v3.5 changelog
quickstart →
Pricing Benchmarks Docs Changelog Blog Troubleshooting GitHub Install in 10 min

Fastest GitHub Actions runners: CPU speed.

GitHub Actions CPU performance on 93 runners from 9 providers, 2 to 12 vCPU, x64 and arm64. Scored inside the job, with the CPU each job got.

Fastest single-thread, x64 at 4 vCPU 4,685 Namespace 4x8 (4 vCPU) · PassMark
Fastest single-thread, arm64 at 4 vCPU 4,860 Avrea 4 vCPU · PassMark
GitHub's fastest runner at 4 vCPU 2,888 x64 single-thread on GitHub 4-core
Most CPU per dollar, x64 at 4 vCPU 155k RunsOn m8a.xlarge · EBS gp3 (4 vCPU) at EC2 spot · CPU Mark per $/hour

How this shows up in real builds

CPU per provider

Each provider's best variant at any size, its vCPU and RAM beside the nameEach provider's best 2 vCPU variantEach provider's best 4 vCPU variantEach provider's best 8 vCPU variant. One thread sets a serial step's speed; all cores, a parallel build's.

Single-threadx64 · PassMark · higher is better
  1. Namespace4 vCPU · 8 GB4x8 · EPYC · model hiddenbest of 54,685, ×1.00
  2. Avrea2 vCPU · 8 GBEPYC 4585PXbest of 34,320, ×0.92
  3. RunsOn12 vCPU · 48 GBm8azn.3xlarge · EPYC 9R05best of 294,266, ×0.91
  4. Blacksmith8 vCPU · 32 GBEPYC · model hiddenbest of 34,234, ×0.90
  5. Warpbuild4 vCPU · 16 GBRyzen 9 9950Xbest of 34,196, ×0.90
  6. Ubicloud8 vCPU · 32 GBpremium · Ryzen 9 7950X3Dbest of 64,038, ×0.86
  7. StarSling4 vCPU · 16 GBEPYC · model hiddenbest of 23,969, ×0.85
  8. GitHub8 vCPU · 32 GBEPYC 9V45best of 32,899, ×0.62
  9. AWS CodeBuild8 vCPU · 16 GBlarge · Xeon 8488Cbest of 32,156, ×0.46
All coresx64 · PassMark CPU Mark · higher is better
  1. RunsOn12 vCPU · 48 GBm8azn.3xlarge · EPYC 9R05best of 2941,116, 12.0 cores, ×1.00
  2. Namespace8 vCPU · 32 GB8x32 · EPYC · model hiddenbest of 530,113, 7.7 cores, ×0.73
  3. Blacksmith8 vCPU · 32 GBEPYC · model hiddenbest of 327,282, 7.9 cores, ×0.66
  4. StarSling8 vCPU · 32 GBEPYC · model hiddenbest of 225,105, 7.7 cores, ×0.61
  5. Ubicloud8 vCPU · 32 GBpremium · Ryzen 9 7950X3Dbest of 624,720, 7.4 cores, ×0.60
  6. Avrea8 vCPU · 32 GBEPYC 4585PXbest of 321,356, 4.3 cores, ×0.52
  7. Warpbuild8 vCPU · 32 GBm8a.2xlarge · Ryzen 9 9950Xbest of 320,991, 4.2 cores, ×0.51
  8. GitHub8 vCPU · 32 GBEPYC 9V45best of 314,420, 4.2 cores, ×0.35
  9. AWS CodeBuild8 vCPU · 16 GBlarge · Xeon 8488Cbest of 39,203, 6.2 cores, ×0.22

It grows with vCPUs: compare at one size. Cores: sysbench on every thread over one.

Single-threadx64 · PassMark · higher is better
  1. Namespace2 vCPU · 8 GB2x8 · EPYC · model hidden4,605, ×1.00
  2. Avrea2 vCPU · 8 GBEPYC 4585PX4,320, ×0.94
  3. RunsOn2 vCPU · 8 GBm8azn.large · EPYC 9R05best of 104,253, ×0.92
  4. Blacksmith2 vCPU · 8 GBEPYC · model hidden4,093, ×0.89
  5. Warpbuild2 vCPU · 8 GBRyzen 9 9950X4,008, ×0.87
  6. Ubicloud2 vCPU · 8 GBpremium · Ryzen 9 7950X3Dbest of 22,933, ×0.64
  7. GitHub2 vCPU · 8 GBEPYC 9V742,316, ×0.50
  8. AWS CodeBuild2 vCPU · 3 GBsmall · Xeon 8488C2,066, ×0.45
All coresx64 · PassMark CPU Mark · higher is better
  1. Namespace2 vCPU · 8 GB2x8 · EPYC · model hidden8,233, 1.9 cores, ×1.00
  2. RunsOn2 vCPU · 8 GBm8azn.large · EPYC 9R05best of 107,722, 2.0 cores, ×0.94
  3. Blacksmith2 vCPU · 8 GBEPYC · model hidden6,944, 1.9 cores, ×0.84
  4. Avrea2 vCPU · 8 GBEPYC 4585PX5,856, 1.1 cores, ×0.71
  5. Warpbuild2 vCPU · 8 GBRyzen 9 9950X5,501, 1.0 cores, ×0.67
  6. Ubicloud2 vCPU · 8 GBstandard · EPYC 9454Pbest of 25,220, 1.9 cores, ×0.63
  7. GitHub2 vCPU · 8 GBEPYC 9V743,100, 1.1 cores, ×0.38
  8. AWS CodeBuild2 vCPU · 3 GBsmall · Xeon 8488C2,357, 1.6 cores, ×0.29

It grows with vCPUs: compare at one size. Cores: sysbench on every thread over one.

Single-threadx64 · PassMark · higher is better
  1. Namespace4 vCPU · 8 GB4x8 · EPYC · model hiddenbest of 24,685, ×1.00
  2. RunsOn4 vCPU · 16 GBm8azn.xlarge · EPYC 9R05best of 94,264, ×0.91
  3. Avrea4 vCPU · 16 GBEPYC 4585PX4,243, ×0.91
  4. Blacksmith4 vCPU · 16 GBEPYC · model hidden4,230, ×0.90
  5. Warpbuild4 vCPU · 16 GBRyzen 9 9950X4,196, ×0.90
  6. StarSling4 vCPU · 16 GBEPYC · model hidden3,969, ×0.85
  7. Ubicloud4 vCPU · 16 GBpremium · Ryzen 9 7950X3Dbest of 23,090, ×0.66
  8. GitHub4 vCPU · 16 GBEPYC 9V452,888, ×0.62
  9. AWS CodeBuild4 vCPU · 8 GBmedium · Xeon 8488C2,148, ×0.46
All coresx64 · PassMark CPU Mark · higher is better
  1. Namespace4 vCPU · 8 GB4x8 · EPYC · model hiddenbest of 216,138, 3.9 cores, ×1.00
  2. RunsOn4 vCPU · 16 GBm8azn.xlarge · EPYC 9R05best of 915,141, 4.0 cores, ×0.94
  3. Blacksmith4 vCPU · 16 GBEPYC · model hidden14,759, 3.8 cores, ×0.91
  4. StarSling4 vCPU · 16 GBEPYC · model hidden13,914, 3.8 cores, ×0.86
  5. Avrea4 vCPU · 16 GBEPYC 4585PX11,174, 2.2 cores, ×0.69
  6. Warpbuild4 vCPU · 16 GBRyzen 9 9950X11,083, 2.1 cores, ×0.69
  7. Ubicloud4 vCPU · 16 GBstandard · EPYC 9454best of 29,973, 3.8 cores, ×0.62
  8. GitHub4 vCPU · 16 GBEPYC 9V457,510, 2.2 cores, ×0.47
  9. AWS CodeBuild4 vCPU · 8 GBmedium · Xeon 8488C4,694, 3.2 cores, ×0.29

It grows with vCPUs: compare at one size. Cores: sysbench on every thread over one.

Single-threadx64 · PassMark · higher is better
  1. Namespace8 vCPU · 32 GB8x32 · EPYC · model hiddenbest of 24,677, ×1.00
  2. Avrea8 vCPU · 32 GBEPYC 4585PX4,273, ×0.91
  3. Blacksmith8 vCPU · 32 GBEPYC · model hidden4,234, ×0.91
  4. Warpbuild8 vCPU · 32 GBm8a.2xlarge · Ryzen 9 9950X4,177, ×0.89
  5. Ubicloud8 vCPU · 32 GBpremium · Ryzen 9 7950X3Dbest of 24,038, ×0.86
  6. StarSling8 vCPU · 32 GBEPYC · model hidden3,849, ×0.82
  7. RunsOn8 vCPU · 16 GBc8a.2xlarge · EPYC 9R45best of 93,817, ×0.82
  8. GitHub8 vCPU · 32 GBEPYC 9V452,899, ×0.62
  9. AWS CodeBuild8 vCPU · 16 GBlarge · Xeon 8488C2,156, ×0.46
All coresx64 · PassMark CPU Mark · higher is better
  1. Namespace8 vCPU · 32 GB8x32 · EPYC · model hiddenbest of 230,113, 7.7 cores, ×1.00
  2. Blacksmith8 vCPU · 32 GBEPYC · model hidden27,282, 7.9 cores, ×0.91
  3. RunsOn8 vCPU · 64 GBr8a.2xlarge · EPYC 9R45best of 925,468, 8.0 cores, ×0.85
  4. StarSling8 vCPU · 32 GBEPYC · model hidden25,105, 7.7 cores, ×0.83
  5. Ubicloud8 vCPU · 32 GBpremium · Ryzen 9 7950X3Dbest of 224,720, 7.4 cores, ×0.82
  6. Avrea8 vCPU · 32 GBEPYC 4585PX21,356, 4.3 cores, ×0.71
  7. Warpbuild8 vCPU · 32 GBm8a.2xlarge · Ryzen 9 9950X20,991, 4.2 cores, ×0.70
  8. GitHub8 vCPU · 32 GBEPYC 9V4514,420, 4.2 cores, ×0.48
  9. AWS CodeBuild8 vCPU · 16 GBlarge · Xeon 8488C9,203, 6.2 cores, ×0.31

It grows with vCPUs: compare at one size. Cores: sysbench on every thread over one.

Single-threadarm64 · PassMark · higher is better
  1. Avrea2 vCPU · 8 GBApple M5 Max · as stated by the providerbest of 35,028, ×1.00
  2. Namespace4 vCPU · 16 GBApple silicon 4x16 · CPU not reportedbest of 74,730, ×0.94
  3. Warpbuild6 vCPU · 14 GBxfast (Apple M4 Pro)best of 53,221, ×0.64
  4. RunsOn2 vCPU · 8 GBm9gd.large · Neoverse-V3best of 132,476, ×0.49
  5. GitHub4 vCPU · 16 GBNeoverse-N2best of 21,877, ×0.37
  6. Blacksmith8 vCPU · 32 GBAmpere-1abest of 31,356, ×0.27
  7. Ubicloud8 vCPU · 32 GBstandard · Neoverse-N1best of 31,316, ×0.26
All coresarm64 · PassMark CPU Mark · higher is better
  1. Avrea8 vCPU · 32 GBApple M5 Max · as stated by the providerbest of 315,458, 7.2 cores, ×1.00
  2. Warpbuild12 vCPU · 28 GBxfast (Apple M4 Pro)best of 514,005, 8.1 cores, ×0.91
  3. RunsOn8 vCPU · 32 GBm9g.2xlarge · Neoverse-V3best of 1310,908, 8.0 cores, ×0.71
  4. Namespace4 vCPU · 16 GBApple silicon 4x16 · CPU not reportedbest of 78,195, 3.7 cores, ×0.53
  5. Blacksmith8 vCPU · 32 GBAmpere-1abest of 37,368, 7.9 cores, ×0.48
  6. Ubicloud8 vCPU · 32 GBstandard · Neoverse-N1best of 36,138, 7.9 cores, ×0.40
  7. GitHub4 vCPU · 16 GBNeoverse-N2best of 23,974, 4.0 cores, ×0.26

It grows with vCPUs: compare at one size. Cores: sysbench on every thread over one.

