matt/ollama37
1
0
Fork 0
mirror of https://github.com/dogkeeper886/ollama37.git synced 2025-12-10 07:46:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

Remove remaining MMA/WMMA template instances for CC 3.7 optimization

Browse Source 
Delete 24 tensor core template instance files that were missed in the initial optimization:
- 19 fattn-mma-f16 template instances (various ncols1/ncols2 combinations)
- 5 fattn-wmma-f16 template instances (kqfloat and kqhalf variants)

These files implement tensor core operations (MMA/WMMA) which require Compute Capability 7.0+
and are not available on Tesla K80 (CC 3.7). Removing them completes the CC 3.7-only optimization.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
Shang Chieh Tseng
2025-10-29 15:24:08 +08:00
parent 771044bead
commit 66fca1b685
24 changed files with 0 additions and 222 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu vendored
View File

@@ -1,5 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(576, 512, 1, 16);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 1, 8);
DECL_FATTN_MMA_F16_CASE(80, 80, 1, 8);
DECL_FATTN_MMA_F16_CASE(96, 96, 1, 8);
DECL_FATTN_MMA_F16_CASE(112, 112, 1, 8);
DECL_FATTN_MMA_F16_CASE(128, 128, 1, 8);
DECL_FATTN_MMA_F16_CASE(256, 256, 1, 8);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 16, 1);
DECL_FATTN_MMA_F16_CASE(80, 80, 16, 1);
DECL_FATTN_MMA_F16_CASE(96, 96, 16, 1);
DECL_FATTN_MMA_F16_CASE(112, 112, 16, 1);
DECL_FATTN_MMA_F16_CASE(128, 128, 16, 1);
DECL_FATTN_MMA_F16_CASE(256, 256, 16, 1);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 16, 2);
DECL_FATTN_MMA_F16_CASE(80, 80, 16, 2);
DECL_FATTN_MMA_F16_CASE(96, 96, 16, 2);
DECL_FATTN_MMA_F16_CASE(112, 112, 16, 2);
DECL_FATTN_MMA_F16_CASE(128, 128, 16, 2);
DECL_FATTN_MMA_F16_CASE(256, 256, 16, 2);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 16, 4);
DECL_FATTN_MMA_F16_CASE(80, 80, 16, 4);
DECL_FATTN_MMA_F16_CASE(96, 96, 16, 4);
DECL_FATTN_MMA_F16_CASE(112, 112, 16, 4);
DECL_FATTN_MMA_F16_CASE(128, 128, 16, 4);
DECL_FATTN_MMA_F16_CASE(256, 256, 16, 4);

5
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu vendored
View File

@@ -1,5 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(576, 512, 2, 16);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 2, 4);
DECL_FATTN_MMA_F16_CASE(80, 80, 2, 4);
DECL_FATTN_MMA_F16_CASE(96, 96, 2, 4);
DECL_FATTN_MMA_F16_CASE(112, 112, 2, 4);
DECL_FATTN_MMA_F16_CASE(128, 128, 2, 4);
DECL_FATTN_MMA_F16_CASE(256, 256, 2, 4);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 2, 8);
DECL_FATTN_MMA_F16_CASE(80, 80, 2, 8);
DECL_FATTN_MMA_F16_CASE(96, 96, 2, 8);
DECL_FATTN_MMA_F16_CASE(112, 112, 2, 8);
DECL_FATTN_MMA_F16_CASE(128, 128, 2, 8);
DECL_FATTN_MMA_F16_CASE(256, 256, 2, 8);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 32, 1);
DECL_FATTN_MMA_F16_CASE(80, 80, 32, 1);
DECL_FATTN_MMA_F16_CASE(96, 96, 32, 1);
DECL_FATTN_MMA_F16_CASE(112, 112, 32, 1);
DECL_FATTN_MMA_F16_CASE(128, 128, 32, 1);
DECL_FATTN_MMA_F16_CASE(256, 256, 32, 1);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 32, 2);
DECL_FATTN_MMA_F16_CASE(80, 80, 32, 2);
DECL_FATTN_MMA_F16_CASE(96, 96, 32, 2);
DECL_FATTN_MMA_F16_CASE(112, 112, 32, 2);
DECL_FATTN_MMA_F16_CASE(128, 128, 32, 2);
DECL_FATTN_MMA_F16_CASE(256, 256, 32, 2);

