gpu	92.1 µs	95.4 µs	157 µs	89.6 µs	7305
query: structural, parents (:parent vs { parent: true })	v3	523 µs	530 µs	638 µs	498 µs	1319
query: structural, parents (:parent vs { parent: true })	gpu	15.5 µs	16.4 µs	19.5 µs	15.1 µs	12
query: structural, childless (:childless vs { parent: false })	v3	808 µs	837 µs	1.12 ms	786 µs	834
query: structural, childless (:childless vs { parent: false })	gpu	60.9 µs	72.9 µs	231 µs	57.4 µs	9347
query: structural, orphans (:orphan vs { child: false })	v3	503 µs	518 µs	986 µs	489 µs	1343
query: structural, orphans (:orphan vs { child: false })	gpu	15.3 µs	16.2 µs	16.5 µs	15.1 µs	12

events N=2,000 · 27.1 sv4 (gpu) v3

emit: position write, no listeners

25×

emit: 1 core listener

8.5×

emit: 1 element (ref-qualified) listener

7.6×

emit: delegated (v3 selector vs v4 predicate)

7.9×

emit: 10 core listeners

6.1×

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

2.3×

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

310 ns

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

869 ns

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

83.9 ns

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

488 ns

anim: start + stop, one element

1.29×

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

1.32 µs

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

898 ns

anim: start + stop, 512-node collection

3.0×

anim: delay() (no channels touched)

1.57×

anim: controls (v4 only) — pause + resume

479 ns

anim: controls (v4 only) — reverse

905 ns

anim: viewport animate + stop

3.0×

10 ns100 ns1 µs10 µs100 µs1 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
emit: position write, no listeners	v3	1.33 µs	1.41 µs	1.99 µs	1.28 µs	111
emit: position write, no listeners	gpu	53.0 ns	56.3 ns	104 ns	50.3 ns	3035
emit: 1 core listener	v3	1.42 µs	1.45 µs	1.88 µs	1.37 µs	105
emit: 1 core listener	gpu	166 ns	174 ns	287 ns	162 ns	906
emit: 1 element (ref-qualified) listener	v3	1.42 µs	1.50 µs	2.38 µs	1.38 µs	102
emit: 1 element (ref-qualified) listener	gpu	186 ns	198 ns	376 ns	180 ns	801
emit: delegated (v3 selector vs v4 predicate)	v3	1.48 µs	1.51 µs	1.76 µs	1.45 µs	103
emit: delegated (v3 selector vs v4 predicate)	gpu	188 ns	206 ns	1.08 µs	181 ns	779
emit: 10 core listeners	v3	1.97 µs	1.98 µs	2.15 µs	1.90 µs	76
emit: 10 core listeners	gpu	323 ns	327 ns	377 ns	309 ns	476
emit: on() + off() round-trip	v3	480 ns	539 ns	1.54 µs	420 ns	302
emit: on() + off() round-trip	gpu	213 ns	256 ns	871 ns	180 ns	641
emit: flat vs phased (compound bubbling, 14.5)	orphan (flat emit)	310 ns	357 ns	2.21 µs	280 ns	966731
emit: flat vs phased (compound bubbling, 14.5)	child of 2 ancestors (phased)	869 ns	929 ns	1.27 µs	846 ns	170
emit: flat vs phased, no listeners	orphan (flat)	83.9 ns	93.0 ns	675 ns	81.3 ns	1816
emit: flat vs phased, no listeners	child (phased)	488 ns	524 ns	1.85 µs	484 ns	316
anim: start + stop, one element	v3	821 ns	1.19 µs	3.47 µs	765 ns	418849
anim: start + stop, one element	gpu	635 ns	1.06 µs	4.53 µs	587 ns	460806
anim: start against a running animation (round-21 eviction)	disjoint channels (composes)	1.32 µs	2.00 µs	4.24 µs	1.09 µs	75
anim: start against a running animation (round-21 eviction)	overlapping channels (evicts)	898 ns	1.49 µs	3.61 µs	796 ns	100
anim: start + stop, 512-node collection	v3	298 µs	452 µs	1.78 ms	287 µs	1446
anim: start + stop, 512-node collection	gpu	99.9 µs	114 µs	210 µs	95.9 µs	5815
anim: delay() (no channels touched)	v3	647 ns	971 ns	3.41 µs	629 ns	169
anim: delay() (no channels touched)	gpu	411 ns	720 ns	2.67 µs	399 ns	224
anim: controls (v4 only)	pause + resume	479 ns	752 ns	2.65 µs	451 ns	203
anim: controls (v4 only)	reverse	905 ns	1.52 µs	3.45 µs	878 ns	100
anim: viewport animate + stop	v3	575 ns	860 ns	2.99 µs	553 ns	188
anim: viewport animate + stop	gpu	189 ns	224 ns	1.38 µs	187 ns	744

store N=2,000 · 20.9 s

id index: set x 2000 (build) — gpu

290 µs

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

63.0 ns

id index: single-key operations — has

54.9 ns

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

8.97 ns

id index: single-key operations — hashAt

7.01 ns

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

232 ns

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

57.2 µs

adjacency: reads — outDegree (O(1))

6.22 ns

adjacency: reads — outEdges

44.6 ns

adjacency: reads — connectedEdges

118 ns

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

135 ns

blob: write x 2000 (build) — gpu

103 µs

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

30.3 ns

blob: single-record operations — offsetOf

4.28 ns

blob: single-record operations — free + rewrite

152 ns

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

13.0 ns

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

18.1 ns

dirty: take (a frame's drain) — gpu

594 ns

images: registry bookkeeping — acquire (existing url)

