Ck tile/gemm perf measure (#1750)

* Finished adding the performance benchmark for ck tile gemm

* Fix the executable rename problem

* fix the executable name error

* delete the unsupported layout combinations

* Update run_full_test.sh

* Update benchmark_mem_pipeline.sh

* Update benchmark_basic.sh

* change the executable of gemm_universal

* change ck_tile_gemm script permissions

* Addressed the comment

* Addressed the comment

* Fixed the comments

* Fixed Comment

* roll back the malfunctioned change

* Fix the Typo

* finalize the tile_gemm_fp16 performance monitoring

* fix the stash names for ck_tile gemm logs

* change the stashing logic

* change stashing syntax

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
Co-authored-by: illsilin <Illia.Silin@amd.com>

[ROCm/composable_kernel commit: 73a076eee1]

script/process_perf_data.py

@@ -149,6 +149,12 @@ def parse_logfile(logfile):
                 lst=line.split()
                 line_dict=dict(zip(lst[1:],lst))
                 res.append(line_dict['TFlops,'])
+    elif 'perf_tile_gemm_basic' in logfile or 'perf_tile_gemm_mem_pipeline' in logfile:
+        for line in open(logfile):
+            if 'TFlops' in line:
+                lst=line.split()
+                line_dict=dict(zip(lst[1:],lst))
+                res.append(line_dict['TFlops,'])
     return res
 
 
@@ -330,6 +336,14 @@ def main():
             for i in range(1,len(results)+1):
                 testlist.append("Test%i"%i)
             table_name="ck_fmha_bwd_tflops"
+        if 'gemm_basic_fp16' in filename:
+            for i in range(1, len(results)+1):
+                testlist.append("Test%i"%i)
+            table_name="ck_tile_gemm_basic_fp16_tflops"
+        if 'gemm_mem_pipeline_fp16' in filename:
+            for i in range(1, len(results)+1):
+                testlist.append("Test%i"%i)
+            table_name="ck_tile_gemm_mem_pipeline_fp16_tflops"
 
         tflops_base = get_baseline(table_name,conn)
         store_new_test_result(table_name, results, testlist, branch_name, node_id, gpu_arch, compute_units, rocm_vers, hip_vers, environment, sqlEngine)

script/process_perf_data.sh

@@ -43,3 +43,19 @@ file=./perf_fmha_bwd_gfx90a.log
 if [ -e "$file" ]; then
     python3 process_perf_data.py perf_fmha_bwd_gfx90a.log
 fi
+file=./perf_tile_gemm_basic_fp16_gfx942.log
+if [ -e "$file" ]; then
+    python3 process_perf_data.py perf_tile_gemm_basic_fp16_gfx942.log
+fi
+file=./perf_tile_gemm_basic_fp16_gfx90a.log
+if [ -e "$file" ]; then
+    python3 process_perf_data.py perf_tile_gemm_basic_fp16_gfx90a.log
+fi
+file=./perf_tile_gemm_mem_pipeline_fp16_gfx942.log
+if [ -e "$file" ]; then
+    python3 process_perf_data.py perf_tile_gemm_mem_pipeline_fp16_gfx942.log
+fi
+file=./perf_tile_gemm_mem_pipeline_fp16_gfx90a.log
+if [ -e "$file" ]; then
+    python3 process_perf_data.py perf_tile_gemm_mem_pipeline_fp16_gfx90a.log
+fi

script/process_qa_data.sh

@@ -52,3 +52,19 @@ file=./perf_fmha_bwd_gfx90a.log
 if [ -e "$file" ]; then
     python3 process_perf_data.py perf_fmha_bwd_gfx90a.log
 fi
+file=./perf_gemm_basic_gfx942.log
+if [ -e "$file" ]; then
+    python3 process_perf_data.py perf_gemm_basic_gfx942.log
+fi
+file=./perf_gemm_basic_gfx90a.log
+if [ -e "$file" ]; then
+    python3 process_perf_data.py perf_gemm_basic_gfx90a.log
+fi
+file=./perf_gemm_mem_pipeline_gfx942.log
+if [ -e "$file" ]; then
+    python3 process_perf_data.py perf_gemm_mem_pipeline_gfx942.log
+fi
+file=./perf_gemm_mem_pipeline_gfx90a.log
+if [ -e "$file" ]; then
+    python3 process_perf_data.py perf_gemm_mem_pipeline_gfx90a.log
+fi