temporarily disable the large a matrix block load

This is not working (silently failing) with some recent drivers, so disable it for now. Ideally we will be able to reenable it shortly.
bashbaug · Jan 19, 2024 · 4caea7b · 4caea7b
1 parent 756d2e9
commit 4caea7b
Showing 1 changed file with 16 additions and 16 deletions.
diff --git a/samples/99_matrixexperiments/matrix_kernel_tiled.cl b/samples/99_matrixexperiments/matrix_kernel_tiled.cl
@@ -202,17 +202,17 @@ kernel void MM_KERNEL_NAME(bfloat16_dpas_blockread_rowmajor_tiled, 8, 16, MM, NN
 
  for (int k = 0; k < K; k += tK) {
  short8 aData[MM];
- if (MM % 2 == 0) {
- for (int mm = 0; mm < MM; mm += 2) {
- short16 aTemp = as_short16(intel_subgroup_block_read_u16_m16k16(A, K * sizeof(ushort), M, K * sizeof(ushort), (int2)(k, m + mm * tM)));
- aData[mm + 0] = aTemp.lo;
- aData[mm + 1] = aTemp.hi;
- }
- } else {
+ //if (MM % 2 == 0) {
+ // for (int mm = 0; mm < MM; mm += 2) {
+ // short16 aTemp = as_short16(intel_subgroup_block_read_u16_m16k16(A, K * sizeof(ushort), M, K * sizeof(ushort), (int2)(k, m + mm * tM)));
+ // aData[mm + 0] = aTemp.lo;
+ // aData[mm + 1] = aTemp.hi;
+ // }
+ //} else {
  for (int mm = 0; mm < MM; mm++) {
  aData[mm] = as_short8(intel_subgroup_block_read_u16_m8k16(A, K * sizeof(ushort), M, K * sizeof(ushort), (int2)(k, m + mm * tM)));
  }
- }
+ //}
 
  int8 bData[NN];
  for (int nn = 0; nn < NN; nn++) {
@@ -252,17 +252,17 @@ kernel void MM_KERNEL_NAME(bfloat16_dpas_blockread_vnni_tiled, 8, 16, MM, NN)(gl
 
  for (int k = 0; k < K; k += tK) {
  short8 aData[MM];
- if (MM % 2 == 0) {
- for (int mm = 0; mm < MM; mm += 2) {
- short16 aTemp = as_short16(intel_subgroup_block_read_u16_m16k16(A, K * sizeof(ushort), M, K * sizeof(ushort), (int2)(k, m + mm * tM)));
- aData[mm + 0] = aTemp.lo;
- aData[mm + 1] = aTemp.hi;
- }
- } else {
+ //if (MM % 2 == 0) {
+ // for (int mm = 0; mm < MM; mm += 2) {
+ // short16 aTemp = as_short16(intel_subgroup_block_read_u16_m16k16(A, K * sizeof(ushort), M, K * sizeof(ushort), (int2)(k, m + mm * tM)));
+ // aData[mm + 0] = aTemp.lo;
+ // aData[mm + 1] = aTemp.hi;
+ // }
+ //} else {
  for (int mm = 0; mm < MM; mm++) {
  aData[mm] = as_short8(intel_subgroup_block_read_u16_m8k16(A, K * sizeof(ushort), M, K * sizeof(ushort), (int2)(k, m + mm * tM)));
  }
- }
+ //}
 
  int8 bData[NN];
  for (int nn = 0; nn < NN; nn++) {