210 ns

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

534 ns

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

10.3 ms

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

10.9 ms

charts: data write refreshing chart-values — gpu

2.48 ms

1 ns10 ns100 ns1 µs10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
id index: set x 2000 (build)	gpu	290 µs	305 µs	480 µs	280 µs	2302
id index: single-key operations	get (probe)	63.0 ns	66.9 ns	124 ns	57.6 ns	2561
id index: single-key operations	has	54.9 ns	56.4 ns	89.4 ns	49.6 ns	3030
id index: single-key operations	idAt (memoized hit)	8.97 ns	9.24 ns	17.2 ns	8.71 ns	18440
id index: single-key operations	hashAt	7.01 ns	7.22 ns	14.0 ns	6.91 ns	23448
id index: remove + re-set (churn, drives the round-11 reclaim)	gpu	232 ns	244 ns	335 ns	228 ns	686
adjacency: rebuild (4000 edges, two counting passes)	gpu	57.2 µs	65.7 µs	234 µs	54.8 µs	10577
adjacency: reads	outDegree (O(1))	6.22 ns	6.37 ns	11.0 ns	5.90 ns	26608
adjacency: reads	outEdges	44.6 ns	45.4 ns	56.8 ns	43.7 ns	3767
adjacency: reads	connectedEdges	118 ns	179 ns	2.04 µs	113 ns	935
adjacency: addEdge + removeEdge (the overlay path)	gpu	135 ns	139 ns	178 ns	134 ns	1224
blob: write x 2000 (build)	gpu	103 µs	124 µs	647 µs	100 µs	5606
blob: single-record operations	rewrite in place (same length)	30.3 ns	30.9 ns	44.9 ns	29.5 ns	5519
blob: single-record operations	offsetOf	4.28 ns	4.40 ns	8.08 ns	4.18 ns	38209
blob: single-record operations	free + rewrite	152 ns	177 ns	582 ns	147 ns	962
dirty: mark (per-write span coalescing)	contiguous marks	13.0 ns	13.5 ns	21.8 ns	12.4 ns	12542
dirty: mark (per-write span coalescing)	scattered marks	18.1 ns	18.5 ns	27.1 ns	17.5 ns	9171
dirty: take (a frame's drain)	gpu	594 ns	695 ns	1.33 µs	575 ns	237
images: registry bookkeeping	acquire (existing url)	210 ns	212 ns	250 ns	206 ns	802
images: registry bookkeeping	acquire (new url) + release	534 ns	611 ns	2.87 µs	523 ns	270
charts: sheet apply, chart vs no chart	no chart	10.3 ms	10.4 ms	11.9 ms	9.93 ms	61
charts: sheet apply, chart vs no chart	pie from data (blob record per node)	10.9 ms	11.0 ms	12.7 ms	10.6 ms	59
charts: data write refreshing chart-values	gpu	2.48 ms	2.55 ms	4.25 ms	2.39 ms	267

surface N=2,000 · 2.9 minv4 (gpu) v3

viewport: zoom() get

4.7×

viewport: zoom( n ) set

1.91×

viewport: pan() get

1.05×

viewport: pan( p ) set

1.96×

viewport: panBy()

1.83×

viewport: extent()

1.52×

viewport: renderedExtent()

1.68×

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

2.8×

viewport: minZoom/maxZoom get (2 calls)

15×

viewport: center( eles )

19×

viewport: reset()

106×

viewport: viewport({ zoom, pan })

59×

viewport: getCenterPan( eles ) — gpu

6.21 µs

core: instanceString + headless + styleEnabled (3 calls)

1.24×

core: hasElementWithId + hasCompoundNodes (2 calls)

1.19×

core: options()

85×

core: mutableElements() (v4 materializes; v3 is O(1))

1.39×

core: gating getters (5 calls)

1.11×

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)

8472×

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

17105×

core: batch( fn ) (empty)

6.3×

core: startBatch + endBatch + batching

5.7×

core: collection()

7.9×

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

16.9 µs

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

1.75×

collection: length + empty + nonempty (3 calls)

4.6×

collection: eq/first/last (3 calls)

2.4×

collection: slice()

2.2×

collection: toArray()

12×

collection: sort( fn )

1.61×

collection: reduce( fn )

1.37×

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

1.95×

collection: some + every (2 calls)

1.61×

collection: same + anySame + contains (3 calls)

77×

collection: allAre (all match: the full walk)

1.50×

collection: is (none match: the full walk)

2.4×

collection: union

5.7×

collection: intersection

29×

collection: difference

29×

collection: symmetricDifference

3.8×

collection: absoluteComplement

14×

collection: byGroup

3.0×

collection: diff

3.3×

collection: indexOf

1.24×

traversal: openNeighborhood + closedNeighborhood (2 calls)

3.0×

traversal: connectedNodes

3.7×

traversal: sources + targets (2 calls)

3.4×

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

1.34×

traversal: edgesWith

1.21×

traversal: edgesTo

1.18×

traversal: parallelEdges + codirectedEdges (2 calls)

2.9×

traversal: roots + leaves (2 calls)

17×

traversal: allAreNeighbors

5.1×

traversal: isLoop + isSimple (2 calls)

2.5×

degree: indegree + outdegree (2 calls)

68×

degree: minDegree + maxDegree (2 calls)

36×

degree: minIndegree + maxOutdegree (2 calls)

21×

dimensions: renderedPosition get

2.6×

dimensions: relativePosition get

11×

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

2.9×

dimensions: rendered dims (4 calls)

3.3×

dimensions: renderedBoundingBox

1.98×

dimensions: midpoint + renderedMidpoint (2 calls) — gpu

1.13 µs

dimensions: source/targetEndpoint (2 calls) — gpu

379 ns

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

389 ns

dimensions: shift + silentShift (round-trip)

61×

dimensions: silentPosition

15×

dimensions: positions( fn ) over 100

45×

dimensions: padding + paddedWidth + paddedHeight (3 calls)

14×

curves: isBundledBezier — gpu

20.5 ns

curves: controlPoints + renderedControlPoints (2 calls) — gpu

540 ns

curves: segmentPoints (undefined for bezier) — gpu

33.1 ns

compound: parent + parents + ancestors (3 calls)

3.9×

compound: children + descendants + siblings (3 calls)

4.9×

compound: orphans + nonorphans (2 calls)

7.0×

compound: commonAncestors

130×

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

1.53×

flags: select + unselect + selected (band)

15×

flags: selectify + unselectify + selectable

98×

flags: lock + unlock + locked

96×

flags: grabify + ungrabify + grabbable

96×

flags: show + hide + visible

239×

flags: active/activate/unactivate (3 calls)

17×

flags: panify + unpanify + pannable

95×

flags: takesUpSpace + interactive + transparent (3 calls)

2.7×

flags: effectiveOpacity

2.3×

element data: scratch get + set

7.8×

element data: json() one element

1.90×

element data: jsons() over 100

1.42×

element data: label() — gpu

15.2 ns

core: batchData()

8.5×

core: animated()

4.9×

core: zoomRange get — gpu

2.20 ns

core: forceRender() (headless no-op)

16×

core: resize() / invalidateSize()

2.3×

core: onRender + offRender

2.5×

core: removeScratch()

9872×

core: selectionType get

1.97×

core: isReady()

2.2×

core: window()

2.8×

core: container()

2.6×

core: destroyed()

2.2×

core: removeAllListeners()

8.6×

collection: cy()

3.6×

collection: indexOfId()

1.53×

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

1.11×

collection: silentPositions( fn ) over 100

31×

collection: animated()

2.8×

collection: delayAnimation()

2.9×

collection: removeScratch()

9.1×

collection: component()

44×

collection: layoutDimensions()

