zhangyiss/mlx
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx.git synced 2025-12-16 01:49:05 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Option to JIT steel gemm / conv (#1139)

Browse Source
This commit is contained in:
Awni Hannun
2024-05-23 18:07:34 -07:00
committed by GitHub
parent eab2685c67
commit 7e26fd8032
31 changed files with 2504 additions and 1540 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
mlx/backend/metal/conv.cpp
View File

@@ -7,6 +7,7 @@
#include "mlx/backend/metal/copy.h"
#include "mlx/backend/metal/device.h"
#include "mlx/backend/metal/kernels.h"
#include "mlx/backend/metal/kernels/defines.h"
#include "mlx/backend/metal/kernels/steel/conv/params.h"
#include "mlx/backend/metal/matmul.h"
@@ -335,7 +336,17 @@ void implicit_gemm_conv_2D_gpu(
// Encode and dispatch kernel
auto& compute_encoder = d.get_command_encoder(s.index);
auto kernel = d.get_kernel(kname.str());
auto kernel = get_steel_conv_kernel(
d,
kname.str(),
out,
bm,
bn,
bk,
wm,
wn,
n_channel_specialization,
small_filter);
compute_encoder->setComputePipelineState(kernel);
// Deduce grid launch dimensions
@@ -488,7 +499,8 @@ void implicit_gemm_conv_2D_general_gpu(
// Encode and dispatch kernel
auto& compute_encoder = d.get_command_encoder(s.index);
auto kernel = d.get_kernel(kname.str());
auto kernel =
get_steel_conv_general_kernel(d, kname.str(), out, bm, bn, bk, wm, wn);
compute_encoder->setComputePipelineState(kernel);
// Deduce grid launch dimensions