On a single 8×B200 node, over about a week in June, KDA generated kernels that took 53 first places on SOL-ExecBench, NVIDIA's leaderboard that scores each kernel against its speed-of-light bound.The June count of 53 is Humanfia-reported; the board does not keep history. The counts below are read from the live board on 5 October. The run never really stopped. On the board today, the entry it submits under, Databricks, holds more per-kernel first places than anyone.
- 1Databricks · KDA71
- 2SF Tensor47
- 3Infinigence AI39
- 4Geometric24
- 5RIAC18
The firsts are not spread evenly. KDA is strongest on fused operations, where it holds 36 of 82, and weakest on the quantization and FlashInfer kernels, where specialists lead. Switch the baseline to set our count against any rival's.
A collection is ranked by one SOL score over all of its kernels, so the most first places do not make the top spot. On those boards, the same entry is second or third everywhere, within 0.007 of the leader:
| L1 · single operations | #3 | 0.7963 | Infinigence AI | 0.8032 |
|---|---|---|---|---|
| L2 · fused operations | #2 | 0.7921 | SF Tensor | 0.7923 |
| Quantization | #3 | 0.8861 | SF Tensor | 0.8916 |
| FlashInfer-Bench | #3 | 0.7962 | SF Tensor | 0.7992 |
One node, a week, nobody watching
The number to notice is not the count, it is the cost. Kernel optimisation has historically priced in a scarce human: someone who knows the architecture, reads the profiler and has the patience for the twentieth variant. This run cost a machine that was already there, and nobody watching it.
Swipe the diagram sideways to see all of it →
1/0 A queue of kernels to beat.
That changes which problems are worth attacking. A kernel that would take an engineer three days and probably yield nothing is not worth an engineer's three days. It is trivially worth a slot in a queue.