3.9×

degree: maxIndegree + minOutdegree (2 calls)

22×

flags: grabbed + hidden + inactive (3 calls)

2.0×

curves: renderedSegmentPoints — gpu

37.3 ns

layout contract: edgeSlots + endpoints + degreeOf — gpu

34.6 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	5.72 ns	5.79 ns	7.80 ns	5.63 ns	29100
viewport: zoom() get	gpu	1.21 ns	1.25 ns	1.94 ns	1.21 ns	127035
viewport: zoom( n ) set	v3	621 ns	631 ns	833 ns	603 ns	262
viewport: zoom( n ) set	gpu	326 ns	333 ns	458 ns	309 ns	504
viewport: pan() get	v3	5.27 ns	5.51 ns	12.7 ns	5.06 ns	30537
viewport: pan() get	gpu	5.01 ns	5.18 ns	11.3 ns	4.85 ns	32528
viewport: pan( p ) set	v3	647 ns	675 ns	1.05 µs	633 ns	247
viewport: pan( p ) set	gpu	330 ns	334 ns	392 ns	317 ns	495
viewport: panBy()	v3	598 ns	617 ns	890 ns	578 ns	272
viewport: panBy()	gpu	326 ns	339 ns	523 ns	315 ns	489
viewport: extent()	v3	29.2 ns	45.7 ns	263 ns	28.2 ns	3684
viewport: extent()	gpu	19.3 ns	26.3 ns	106 ns	18.3 ns	6414
viewport: renderedExtent()	v3	27.4 ns	28.7 ns	50.8 ns	26.4 ns	5894
viewport: renderedExtent()	gpu	16.3 ns	23.2 ns	99.7 ns	15.1 ns	7280
viewport: width() + height() + size() (3 calls)	v3	20.9 ns	21.6 ns	42.8 ns	20.0 ns	7893
viewport: width() + height() + size() (3 calls)	gpu	7.39 ns	7.84 ns	15.9 ns	7.04 ns	21500
viewport: minZoom/maxZoom get (2 calls)	v3	15.0 ns	15.5 ns	20.5 ns	14.5 ns	11006
viewport: minZoom/maxZoom get (2 calls)	gpu	0.999 ns	1.04 ns	1.44 ns	0.997 ns	151502
viewport: center( eles )	v3	120 µs	125 µs	202 µs	113 µs	5558
viewport: center( eles )	gpu	6.20 µs	6.28 µs	6.66 µs	6.07 µs	22
viewport: reset()	v3	931 ns	1.02 µs	1.67 µs	905 ns	162
viewport: reset()	gpu	8.76 ns	9.06 ns	16.7 ns	8.39 ns	18517
viewport: viewport({ zoom, pan })	v3	928 ns	1.02 µs	1.66 µs	906 ns	163
viewport: viewport({ zoom, pan })	gpu	15.9 ns	16.8 ns	38.9 ns	15.1 ns	10132
viewport: getCenterPan( eles )	gpu	6.21 µs	6.23 µs	6.36 µs	6.12 µs	24
core: instanceString + headless + styleEnabled (3 calls)	v3	32.2 ns	33.3 ns	54.2 ns	30.9 ns	5120
core: instanceString + headless + styleEnabled (3 calls)	gpu	26.0 ns	26.9 ns	44.8 ns	24.9 ns	6341
core: hasElementWithId + hasCompoundNodes (2 calls)	v3	37.5 ns	38.7 ns	59.3 ns	35.7 ns	4425
core: hasElementWithId + hasCompoundNodes (2 calls)	gpu	31.5 ns	32.8 ns	55.8 ns	30.8 ns	5200
core: options()	v3	90.6 ns	93.0 ns	128 ns	87.4 ns	1836
core: options()	gpu	1.07 ns	1.15 ns	2.24 ns	1.03 ns	138645
core: mutableElements() (v4 materializes; v3 is O(1))	v3	3.26 ns	3.36 ns	4.64 ns	3.21 ns	49646
core: mutableElements() (v4 materializes; v3 is O(1))	gpu	2.35 ns	2.43 ns	3.92 ns	2.25 ns	68259
core: gating getters (5 calls)	v3	15.2 ns	15.6 ns	24.1 ns	14.8 ns	10865
core: gating getters (5 calls)	gpu	13.6 ns	14.3 ns	34.6 ns	12.3 ns	11886
core: gating setters round-trip (autolock/autoungrabify/autounselectify)	v3	11.3 ns	11.8 ns	21.5 ns	10.7 ns	14392
core: gating setters round-trip (autolock/autoungrabify/autounselectify)	gpu	6.01 ns	6.50 ns	16.2 ns	5.73 ns	26009
core: interaction tuning getters (5 calls)	gpu	7.30 ns	7.57 ns	14.2 ns	7.00 ns	22343
core: graph data() set + get (v3 restyles the graph)	v3	2.84 ms	2.91 ms	4.00 ms	2.76 ms	237
core: graph data() set + get (v3 restyles the graph)	gpu	335 ns	345 ns	538 ns	327 ns	491
core: graph scratch() set + get (v3 restyles the graph)	v3	2.86 ms	2.91 ms	3.43 ms	2.74 ms	237
core: graph scratch() set + get (v3 restyles the graph)	gpu	167 ns	172 ns	225 ns	166 ns	994
core: batch( fn ) (empty)	v3	532 ns	644 ns	1.63 µs	510 ns	261
core: batch( fn ) (empty)	gpu	83.9 ns	86.8 ns	114 ns	82.2 ns	1956
core: startBatch + endBatch + batching	v3	516 ns	634 ns	1.60 µs	494 ns	265
core: startBatch + endBatch + batching	gpu	90.1 ns	164 ns	2.89 µs	86.8 ns	1009
core: collection()	v3	476 ns	575 ns	1.64 µs	447 ns	291
core: collection()	gpu	59.9 ns	62.6 ns	117 ns	57.4 ns	2709
core: compact() (no dead slots — the early-out)	gpu	16.9 µs	21.3 µs	98.5 µs	16.4 µs	32140
collection: element accessors (id/group/isNode/isEdge, 4 calls)	v3	60.6 ns	64.5 ns	189 ns	55.8 ns	2626
collection: element accessors (id/group/isNode/isEdge, 4 calls)	gpu	34.7 ns	37.8 ns	105 ns	33.6 ns	4500
collection: length + empty + nonempty (3 calls)	v3	6.72 ns	6.88 ns	11.4 ns	6.34 ns	24629
