Skip to content

hrx: fence both global and LDS at the decode-split barriers (engine#123 follow-up) #29

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
bong-water-water-bong merged 1 commit into from
Sep 27, 2026
Merged

hrx: fence both global and LDS at the decode-split barriers (engine#123 follow-up) #29

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions ...ml-hrx/kernel-corpus/kernels/loom-libs/ops/flash_attention_decode_split_f32_f16_wmma.loom
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -968,6 +968,7 @@ template.def<@ggml.flash_attention.decode_split.reduce_fused> device priority(20
template.apply<@ggml.flash_attention.decode_split.reduce_completed.direct>(%partial_block_capacity, %producer_block_count, %partial_max_noalias, %partial_sum_noalias, %partial_output_noalias, %output_noalias) : (index, index, buffer, buffer, buffer, buffer)
scf.if %publish_q8 {
kernel.barrier<global> scope(workgroup) ordering(acq_rel)
kernel.barrier<workgroup> scope(workgroup) ordering(acq_rel)
template.apply<@ggml.flash_attention.decode_split.pack_completed_q8>(%key_value_head, %output_noalias, %q8_output) : (index, buffer, buffer)
}
// Do not expose the reset until every final F32 and Q8 store completes.
Expand Down Expand Up @@ -1016,6 +1017,7 @@ template.def<@ggml.flash_attention.decode_split.reduce_fused> device priority(10
template.apply<@ggml.flash_attention.decode_split.reduce_completed.cooperative>(%partial_block_capacity, %producer_block_count, %partial_max_noalias, %partial_sum_noalias, %partial_output_noalias, %output_noalias) : (index, index, buffer, buffer, buffer, buffer)
scf.if %publish_q8 {
kernel.barrier<global> scope(workgroup) ordering(acq_rel)
kernel.barrier<workgroup> scope(workgroup) ordering(acq_rel)
template.apply<@ggml.flash_attention.decode_split.pack_completed_q8>(%key_value_head, %output_noalias, %q8_output) : (index, buffer, buffer)
}
// Do not expose the reset until every final F32 and Q8 store completes.
Expand Down Expand Up @@ -1064,6 +1066,7 @@ template.def<@ggml.flash_attention.decode_split.reduce_fused> device priority(5)
template.apply<@ggml.flash_attention.decode_split.reduce_completed.multipass>(%partial_block_capacity, %producer_block_count, %partial_max_noalias, %partial_sum_noalias, %partial_output_noalias, %output_noalias) : (index, index, buffer, buffer, buffer, buffer)
scf.if %publish_q8 {
kernel.barrier<global> scope(workgroup) ordering(acq_rel)
kernel.barrier<workgroup> scope(workgroup) ordering(acq_rel)
template.apply<@ggml.flash_attention.decode_split.pack_completed_q8>(%key_value_head, %output_noalias, %q8_output) : (index, buffer, buffer)
}
kernel.barrier<global> scope(workgroup) ordering(release)
Expand Down Expand Up @@ -1233,6 +1236,8 @@ kernel.def target(@ggml_flash_attention_decode_split_gfx11_wave64) @ggml_flash_a
scf.if %workitem_is_zero {
view.store %subgroup_sum, %reduction_stage_view[%c1] : f32, view<2xf32>
}
// Wave 0 rewrote partial_max in global memory; every wave reads it below.
kernel.barrier<global> scope(workgroup) ordering(acq_rel)
kernel.barrier<workgroup> scope(workgroup) ordering(acq_rel)
%sum = view.load %reduction_stage_view[%c1] : view<2xf32> -> f32
scf.if %output_channel_valid {
Expand Down
4 changes: 4 additions & 0 deletions ...x/kernel-corpus/kernels/qwen_moe/qwen3_moe/flash_attention_decode_split_f32_f16_wmma.loom
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -830,6 +830,7 @@ template.def<@qwen3_moe.attention.decode_split.reduce_fused> device priority(20)
template.apply<@qwen3_moe.attention.decode_split.reduce_completed.direct>(%partial_block_capacity, %producer_block_count, %partial_max_noalias, %partial_sum_noalias, %partial_output_noalias, %output_noalias) : (index, index, buffer, buffer, buffer, buffer)
scf.if %publish_q8 {
kernel.barrier<global> scope(workgroup) ordering(acq_rel)
kernel.barrier<workgroup> scope(workgroup) ordering(acq_rel)
template.apply<@qwen3_moe.attention.decode_split.pack_completed_q8>(%key_value_head, %output_noalias, %q8_output) : (index, buffer, buffer)
}
// Do not expose the reset until every final F32 and Q8 store completes.
Expand Down Expand Up @@ -878,6 +879,7 @@ template.def<@qwen3_moe.attention.decode_split.reduce_fused> device priority(10)
template.apply<@qwen3_moe.attention.decode_split.reduce_completed.cooperative>(%partial_block_capacity, %producer_block_count, %partial_max_noalias, %partial_sum_noalias, %partial_output_noalias, %output_noalias) : (index, index, buffer, buffer, buffer, buffer)
scf.if %publish_q8 {
kernel.barrier<global> scope(workgroup) ordering(acq_rel)
kernel.barrier<workgroup> scope(workgroup) ordering(acq_rel)
template.apply<@qwen3_moe.attention.decode_split.pack_completed_q8>(%key_value_head, %output_noalias, %q8_output) : (index, buffer, buffer)
}
// Do not expose the reset until every final F32 and Q8 store completes.
Expand Down Expand Up @@ -1045,6 +1047,8 @@ kernel.def target(@qwen3_moe_decode_split_gfx11_wave64) @qwen3_moe_flash_attenti
scf.if %workitem_is_zero {
view.store %subgroup_sum, %reduction_stage_view[%c1] : f32, view<2xf32>
}
// Wave 0 rewrote partial_max in global memory; every wave reads it below.
kernel.barrier<global> scope(workgroup) ordering(acq_rel)
kernel.barrier<workgroup> scope(workgroup) ordering(acq_rel)
%sum = view.load %reduction_stage_view[%c1] : view<2xf32> -> f32
%unnormalized_output = scf.for %block = [%c0 to %key_value_block_count step %c1](%running_output = %c0_f32 : f32) -> (f32) unroll(%c4) schedule(interleaved) {
Expand Down