Skip to content
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

llm decode shapes fp8 rowwise gemm tuning #3565

Closed
wants to merge 1 commit into from
Closed

llm decode shapes fp8 rowwise gemm tuning #3565

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
18 changes: 18 additions & 0 deletions fbgemm_gpu/experimental/gen_ai/src/quantize/ck_extensions/fp8_rowwise/fp8_rowwise_gemm.hip
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,24 @@ struct IntTupleHash {
// For certain high priority shapes, we directly map to the best kernel rather
// than use heuristics.
static const std::unordered_map<std::tuple<int, int, int>, RowwiseKernel, IntTupleHash> rowwise_lookup_dispatch = {
// Support for decode for [1024, 5120]
{{16, 1024, 5120},
fp8_rowwise_128x16x32x128_16x16_1x1_8x16x1_8x16x1_1x16x1x8_4x4x1_1x1_interwave_v2},
{{32, 1024, 5120},
fp8_rowwise_128x16x32x128_16x16_1x1_8x16x1_8x16x1_1x16x1x8_4x4x1_1x1_interwave_v2},
{{64, 1024, 5120},
fp8_rowwise_128x16x32x128_16x16_1x1_8x16x1_8x16x1_1x16x1x8_4x4x1_1x1_interwave_v2},
{{128, 1024, 5120},
fp8_rowwise_128x32x16x128_16x16_1x1_8x16x1_8x16x1_1x16x1x8_2x2x1_1x1_interwave_v2},
// Support for decode for [5120, 1024]
{{16, 5120, 1024},
fp8_rowwise_128x32x16x128_16x16_1x1_8x16x1_8x16x1_1x16x1x8_2x2x1_1x1_interwave_v2},
{{32, 5120, 1024},
fp8_rowwise_64x16x16x128_16x16_1x1_8x8x1_8x8x1_1x16x1x4_4x4x1_1x1_interwave_v2},
{{64, 5120, 1024},
fp8_rowwise_128x32x16x128_16x16_1x1_8x16x1_8x16x1_1x16x1x8_2x2x1_1x1_interwave_v2},
{{128, 5120, 1024},
fp8_rowwise_128x32x64x128_32x32_1x1_8x16x1_8x16x1_1x16x1x8_8x8x1_1x1_intrawave_v2},
// LLama 70B Decode shapes.
// Support for decode across batch sizes for [1280, 8192]
{{16, 1280, 8192},
Expand Down
Loading