collection: length + empty + nonempty (3 calls)	gpu	1.47 ns	1.52 ns	2.00 ns	1.46 ns	106697
collection: eq/first/last (3 calls)	v3	15.4 ns	15.8 ns	25.7 ns	14.9 ns	10843
collection: eq/first/last (3 calls)	gpu	6.50 ns	6.76 ns	14.0 ns	6.20 ns	25173
collection: slice()	v3	3.87 µs	3.92 µs	4.35 µs	3.64 µs	39
collection: slice()	gpu	1.72 µs	1.74 µs	1.95 µs	1.68 µs	92
collection: toArray()	v3	8.79 µs	8.93 µs	9.33 µs	8.63 µs	15
collection: toArray()	gpu	746 ns	759 ns	870 ns	722 ns	216
collection: sort( fn )	v3	14.6 µs	15.1 µs	15.3 µs	14.3 µs	12
collection: sort( fn )	gpu	9.10 µs	9.12 µs	9.31 µs	8.90 µs	15
collection: reduce( fn )	v3	157 ns	160 ns	203 ns	113 ns	1065
collection: reduce( fn )	gpu	114 ns	116 ns	202 ns	113 ns	1474
collection: min + max ( fn ) (2 calls)	v3	11.1 µs	11.3 µs	11.8 µs	10.9 µs	12
collection: min + max ( fn ) (2 calls)	gpu	5.67 µs	5.74 µs	6.12 µs	5.59 µs	25
collection: some + every (2 calls)	v3	885 ns	903 ns	1.24 µs	856 ns	184
collection: some + every (2 calls)	gpu	551 ns	558 ns	602 ns	538 ns	304
collection: same + anySame + contains (3 calls)	v3	7.71 µs	7.73 µs	7.87 µs	7.60 µs	18
collection: same + anySame + contains (3 calls)	gpu	100 ns	102 ns	127 ns	95.1 ns	1678
collection: allAre (all match: the full walk)	v3	3.68 µs	3.68 µs	3.98 µs	3.45 µs	42
collection: allAre (all match: the full walk)	gpu	2.45 µs	2.50 µs	2.96 µs	2.40 µs	63
collection: is (none match: the full walk)	v3	5.58 µs	5.57 µs	5.76 µs	5.29 µs	27
collection: is (none match: the full walk)	gpu	2.37 µs	2.45 µs	3.22 µs	2.33 µs	64
collection: union	v3	92.8 µs	104 µs	163 µs	91.0 µs	6642
collection: union	gpu	16.2 µs	16.6 µs	16.8 µs	15.9 µs	12
collection: intersection	v3	90.9 µs	100 µs	140 µs	87.6 µs	6891
collection: intersection	gpu	3.17 µs	3.20 µs	3.48 µs	3.09 µs	49
collection: difference	v3	92.1 µs	103 µs	164 µs	88.5 µs	6734
collection: difference	gpu	3.20 µs	3.37 µs	5.25 µs	3.15 µs	45
collection: symmetricDifference	v3	25.3 µs	25.6 µs	25.8 µs	24.7 µs	12
collection: symmetricDifference	gpu	6.62 µs	6.64 µs	6.81 µs	6.52 µs	21
collection: absoluteComplement	v3	6.04 ms	6.55 ms	10.2 ms	5.86 ms	98
collection: absoluteComplement	gpu	417 µs	425 µs	606 µs	411 µs	1644
collection: byGroup	v3	1.01 ms	1.02 ms	1.29 ms	988 µs	684
collection: byGroup	gpu	334 µs	345 µs	572 µs	322 µs	2026
collection: diff	v3	37.2 µs	37.7 µs	39.1 µs	36.5 µs	12
collection: diff	gpu	11.4 µs	11.4 µs	11.6 µs	11.2 µs	11
collection: indexOf	v3	28.7 ns	29.4 ns	39.9 ns	28.0 ns	5844
collection: indexOf	gpu	23.2 ns	23.9 ns	35.1 ns	22.5 ns	7152
traversal: openNeighborhood + closedNeighborhood (2 calls)	v3	9.71 µs	9.78 µs	10.2 µs	9.41 µs	14
traversal: openNeighborhood + closedNeighborhood (2 calls)	gpu	3.28 µs	3.41 µs	3.86 µs	3.17 µs	46
traversal: connectedNodes	v3	1.13 µs	1.20 µs	1.99 µs	999 ns	138
traversal: connectedNodes	gpu	303 ns	314 ns	517 ns	295 ns	525
traversal: sources + targets (2 calls)	v3	1.74 µs	1.89 µs	2.74 µs	1.67 µs	87
traversal: sources + targets (2 calls)	gpu	514 ns	676 ns	2.16 µs	501 ns	244
traversal: source() + target() (2 calls)	v3	75.6 ns	77.7 ns	113 ns	60.9 ns	2211
traversal: source() + target() (2 calls)	gpu	56.5 ns	58.4 ns	84.5 ns	54.0 ns	2930
traversal: edgesWith	v3	69.9 µs	72.7 µs	127 µs	68.9 µs	9582
traversal: edgesWith	gpu	57.7 µs	61.8 µs	116 µs	55.8 µs	11189
traversal: edgesTo	v3	70.2 µs	72.6 µs	123 µs	69.2 µs	9603
traversal: edgesTo	gpu	59.5 µs	60.3 µs	62.4 µs	58.9 µs	12
traversal: parallelEdges + codirectedEdges (2 calls)	v3	3.52 µs	3.81 µs	4.47 µs	3.41 µs	41
traversal: parallelEdges + codirectedEdges (2 calls)	gpu	1.21 µs	1.36 µs	1.95 µs	1.17 µs	121
traversal: roots + leaves (2 calls)	v3	5.75 ms	6.13 ms	9.61 ms	5.24 ms	110
traversal: roots + leaves (2 calls)	gpu	342 µs	356 µs	517 µs	329 µs	1965
traversal: allAreNeighbors	v3	199 µs	212 µs	288 µs	192 µs	3282
traversal: allAreNeighbors	gpu	38.6 µs	39.4 µs	41.3 µs	38.1 µs	12
traversal: isLoop + isSimple (2 calls)	v3	132 ns	135 ns	208 ns	127 ns	1263
traversal: isLoop + isSimple (2 calls)	gpu	53.5 ns	55.1 ns	71.5 ns	51.6 ns	3102
degree: indegree + outdegree (2 calls)	v3	3.35 µs	3.65 µs	4.47 µs	3.29 µs	43
degree: indegree + outdegree (2 calls)	gpu	49.0 ns	50.6 ns	79.3 ns	47.4 ns	3284