Single-threadarm64 · PassMark · higher is better
  1. Avrea2 vCPU · 8 GBApple M5 Max · as stated by the provider5,028, ×1.00
  2. RunsOn2 vCPU · 8 GBm9gd.large · Neoverse-V3best of 42,476, ×0.49
  3. Warpbuild2 vCPU · 8 GBm8g.large · Neoverse-V21,941, ×0.39
  4. GitHub2 vCPU · 8 GBNeoverse-N21,874, ×0.37
  5. Blacksmith2 vCPU · 6 GBAmpere-1a1,341, ×0.27
  6. Ubicloud2 vCPU · 8 GBstandard · Neoverse-N11,314, ×0.26
  7. Namespace2 vCPU · 8 GB2x8 · Ampere-1abest of 21,171, ×0.23
All coresarm64 · PassMark CPU Mark · higher is better
  1. Avrea2 vCPU · 8 GBApple M5 Max · as stated by the provider4,413, 1.9 cores, ×1.00
  2. RunsOn2 vCPU · 8 GBm9gd.large · Neoverse-V3best of 42,855, 2.0 cores, ×0.65
  3. GitHub2 vCPU · 8 GBNeoverse-N22,008, 2.0 cores, ×0.45
  4. Warpbuild2 vCPU · 8 GBm8g.large · Neoverse-V21,977, 2.0 cores, ×0.45
  5. Blacksmith2 vCPU · 6 GBAmpere-1a1,905, 2.0 cores, ×0.43
  6. Namespace2 vCPU · 8 GB2x8 · Ampere-1abest of 21,607, 1.9 cores, ×0.36
  7. Ubicloud2 vCPU · 8 GBstandard · Neoverse-N11,528, 2.0 cores, ×0.35

It grows with vCPUs: compare at one size. Cores: sysbench on every thread over one.

Single-threadarm64 · PassMark · higher is better
  1. Avrea4 vCPU · 16 GBApple M5 Max · as stated by the provider4,860, ×1.00
  2. Namespace4 vCPU · 16 GBApple silicon 4x16 · CPU not reportedbest of 34,730, ×0.97
  3. RunsOn4 vCPU · 16 GBm9g.xlarge · Neoverse-V3best of 52,476, ×0.51
  4. Warpbuild4 vCPU · 16 GBm8g.xlarge · Neoverse-V21,942, ×0.40
  5. GitHub4 vCPU · 16 GBNeoverse-N21,877, ×0.39
  6. Blacksmith4 vCPU · 16 GBAmpere-1a1,346, ×0.28
  7. Ubicloud4 vCPU · 16 GBstandard · Neoverse-N11,311, ×0.27
All coresarm64 · PassMark CPU Mark · higher is better
  1. Avrea4 vCPU · 16 GBApple M5 Max · as stated by the provider8,388, 3.7 cores, ×1.00
  2. Namespace4 vCPU · 16 GBApple silicon 4x16 · CPU not reportedbest of 38,195, 3.7 cores, ×0.98
  3. RunsOn4 vCPU · 16 GBm9gd.xlarge · Neoverse-V3best of 55,631, 4.0 cores, ×0.67
  4. GitHub4 vCPU · 16 GBNeoverse-N23,974, 4.0 cores, ×0.47
  5. Warpbuild4 vCPU · 16 GBm8g.xlarge · Neoverse-V23,925, 4.0 cores, ×0.47
  6. Blacksmith4 vCPU · 16 GBAmpere-1a3,778, 3.9 cores, ×0.45
  7. Ubicloud4 vCPU · 16 GBstandard · Neoverse-N13,099, 3.9 cores, ×0.37

It grows with vCPUs: compare at one size. Cores: sysbench on every thread over one.

Single-threadarm64 · PassMark · higher is better
  1. Avrea8 vCPU · 32 GBApple M5 Max · as stated by the provider4,842, ×1.00
  2. RunsOn8 vCPU · 32 GBm9gd.2xlarge · Neoverse-V3best of 42,472, ×0.51
  3. Warpbuild8 vCPU · 32 GBm8g.2xlarge · Neoverse-V21,941, ×0.40
  4. Blacksmith8 vCPU · 32 GBAmpere-1a1,356, ×0.28
  5. Ubicloud8 vCPU · 32 GBstandard · Neoverse-N11,316, ×0.27
  6. Namespace8 vCPU · 16 GB8x16 · Ampere-1abest of 21,182, ×0.24
All coresarm64 · PassMark CPU Mark · higher is better
  1. Avrea8 vCPU · 32 GBApple M5 Max · as stated by the provider15,458, 7.2 cores, ×1.00
  2. RunsOn8 vCPU · 32 GBm9g.2xlarge · Neoverse-V3best of 410,908, 8.0 cores, ×0.71
  3. Warpbuild8 vCPU · 32 GBm8g.2xlarge · Neoverse-V27,642, 8.0 cores, ×0.49
  4. Blacksmith8 vCPU · 32 GBAmpere-1a7,368, 7.9 cores, ×0.48
  5. Namespace8 vCPU · 16 GB8x16 · Ampere-1abest of 26,420, 7.9 cores, ×0.42
  6. Ubicloud8 vCPU · 32 GBstandard · Neoverse-N16,138, 7.9 cores, ×0.40

It grows with vCPUs: compare at one size. Cores: sysbench on every thread over one.

CPU per dollar

All-core score over the price of an hour: how much CPU a dollar buys.

CPU Mark per $/hourx64 · RunsOn at EC2 spot · higher is better
  1. RunsOn4 vCPU · 16 GBm8a.xlarge · EPYC 9R45best of 28155k, spot $0.0015/min, ×1.00
  2. Ubicloud2 vCPU · 8 GBstandard · EPYC 9454Pbest of 670k, $0.0013/min, ×0.45
  3. Namespace4 vCPU · 8 GB4x8 · EPYC · model hiddenbest of 545k, $0.0060/min, ×0.29
  4. Blacksmith4 vCPU · 16 GBEPYC · model hiddenbest of 331k, $0.0080/min, ×0.20
  5. StarSling4 vCPU · 16 GBEPYC · model hiddenbest of 229k, $0.0080/min, ×0.19
  6. Avrea2 vCPU · 8 GBEPYC 4585PXbest of 324k, $0.0040/min, ×0.16
  7. Warpbuild4 vCPU · 16 GBRyzen 9 9950Xbest of 323k, $0.0080/min, ×0.15
  8. GitHub8 vCPU · 32 GBEPYC 9V45best of 311k, $0.0220/min, ×0.07
  9. AWS CodeBuild2 vCPU · 3 GBsmall · Xeon 8488Cbest of 37.9k, $0.0050/min, ×0.05

RunsOn m8a.xlarge · EBS gp3 buys the most CPU per dollar on x64: 155k CPU Mark per $/hour, 2.2× Ubicloud.

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

Left out: t8i.medium (RunsOn), burstable. Its score is a burst on CPU credits: over an hour, a burstable type slows to its baseline or bills the surplus credits.

CPU Mark per $/hourx64 · RunsOn at EC2 on-demand · higher is better
  1. Ubicloud2 vCPU · 8 GBstandard · EPYC 9454Pbest of 670k, $0.0013/min, ×1.00
  2. RunsOn4 vCPU · 8 GBc8a.xlarge · EPYC 9R45best of 2858k, on-demand $0.0039/min, ×0.83
  3. Namespace4 vCPU · 8 GB4x8 · EPYC · model hiddenbest of 545k, $0.0060/min, ×0.64
  4. Blacksmith4 vCPU · 16 GBEPYC · model hiddenbest of 331k, $0.0080/min, ×0.44
  5. StarSling4 vCPU · 16 GBEPYC · model hiddenbest of 229k, $0.0080/min, ×0.42
  6. Avrea2 vCPU · 8 GBEPYC 4585PXbest of 324k, $0.0040/min, ×0.35
  7. Warpbuild4 vCPU · 16 GBRyzen 9 9950Xbest of 323k, $0.0080/min, ×0.33
  8. GitHub8 vCPU · 32 GBEPYC 9V45best of 311k, $0.0220/min, ×0.16
  9. AWS CodeBuild2 vCPU · 3 GBsmall · Xeon 8488Cbest of 37.9k, $0.0050/min, ×0.11

Ubicloud standard 2 vCPU buys the most CPU per dollar on x64: 70k CPU Mark per $/hour, 1.2× RunsOn.

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

Left out: t8i.medium (RunsOn), burstable. Its score is a burst on CPU credits: over an hour, a burstable type slows to its baseline or bills the surplus credits.

CPU Mark per $/hourx64 · RunsOn at EC2 spot · higher is better
  1. RunsOn2 vCPU · 4 GBc8a.large · EPYC 9R45best of 9126k, spot $0.0009/min, ×1.00
  2. Ubicloud2 vCPU · 8 GBstandard · EPYC 9454Pbest of 270k, $0.0013/min, ×0.55
  3. Blacksmith2 vCPU · 8 GBEPYC · model hidden29k, $0.0040/min, ×0.23
  4. Avrea2 vCPU · 8 GBEPYC 4585PX24k, $0.0040/min, ×0.19
  5. Warpbuild2 vCPU · 8 GBRyzen 9 9950X23k, $0.0040/min, ×0.18
  6. Namespace2 vCPU · 8 GB2x8 · EPYC · model hidden23k, $0.0060/min, ×0.18
  7. GitHub2 vCPU · 8 GBEPYC 9V748.6k, $0.0060/min, ×0.07
  8. AWS CodeBuild2 vCPU · 3 GBsmall · Xeon 8488C7.9k, $0.0050/min, ×0.06

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

Left out: t8i.medium (RunsOn), burstable. Its score is a burst on CPU credits: over an hour, a burstable type slows to its baseline or bills the surplus credits.

CPU Mark per $/hourx64 · RunsOn at EC2 on-demand · higher is better
  1. Ubicloud2 vCPU · 8 GBstandard · EPYC 9454Pbest of 270k, $0.0013/min, ×1.00
  2. RunsOn2 vCPU · 4 GBc8a.large · EPYC 9R45best of 955k, on-demand $0.0021/min, ×0.78
  3. Blacksmith2 vCPU · 8 GBEPYC · model hidden29k, $0.0040/min, ×0.42
  4. Avrea2 vCPU · 8 GBEPYC 4585PX24k, $0.0040/min, ×0.35
  5. Warpbuild2 vCPU · 8 GBRyzen 9 9950X23k, $0.0040/min, ×0.33
  6. Namespace2 vCPU · 8 GB2x8 · EPYC · model hidden23k, $0.0060/min, ×0.33
  7. GitHub2 vCPU · 8 GBEPYC 9V748.6k, $0.0060/min, ×0.12
  8. AWS CodeBuild2 vCPU · 3 GBsmall · Xeon 8488C7.9k, $0.0050/min, ×0.11

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

Left out: t8i.medium (RunsOn), burstable. Its score is a burst on CPU credits: over an hour, a burstable type slows to its baseline or bills the surplus credits.

