[CK_TILE] Refine ck_tile test script

1. Add  test script for img2col, batched_gemm, grouped_gemm and flatmm
2. Fix typos in ck tile script to match test in interface change.
3. Update gemm test max dim size to 8K and use GPU validation by default
4. Add run_ck_tile_full_performance_tests.sh to run all ck_tile examples

Author: linqun

example/ck_tile/03_gemm/script/benchmark_basic_bf16.sh

@@ -1,12 +1,12 @@
 #!/bin/sh
 EXE="$(find . -name tile_example_gemm_basic -type f | head -n 1)"
-VALID=1
+VALID=2
 
 
 for b_matrix_layout in "C"; do
-    for m in "64" "512" "1024" "2048"; do
-        for n in "512" "1024" "2048"; do
-            for k in "64" "512" "1024" "2048"; do
+    for m in "64" "512" "1024" "8192"; do
+        for n in "512" "1024" "8192"; do
+            for k in "64" "512" "1024" "8192"; do
                 $EXE -prec=bf16 -m=$m -n=$n -k=$k -a_layout="R" -b_layout="$b_matrix_layout" -c_layout="R" -v=$VALID
             done
         done

example/ck_tile/03_gemm/script/benchmark_basic_bf8.sh

@@ -1,12 +1,12 @@
 #!/bin/sh
 EXE="$(find . -name tile_example_gemm_basic -type f | head -n 1)"
-VALID=1
+VALID=2
 
 
 for b_matrix_layout in "C"; do
-    for m in "64" "512" "1024" "2048"; do
-        for n in "512" "1024" "2048"; do
-            for k in "64" "512" "1024" "2048"; do
+    for m in "64" "512" "1024" "8192"; do
+        for n in "512" "1024" "8192"; do
+            for k in "64" "512" "1024" "8192"; do
                 $EXE -prec=bf8 -m=$m -n=$n -k=$k -a_layout="R" -b_layout="$b_matrix_layout" -c_layout="R" -v=$VALID
             done
         done

example/ck_tile/03_gemm/script/benchmark_basic_fp16.sh

@@ -1,11 +1,11 @@
 #!/bin/sh
 EXE="$(find . -name tile_example_gemm_basic -type f | head -n 1)"
-VALID=1
+VALID=2
 
 for b_matrix_layout in "C"; do
-    for m in "64" "512" "1024" "2048"; do
-        for n in "512" "1024" "2048"; do
-            for k in "64" "512" "1024" "2048"; do
+    for m in "64" "512" "1024" "8192"; do
+        for n in "512" "1024" "8192"; do
+            for k in "64" "512" "1024" "8192"; do
                 $EXE -prec=fp16 -m=$m -n=$n -k=$k -a_layout="R" -b_layout="$b_matrix_layout" -c_layout="R" -v=$VALID
             done
         done

example/ck_tile/03_gemm/script/benchmark_basic_fp8.sh

@@ -1,12 +1,12 @@
 #!/bin/sh
 EXE="$(find . -name tile_example_gemm_basic -type f | head -n 1)"
-VALID=1
+VALID=2
 
 
 for b_matrix_layout in "C"; do
-    for m in "64" "512" "1024" "2048"; do
-        for n in "512" "1024" "2048"; do
-            for k in "64" "512" "1024" "2048"; do
+    for m in "64" "512" "1024" "8192"; do
+        for n in "512" "1024" "8192"; do
+            for k in "64" "512" "1024" "8192"; do
                 $EXE -prec=fp8 -m=$m -n=$n -k=$k -a_layout="R" -b_layout="$b_matrix_layout" -c_layout="R" -v=$VALID
             done
         done

example/ck_tile/03_gemm/script/benchmark_mem_pipeline_bf16.sh

@@ -1,11 +1,11 @@
 #!/bin/sh
 EXE="$(find . -name tile_example_gemm_universal -type f | head -n 1)"
-VALID=1
+VALID=2
 
 for b_matrix_layout in "C"; do
-    for m in "512" "1024" "2048" "4096"; do
-        for n in "512" "1024" "2048"; do
-            for k in "512" "1024" "2048"; do
+    for m in "512" "1024" "2048" "8192"; do
+        for n in "512" "1024" "8192"; do
+            for k in "512" "1024" "8192"; do
                 $EXE -prec=bf16 -m=$m -n=$n -k=$k -a_layout="R" -b_layout="$b_matrix_layout" -c_layout="R" -v=$VALID
             done
         done

example/ck_tile/03_gemm/script/benchmark_mem_pipeline_bf8.sh

@@ -1,11 +1,11 @@
 #!/bin/sh
 EXE="$(find . -name tile_example_gemm_universal -type f | head -n 1)"
-VALID=1
+VALID=2
 
 for b_matrix_layout in "C"; do
-    for m in "512" "1024" "2048" "4096"; do
-        for n in "512" "1024" "2048"; do
-            for k in "512" "1024" "2048"; do
+    for m in "512" "1024" "2048" "8192"; do
+        for n in "512" "1024" "8192"; do
+            for k in "512" "1024" "8192"; do
                 $EXE -prec=bf8 -m=$m -n=$n -k=$k -a_layout="R" -b_layout="$b_matrix_layout" -c_layout="R" -v=$VALID
             done
         done

example/ck_tile/03_gemm/script/benchmark_mem_pipeline_fp16.sh

@@ -1,11 +1,11 @@
 #!/bin/sh
 EXE="$(find . -name tile_example_gemm_universal -type f | head -n 1)"
-VALID=1
+VALID=2
 
 for b_matrix_layout in "C"; do
-    for m in "512" "1024" "2048" "4096"; do
-        for n in "512" "1024" "2048"; do
-            for k in "512" "1024" "2048"; do
+    for m in "512" "1024" "2048" "8192"; do
+        for n in "512" "1024" "8192"; do
+            for k in "512" "1024" "8192"; do
                 $EXE -prec=fp16 -m=$m -n=$n -k=$k -a_layout="R" -b_layout="$b_matrix_layout" -c_layout="R" -v=$VALID
             done
         done

example/ck_tile/03_gemm/script/benchmark_mem_pipeline_fp8.sh

@@ -1,11 +1,11 @@
 #!/bin/sh
 EXE="$(find . -name tile_example_gemm_universal -type f | head -n 1)"
-VALID=1
+VALID=2
 
 for b_matrix_layout in "C"; do
-    for m in "512" "1024" "2048" "4096"; do
-        for n in "512" "1024" "2048"; do
-            for k in "512" "1024" "2048"; do
+    for m in "512" "1024" "2048" "8192"; do
+        for n in "512" "1024" "8192"; do
+            for k in "512" "1024" "8192"; do
                 $EXE -prec=fp8 -m=$m -n=$n -k=$k -a_layout="R" -b_layout="$b_matrix_layout" -c_layout="R" -v=$VALID
             done
         done

example/ck_tile/04_img2col/script/smoke_test.sh

@@ -0,0 +1,14 @@
+#!/bin/sh
+# TODO: run this script from CK root
+BUILD=build
+EXE=$BUILD/bin/tile_example_img2col 
+
+
+N=256
+######    Dim G N__ K___ C___ Y X Hi__ Wi__ Strides Dilations LeftPads RightPads
+$EXE 1 2 1 2  1 $N  256  64   1 1   64   64    1 1       1 1      0 0       0 0
+$EXE 1 2 1 2  1 $N  512  64   1 1   64   64    1 1       1 1      0 0       0 0
+$EXE 1 2 1 2  1 $N  128  64   3 3   64   64    1 1       1 1      1 1       1 1
+
+
+