degree: minDegree + maxDegree (2 calls)	v3	15.9 ms	17.2 ms	21.1 ms	15.3 ms	36
degree: minDegree + maxDegree (2 calls)	gpu	441 µs	470 µs	953 µs	435 µs	1488
degree: minIndegree + maxOutdegree (2 calls)	v3	8.91 ms	9.59 ms	13.7 ms	8.65 ms	68
degree: minIndegree + maxOutdegree (2 calls)	gpu	423 µs	435 µs	609 µs	418 µs	1605
dimensions: renderedPosition get	v3	74.8 ns	80.5 ns	181 ns	73.5 ns	2107
dimensions: renderedPosition get	gpu	29.3 ns	33.4 ns	119 ns	28.4 ns	5041
dimensions: relativePosition get	v3	858 ns	878 ns	1.17 µs	841 ns	190
dimensions: relativePosition get	gpu	76.1 ns	81.4 ns	159 ns	73.5 ns	2082
dimensions: width + height + outerWidth + outerHeight (4 calls)	v3	617 ns	620 ns	698 ns	593 ns	273
dimensions: width + height + outerWidth + outerHeight (4 calls)	gpu	210 ns	214 ns	301 ns	201 ns	795
dimensions: rendered dims (4 calls)	v3	717 ns	723 ns	794 ns	699 ns	234
dimensions: rendered dims (4 calls)	gpu	216 ns	221 ns	275 ns	211 ns	773
dimensions: renderedBoundingBox	v3	1.81 µs	1.83 µs	2.25 µs	1.74 µs	87
dimensions: renderedBoundingBox	gpu	914 ns	921 ns	1.02 µs	891 ns	180
dimensions: midpoint + renderedMidpoint (2 calls)	gpu	1.13 µs	1.13 µs	1.20 µs	1.10 µs	143
dimensions: source/targetEndpoint (2 calls)	gpu	379 ns	383 ns	428 ns	371 ns	443
dimensions: rendered source/targetEndpoint (2 calls)	gpu	389 ns	396 ns	447 ns	385 ns	429
dimensions: shift + silentShift (round-trip)	v3	297 µs	307 µs	610 µs	291 µs	2281
dimensions: shift + silentShift (round-trip)	gpu	4.84 µs	4.84 µs	4.95 µs	4.74 µs	32
dimensions: silentPosition	v3	850 ns	858 ns	927 ns	827 ns	195
dimensions: silentPosition	gpu	55.8 ns	58.1 ns	83.3 ns	53.9 ns	2928
dimensions: positions( fn ) over 100	v3	177 µs	182 µs	242 µs	172 µs	3828
dimensions: positions( fn ) over 100	gpu	3.89 µs	3.86 µs	4.07 µs	3.58 µs	40
dimensions: padding + paddedWidth + paddedHeight (3 calls)	v3	2.01 µs	2.07 µs	2.84 µs	1.99 µs	77
dimensions: padding + paddedWidth + paddedHeight (3 calls)	gpu	142 ns	149 ns	249 ns	140 ns	1151
curves: isBundledBezier	gpu	20.5 ns	21.3 ns	33.0 ns	20.0 ns	7995
curves: controlPoints + renderedControlPoints (2 calls)	gpu	540 ns	552 ns	768 ns	527 ns	297
curves: segmentPoints (undefined for bezier)	gpu	33.1 ns	34.2 ns	48.4 ns	31.8 ns	5001
compound: parent + parents + ancestors (3 calls)	v3	3.12 µs	3.43 µs	4.57 µs	2.98 µs	46
compound: parent + parents + ancestors (3 calls)	gpu	802 ns	1.05 µs	2.30 µs	772 ns	156
compound: children + descendants + siblings (3 calls)	v3	8.90 µs	8.85 µs	9.14 µs	7.93 µs	16
compound: children + descendants + siblings (3 calls)	gpu	1.83 µs	2.21 µs	3.34 µs	1.76 µs	72
compound: orphans + nonorphans (2 calls)	v3	3.71 µs	3.98 µs	4.74 µs	3.53 µs	39
compound: orphans + nonorphans (2 calls)	gpu	528 ns	668 ns	2.62 µs	514 ns	245
compound: commonAncestors	v3	9.72 µs	9.75 µs	9.91 µs	9.51 µs	14
compound: commonAncestors	gpu	74.8 ns	78.1 ns	121 ns	73.4 ns	2183
compound: isParent/isChild/isChildless/isOrphan (4 calls)	v3	70.8 ns	72.8 ns	96.9 ns	61.1 ns	2359
compound: isParent/isChild/isChildless/isOrphan (4 calls)	gpu	46.2 ns	48.3 ns	91.5 ns	44.4 ns	3543
flags: select + unselect + selected (band)	v3	201 µs	207 µs	461 µs	196 µs	3383
flags: select + unselect + selected (band)	gpu	13.1 µs	13.2 µs	13.3 µs	13.0 µs	9
flags: selectify + unselectify + selectable	v3	198 µs	204 µs	353 µs	193 µs	3387
flags: selectify + unselectify + selectable	gpu	2.03 µs	2.04 µs	2.12 µs	1.99 µs	80
flags: lock + unlock + locked	v3	196 µs	204 µs	362 µs	191 µs	3427
flags: lock + unlock + locked	gpu	2.05 µs	2.07 µs	2.47 µs	2.02 µs	79
flags: grabify + ungrabify + grabbable	v3	196 µs	202 µs	310 µs	191 µs	3464
flags: grabify + ungrabify + grabbable	gpu	2.04 µs	2.05 µs	2.15 µs	2.00 µs	80
flags: show + hide + visible	v3	2.70 ms	2.86 ms	4.52 ms	2.60 ms	241
flags: show + hide + visible	gpu	11.3 µs	11.4 µs	11.8 µs	11.2 µs	11
flags: active/activate/unactivate (3 calls)	v3	196 µs	202 µs	315 µs	191 µs	3466
flags: active/activate/unactivate (3 calls)	gpu	11.9 µs	11.9 µs	12.0 µs	11.7 µs	11
flags: panify + unpanify + pannable	v3	195 µs	203 µs	340 µs	190 µs	3444
flags: panify + unpanify + pannable	gpu	2.05 µs	2.07 µs	2.42 µs	2.00 µs	79