CPU Mark per $/hourx64 · RunsOn at EC2 spot · higher is better
  1. RunsOn4 vCPU · 16 GBm8a.xlarge · EPYC 9R45best of 9155k, spot $0.0015/min, ×1.00
  2. Ubicloud4 vCPU · 16 GBstandard · EPYC 9454best of 266k, $0.0025/min, ×0.43
  3. Namespace4 vCPU · 8 GB4x8 · EPYC · model hiddenbest of 245k, $0.0060/min, ×0.29
  4. Blacksmith4 vCPU · 16 GBEPYC · model hidden31k, $0.0080/min, ×0.20
  5. StarSling4 vCPU · 16 GBEPYC · model hidden29k, $0.0080/min, ×0.19
  6. Avrea4 vCPU · 16 GBEPYC 4585PX23k, $0.0080/min, ×0.15
  7. Warpbuild4 vCPU · 16 GBRyzen 9 9950X23k, $0.0080/min, ×0.15
  8. GitHub4 vCPU · 16 GBEPYC 9V4510k, $0.0120/min, ×0.07
  9. AWS CodeBuild4 vCPU · 8 GBmedium · Xeon 8488C7.8k, $0.0100/min, ×0.05

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourx64 · RunsOn at EC2 on-demand · higher is better
  1. Ubicloud4 vCPU · 16 GBstandard · EPYC 9454best of 266k, $0.0025/min, ×1.00
  2. RunsOn4 vCPU · 8 GBc8a.xlarge · EPYC 9R45best of 958k, on-demand $0.0039/min, ×0.87
  3. Namespace4 vCPU · 8 GB4x8 · EPYC · model hiddenbest of 245k, $0.0060/min, ×0.67
  4. Blacksmith4 vCPU · 16 GBEPYC · model hidden31k, $0.0080/min, ×0.46
  5. StarSling4 vCPU · 16 GBEPYC · model hidden29k, $0.0080/min, ×0.44
  6. Avrea4 vCPU · 16 GBEPYC 4585PX23k, $0.0080/min, ×0.35
  7. Warpbuild4 vCPU · 16 GBRyzen 9 9950X23k, $0.0080/min, ×0.35
  8. GitHub4 vCPU · 16 GBEPYC 9V4510k, $0.0120/min, ×0.16
  9. AWS CodeBuild4 vCPU · 8 GBmedium · Xeon 8488C7.8k, $0.0100/min, ×0.12

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourx64 · RunsOn at EC2 spot · higher is better
  1. RunsOn8 vCPU · 32 GBm8a.2xlarge · EPYC 9R45best of 9141k, spot $0.0030/min, ×1.00
  2. Ubicloud8 vCPU · 32 GBstandard · EPYC 9454Pbest of 254k, $0.0050/min, ×0.38
  3. Namespace8 vCPU · 16 GB8x16 · EPYC · model hiddenbest of 241k, $0.0120/min, ×0.29
  4. Blacksmith8 vCPU · 32 GBEPYC · model hidden28k, $0.0160/min, ×0.20
  5. StarSling8 vCPU · 32 GBEPYC · model hidden26k, $0.0160/min, ×0.19
  6. Avrea8 vCPU · 32 GBEPYC 4585PX22k, $0.0160/min, ×0.16
  7. Warpbuild8 vCPU · 32 GBm8a.2xlarge · Ryzen 9 9950X22k, $0.0160/min, ×0.16
  8. GitHub8 vCPU · 32 GBEPYC 9V4511k, $0.0220/min, ×0.08
  9. AWS CodeBuild8 vCPU · 16 GBlarge · Xeon 8488C7.7k, $0.0200/min, ×0.05

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourx64 · RunsOn at EC2 on-demand · higher is better
  1. RunsOn8 vCPU · 16 GBc8a.2xlarge · EPYC 9R45best of 957k, on-demand $0.0075/min, ×1.00
  2. Ubicloud8 vCPU · 32 GBstandard · EPYC 9454Pbest of 254k, $0.0050/min, ×0.95
  3. Namespace8 vCPU · 16 GB8x16 · EPYC · model hiddenbest of 241k, $0.0120/min, ×0.72
  4. Blacksmith8 vCPU · 32 GBEPYC · model hidden28k, $0.0160/min, ×0.50
  5. StarSling8 vCPU · 32 GBEPYC · model hidden26k, $0.0160/min, ×0.46
  6. Avrea8 vCPU · 32 GBEPYC 4585PX22k, $0.0160/min, ×0.39
  7. Warpbuild8 vCPU · 32 GBm8a.2xlarge · Ryzen 9 9950X22k, $0.0160/min, ×0.39
  8. GitHub8 vCPU · 32 GBEPYC 9V4511k, $0.0220/min, ×0.19
  9. AWS CodeBuild8 vCPU · 16 GBlarge · Xeon 8488C7.7k, $0.0200/min, ×0.14

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourarm64 · RunsOn at EC2 spot · higher is better
  1. RunsOn8 vCPU · 32 GBm9gd.2xlarge · Neoverse-V3best of 1358k, spot $0.0031/min, ×1.00
  2. Ubicloud4 vCPU · 16 GBstandard · Neoverse-N1best of 321k, $0.0025/min, ×0.36
  3. Blacksmith2 vCPU · 6 GBAmpere-1abest of 313k, $0.0025/min, ×0.22
  4. Warpbuild2 vCPU · 8 GBm8g.large · Neoverse-V2best of 511k, $0.0030/min, ×0.19
  5. Namespace4 vCPU · 8 GB4x8 · Ampere-1abest of 79.1k, $0.0060/min, ×0.16
  6. GitHub4 vCPU · 16 GBNeoverse-N2best of 28.3k, $0.0080/min, ×0.14
  7. Avrea2 vCPU · 8 GBApple M5 Max · as stated by the providerbest of 36.1k, $0.0120/min, ×0.11

RunsOn m9gd.2xlarge · local NVMe buys the most CPU per dollar on arm64: 58k CPU Mark per $/hour, 2.8× Ubicloud.

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourarm64 · RunsOn at EC2 on-demand · higher is better
  1. RunsOn8 vCPU · 32 GBm9g.2xlarge · Neoverse-V3best of 1327k, on-demand $0.0068/min, ×1.00
  2. Ubicloud4 vCPU · 16 GBstandard · Neoverse-N1best of 321k, $0.0025/min, ×0.78
  3. Blacksmith2 vCPU · 6 GBAmpere-1abest of 313k, $0.0025/min, ×0.48
  4. Warpbuild2 vCPU · 8 GBm8g.large · Neoverse-V2best of 511k, $0.0030/min, ×0.41
  5. Namespace4 vCPU · 8 GB4x8 · Ampere-1abest of 79.1k, $0.0060/min, ×0.34
  6. GitHub4 vCPU · 16 GBNeoverse-N2best of 28.3k, $0.0080/min, ×0.31
  7. Avrea2 vCPU · 8 GBApple M5 Max · as stated by the providerbest of 36.1k, $0.0120/min, ×0.23

RunsOn m9g.2xlarge · EBS gp3 buys the most CPU per dollar on arm64: 27k CPU Mark per $/hour, 1.3× Ubicloud.

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourarm64 · RunsOn at EC2 spot · higher is better
  1. RunsOn2 vCPU · 8 GBm9gd.large · Neoverse-V3best of 449k, spot $0.0010/min, ×1.00
  2. Ubicloud2 vCPU · 8 GBstandard · Neoverse-N120k, $0.0013/min, ×0.42
  3. Blacksmith2 vCPU · 6 GBAmpere-1a13k, $0.0025/min, ×0.26
  4. Warpbuild2 vCPU · 8 GBm8g.large · Neoverse-V211k, $0.0030/min, ×0.22
  5. GitHub2 vCPU · 8 GBNeoverse-N26.7k, $0.0050/min, ×0.14
  6. Avrea2 vCPU · 8 GBApple M5 Max · as stated by the provider6.1k, $0.0120/min, ×0.13
  7. Namespace2 vCPU · 8 GB2x8 · Ampere-1abest of 24.5k, $0.0060/min, ×0.09

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourarm64 · RunsOn at EC2 on-demand · higher is better
  1. RunsOn2 vCPU · 8 GBm9g.large · Neoverse-V3best of 424k, on-demand $0.0019/min, ×1.00
  2. Ubicloud2 vCPU · 8 GBstandard · Neoverse-N120k, $0.0013/min, ×0.83
  3. Blacksmith2 vCPU · 6 GBAmpere-1a13k, $0.0025/min, ×0.52
  4. Warpbuild2 vCPU · 8 GBm8g.large · Neoverse-V211k, $0.0030/min, ×0.45
  5. GitHub2 vCPU · 8 GBNeoverse-N26.7k, $0.0050/min, ×0.27
  6. Avrea2 vCPU · 8 GBApple M5 Max · as stated by the provider6.1k, $0.0120/min, ×0.25
  7. Namespace2 vCPU · 8 GB2x8 · Ampere-1abest of 24.5k, $0.0060/min, ×0.18

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourarm64 · RunsOn at EC2 spot · higher is better
  1. RunsOn4 vCPU · 16 GBm9gd.xlarge · Neoverse-V3best of 555k, spot $0.0017/min, ×1.00
  2. Ubicloud4 vCPU · 16 GBstandard · Neoverse-N121k, $0.0025/min, ×0.37
  3. Blacksmith4 vCPU · 16 GBAmpere-1a13k, $0.0050/min, ×0.23
  4. Warpbuild4 vCPU · 16 GBm8g.xlarge · Neoverse-V211k, $0.0060/min, ×0.20
  5. Namespace4 vCPU · 8 GB4x8 · Ampere-1abest of 39.1k, $0.0060/min, ×0.16
  6. GitHub4 vCPU · 16 GBNeoverse-N28.3k, $0.0080/min, ×0.15
  7. Avrea4 vCPU · 16 GBApple M5 Max · as stated by the provider5.8k, $0.0240/min, ×0.10

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourarm64 · RunsOn at EC2 on-demand · higher is better
  1. RunsOn4 vCPU · 16 GBm9g.xlarge · Neoverse-V3best of 526k, on-demand $0.0036/min, ×1.00
  2. Ubicloud4 vCPU · 16 GBstandard · Neoverse-N121k, $0.0025/min, ×0.79
  3. Blacksmith4 vCPU · 16 GBAmpere-1a13k, $0.0050/min, ×0.48
  4. Warpbuild4 vCPU · 16 GBm8g.xlarge · Neoverse-V211k, $0.0060/min, ×0.41
  5. Namespace4 vCPU · 8 GB4x8 · Ampere-1abest of 39.1k, $0.0060/min, ×0.35
  6. GitHub4 vCPU · 16 GBNeoverse-N28.3k, $0.0080/min, ×0.32
  7. Avrea4 vCPU · 16 GBApple M5 Max · as stated by the provider5.8k, $0.0240/min, ×0.22

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourarm64 · RunsOn at EC2 spot · higher is better
  1. RunsOn8 vCPU · 32 GBm9gd.2xlarge · Neoverse-V3best of 458k, spot $0.0031/min, ×1.00
  2. Ubicloud8 vCPU · 32 GBstandard · Neoverse-N120k, $0.0050/min, ×0.35
  3. Blacksmith8 vCPU · 32 GBAmpere-1a12k, $0.0100/min, ×0.21
  4. Warpbuild8 vCPU · 32 GBm8g.2xlarge · Neoverse-V211k, $0.0120/min, ×0.18
  5. Namespace8 vCPU · 16 GB8x16 · Ampere-1abest of 28.9k, $0.0120/min, ×0.15
  6. Avrea8 vCPU · 32 GBApple M5 Max · as stated by the provider5.4k, $0.0480/min, ×0.09

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

CPU Mark per $/hourarm64 · RunsOn at EC2 on-demand · higher is better
  1. RunsOn8 vCPU · 32 GBm9g.2xlarge · Neoverse-V3best of 427k, on-demand $0.0068/min, ×1.00
  2. Ubicloud8 vCPU · 32 GBstandard · Neoverse-N120k, $0.0050/min, ×0.77
  3. Blacksmith8 vCPU · 32 GBAmpere-1a12k, $0.0100/min, ×0.46
  4. Warpbuild8 vCPU · 32 GBm8g.2xlarge · Neoverse-V211k, $0.0120/min, ×0.40
  5. Namespace8 vCPU · 16 GB8x16 · Ampere-1abest of 28.9k, $0.0120/min, ×0.33
  6. Avrea8 vCPU · 32 GBApple M5 Max · as stated by the provider5.4k, $0.0480/min, ×0.20

List price per minute for the machine behind each bar, matched to its exact shape. RunsOn: EC2 and EBS, per minute. Licenses and plan fees are fixed costs, listed on each provider's page.

