mirror of
https://github.com/tinygrad/tinygrad.git
synced 2026-08-30 03:06:06 +00:00
llm: use two waves for IQ3 prefill
This commit is contained in:
@@ -312,7 +312,7 @@ def _packed_expert_kernel(out:UOp, raw:UOp, sel:UOp, xq:UOp, xd:UOp, lut:UOp,
|
||||
prefill = out.shape[0] > routes_per_input
|
||||
output_tile = 2 if ggml_type == 21 and not prefill else \
|
||||
8 if ggml_type == 21 else 4
|
||||
wave_count = 2 if prefill and ggml_type != 21 else 1
|
||||
wave_count = 2 if prefill else 1
|
||||
route, output_block = UOp.range(out.shape[0], 0), UOp.range(out_features // (output_tile * wave_count), 1)
|
||||
group_count, lane_count = in_features // 32, min(32, in_features // 32)
|
||||
lane = UOp.range(lane_count, 2, axis_type=AxisType.LOCAL)
|
||||
|
||||
Reference in New Issue
Block a user