WebGPU Micro-Benchmark Result
Per-operation GPU timings for the profiled device. Benchmark details available in the webgpu-bench open source project.
Device
Recorded 10/3/2026, 4:01:21 PM
WebGPU device
Vendornvidia
Device—
Architectureblackwell
Description—
WebGL device
VendorGoogle Inc. (NVIDIA)
RendererANGLE (NVIDIA, NVIDIA GeForce RTX 5080 (0x00002C02) Direct3D11 vs_5_0 ps_5_0, D3D11)
Browser
NameEdge
Version154.0.0.0
Operating system
NameWindows
Version10
Benchmark Results
Throughput from this run, in bytes or operations per second. Higher is faster.
| Benchmark | Throughput |
|---|---|
atomic direct | 3.17 Gupdate/s |
atomic sharded | 4.82 Gupdate/s |
atomic workgroup | 17.1 Gupdate/s |
branch coherent | 17.1 TFLOP/s |
branch divergent | 10.4 TFLOP/s |
branch none | 25.4 TFLOP/s |
branch uniform | 19.8 TFLOP/s |
branch vec4 if | 2.34 Tchoice/s |
branch vec4 select | 2.34 Tchoice/s |
f32<->f16 convert | 13.4 TOP/s |
fp16 div | 7.26 TFLOP/s |
fp16 ln | 7.44 TFLOP/s |
fp16 mat4 FMA | 29.8 TFLOP/s |
fp16 matvec FMA | 16.8 TFLOP/s |
fp16 pow | 3.78 TFLOP/s |
fp16 rsqrt | 7.45 TFLOP/s |
fp16 scalar FMA | 29.5 TFLOP/s |
fp16 sin/cos | 3.69 TFLOP/s |
fp16 sqrt | 7.43 TFLOP/s |
fp16 vec4 FMA | 28.1 TFLOP/s |
fp32 clamp | 12.8 TOP/s |
fp32 div | 7.4 TFLOP/s |
fp32 fma() builtin | 43.9 TFLOP/s |
fp32 ln | 7.45 TFLOP/s |
fp32 mat4 FMA | 24.7 TFLOP/s |
fp32 matvec FMA | 19.6 TFLOP/s |
fp32 min/max | 25.5 TOP/s |
fp32 pow | 3.79 TFLOP/s |
fp32 rsqrt | 7.57 TFLOP/s |
fp32 scalar FMA | 25.5 TFLOP/s |
fp32 select | 8.65 TOP/s |
fp32 sin/cos | 3.81 TFLOP/s |
fp32 sqrt | 7.58 TFLOP/s |
fp32 vec4 FMA | 27 TFLOP/s |
i32 div | 2.04 TOP/s |
i32 mat4 multiply-add | 26.3 TOP/s |
i32 matvec multiply-add | 20.5 TOP/s |
i32 scalar multiply-add | 29.5 TOP/s |
i32 vec4 multiply-add | 29.3 TOP/s |
i32<->f32 convert | 6.66 TOP/s |
int8 dp4a | 50.3 TOP/s |
int8 dp4a matvec | 54 TOP/s |
layout aos | 40 Gparticle/s |
layout aosoa | 53.4 Gparticle/s |
layout soa | 53.3 Gparticle/s |
read dependent chain | 7.55 Mhop/s |
read gather 16kb | 2.54 TB/s |
read gather 4mb | 1.02 TB/s |
read gather 64mb | 995 GB/s |
read linear | 3.87 TB/s |
reduction serial | 6.9 Gelement/s |
reduction subgroup | 64.1 Gelement/s |
reduction workgroup | 45.7 Gelement/s |
texture interp (built-in) | 6.98 TB/s |
texture interp (manual) | 9.29 TB/s |
tile direct | 64 Goutput/s |
tile shared | 63.9 Goutput/s |
tile shared padded | 62.6 Goutput/s |
u32 byte pack/unpack | 12.9 TOP/s |
u32 countOneBits | 3.77 TOP/s |
u32 firstLeadingBit | 1.03 TOP/s |
u32 reverseBits | 67.3 TOP/s |
u32 variable shift | 7.28 TOP/s |
uint8 dp4a | 50.5 TOP/s |
workgroup 128 | 45.7 Gelement/s |
workgroup 256 | 42.1 Gelement/s |
workgroup 64 | 35.6 Gelement/s |
write linear | 906 GB/s |
write scatter 16kb | 8.78 GB/s |
write scatter 4mb | 1.06 TB/s |
write scatter 64mb | 151 GB/s |