Price covers the runner only: a faster core that finishes sooner costs fewer minutes, which this chart does not count. Builds do: cost per build.

What the numbers say

Measured, not listed: the CPU each job reported.

  • On x64, each provider's best core scores from 2,156 single-thread on AWS CodeBuild large (Xeon 8488C) to 4,685 on Namespace 4x8 (EPYC · model hidden): 2.2× the work per core.
  • On arm64, each provider's best core scores from 1,316 single-thread on Ubicloud standard 8 vCPU (Neoverse-N1) to 5,028 on Avrea 2 vCPU (Apple M5 Max · as stated by the provider): 3.8× the work per core.
  • At 4 vCPU on x64, every thread together does 2.1× the work of one on Warpbuild 4 vCPU, and 4.0× on RunsOn m8azn.xlarge · EBS gp3: not every vCPU is a full core.

Over time

PassMark single-thread on 2 vCPU runners since 27 Jul 2026, one point per day measured.

One panel per provider

  • Namespace4,669median of 14 days

    02k4k6k4,580 on 27 Jul 20264,084 on 28 Jul 20264,472 on 29 Jul 20264,412 on 30 Jul 20264,641 on 31 Jul 20264,676 on 1 Aug 20264,748 on 2 Aug 20264,474 on 3 Aug 20264,544 on 4 Aug 20264,617 on 5 Aug 20264,702 on 6 Aug 20264,699 on 7 Aug 20264,637 on 8 Aug 20264,684 on 9 Aug 20264,663 on 10 Aug 20264,715 on 11 Aug 20264,722 on 12 Aug 20264,347 on 13 Aug 20264,688 on 14 Aug 20264,683 on 15 Aug 20264,722 on 16 Aug 20264,303 on 18 Aug 20264,707 on 19 Aug 20264,384 on 20 Aug 20264,710 on 21 Aug 20264,712 on 22 Aug 20264,669 on 23 Aug 20264,682 on 24 Aug 20264,683 on 25 Aug 20264,693 on 26 Aug 20264,478 on 27 Aug 20264,681 on 28 Aug 20264,513 on 29 Aug 20264,716 on 30 Aug 20264,686 on 31 Aug 20264,709 on 1 Sep 20264,692 on 2 Sep 20264,717 on 3 Sep 20264,158 on 4 Sep 20264,732 on 5 Sep 20264,534 on 6 Sep 20264,394 on 7 Sep 20264,691 on 8 Sep 20264,699 on 9 Sep 20264,686 on 10 Sep 20264,687 on 11 Sep 20264,676 on 12 Sep 20264,672 on 13 Sep 20264,680 on 14 Sep 20264,719 on 15 Sep 20264,712 on 16 Sep 20264,687 on 17 Sep 20264,682 on 18 Sep 20264,714 on 19 Sep 20264,674 on 20 Sep 20264,502 on 21 Sep 20264,714 on 22 Sep 20264,679 on 23 Sep 20264,733 on 30 Sep 20264,678 on 1 Oct 2026; samples 4,675 to 4,6804,668 on 2 Oct 2026; samples 4,605 to 4,6724,709 on 3 Oct 20264,665 on 4 Oct 20264,690 on 5 Oct 2026; samples 4,672 to 4,7084,719 on 6 Oct 20264,670 on 7 Oct 20264,664 on 8 Oct 2026; samples 4,650 to 4,6784,706 on 9 Oct 20264,595 on 10 Oct 20264,581 on 11 Oct 2026; samples 4,562 to 4,5994,568 on 12 Oct 20264,685 on 13 Oct 20264,381 on 14 Oct 2026; samples 4,095 to 4,667

    EPYC · nscloud-ubuntu-24.04-amd64-2x8

  • Avrea4,114median of 14 days

    02k4k6k4,536 on 27 Jul 20264,435 on 28 Jul 20264,397 on 29 Jul 20264,490 on 30 Jul 20264,346 on 31 Jul 20264,516 on 1 Aug 20264,530 on 2 Aug 20264,415 on 3 Aug 20264,465 on 4 Aug 20264,427 on 5 Aug 20264,455 on 6 Aug 20264,525 on 7 Aug 20264,468 on 8 Aug 20264,379 on 9 Aug 20264,492 on 10 Aug 20264,439 on 11 Aug 20264,393 on 12 Aug 20264,406 on 13 Aug 20264,134 on 14 Aug 20264,453 on 15 Aug 20264,551 on 16 Aug 20264,331 on 17 Aug 20264,418 on 18 Aug 20264,459 on 19 Aug 20264,481 on 20 Aug 20264,443 on 21 Aug 20264,373 on 22 Aug 20264,476 on 23 Aug 20264,194 on 24 Aug 20264,423 on 25 Aug 20264,420 on 26 Aug 20264,427 on 27 Aug 20264,538 on 28 Aug 20264,536 on 29 Aug 20264,417 on 30 Aug 20264,355 on 31 Aug 20264,391 on 1 Sep 20264,374 on 2 Sep 20264,397 on 3 Sep 20264,415 on 4 Sep 20264,484 on 5 Sep 20264,354 on 7 Sep 20264,457 on 8 Sep 20264,290 on 9 Sep 20264,152 on 10 Sep 20262,918 on 11 Sep 20263,299 on 12 Sep 20264,093 on 13 Sep 20263,254 on 14 Sep 20263,483 on 15 Sep 20263,261 on 16 Sep 20263,597 on 17 Sep 20263,895 on 18 Sep 20264,446 on 19 Sep 20263,455 on 20 Sep 20263,978 on 21 Sep 20263,830 on 22 Sep 20264,167 on 23 Sep 20264,403 on 30 Sep 20263,912 on 1 Oct 2026; samples 3,700 to 4,1244,190 on 2 Oct 2026; samples 4,189 to 4,2914,179 on 3 Oct 20264,018 on 4 Oct 20264,069 on 5 Oct 2026; samples 3,703 to 4,4354,264 on 6 Oct 20264,160 on 7 Oct 20263,928 on 8 Oct 2026; samples 3,862 to 3,9933,846 on 9 Oct 20264,332 on 10 Oct 20264,014 on 11 Oct 2026; samples 3,708 to 4,3203,967 on 12 Oct 20264,217 on 13 Oct 20264,399 on 14 Oct 2026; samples 4,278 to 4,520

    EPYC 4585PX · avrea-ubuntu-24.04-2-vcpu

  • RunsOn4,264median of 14 days

    02k4k6k4,263 on 27 Jul 20264,267 on 28 Jul 20264,262 on 29 Jul 20264,267 on 30 Jul 20264,273 on 31 Jul 20264,272 on 1 Aug 20264,277 on 2 Aug 20264,226 on 3 Aug 20264,260 on 4 Aug 20264,263 on 5 Aug 20264,285 on 6 Aug 20264,282 on 7 Aug 20264,269 on 8 Aug 20264,268 on 9 Aug 20264,259 on 10 Aug 20264,270 on 11 Aug 20264,270 on 12 Aug 20264,271 on 13 Aug 20264,267 on 14 Aug 20264,255 on 15 Aug 20264,289 on 16 Aug 20264,266 on 17 Aug 20264,259 on 18 Aug 20264,253 on 19 Aug 20264,269 on 20 Aug 20264,282 on 21 Aug 20264,259 on 22 Aug 20264,258 on 23 Aug 20264,258 on 24 Aug 20264,270 on 25 Aug 20264,272 on 26 Aug 20264,255 on 27 Aug 20264,264 on 28 Aug 20264,270 on 29 Aug 20264,265 on 30 Aug 20264,263 on 31 Aug 20264,267 on 1 Sep 20264,264 on 2 Sep 20264,263 on 3 Sep 20264,255 on 4 Sep 20264,277 on 5 Sep 20264,255 on 6 Sep 20264,265 on 7 Sep 20264,264 on 8 Sep 20264,240 on 9 Sep 20264,263 on 10 Sep 20264,271 on 11 Sep 20264,255 on 12 Sep 20264,265 on 13 Sep 20264,268 on 14 Sep 20264,252 on 15 Sep 20264,261 on 16 Sep 20264,278 on 17 Sep 20264,256 on 18 Sep 20264,263 on 19 Sep 20264,288 on 20 Sep 20264,259 on 21 Sep 20264,253 on 22 Sep 20264,252 on 30 Sep 20264,265 on 1 Oct 2026; samples 4,259 to 4,2704,252 on 2 Oct 2026; samples 4,250 to 4,2684,268 on 3 Oct 20264,262 on 4 Oct 20264,263 on 5 Oct 2026; samples 4,262 to 4,2654,274 on 6 Oct 20264,266 on 7 Oct 20264,254 on 8 Oct 2026; samples 4,253 to 4,2554,246 on 9 Oct 20264,267 on 10 Oct 20264,270 on 11 Oct 2026; samples 4,265 to 4,2744,282 on 12 Oct 20264,240 on 13 Oct 20264,235 on 14 Oct 2026; samples 4,217 to 4,253

    EPYC 9R05 · m8azn

  • Blacksmith4,238median of 14 days

    02k4k6k4,225 on 27 Jul 20264,217 on 28 Jul 20264,025 on 29 Jul 20264,435 on 30 Jul 20264,048 on 31 Jul 20264,430 on 1 Aug 20264,299 on 2 Aug 20263,739 on 3 Aug 20263,385 on 5 Aug 20264,351 on 6 Aug 20263,749 on 7 Aug 20264,544 on 8 Aug 20264,621 on 9 Aug 20264,183 on 10 Aug 20264,103 on 11 Aug 20263,875 on 12 Aug 20263,891 on 13 Aug 20264,204 on 14 Aug 20264,429 on 15 Aug 20264,426 on 16 Aug 20264,529 on 17 Aug 20263,789 on 18 Aug 20263,629 on 19 Aug 20264,220 on 20 Aug 20263,598 on 21 Aug 20264,506 on 22 Aug 20264,574 on 23 Aug 20264,426 on 24 Aug 20264,111 on 25 Aug 20264,423 on 26 Aug 20263,745 on 27 Aug 20264,357 on 28 Aug 20264,419 on 29 Aug 20264,429 on 30 Aug 20264,153 on 31 Aug 20264,106 on 1 Sep 20264,482 on 2 Sep 20264,184 on 3 Sep 20263,792 on 4 Sep 20264,432 on 5 Sep 20263,901 on 6 Sep 20264,430 on 7 Sep 20264,282 on 8 Sep 20264,259 on 9 Sep 20264,372 on 10 Sep 20264,489 on 12 Sep 20264,330 on 13 Sep 20263,431 on 14 Sep 20263,698 on 15 Sep 20264,159 on 16 Sep 20263,732 on 17 Sep 20264,434 on 18 Sep 20264,554 on 19 Sep 20264,419 on 20 Sep 20263,739 on 21 Sep 20263,808 on 22 Sep 20263,914 on 23 Sep 20263,675 on 30 Sep 20263,821 on 1 Oct 20264,098 on 2 Oct 2026; samples 3,902 to 4,3604,420 on 3 Oct 20263,955 on 4 Oct 20263,729 on 5 Oct 2026; samples 3,583 to 3,8763,898 on 6 Oct 20264,331 on 7 Oct 20264,242 on 8 Oct 2026; samples 4,093 to 4,3914,249 on 9 Oct 20264,501 on 10 Oct 20264,234 on 11 Oct 2026; samples 4,217 to 4,2504,347 on 12 Oct 20264,118 on 13 Oct 20264,323 on 14 Oct 2026; samples 4,186 to 4,461