flags: takesUpSpace + interactive + transparent (3 calls)	v3	187 ns	193 ns	385 ns	178 ns	883
flags: takesUpSpace + interactive + transparent (3 calls)	gpu	68.2 ns	72.4 ns	127 ns	63.9 ns	2356
flags: effectiveOpacity	v3	49.4 ns	52.0 ns	93.2 ns	47.1 ns	3292
flags: effectiveOpacity	gpu	21.8 ns	22.6 ns	36.4 ns	20.9 ns	7580
element data: scratch get + set	v3	1.36 µs	1.37 µs	1.52 µs	1.30 µs	121
element data: scratch get + set	gpu	175 ns	180 ns	262 ns	167 ns	947
element data: json() one element	v3	286 ns	288 ns	340 ns	274 ns	582
element data: json() one element	gpu	150 ns	187 ns	537 ns	148 ns	862
element data: jsons() over 100	v3	28.3 µs	28.7 µs	29.0 µs	27.8 µs	12
element data: jsons() over 100	gpu	20.0 µs	21.2 µs	24.4 µs	19.0 µs	12
element data: label()	gpu	15.2 ns	15.9 ns	27.6 ns	14.3 ns	10580
core: batchData()	v3	2.40 µs	2.51 µs	2.94 µs	2.32 µs	64
core: batchData()	gpu	283 ns	377 ns	2.16 µs	276 ns	437
core: animated()	v3	7.52 ns	7.87 ns	14.4 ns	7.22 ns	21510
core: animated()	gpu	1.53 ns	1.60 ns	2.82 ns	1.53 ns	102960
core: zoomRange get	gpu	2.20 ns	2.37 ns	6.27 ns	2.09 ns	70414
core: forceRender() (headless no-op)	v3	23.8 ns	24.6 ns	47.7 ns	23.0 ns	6964
core: forceRender() (headless no-op)	gpu	1.48 ns	1.56 ns	2.80 ns	1.45 ns	105708
core: resize() / invalidateSize()	v3	287 ns	294 ns	385 ns	279 ns	578
core: resize() / invalidateSize()	gpu	126 ns	131 ns	211 ns	121 ns	1305
core: onRender + offRender	v3	468 ns	528 ns	1.08 µs	455 ns	318
core: onRender + offRender	gpu	190 ns	230 ns	671 ns	185 ns	729
core: removeScratch()	v3	2.81 ms	2.85 ms	3.36 ms	2.73 ms	241
core: removeScratch()	gpu	284 ns	291 ns	390 ns	273 ns	584
core: selectionType get	v3	4.46 ns	4.74 ns	10.7 ns	4.26 ns	35584
core: selectionType get	gpu	2.26 ns	2.44 ns	6.22 ns	2.11 ns	68246
core: isReady()	v3	3.26 ns	3.38 ns	6.32 ns	3.25 ns	49712
core: isReady()	gpu	1.48 ns	1.54 ns	1.94 ns	1.45 ns	106983
core: window()	v3	4.10 ns	4.23 ns	7.60 ns	4.01 ns	39924
core: window()	gpu	1.48 ns	1.52 ns	1.93 ns	1.45 ns	108408
core: container()	v3	3.99 ns	4.22 ns	9.50 ns	3.75 ns	39844
core: container()	gpu	1.51 ns	1.59 ns	2.98 ns	1.51 ns	103292
core: destroyed()	v3	3.30 ns	3.40 ns	6.71 ns	3.18 ns	49365
core: destroyed()	gpu	1.48 ns	1.53 ns	2.21 ns	1.45 ns	107365
core: removeAllListeners()	v3	13.7 ns	14.4 ns	27.4 ns	13.2 ns	11812
core: removeAllListeners()	gpu	1.58 ns	1.66 ns	2.00 ns	1.58 ns	98759
collection: cy()	v3	7.73 ns	8.18 ns	15.2 ns	7.17 ns	20879
collection: cy()	gpu	2.16 ns	2.36 ns	6.12 ns	2.06 ns	70595
collection: indexOfId()	v3	59.4 ns	60.8 ns	85.2 ns	58.4 ns	2814
collection: indexOfId()	gpu	38.9 ns	40.0 ns	57.7 ns	38.1 ns	4283
collection: removed() + inside() (2 calls)	v3	16.7 ns	17.4 ns	28.8 ns	16.1 ns	9839
collection: removed() + inside() (2 calls)	gpu	15.1 ns	15.6 ns	25.4 ns	14.6 ns	10992
collection: silentPositions( fn ) over 100	v3	109 µs	113 µs	213 µs	105 µs	6148
collection: silentPositions( fn ) over 100	gpu	3.53 µs	3.55 µs	3.69 µs	3.48 µs	44
collection: animated()	v3	12.1 ns	12.4 ns	21.7 ns	11.7 ns	13764
collection: animated()	gpu	4.28 ns	4.55 ns	11.2 ns	4.00 ns	37042
collection: delayAnimation()	v3	482 ns	663 ns	2.11 µs	454 ns	254
collection: delayAnimation()	gpu	164 ns	171 ns	300 ns	158 ns	980
collection: removeScratch()	v3	1.87 µs	1.88 µs	2.06 µs	1.81 µs	86
collection: removeScratch()	gpu	205 ns	209 ns	285 ns	200 ns	814
collection: component()	v3	86.3 ms	87.4 ms	90.2 ms	81.7 ms	12
collection: component()	gpu	1.97 ms	2.07 ms	3.04 ms	1.90 ms	327
collection: layoutDimensions()	v3	650 ns	669 ns	988 ns	632 ns	252
collection: layoutDimensions()	gpu	169 ns	174 ns	296 ns	161 ns	979
degree: maxIndegree + minOutdegree (2 calls)	v3	9.57 ms	10.2 ms	14.4 ms	9.18 ms	65
degree: maxIndegree + minOutdegree (2 calls)	gpu	427 µs	437 µs	607 µs	411 µs	1603
flags: grabbed + hidden + inactive (3 calls)	v3	84.9 ns	86.9 ns	113 ns	80.7 ns	1975
flags: grabbed + hidden + inactive (3 calls)	gpu	42.4 ns	43.6 ns	64.5 ns	40.9 ns	3937
curves: renderedSegmentPoints	gpu	37.3 ns	38.8 ns	77.4 ns	34.6 ns	4410
layout contract: edgeSlots + endpoints + degreeOf	gpu	34.6 µs	36.6 µs	62.4 µs	29.5 µs	18849
formats: isColumnarElements + isSerializedElements + isPackedIds	gpu	4.61 ns	4.86 ns	9.81 ns	4.60 ns	34785

