provenance string | rows list |
|---|---|
Hand-written example kernels in examples/ scored with `cakebench suite --device cuda --full` on one shared A100-SXM4-40GB, as recorded in this repo's README at commit 7631683. Timing is median GPU span with L2 flushed; shared box, so timings are noisy. | [
{
"task": "kmeans_assign",
"shapes_correct": 3,
"shapes_total": 3,
"geomean_speedup": 1.69
},
{
"task": "knn_search",
"shapes_correct": 3,
"shapes_total": 3,
"geomean_speedup": 1.13
},
{
"task": "tinygemm",
"shapes_correct": 8,
"shapes_total": 8,
"geomean_spee... |
cakebench
A kernel benchmark modelled on CAKE (arXiv 2608.12629). It scores a candidate kernel on 5 kernel families (21 shapes in the full GPU portfolio) by first checking correctness against an fp32 oracle on every shape and 3 input distributions (normal / heavy-tailed / tiny), then timing it (median GPU span, L2 flushed) against a torch baseline.
No model weights and no LLM-generated kernels are released here. This repo holds the recorded scores of the five hand-written example kernels (plain torch) on one shared A100-SXM4-40GB, so you can compare your own harness runs.
| task | correct shapes | geomean vs torch baseline |
|---|---|---|
| kmeans_assign | 3/3 | 1.69x |
| knn_search | 3/3 | 1.13x |
| tinygemm | 8/8 | 1.00x |
| kda_decode | 4/4 | 1.03x |
| kda_prefill | 3/3 | 1.06x |
Shared box, noisy timings; the Cake compiler is not public, so this benches kernels, not the compiler; baselines are torch, not the paper's FlashML/FlashKDA/FlashInfer.
- Code and usage: https://github.com/Bobliuuu/cakebench
- Explainer video: https://github.com/Bobliuuu/cakebench/blob/main/video/cakebench_explainer.mp4
- Release: https://github.com/Bobliuuu/cakebench/releases/tag/v0.1.0
- Downloads last month
- 37