    EPYC · blacksmith-2vcpu-ubuntu-2404

  • Warpbuild4,148median of 14 days

    02k4k6k3,594 on 27 Jul 20263,636 on 28 Jul 20263,595 on 29 Jul 20263,692 on 30 Jul 20263,613 on 31 Jul 20263,697 on 1 Aug 20263,635 on 2 Aug 20263,815 on 3 Aug 20263,792 on 4 Aug 20263,119 on 5 Aug 20263,820 on 6 Aug 20263,842 on 7 Aug 20263,881 on 8 Aug 20263,881 on 9 Aug 20262,703 on 10 Aug 20262,308 on 11 Aug 20263,868 on 12 Aug 20263,781 on 13 Aug 20263,851 on 14 Aug 20263,850 on 15 Aug 20263,858 on 16 Aug 20263,882 on 17 Aug 20263,918 on 18 Aug 20262,635 on 19 Aug 20262,577 on 20 Aug 20262,597 on 21 Aug 20263,841 on 22 Aug 20263,892 on 23 Aug 20263,795 on 24 Aug 20263,853 on 25 Aug 20263,788 on 26 Aug 20263,638 on 27 Aug 20263,873 on 28 Aug 20263,710 on 29 Aug 20263,702 on 30 Aug 20263,793 on 31 Aug 20263,754 on 1 Sep 20264,165 on 2 Sep 20264,482 on 3 Sep 20264,468 on 4 Sep 20264,515 on 5 Sep 20264,506 on 6 Sep 20264,465 on 7 Sep 20264,542 on 8 Sep 20264,539 on 9 Sep 20264,451 on 10 Sep 20264,042 on 11 Sep 20264,473 on 12 Sep 20264,119 on 13 Sep 20263,898 on 14 Sep 20264,284 on 15 Sep 20264,138 on 16 Sep 20264,237 on 17 Sep 20264,006 on 18 Sep 20264,124 on 19 Sep 20264,433 on 20 Sep 20264,113 on 21 Sep 20264,131 on 22 Sep 20264,067 on 23 Sep 20264,035 on 30 Sep 20264,181 on 1 Oct 2026; samples 4,144 to 4,2194,124 on 2 Oct 2026; samples 3,975 to 4,3363,944 on 3 Oct 20264,207 on 4 Oct 20264,106 on 5 Oct 2026; samples 3,826 to 4,3854,230 on 6 Oct 20263,812 on 7 Oct 20264,048 on 8 Oct 2026; samples 4,008 to 4,0884,219 on 9 Oct 20264,121 on 10 Oct 20264,251 on 11 Oct 2026; samples 3,987 to 4,5164,129 on 12 Oct 20264,168 on 13 Oct 20264,181 on 14 Oct 2026; samples 4,148 to 4,214

    Ryzen 9 7950X3D and Ryzen 9 9950X +1 · warp-ubuntu-2404-x64-2x

  • Ubicloud2,936median of 14 days

    02k4k6k3,836 on 27 Jul 20263,315 on 28 Jul 20263,444 on 29 Jul 20263,085 on 30 Jul 20263,393 on 31 Jul 20263,399 on 1 Aug 20263,680 on 2 Aug 20263,618 on 3 Aug 20262,721 on 4 Aug 20263,720 on 5 Aug 20263,749 on 6 Aug 20263,687 on 7 Aug 20263,843 on 8 Aug 20263,908 on 9 Aug 20263,607 on 10 Aug 20263,633 on 11 Aug 20263,468 on 12 Aug 20263,730 on 13 Aug 20263,561 on 14 Aug 20263,624 on 15 Aug 20262,743 on 16 Aug 20263,760 on 17 Aug 20263,122 on 18 Aug 20263,581 on 19 Aug 20263,649 on 20 Aug 20263,712 on 21 Aug 20263,849 on 22 Aug 20263,520 on 23 Aug 20263,668 on 24 Aug 20263,560 on 25 Aug 20263,722 on 26 Aug 20263,529 on 27 Aug 20263,137 on 28 Aug 20263,162 on 29 Aug 20263,550 on 30 Aug 20263,628 on 31 Aug 20263,472 on 1 Sep 20263,657 on 2 Sep 20263,341 on 3 Sep 20263,818 on 4 Sep 20263,644 on 5 Sep 20262,942 on 6 Sep 20263,589 on 7 Sep 20263,012 on 8 Sep 20263,032 on 9 Sep 20263,698 on 10 Sep 20263,674 on 11 Sep 20263,174 on 12 Sep 20263,492 on 13 Sep 20263,271 on 14 Sep 20263,104 on 15 Sep 20263,738 on 16 Sep 20262,923 on 17 Sep 20263,773 on 18 Sep 20262,546 on 19 Sep 20263,735 on 20 Sep 20263,273 on 21 Sep 20263,495 on 22 Sep 20263,255 on 23 Sep 20262,959 on 30 Sep 20263,451 on 1 Oct 2026; samples 3,342 to 3,5603,397 on 2 Oct 2026; samples 2,994 to 3,7131,906 on 3 Oct 20262,196 on 4 Oct 20262,790 on 5 Oct 2026; samples 2,646 to 2,9332,430 on 6 Oct 20263,661 on 7 Oct 20263,082 on 8 Oct 2026; samples 2,834 to 3,3292,458 on 9 Oct 20263,829 on 10 Oct 20262,425 on 11 Oct 2026; samples 2,409 to 2,4422,669 on 12 Oct 20263,739 on 13 Oct 20263,224 on 14 Oct 2026; samples 2,701 to 3,746

    Ryzen 9 7950X3D · ubicloud-premium-2-ubuntu-2404

  • StarSling3,934median of 14 days

    02k4k6k3,321 on 27 Jul 20264,085 on 28 Jul 20263,276 on 29 Jul 20263,294 on 30 Jul 20263,271 on 31 Jul 20263,342 on 1 Aug 20264,101 on 2 Aug 20263,316 on 3 Aug 20263,315 on 4 Aug 20264,055 on 5 Aug 20263,257 on 6 Aug 20264,082 on 7 Aug 20263,715 on 8 Aug 20263,345 on 9 Aug 20264,107 on 10 Aug 20264,099 on 11 Aug 20264,090 on 12 Aug 20264,068 on 13 Aug 20263,336 on 14 Aug 20263,724 on 15 Aug 20263,343 on 16 Aug 20263,362 on 17 Aug 20263,269 on 18 Aug 20263,324 on 19 Aug 20264,057 on 20 Aug 20263,304 on 21 Aug 20263,347 on 22 Aug 20264,086 on 23 Aug 20263,996 on 24 Aug 20264,085 on 25 Aug 20263,310 on 26 Aug 20263,374 on 27 Aug 20264,093 on 28 Aug 20263,327 on 29 Aug 20263,341 on 30 Aug 20263,325 on 31 Aug 20263,337 on 1 Sep 20263,976 on 2 Sep 20263,323 on 3 Sep 20264,093 on 4 Sep 20263,278 on 5 Sep 20263,355 on 6 Sep 20263,356 on 7 Sep 20263,355 on 8 Sep 20263,339 on 9 Sep 20263,947 on 10 Sep 20263,345 on 11 Sep 20263,313 on 12 Sep 20263,356 on 13 Sep 20263,337 on 14 Sep 20263,302 on 15 Sep 20263,334 on 16 Sep 20263,329 on 17 Sep 20264,081 on 18 Sep 20264,072 on 19 Sep 20264,089 on 20 Sep 20263,711 on 21 Sep 20263,701 on 22 Sep 20263,737 on 23 Sep 20263,902 on 30 Sep 20264,089 on 1 Oct 2026; samples 4,081 to 4,0973,755 on 2 Oct 2026; samples 3,729 to 3,9154,106 on 3 Oct 20264,069 on 4 Oct 20263,969 on 5 Oct 2026; samples 3,932 to 4,0074,034 on 6 Oct 20264,115 on 7 Oct 20263,884 on 8 Oct 2026; samples 3,673 to 4,0963,898 on 9 Oct 20263,667 on 10 Oct 20264,091 on 11 Oct 2026; samples 4,082 to 4,1003,614 on 12 Oct 20263,699 on 13 Oct 20263,781 on 14 Oct 2026; samples 3,594 to 3,969

    EPYC · 4 vCPU · starsling-ubuntu-24.04

  • GitHub2,576median of 14 days

    02k4k6k2,845 on 27 Jul 20262,252 on 28 Jul 20262,374 on 29 Jul 20262,192 on 30 Jul 20262,256 on 31 Jul 20262,339 on 1 Aug 20262,269 on 2 Aug 20262,258 on 3 Aug 20262,270 on 5 Aug 20262,867 on 6 Aug 20262,271 on 7 Aug 20262,864 on 8 Aug 20262,275 on 9 Aug 20262,265 on 10 Aug 20262,200 on 11 Aug 20262,266 on 12 Aug 20263,479 on 13 Aug 20262,283 on 14 Aug 20262,314 on 15 Aug 20262,674 on 16 Aug 20262,282 on 17 Aug 20262,678 on 18 Aug 20262,268 on 19 Aug 20262,198 on 20 Aug 20262,195 on 21 Aug 20262,269 on 22 Aug 20262,161 on 23 Aug 20262,195 on 24 Aug 20262,199 on 25 Aug 20262,279 on 26 Aug 20262,272 on 27 Aug 20262,269 on 28 Aug 20262,253 on 29 Aug 20262,275 on 30 Aug 20262,269 on 31 Aug 20262,190 on 1 Sep 20262,786 on 2 Sep 20262,266 on 3 Sep 20262,277 on 4 Sep 20262,272 on 5 Sep 20262,280 on 6 Sep 20261,836 on 7 Sep 20262,280 on 8 Sep 20262,349 on 9 Sep 20263,596 on 10 Sep 20263,358 on 11 Sep 20263,346 on 12 Sep 20262,319 on 13 Sep 20262,267 on 14 Sep 20262,266 on 15 Sep 20262,265 on 16 Sep 20262,255 on 17 Sep 20262,272 on 18 Sep 20262,272 on 19 Sep 20262,828 on 20 Sep 20262,777 on 21 Sep 20262,267 on 22 Sep 20262,781 on 23 Sep 20262,311 on 30 Sep 20262,573 on 1 Oct 2026; samples 2,311 to 2,8342,316 on 2 Oct 2026; samples 2,253 to 3,4893,490 on 3 Oct 20262,306 on 4 Oct 20262,579 on 5 Oct 2026; samples 2,320 to 2,8383,402 on 6 Oct 20263,568 on 7 Oct 20262,933 on 8 Oct 2026; samples 2,309 to 3,5572,314 on 9 Oct 20262,306 on 10 Oct 20262,572 on 11 Oct 2026; samples 2,314 to 2,8303,474 on 12 Oct 20263,544 on 13 Oct 20262,276 on 14 Oct 2026; samples 2,247 to 2,305