compaction N=20,000 · 18.6 s

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

222 µs

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

49.6 µs

cpu pick: node hit — peak

223 µs

cpu pick: node hit — compacted

49.9 µs

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

6.02 ns

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

5.90 ns

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

8.86 ns

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

44.7 ns

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

25.4 µs

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

21.0 µs

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

96.5 µs

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

97.3 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	222 µs	227 µs	280 µs	220 µs	3097
cpu pick: background miss (full highWater walk)	compacted	49.6 µs	50.0 µs	50.4 µs	49.1 µs	12
cpu pick: node hit	peak	223 µs	230 µs	324 µs	217 µs	3058
cpu pick: node hit	compacted	49.9 µs	50.4 µs	50.8 µs	49.2 µs	12
isCurrent: current ref (forwards present vs absent)	no forwards (peak)	6.02 ns	6.13 ns	7.37 ns	5.74 ns	27505
isCurrent: current ref (forwards present vs absent)	forwards present (compacted)	5.90 ns	6.07 ns	10.9 ns	5.62 ns	27879
isCurrent: stale-ref repair (chase + rewrite, alloc included)	current (baseline alloc)	8.86 ns	9.58 ns	22.3 ns	8.45 ns	17592
isCurrent: stale-ref repair (chase + rewrite, alloc included)	stale → repair	44.7 ns	46.0 ns	59.3 ns	43.7 ns	3715
control: nodes({ selected }) scan (order-list, expected parity)	peak	25.4 µs	27.2 µs	36.2 µs	21.6 µs	12
control: nodes({ selected }) scan (order-list, expected parity)	compacted	21.0 µs	22.5 µs	24.6 µs	20.5 µs	12
control: whole-graph boundingBox (expected parity)	peak	96.5 µs	101 µs	190 µs	92.4 µs	6858
control: whole-graph boundingBox (expected parity)	compacted	97.3 µs	100 µs	188 µs	92.3 µs	6949

compound N=2,000 · 9.2 sv4 (gpu) v3

parent drag (shift subtree + bb settle)

191×

child drag (move + parent re-derive)

51×

reparent round-trip

67×

1 µs10 µs100 µs1 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
parent drag (shift subtree + bb settle)	v3	266 µs	304 µs	757 µs	249 µs	2290
parent drag (shift subtree + bb settle)	gpu	1.39 µs	1.40 µs	1.55 µs	1.34 µs	113
child drag (move + parent re-derive)	v3	83.9 µs	92.0 µs	175 µs	79.7 µs	7376
child drag (move + parent re-derive)	gpu	1.66 µs	1.67 µs	1.99 µs	1.59 µs	96
reparent round-trip	v3	85.0 µs	94.0 µs	209 µs	82.4 µs	7332
reparent round-trip	gpu	1.26 µs	1.28 µs	1.65 µs	1.23 µs	124

curves N=2,000 · 0.7 s

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

26.3 ms

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

23.1 ms

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

386 µs

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

348 µs

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

305 µs

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

288 µs

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

7.58 ms

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

3.27 ms

curve premium: same read again (warm) — curved

2.72 ms

curve premium: same read again (warm) — straight

1.58 ms

curve premium: edge.midpoint x 4000 — curved

1.88 ms

curve premium: edge.midpoint x 4000 — straight

3.06 ms

curve premium: edge.controlPoints x 4000 — curved

1.95 ms

curve premium: edge.controlPoints x 4000 — straight

1.21 ms

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

34.7 ms

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

34.6 ms

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

4.01 ms

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

3.65 ms

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

23.9 ms

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

6.14 ms

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

3.68 ms

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

2.01 ms

100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
curve premium: build (style apply + derivation)	curved	26.3 ms	26.3 ms	26.3 ms	26.3 ms	1
curve premium: build (style apply + derivation)	straight	23.1 ms	23.1 ms	23.1 ms	23.1 ms	1
curve premium: node.position x 1000 (drag)	curved	386 µs	386 µs	386 µs	386 µs	1
curve premium: node.position x 1000 (drag)	straight	348 µs	348 µs	348 µs	348 µs	1
curve premium: nodes().positions() (bulk move)	curved	305 µs	305 µs	305 µs	305 µs	1
curve premium: nodes().positions() (bulk move)	straight	288 µs	288 µs	288 µs	288 µs	1
curve premium: first read after a bulk move (cold)	curved	7.58 ms	7.58 ms	7.58 ms	7.58 ms	1
curve premium: first read after a bulk move (cold)	straight	3.27 ms	3.27 ms	3.27 ms	3.27 ms	1
curve premium: same read again (warm)	curved	2.72 ms	2.72 ms	2.72 ms	2.72 ms	1
curve premium: same read again (warm)	straight	1.58 ms	1.58 ms	1.58 ms	1.58 ms	1
curve premium: edge.midpoint x 4000	curved	1.88 ms	1.88 ms	1.88 ms	1.88 ms	1
curve premium: edge.midpoint x 4000	straight	3.06 ms	3.06 ms	3.06 ms	3.06 ms	1
curve premium: edge.controlPoints x 4000	curved	1.95 ms	1.95 ms	1.95 ms	1.95 ms	1
curve premium: edge.controlPoints x 4000	straight	1.21 ms	1.21 ms	1.21 ms	1.21 ms	1
curve premium: cy.elements().boundingBox() x 20	curved	34.7 ms	34.7 ms	34.7 ms	34.7 ms	1
curve premium: cy.elements().boundingBox() x 20	straight	34.6 ms	34.6 ms	34.6 ms	34.6 ms	1
curve premium: cy.fit() x 20 (store scan)	curved	4.01 ms	4.01 ms	4.01 ms	4.01 ms	1
curve premium: cy.fit() x 20 (store scan)	straight	3.65 ms	3.65 ms	3.65 ms	3.65 ms	1
curve premium: elementsInBox x 20 (half the graph)	curved	23.9 ms	23.9 ms	23.9 ms	23.9 ms	1
curve premium: elementsInBox x 20 (half the graph)	straight	6.14 ms	6.14 ms	6.14 ms	6.14 ms	1
curve premium: edge.hide/show x 1000 (bundle re-fan)	curved	3.68 ms	3.68 ms	3.68 ms	3.68 ms	1
curve premium: edge.hide/show x 1000 (bundle re-fan)	straight	2.01 ms	2.01 ms	2.01 ms	2.01 ms	1

