jgrusewski
74f63f80b2
perf: attention + IQL block 32→256 (occupancy 3.1%→25%)
Same optimization pattern as IQN: increase block size from 32 (1 warp)
to 256 (8 warps) for all attention and IQL kernels.
Attention (forward + backward):
- Loop strides 32→256, shared_concat dynamic shmem
- Added bf16_block_max/bf16_block_sum for 8-warp reduction
- Replaces warp-only shuffle with shared memory cross-warp reduce
IQL (forward+loss, backward, forward-only):
- Loop strides 32→256, added iql_bf16_block_sum
- grad_norm + adam kernels already at 256, unchanged
Expected: ~15-20ms/step savings from attention+IQL combined.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-06 23:37:36 +02:00
..
2026-03-13 10:18:35 +01:00
2026-04-02 16:29:55 +02:00
2026-03-12 19:10:18 +01:00
2026-03-14 11:35:15 +01:00
2026-03-13 10:18:35 +01:00
2026-03-13 10:18:35 +01:00
2026-03-01 22:47:39 +01:00
2026-04-06 23:37:36 +02:00
2026-03-13 10:18:35 +01:00
2026-03-13 10:18:35 +01:00
2026-03-13 10:18:35 +01:00
2026-04-04 01:24:19 +02:00
2026-03-01 22:47:39 +01:00
2026-03-13 10:18:35 +01:00
2026-04-06 19:38:50 +02:00
2026-03-29 22:35:37 +02:00
2026-03-29 22:35:37 +02:00
2026-03-31 12:37:25 +02:00
2026-03-29 19:18:47 +02:00
2026-03-19 00:39:03 +01:00
2026-03-13 10:18:35 +01:00
2026-03-13 10:18:35 +01:00
2026-03-31 00:39:59 +02:00
2026-03-16 21:01:28 +01:00
2026-03-13 10:18:35 +01:00
2026-03-13 10:18:35 +01:00
2026-03-13 10:18:35 +01:00
2026-03-10 13:18:57 +01:00
2026-03-28 01:51:45 +01:00
2026-03-13 10:18:35 +01:00
2026-03-13 10:18:35 +01:00
2026-03-13 10:18:35 +01:00
2026-03-18 00:53:47 +01:00
2026-03-14 11:35:15 +01:00
2026-03-01 22:47:39 +01:00
2026-03-15 11:59:31 +01:00
2026-03-01 22:47:39 +01:00
2026-03-03 22:16:35 +01:00