    EPYC 7763 and EPYC 9V74 +4 · github24-2cpu-x64

  • AWS CodeBuild2,055median of 14 days

    02k4k6k2,013 on 27 Jul 20262,003 on 28 Jul 20262,007 on 29 Jul 20261,876 on 30 Jul 20261,871 on 31 Jul 20261,888 on 1 Aug 20262,014 on 2 Aug 20261,914 on 3 Aug 20262,014 on 5 Aug 20261,998 on 6 Aug 20262,007 on 7 Aug 20262,007 on 8 Aug 20261,880 on 9 Aug 20262,006 on 10 Aug 20261,873 on 11 Aug 20262,013 on 12 Aug 20262,008 on 13 Aug 20262,015 on 14 Aug 20262,011 on 15 Aug 20262,000 on 16 Aug 20262,009 on 17 Aug 20261,997 on 18 Aug 20261,877 on 19 Aug 20261,915 on 20 Aug 20261,993 on 21 Aug 20261,908 on 22 Aug 20261,878 on 23 Aug 20262,000 on 24 Aug 20262,015 on 25 Aug 20262,001 on 26 Aug 20262,017 on 27 Aug 20262,006 on 28 Aug 20262,006 on 29 Aug 20261,861 on 30 Aug 20261,903 on 31 Aug 20261,995 on 1 Sep 20261,735 on 2 Sep 20261,874 on 3 Sep 20261,869 on 4 Sep 20261,987 on 5 Sep 20262,012 on 6 Sep 20261,892 on 7 Sep 20261,869 on 8 Sep 20261,885 on 9 Sep 20261,872 on 10 Sep 20261,874 on 11 Sep 20261,866 on 12 Sep 20261,732 on 13 Sep 20262,016 on 14 Sep 20261,870 on 15 Sep 20261,861 on 16 Sep 20262,006 on 17 Sep 20262,008 on 18 Sep 20262,010 on 19 Sep 20262,022 on 20 Sep 20261,894 on 21 Sep 20261,883 on 22 Sep 20262,003 on 23 Sep 20261,974 on 1 Oct 2026; samples 1,973 to 1,9751,972 on 2 Oct 2026; samples 1,963 to 2,0411,983 on 3 Oct 20261,984 on 4 Oct 20262,056 on 5 Oct 2026; samples 2,045 to 2,0662,055 on 6 Oct 20262,049 on 7 Oct 20261,967 on 8 Oct 2026; samples 1,967 to 1,9672,055 on 9 Oct 20262,788 on 10 Oct 20262,589 on 11 Oct 2026; samples 2,586 to 2,5932,599 on 12 Oct 20262,601 on 13 Oct 20262,786 on 14 Oct 2026; samples 2,599 to 2,972

    Xeon 8275CL and Xeon 8124M +2 · codebuild-ubuntu-8.0-small

  • Avrea4,973median of 14 days

    02k4k6k4,892 on 27 Jul 20264,903 on 28 Jul 20264,816 on 29 Jul 20264,830 on 30 Jul 20264,882 on 31 Jul 20264,907 on 1 Aug 20264,886 on 2 Aug 20264,895 on 3 Aug 20264,914 on 4 Aug 20264,911 on 5 Aug 20264,912 on 7 Aug 20264,914 on 8 Aug 20264,914 on 9 Aug 20264,836 on 10 Aug 20264,744 on 11 Aug 20264,887 on 12 Aug 20264,770 on 13 Aug 20264,787 on 14 Aug 20264,736 on 15 Aug 20264,896 on 16 Aug 20264,787 on 18 Aug 20264,885 on 19 Aug 20264,905 on 20 Aug 20264,565 on 21 Aug 20264,907 on 22 Aug 20264,837 on 23 Aug 20264,874 on 24 Aug 20264,870 on 25 Aug 20264,895 on 26 Aug 20264,867 on 27 Aug 20264,885 on 28 Aug 20264,791 on 29 Aug 20264,894 on 30 Aug 20264,844 on 31 Aug 20264,742 on 1 Sep 20264,765 on 2 Sep 20264,839 on 3 Sep 20264,895 on 4 Sep 20264,840 on 5 Sep 20264,897 on 6 Sep 20264,880 on 7 Sep 20264,769 on 8 Sep 20264,898 on 9 Sep 20264,875 on 10 Sep 20264,178 on 11 Sep 20264,790 on 12 Sep 20264,713 on 13 Sep 20264,337 on 14 Sep 20264,782 on 15 Sep 20264,217 on 16 Sep 20264,002 on 17 Sep 20264,859 on 18 Sep 20265,028 on 19 Sep 20264,205 on 20 Sep 20265,023 on 21 Sep 20264,465 on 22 Sep 2026; samples 4,306 to 4,6234,587 on 23 Sep 20264,943 on 30 Sep 20264,875 on 1 Oct 2026; samples 4,721 to 5,0295,015 on 2 Oct 2026; samples 5,000 to 5,0205,037 on 3 Oct 20265,038 on 4 Oct 20264,899 on 5 Oct 2026; samples 4,790 to 5,0095,034 on 6 Oct 20264,478 on 7 Oct 20264,882 on 8 Oct 2026; samples 4,734 to 5,0314,879 on 9 Oct 20264,996 on 10 Oct 20264,950 on 11 Oct 2026; samples 4,872 to 5,0284,828 on 12 Oct 20265,010 on 13 Oct 20265,013 on 14 Oct 2026; samples 4,993 to 5,032

    Apple M5 Max · avrea-ubuntu-24.04-arm-2-vcpu

  • Namespace4,835median of 14 days

    02k4k6k4,948 on 27 Jul 20264,712 on 28 Jul 20263,871 on 29 Jul 20264,682 on 30 Jul 20264,313 on 31 Jul 20264,620 on 1 Aug 20264,600 on 2 Aug 20264,718 on 3 Aug 20264,776 on 4 Aug 20264,134 on 5 Aug 20264,850 on 7 Aug 2026; samples 4,839 to 4,8614,944 on 8 Aug 20264,941 on 9 Aug 20264,826 on 10 Aug 20264,900 on 11 Aug 20264,832 on 12 Aug 20264,805 on 13 Aug 20264,796 on 14 Aug 20264,922 on 15 Aug 20264,963 on 16 Aug 20264,220 on 18 Aug 20263,742 on 19 Aug 20264,242 on 20 Aug 20264,806 on 21 Aug 20264,705 on 22 Aug 20264,982 on 23 Aug 20264,779 on 24 Aug 20264,747 on 25 Aug 20264,946 on 26 Aug 20264,666 on 27 Aug 20264,760 on 28 Aug 20264,811 on 29 Aug 20264,802 on 30 Aug 20264,752 on 31 Aug 20264,736 on 1 Sep 20264,884 on 2 Sep 20264,820 on 3 Sep 20263,700 on 4 Sep 20264,924 on 5 Sep 20264,876 on 6 Sep 20264,753 on 7 Sep 20264,794 on 8 Sep 20264,769 on 10 Sep 20264,695 on 11 Sep 20264,888 on 12 Sep 20264,910 on 13 Sep 20264,868 on 14 Sep 20264,594 on 15 Sep 20264,819 on 16 Sep 20264,817 on 17 Sep 20264,692 on 18 Sep 20264,914 on 19 Sep 20263,969 on 20 Sep 20264,851 on 21 Sep 20264,893 on 22 Sep 2026; samples 4,799 to 4,9874,919 on 23 Sep 20264,784 on 13 Oct 20264,638 on 14 Oct 2026; samples 4,459 to 4,817

    Apple M4 Pro / M5 Max · nscloud-ubuntu-24.04-arm64-2x8,namespace-features:linux-on-apple-silicon=true

  • Warpbuild1,941median of 14 days

    02k4k6k1,943 on 27 Jul 20261,945 on 28 Jul 20261,944 on 29 Jul 20261,946 on 30 Jul 20261,556 on 31 Jul 20261,942 on 1 Aug 20261,944 on 2 Aug 20261,944 on 3 Aug 20261,945 on 4 Aug 20261,944 on 5 Aug 20261,946 on 7 Aug 20261,945 on 8 Aug 20261,944 on 9 Aug 20261,944 on 10 Aug 20261,946 on 11 Aug 20261,943 on 12 Aug 20261,943 on 13 Aug 20261,945 on 14 Aug 20261,945 on 15 Aug 20261,944 on 16 Aug 20261,945 on 18 Aug 20261,946 on 19 Aug 20261,943 on 20 Aug 20261,942 on 21 Aug 20261,941 on 22 Aug 20261,944 on 23 Aug 20261,946 on 24 Aug 20261,943 on 25 Aug 20261,944 on 26 Aug 20261,943 on 27 Aug 20261,944 on 28 Aug 20261,944 on 29 Aug 20261,944 on 30 Aug 20261,945 on 31 Aug 20261,942 on 1 Sep 20261,945 on 2 Sep 20261,945 on 3 Sep 20261,945 on 4 Sep 20261,943 on 5 Sep 20261,944 on 6 Sep 20261,945 on 7 Sep 20261,944 on 8 Sep 20261,943 on 9 Sep 20261,944 on 10 Sep 20261,944 on 11 Sep 20261,944 on 12 Sep 20261,943 on 13 Sep 20261,944 on 14 Sep 20261,942 on 15 Sep 20261,944 on 16 Sep 20261,937 on 17 Sep 20261,945 on 18 Sep 20261,943 on 19 Sep 20261,943 on 20 Sep 20261,943 on 21 Sep 20261,749 on 22 Sep 2026; samples 1,555 to 1,9431,944 on 23 Sep 20261,941 on 30 Sep 20261,941 on 1 Oct 2026; samples 1,941 to 1,9411,941 on 2 Oct 2026; samples 1,940 to 1,9411,941 on 3 Oct 20261,942 on 4 Oct 20261,939 on 5 Oct 2026; samples 1,937 to 1,9411,939 on 6 Oct 20261,940 on 7 Oct 20261,941 on 8 Oct 2026; samples 1,941 to 1,9411,943 on 9 Oct 20261,941 on 10 Oct 20261,941 on 11 Oct 2026; samples 1,940 to 1,9411,942 on 12 Oct 20261,942 on 13 Oct 20261,940 on 14 Oct 2026; samples 1,939 to 1,940

    Neoverse-V2 · warp-ubuntu-latest-arm64-2x

  • RunsOn2,476median of 14 days

    02k4k6k2,468 on 30 Sep 20262,463 on 1 Oct 2026; samples 2,460 to 2,4672,477 on 2 Oct 2026; samples 2,474 to 2,4772,475 on 3 Oct 20262,476 on 4 Oct 20262,476 on 5 Oct 2026; samples 2,475 to 2,4782,474 on 6 Oct 20262,478 on 7 Oct 20262,474 on 8 Oct 2026; samples 2,472 to 2,4762,478 on 9 Oct 20262,477 on 10 Oct 20262,474 on 11 Oct 2026; samples 2,470 to 2,4782,478 on 12 Oct 20262,469 on 13 Oct 20262,476 on 14 Oct 2026