arrows N=2,000 · 3.3 s

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

24.2 ms

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

20.6 ms

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

41.8 ms

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

40.9 ms

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

23.5 ms

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

23.5 ms

arrow premium: edge.sourceEndpoint x 200000 — arrowed

61.0 ms

arrow premium: edge.sourceEndpoint x 200000 — none

61.9 ms

arrow premium: edge.targetEndpoint x 200000 — arrowed

61.8 ms

arrow premium: edge.targetEndpoint x 200000 — none

61.6 ms

arrow premium: edge.midpoint x 200000 — arrowed

146 ms

arrow premium: edge.midpoint x 200000 — none

146 ms

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

35.0 ms

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

29.1 ms

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

20.9 ms

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

21.3 ms

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

5.61 ms

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

5.52 ms

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

76.4 ms

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

60.9 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	24.2 ms	24.2 ms	24.2 ms	24.2 ms	1
arrow premium: build (style apply + arrow fold)	none	20.6 ms	20.6 ms	20.6 ms	20.6 ms	1
arrow premium: cy.style() re-apply x 4	arrowed	41.8 ms	41.8 ms	41.8 ms	41.8 ms	1
arrow premium: cy.style() re-apply x 4	none	40.9 ms	40.9 ms	40.9 ms	40.9 ms	1
arrow premium: node.position x 200000 (drag)	arrowed	23.5 ms	23.5 ms	23.5 ms	23.5 ms	1
arrow premium: node.position x 200000 (drag)	none	23.5 ms	23.5 ms	23.5 ms	23.5 ms	1
arrow premium: edge.sourceEndpoint x 200000	arrowed	61.0 ms	61.0 ms	61.0 ms	61.0 ms	1
arrow premium: edge.sourceEndpoint x 200000	none	61.9 ms	61.9 ms	61.9 ms	61.9 ms	1
arrow premium: edge.targetEndpoint x 200000	arrowed	61.8 ms	61.8 ms	61.8 ms	61.8 ms	1
arrow premium: edge.targetEndpoint x 200000	none	61.6 ms	61.6 ms	61.6 ms	61.6 ms	1
arrow premium: edge.midpoint x 200000	arrowed	146 ms	146 ms	146 ms	146 ms	1
arrow premium: edge.midpoint x 200000	none	146 ms	146 ms	146 ms	146 ms	1
arrow premium: edge.style('target-arrow-shape') x 200000	arrowed	35.0 ms	35.0 ms	35.0 ms	35.0 ms	1
arrow premium: edge.style('target-arrow-shape') x 200000	none	29.1 ms	29.1 ms	29.1 ms	29.1 ms	1
arrow premium: cy.elements().boundingBox() x 20	arrowed	20.9 ms	20.9 ms	20.9 ms	20.9 ms	1
arrow premium: cy.elements().boundingBox() x 20	none	21.3 ms	21.3 ms	21.3 ms	21.3 ms	1
arrow premium: elementsInBox x 20 (half the graph)	arrowed	5.61 ms	5.61 ms	5.61 ms	5.61 ms	1
arrow premium: elementsInBox x 20 (half the graph)	none	5.52 ms	5.52 ms	5.52 ms	5.52 ms	1
arrow premium: control: sourceEndpoint, bezier vs straight	arrowed	76.4 ms	76.4 ms	76.4 ms	76.4 ms	1
arrow premium: control: sourceEndpoint, bezier vs straight	none	60.9 ms	60.9 ms	60.9 ms	60.9 ms	1
arrow premium: edge column bytes	B/edge	156 ms	156 ms	156 ms	156 ms	1

labels N=20,000 · 0.5 s

labels: breakLines x 20000 — gpu

79.1 ms

labels: estimateBlock x 20000 — gpu

93.4 ms

labels: addNode x 20000 — gpu

28.8 ms

labels: setLabel (build) x 20000 — gpu

104 ms

labels: setLabel (rewrite) x 20000 — gpu

102 ms

labels: boundingBox with 20000 label terms — gpu

3.42 ms

1 ms10 ms100 ms1 s

All stats (table view)

group	bench	p50	avg	p99	min	samples
labels: breakLines x 20000	gpu	79.1 ms	79.1 ms	79.1 ms	79.1 ms	1
labels: estimateBlock x 20000	gpu	93.4 ms	93.4 ms	93.4 ms	93.4 ms	1
labels: addNode x 20000	gpu	28.8 ms	28.8 ms	28.8 ms	28.8 ms	1
labels: setLabel (build) x 20000	gpu	104 ms	104 ms	104 ms	104 ms	1
labels: setLabel (rewrite) x 20000	gpu	102 ms	102 ms	102 ms	102 ms	1
labels: boundingBox with 20000 label terms	gpu	3.42 ms	3.42 ms	3.42 ms	3.42 ms	1

transitions N=2,000 · 9.5 s

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

2.29 ms

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

4.69 ms

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

139 µs

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

2.36 µs

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

6.28 µs

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

10.1 ms

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

12.7 ms

1 µs10 µs100 µs1 ms10 ms100 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
transition: auto-extent shift, whole-channel re-derive (2000 slots)	transitions off	2.29 ms	2.38 ms	4.39 ms	2.23 ms	288
transition: auto-extent shift, whole-channel re-derive (2000 slots)	transitions on	4.69 ms	4.91 ms	7.02 ms	4.44 ms	138
transition: bulk tween tick (2000 slots, one manager tick)	tick	139 µs	143 µs	229 µs	134 µs	4874
transition: explicit-domain write (O(changed) — one element)	transitions off	2.36 µs	2.41 µs	3.22 µs	2.30 µs	65
transition: explicit-domain write (O(changed) — one element)	transitions on	6.28 µs	6.21 µs	6.80 µs	5.25 µs	24
transition: sheet swap, whole-graph apply (2000 slots diff + tween)	transitions off	10.1 ms	10.1 ms	10.6 ms	9.87 ms	65
transition: sheet swap, whole-graph apply (2000 slots diff + tween)	transitions on	12.7 ms	12.6 ms	14.4 ms	11.4 ms	51

geometry-tween N=2,000 · 11.0 s

size tick: unlabelled (2000 slots) — tick

541 µs

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

595 µs

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

1.22 ms

edge width tick (4000 slots) — bare

281 µs

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

717 µs

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

177 µs

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

364 µs

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

6.61 ms

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

144 µs

100 µs1 ms10 ms

All stats (table view)

group	bench	p50	avg	p99	min	samples
size tick: unlabelled (2000 slots)	tick	541 µs	563 µs	867 µs	525 µs	1221
size tick: labelled, center/center (re-anchor early-out) (2000 slots)	tick	595 µs	619 µs	930 µs	578 µs	1111
size tick: labelled, hanging (sidecar rewrite per tick) (2000 slots)	tick	1.22 ms	1.31 ms	2.41 ms	1.13 ms	514
edge width tick (4000 slots)	bare	281 µs	289 µs	405 µs	269 µs	2410
edge width tick (4000 slots)	with rides (casing + match-line arrows)	717 µs	734 µs	899 µs	705 µs	954
padding tick + auto-bounds flush (250 parents × 8 children)	tick + flush	177 µs	186 µs	385 µs	172 µs	3756
font-size tick (2000 labels)	wrap none (dims scale-patch)	364 µs	405 µs	916 µs	345 µs	1710
font-size tick (2000 labels)	wrapped (re-break per tick)	6.61 ms	6.68 ms	7.36 ms	6.36 ms	99
baseline: paint (opacity) tick (2000 slots)	tick	144 µs	150 µs	254 µs	141 µs	4665