5
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu vendored
View File

@@ -1,5 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(576, 512, 4, 16);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 4, 2);
DECL_FATTN_MMA_F16_CASE(80, 80, 4, 2);
DECL_FATTN_MMA_F16_CASE(96, 96, 4, 2);
DECL_FATTN_MMA_F16_CASE(112, 112, 4, 2);
DECL_FATTN_MMA_F16_CASE(128, 128, 4, 2);
DECL_FATTN_MMA_F16_CASE(256, 256, 4, 2);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 4, 4);
DECL_FATTN_MMA_F16_CASE(80, 80, 4, 4);
DECL_FATTN_MMA_F16_CASE(96, 96, 4, 4);
DECL_FATTN_MMA_F16_CASE(112, 112, 4, 4);
DECL_FATTN_MMA_F16_CASE(128, 128, 4, 4);
DECL_FATTN_MMA_F16_CASE(256, 256, 4, 4);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 4, 8);
DECL_FATTN_MMA_F16_CASE(80, 80, 4, 8);
DECL_FATTN_MMA_F16_CASE(96, 96, 4, 8);
DECL_FATTN_MMA_F16_CASE(112, 112, 4, 8);
DECL_FATTN_MMA_F16_CASE(128, 128, 4, 8);
DECL_FATTN_MMA_F16_CASE(256, 256, 4, 8);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 64, 1);
DECL_FATTN_MMA_F16_CASE(80, 80, 64, 1);
DECL_FATTN_MMA_F16_CASE(96, 96, 64, 1);
DECL_FATTN_MMA_F16_CASE(112, 112, 64, 1);
DECL_FATTN_MMA_F16_CASE(128, 128, 64, 1);
DECL_FATTN_MMA_F16_CASE(256, 256, 64, 1);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 8, 1);
DECL_FATTN_MMA_F16_CASE(80, 80, 8, 1);
DECL_FATTN_MMA_F16_CASE(96, 96, 8, 1);
DECL_FATTN_MMA_F16_CASE(112, 112, 8, 1);
DECL_FATTN_MMA_F16_CASE(128, 128, 8, 1);
DECL_FATTN_MMA_F16_CASE(256, 256, 8, 1);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 8, 2);
DECL_FATTN_MMA_F16_CASE(80, 80, 8, 2);
DECL_FATTN_MMA_F16_CASE(96, 96, 8, 2);
DECL_FATTN_MMA_F16_CASE(112, 112, 8, 2);
DECL_FATTN_MMA_F16_CASE(128, 128, 8, 2);
DECL_FATTN_MMA_F16_CASE(256, 256, 8, 2);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 8, 4);
DECL_FATTN_MMA_F16_CASE(80, 80, 8, 4);
DECL_FATTN_MMA_F16_CASE(96, 96, 8, 4);
DECL_FATTN_MMA_F16_CASE(112, 112, 8, 4);
DECL_FATTN_MMA_F16_CASE(128, 128, 8, 4);
DECL_FATTN_MMA_F16_CASE(256, 256, 8, 4);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-mma-f16.cuh"
DECL_FATTN_MMA_F16_CASE(64, 64, 8, 8);
DECL_FATTN_MMA_F16_CASE(80, 80, 8, 8);
DECL_FATTN_MMA_F16_CASE(96, 96, 8, 8);
DECL_FATTN_MMA_F16_CASE(112, 112, 8, 8);
DECL_FATTN_MMA_F16_CASE(128, 128, 8, 8);
DECL_FATTN_MMA_F16_CASE(256, 256, 8, 8);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-wmma-f16.cuh"
DECL_FATTN_WMMA_F16_CASE(64, 16, float);
DECL_FATTN_WMMA_F16_CASE(80, 16, float);
DECL_FATTN_WMMA_F16_CASE(96, 16, float);
DECL_FATTN_WMMA_F16_CASE(112, 16, float);
DECL_FATTN_WMMA_F16_CASE(128, 16, float);
DECL_FATTN_WMMA_F16_CASE(256, 16, float);

9
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu vendored
View File

@@ -1,9 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-wmma-f16.cuh"
DECL_FATTN_WMMA_F16_CASE(64, 32, float);
DECL_FATTN_WMMA_F16_CASE(80, 32, float);
DECL_FATTN_WMMA_F16_CASE(96, 32, float);
DECL_FATTN_WMMA_F16_CASE(112, 32, float);
DECL_FATTN_WMMA_F16_CASE(128, 32, float);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-wmma-f16.cuh"
DECL_FATTN_WMMA_F16_CASE(64, 16, half);
DECL_FATTN_WMMA_F16_CASE(80, 16, half);
DECL_FATTN_WMMA_F16_CASE(96, 16, half);
DECL_FATTN_WMMA_F16_CASE(112, 16, half);
DECL_FATTN_WMMA_F16_CASE(128, 16, half);
DECL_FATTN_WMMA_F16_CASE(256, 16, half);

10
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu vendored
View File

@@ -1,10 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-wmma-f16.cuh"
DECL_FATTN_WMMA_F16_CASE(64, 32, half);
DECL_FATTN_WMMA_F16_CASE(80, 32, half);
DECL_FATTN_WMMA_F16_CASE(96, 32, half);
DECL_FATTN_WMMA_F16_CASE(112, 32, half);
DECL_FATTN_WMMA_F16_CASE(128, 32, half);
DECL_FATTN_WMMA_F16_CASE(256, 32, half);

8
ml/backend/ggml/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu vendored
View File

@@ -1,8 +0,0 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../fattn-wmma-f16.cuh"
DECL_FATTN_WMMA_F16_CASE(64, 8, half);
DECL_FATTN_WMMA_F16_CASE(96, 8, half);
DECL_FATTN_WMMA_F16_CASE(128, 8, half);
DECL_FATTN_WMMA_F16_CASE(256, 8, half);