    Neoverse-V3 · m9gd

  • GitHub1,875median of 14 days

    02k4k6k1,872 on 27 Jul 20261,871 on 28 Jul 20261,874 on 29 Jul 20261,873 on 30 Jul 20261,873 on 31 Jul 20261,876 on 1 Aug 20261,877 on 2 Aug 20261,875 on 3 Aug 20261,876 on 4 Aug 20261,872 on 5 Aug 20261,874 on 6 Aug 20261,875 on 7 Aug 20261,878 on 8 Aug 20261,878 on 9 Aug 20261,875 on 10 Aug 20261,500 on 11 Aug 20261,870 on 12 Aug 20261,870 on 13 Aug 20261,871 on 14 Aug 20261,875 on 15 Aug 20261,875 on 16 Aug 20261,876 on 17 Aug 20261,875 on 18 Aug 20261,873 on 19 Aug 20261,872 on 20 Aug 20261,874 on 21 Aug 20261,872 on 22 Aug 20261,878 on 23 Aug 20261,873 on 24 Aug 20261,874 on 25 Aug 20261,873 on 26 Aug 20261,875 on 27 Aug 20261,877 on 28 Aug 20261,873 on 29 Aug 20261,875 on 30 Aug 20261,874 on 31 Aug 20261,870 on 1 Sep 20261,874 on 2 Sep 20261,873 on 3 Sep 20261,867 on 4 Sep 20261,878 on 5 Sep 20261,874 on 6 Sep 20261,873 on 7 Sep 20261,879 on 8 Sep 20261,872 on 9 Sep 20261,878 on 10 Sep 20261,872 on 11 Sep 20261,879 on 12 Sep 20261,874 on 13 Sep 20261,874 on 14 Sep 20261,873 on 15 Sep 20261,872 on 16 Sep 20261,871 on 17 Sep 20261,873 on 18 Sep 20261,877 on 19 Sep 20261,872 on 20 Sep 20261,874 on 21 Sep 20261,873 on 22 Sep 2026; samples 1,872 to 1,8731,877 on 23 Sep 20261,873 on 30 Sep 20261,879 on 1 Oct 2026; samples 1,878 to 1,8791,875 on 2 Oct 2026; samples 1,873 to 1,8791,877 on 3 Oct 20261,874 on 4 Oct 20261,878 on 5 Oct 2026; samples 1,878 to 1,8791,872 on 6 Oct 20261,880 on 7 Oct 20261,875 on 8 Oct 2026; samples 1,874 to 1,8761,874 on 9 Oct 20261,875 on 10 Oct 20261,873 on 11 Oct 2026; samples 1,873 to 1,8741,875 on 12 Oct 20261,875 on 13 Oct 20261,874 on 14 Oct 2026; samples 1,869 to 1,879

    Neoverse-N2 · github24-2cpu-arm64

  • Blacksmith1,344median of 14 days

    02k4k6k1,309 on 27 Jul 20261,326 on 28 Jul 20261,310 on 29 Jul 20261,340 on 30 Jul 20261,342 on 31 Jul 20261,366 on 1 Aug 20261,365 on 2 Aug 20261,355 on 3 Aug 20261,347 on 4 Aug 20261,276 on 5 Aug 20261,342 on 7 Aug 20261,369 on 8 Aug 20261,355 on 9 Aug 20261,345 on 10 Aug 20261,352 on 11 Aug 20261,340 on 12 Aug 20261,355 on 13 Aug 20261,349 on 14 Aug 20261,374 on 15 Aug 20261,379 on 16 Aug 20261,355 on 18 Aug 20261,354 on 19 Aug 20261,351 on 20 Aug 20261,335 on 21 Aug 20261,329 on 22 Aug 20261,369 on 23 Aug 20261,344 on 24 Aug 20261,344 on 25 Aug 20261,332 on 26 Aug 20261,347 on 27 Aug 20261,361 on 28 Aug 20261,361 on 29 Aug 20261,375 on 30 Aug 20261,337 on 31 Aug 20261,341 on 1 Sep 20261,348 on 2 Sep 20261,331 on 3 Sep 20261,344 on 4 Sep 20261,381 on 5 Sep 20261,373 on 6 Sep 20261,359 on 7 Sep 20261,336 on 8 Sep 20261,325 on 9 Sep 20261,322 on 10 Sep 20261,333 on 11 Sep 20261,368 on 12 Sep 20261,373 on 13 Sep 20261,344 on 14 Sep 20261,332 on 15 Sep 20261,321 on 16 Sep 20261,350 on 17 Sep 20261,353 on 18 Sep 20261,376 on 19 Sep 20261,330 on 20 Sep 20261,340 on 21 Sep 20261,331 on 22 Sep 2026; samples 1,328 to 1,3341,328 on 23 Sep 20261,319 on 30 Sep 20261,329 on 1 Oct 2026; samples 1,325 to 1,3321,338 on 2 Oct 2026; samples 1,331 to 1,3421,357 on 3 Oct 20261,351 on 4 Oct 20261,352 on 5 Oct 2026; samples 1,352 to 1,3521,346 on 6 Oct 20261,337 on 7 Oct 20261,339 on 8 Oct 2026; samples 1,337 to 1,3411,343 on 9 Oct 20261,357 on 10 Oct 20261,358 on 11 Oct 2026; samples 1,356 to 1,3611,349 on 12 Oct 20261,339 on 13 Oct 20261,340 on 14 Oct 2026; samples 1,338 to 1,342

    Ampere-1a and Neoverse-N1 · blacksmith-2vcpu-ubuntu-2404-arm

  • Ubicloud1,313median of 14 days

    02k4k6k1,313 on 27 Jul 20261,316 on 28 Jul 20261,316 on 29 Jul 20261,313 on 30 Jul 20261,318 on 31 Jul 20261,314 on 1 Aug 20261,316 on 2 Aug 20261,312 on 3 Aug 20261,300 on 4 Aug 20261,312 on 5 Aug 20261,312 on 7 Aug 20261,311 on 8 Aug 20261,317 on 9 Aug 20261,316 on 10 Aug 20261,309 on 11 Aug 20261,312 on 12 Aug 20261,303 on 13 Aug 20261,307 on 14 Aug 20261,308 on 15 Aug 20261,309 on 16 Aug 20261,315 on 18 Aug 20261,292 on 19 Aug 20261,309 on 20 Aug 20261,307 on 21 Aug 20261,314 on 22 Aug 20261,314 on 23 Aug 20261,312 on 24 Aug 20261,304 on 25 Aug 20261,309 on 26 Aug 20261,310 on 27 Aug 20261,307 on 28 Aug 20261,293 on 29 Aug 20261,304 on 30 Aug 20261,301 on 31 Aug 20261,310 on 1 Sep 20261,313 on 2 Sep 20261,310 on 3 Sep 20261,310 on 4 Sep 20261,314 on 5 Sep 20261,306 on 6 Sep 20261,308 on 7 Sep 20261,298 on 8 Sep 20261,306 on 9 Sep 20261,294 on 10 Sep 20261,313 on 11 Sep 20261,315 on 12 Sep 20261,316 on 13 Sep 20261,305 on 14 Sep 20261,310 on 15 Sep 20261,312 on 16 Sep 20261,309 on 17 Sep 20261,305 on 18 Sep 20261,320 on 19 Sep 20261,317 on 20 Sep 20261,308 on 21 Sep 20261,306 on 22 Sep 2026; samples 1,299 to 1,3131,315 on 23 Sep 20261,317 on 30 Sep 20261,314 on 1 Oct 2026; samples 1,313 to 1,3151,316 on 2 Oct 2026; samples 1,315 to 1,3171,316 on 3 Oct 20261,312 on 4 Oct 20261,311 on 5 Oct 2026; samples 1,308 to 1,3151,309 on 6 Oct 20261,315 on 7 Oct 20261,313 on 8 Oct 2026; samples 1,312 to 1,3141,316 on 9 Oct 20261,314 on 10 Oct 20261,308 on 11 Oct 2026; samples 1,305 to 1,3121,313 on 12 Oct 20261,298 on 13 Oct 20261,306 on 14 Oct 2026; samples 1,300 to 1,312

    Neoverse-N1 · ubicloud-standard-2-arm-ubuntu-2404

Line: each provider's fastest 2 vCPU runner label, the median of each day's samples. A provider without a 2 vCPU label shows its smallest runner, with the size on its panel. Shaded band: that day's lowest to highest sample. One scale for every panel.

Every runner

Best single-thread runner first; the others behind +N more.

Show every runner93 runners · x64, arm64
PassMark sysbench
Runner 1 threadall cores1 threadall threadscores7-Zip runs $/min
x6457 runners, 9 providers
Namespace 4x8 4 vCPU · 8 GB EPYC · model hidden 4,68516,1387,02027,5333.933,179 5 $0.0060
8x32 8 vCPU · 32 GB EPYC · model hidden 4,67730,1137,01154,1587.760,979 5 $0.0240
8x16 8 vCPU · 16 GB EPYC · model hidden 4,64129,3237,02853,3117.660,156 5 $0.0120
4x16 4 vCPU · 16 GB EPYC · model hidden 4,64716,1347,01127,4393.933,114 5 $0.0120
2x8 2 vCPU · 8 GB EPYC · model hidden 4,6058,2336,89413,4421.916,254 5 $0.0060
Avrea 2 vCPU · 8 GB EPYC 4585PX 4,3205,8566,2416,6371.113,004 5 $0.0040
8 vCPU 8 vCPU · 32 GB EPYC 4585PX 4,27321,3566,21226,4144.349,101 5 $0.0160
4 vCPU 4 vCPU · 16 GB EPYC 4585PX 4,24311,1746,11313,3152.223,093 5 $0.0080
RunsOn m8azn.3xlarge 12 vCPU · 48 GB EPYC 9R05 4,26641,1165,63267,57312.0120,314 5 $0.0076 od $0.0210
c8a.2xlarge 8 vCPU · 16 GB EPYC 9R45 3,81725,4515,04540,1378.074,021 5 $0.0032 od $0.0075
m8a.2xlarge 8 vCPU · 32 GB EPYC 9R45 3,79325,4255,02940,1658.073,907 5 $0.0030 od $0.0084
m8a.2xlarge 8 vCPU · 32 GB EPYC 9R45 3,80925,4215,03540,0448.073,826 5 $0.0051 od $0.0105
m8a.2xlarge 8 vCPU · 32 GB EPYC 9R45 3,80225,3895,03340,1328.073,775 5 $0.0030 od $0.0084
r8a.2xlarge 8 vCPU · 64 GB EPYC 9R45 3,81225,4685,04940,2348.074,083 5 $0.0046 od $0.0110
m8azn.xlarge 4 vCPU · 16 GB EPYC 9R05 4,26415,1415,66122,6544.042,503 5 $0.0030 od $0.0072
m8azn.xlarge 4 vCPU · 16 GB EPYC 9R05 4,26015,1305,66622,5814.042,421 5 $0.0051 od $0.0093
m8i.2xlarge 8 vCPU · 32 GB Xeon 6975P-C 3,25815,4933,07012,8214.246,774 5 $0.0033 od $0.0074
m8i-flex.2xlarge 8 vCPU · 32 GB Xeon 6975P-C 3,27015,5403,07812,8584.245,886 4 $0.0024 od $0.0070
m8a.xlarge 4 vCPU · 16 GB EPYC 9R45 3,82813,5585,05420,2344.039,052 3 $0.0015 od $0.0044
i7i.2xlarge 8 vCPU · 64 GB Xeon 8559C 3,12615,1063,04812,1114.043,948 5 $0.0055 od $0.0129
m8id.2xlarge 8 vCPU · 32 GB Xeon 6975P-C 3,26515,5413,07812,8504.246,733 5 $0.0035 od $0.0090
c8a.xlarge 4 vCPU · 8 GB EPYC 9R45 3,80313,4775,03920,2634.039,152 5 $0.0016 od $0.0039
r8a.xlarge 4 vCPU · 32 GB EPYC 9R45 3,80513,4855,03620,1474.038,874 5 $0.0020 od $0.0056
m8azn.large 2 vCPU · 8 GB EPYC 9R05 4,2537,7225,63511,3162.022,156 5 $0.0012 od $0.0037
i7i.xlarge 4 vCPU · 32 GB Xeon 8559C 3,0947,6192,9025,8652.022,329 5 $0.0019 od $0.0066
m8id.xlarge 4 vCPU · 16 GB Xeon 6975P-C 3,2678,0563,0776,4212.124,480 4 $0.0015 od $0.0047
m8i-flex.xlarge 4 vCPU · 16 GB Xeon 6975P-C 3,2598,0093,0726,4022.123,136 5 $0.0017 od $0.0037
r8a.large 2 vCPU · 16 GB EPYC 9R45 3,7896,8635,03910,1012.019,914 5 $0.0010 od $0.0030
m8i.xlarge 4 vCPU · 16 GB Xeon 6975P-C 3,2568,0173,0786,4122.123,780 5 $0.0018 od $0.0038
m8a.large 2 vCPU · 8 GB EPYC 9R45 3,8066,8995,06210,1052.020,215 5 $0.0009 od $0.0023
r8a.large 2 vCPU · 16 GB EPYC 9R45 3,8276,9555,04810,1842.020,212 1 $0.0010 od $0.0030
c8a.large 2 vCPU · 4 GB EPYC 9R45 3,8026,8775,02610,1102.019,740 5 $0.0009 od $0.0021
m8id.large 2 vCPU · 8 GB Xeon 6975P-C 3,2244,0563,0643,1961.011,866 4 $0.0009 od $0.0025
m8i.large 2 vCPU · 8 GB Xeon 6975P-C 3,2354,0723,0693,1961.012,254 5 $0.0009 od $0.0021
m8i-flex.large 2 vCPU · 8 GB Xeon 6975P-C 3,2474,0743,0823,2131.011,875 5 $0.0009 od $0.0020
i7i.large 2 vCPU · 16 GB Xeon 8559C 2,8533,6612,8292,9071.011,143 5 $0.0010 od $0.0035
t8i.medium 2 vCPU · 4 GB Xeon 6975P-C 3,2384,0833,0813,2141.011,504 5 $0.0007 od $0.0011
Blacksmith 8 vCPU · 32 GB EPYC · model hidden 4,23427,2826,23649,2557.955,188 5 $0.0160
4 vCPU 4 vCPU · 16 GB EPYC · model hidden 4,23014,7596,32123,9723.829,161 5 $0.0080
2 vCPU 2 vCPU · 8 GB EPYC · model hidden 4,0936,9445,88511,2191.913,815 5 $0.0040
Warpbuild 4 vCPU · 16 GB Ryzen 9 9950X 4,19611,0836,28813,0802.124,753 5 $0.0080
8 vCPU 8 vCPU · 32 GB Ryzen 9 9950X 4,17720,9916,31526,3194.243,585 5 $0.0160
2 vCPU 2 vCPU · 8 GB Ryzen 9 9950X 4,0085,5016,1236,4071.012,367 5 $0.0040
Ubicloud premium 8 vCPU · 32 GB Ryzen 9 7950X3D 4,03824,7205,97144,2537.451,305 5 $0.0080
standard 8 vCPU 8 vCPU · 32 GB EPYC 9454P 2,75916,0584,22728,3706.728,398 5 $0.0050
premium 4 vCPU 4 vCPU · 16 GB Ryzen 9 7950X3D 3,0909,0415,66320,2503.623,201 5 $0.0040
standard 4 vCPU 4 vCPU · 16 GB EPYC 9454 2,8789,9734,53517,1983.823,200 5 $0.0025
standard 2 vCPU 2 vCPU · 8 GB EPYC 9454P 2,9165,2204,2657,8941.912,371 5 $0.0013
premium 2 vCPU 2 vCPU · 8 GB Ryzen 9 7950X3D 2,9334,5835,3417,5381.410,514 5 $0.0020
StarSling 4 vCPU · 16 GB EPYC · model hidden 3,96913,9145,94522,8753.829,786 5 $0.0080
8 vCPU 8 vCPU · 32 GB EPYC · model hidden 3,84925,1055,59743,2907.752,322 5 $0.0160
GitHub 8 vCPU · 32 GB EPYC 9V45 2,89914,4204,53819,0824.237,865 5 $0.0220
4-core 4 vCPU · 16 GB EPYC 9V45 2,8887,5103,6728,2062.219,087 5 $0.0120
2-core 2 vCPU · 8 GB EPYC 9V74 2,3163,1003,6664,0991.18,312 5 $0.0060
AWS CodeBuild large 8 vCPU · 16 GB Xeon 8488C 2,1569,2031,2367,7076.225,736 5 $0.0200
medium 4 vCPU · 8 GB Xeon 8488C 2,1484,6941,1813,7883.213,218 5 $0.0100
small 2 vCPU · 3 GB Xeon 8488C 2,0662,3571,2241,9101.66,708 5 $0.0050
arm6436 runners, 7 providers
Avrea 2 vCPU · 8 GB Apple M5 Max · as stated by the provider 5,0284,41313,47826,1011.922,603 5 $0.0120
8 vCPU 8 vCPU · 32 GB Apple M5 Max · as stated by the provider 4,84215,45813,43196,3807.274,252 5 $0.0480
4 vCPU 4 vCPU · 16 GB Apple M5 Max · as stated by the provider 4,8608,38813,32249,4403.741,298 5 $0.0240
Namespace Apple silicon 2x8 2 vCPU · 8 GB CPU not reported 4,8174,25013,17225,5561.921,594 1 $0.0420
Apple silicon 4x16 4 vCPU · 16 GB CPU not reported 4,7308,19513,20748,7133.740,223 5 $0.0840
8x16 8 vCPU · 16 GB Ampere-1a 1,1826,4208216,4917.929,019 5 $0.0120
8x32 8 vCPU · 32 GB Ampere-1a 1,1716,3908186,4787.928,989 5 $0.0240
4x8 4 vCPU · 8 GB Ampere-1a 1,1843,2798203,2033.914,335 5 $0.0060
4x16 4 vCPU · 16 GB Ampere-1a 1,1813,2888203,2013.914,835 5 $0.0120
2x8 2 vCPU · 8 GB Ampere-1a 1,1711,6078131,5591.97,117 5 $0.0060
Warpbuild xfast (Apple M4 Pro) 6 vCPU · 14 GB Apple M4 Pro · as stated by the provider 3,2218,98810,71160,4235.652,724 5 $0.0800
xfast (Apple M4 Pro) 12 vCPU 12 vCPU · 28 GB Apple M4 Pro · as stated by the provider 3,20714,00510,71687,3278.184,480 5 $0.1600
8 vCPU 8 vCPU · 32 GB Neoverse-V2 1,9417,6423,33426,5988.050,124 5 $0.0120
4 vCPU 4 vCPU · 16 GB Neoverse-V2 1,9423,9253,33413,2664.025,991 5 $0.0060
2 vCPU 2 vCPU · 8 GB Neoverse-V2 1,9411,9773,3336,6062.013,270 5 $0.0030
RunsOn m9gd.large 2 vCPU · 8 GB Neoverse-V3 2,4762,8553,7717,5212.021,337 5 $0.0010 od $0.0024
m9g.2xlarge 8 vCPU · 32 GB Neoverse-V3 2,47110,8893,76730,0998.080,525 5 $0.0033 od $0.0068
m9gd.2xlarge 8 vCPU · 32 GB Neoverse-V3 2,47210,8913,76730,0888.079,269 5 $0.0031 od $0.0087
m9g.2xlarge 8 vCPU · 32 GB Neoverse-V3 2,47110,9083,76930,0948.078,684 5 $0.0033 od $0.0068
c8g.2xlarge 8 vCPU · 16 GB Neoverse-V2 1,9477,6723,17525,3568.063,997 5 $0.0028 od $0.0056
m9gd.xlarge 4 vCPU · 16 GB Neoverse-V3 2,4755,6313,76615,0514.041,579 5 $0.0017 od $0.0045
m9g.xlarge 4 vCPU · 16 GB Neoverse-V3 2,4765,6163,76715,0564.041,751 5 $0.0017 od $0.0036
m9g.xlarge 4 vCPU · 16 GB Neoverse-V3 2,4765,6243,77015,0564.041,482 5 $0.0017 od $0.0036
r9g.xlarge 4 vCPU · 32 GB Neoverse-V3 2,4745,6243,76815,0564.041,278 2 $0.0021 od $0.0046
c8g.xlarge 4 vCPU · 8 GB Neoverse-V2 1,9473,9313,17312,6644.032,148 5 $0.0013 od $0.0030
r9g.large 2 vCPU · 16 GB Neoverse-V3 2,4692,8403,7667,5182.020,371 1 $0.0010 od $0.0024
m9g.large 2 vCPU · 8 GB Neoverse-V3 2,4662,8403,7637,5012.020,879 5 $0.0010 od $0.0019
c8g.large 2 vCPU · 4 GB Neoverse-V2 1,9471,9823,1746,3152.016,191 5 $0.0008 od $0.0016
GitHub 4 vCPU · 16 GB Neoverse-N2 1,8773,9743,95515,7634.025,864 5 $0.0080
2-core 2 vCPU · 8 GB Neoverse-N2 1,8742,0083,9557,8432.012,904 5 $0.0050
Blacksmith 8 vCPU · 32 GB Ampere-1a 1,3567,3689317,3717.932,078 5 $0.0100
4 vCPU 4 vCPU · 16 GB Ampere-1a 1,3463,7789313,6703.915,699 5 $0.0050
2 vCPU 2 vCPU · 6 GB Ampere-1a 1,3411,9059311,8212.08,048 5 $0.0025
Ubicloud standard 8 vCPU · 32 GB Neoverse-N1 1,3166,1383,34926,3337.934,243 5 $0.0050
standard 4 vCPU 4 vCPU · 16 GB Neoverse-N1 1,3113,0993,34913,0103.916,588 5 $0.0025
standard 2 vCPU 2 vCPU · 8 GB Neoverse-N1 1,3141,5283,3486,6222.08,728 5 $0.0013
  • medians of finished jobs
  • bold best on the arch
  • cores: sysbench, every thread over one
  • od: RunsOn at EC2 on-demand

How it's measured

Methodology

The hardware suite runs PassMark PerformanceTest, sysbench cpu and the 7-Zip benchmark inside the job, on the runner the label picked. The CPU model is the one the job's host reported; a provider that hides it shows the family.

Detailsscores, sizes, over time
  • CPU (sysbench, 7-Zip, PassMark), memory bandwidth, disk (fio in the job workspace), network downloads, a docker pull, and pgbench against a PostgreSQL service container.
  • PassMark single-thread and CPU Mark are PassMark's own scores. sysbench counts prime-number events per second on one thread, then on every thread.
  • Every runner is compared, whatever its size; the Size pill keeps one vCPU count.
  • A variant needs 3 finished jobs to be its provider's bar; until then its best one shows with its run count.
  • Over time: PassMark single-thread, the test of the chart above, on 2 vCPU runner labels: one sample a day at first, then one per hardware run. A runner label keeps one line throughout.
  • Raw numbers: runners.json.