ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-01 04:07:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

new layout sanity checked. not correct instruction generated

Browse Source
This commit is contained in:
aska-0096
2025-04-18 02:07:32 +00:00
parent e95db65de3
commit 1edd7a331e
9 changed files with 770 additions and 159 deletions
Download Patch File Download Diff File Expand all files Collapse all files

55
example/ck_tile/01_fmha/example_bwd_fmha_bf16.cpp
View File

@@ -600,6 +600,54 @@ bool run(const ck_tile::ArgParser& arg_parser)
ck_tile::FillTrigValue<BiasDataType>{}(bias_host);
ck_tile::FillTrigValue<OGradDataType>{}(do_host);
}
else if(init_method == 3)
{
ck_tile::FillUniformDistribution<QDataType>{1.f, 1.f, seed}(q_host);
ck_tile::FillUniformDistribution<KDataType>{1.f, 1.f, seed}(k_host);
ck_tile::FillUniformDistribution<VDataType>{1.f, 1.f, seed}(v_host);
ck_tile::FillUniformDistribution<BiasDataType>{1.f, 1.f, seed}(bias_host);
ck_tile::FillUniformDistribution<OGradDataType>{1.f, 1.f, seed}(do_host);
}
else if(init_method == 4)
{
ck_tile::FillUniformDistribution<QDataType>{0.f, 1.f, seed}(q_host);
ck_tile::FillUniformDistribution<KDataType>{1.f, 1.f, seed}(k_host);
ck_tile::FillUniformDistribution<VDataType>{1.f, 1.f, seed}(v_host);
ck_tile::FillUniformDistribution<BiasDataType>{0.f, 1.f, seed}(bias_host);
ck_tile::FillUniformDistribution<OGradDataType>{0.f, 1.f, seed}(do_host);
}
else if(init_method == 5)
{
ck_tile::FillUniformDistribution<QDataType>{0.f, 1.f, seed}(q_host);
ck_tile::FillUniformDistribution<KDataType>{1.f, 1.f, seed}(k_host);
ck_tile::FillUniformDistribution<VDataType>{1.f, 1.f, seed}(v_host);
ck_tile::FillUniformDistribution<BiasDataType>{1.f, 1.f, seed}(bias_host);
ck_tile::FillUniformDistribution<OGradDataType>{1.f, 1.f, seed}(do_host);
}
else if(init_method == 6)
{
ck_tile::FillUniformDistribution<QDataType>{1.f, 1.f, seed}(q_host);
ck_tile::FillUniformDistribution<KDataType>{0.f, 1.f, seed}(k_host);
ck_tile::FillUniformDistribution<VDataType>{0.f, 1.f, seed}(v_host);
ck_tile::FillUniformDistribution<BiasDataType>{1.f, 1.f, seed}(bias_host);
ck_tile::FillUniformDistribution<OGradDataType>{1.f, 1.f, seed}(do_host);
}
else if(init_method == 7)
{
ck_tile::FillUniformDistribution<QDataType>{1.f, 1.f, seed}(q_host);
ck_tile::FillUniformDistribution<KDataType>{1.f, 1.f, seed}(k_host);
ck_tile::FillUniformDistribution<VDataType>{0.f, 1.f, seed}(v_host);
ck_tile::FillUniformDistribution<BiasDataType>{1.f, 1.f, seed}(bias_host);
ck_tile::FillUniformDistribution<OGradDataType>{1.f, 1.f, seed}(do_host);
}
else if(init_method == 8)
{
ck_tile::FillUniformDistribution<QDataType>{1.f, 1.f, seed}(q_host);
ck_tile::FillUniformDistribution<KDataType>{0.f, 1.f, seed}(k_host);
ck_tile::FillUniformDistribution<VDataType>{1.f, 1.f, seed}(v_host);
ck_tile::FillUniformDistribution<BiasDataType>{1.f, 1.f, seed}(bias_host);
ck_tile::FillUniformDistribution<OGradDataType>{1.f, 1.f, seed}(do_host);
}
if(bias.type == bias_enum::alibi)
{
auto slopes = ck_tile::get_alibi_slopes<AccDataType>(nhead);
@@ -619,9 +667,12 @@ bool run(const ck_tile::ArgParser& arg_parser)
}
}
// for(int iM=0; iM<128; iM++){
// for(int iK=0; iK<16; iK++){
// printf("%04x ", *(reinterpret_cast<uint16_t*>(&(q_host(0, 0, iK, iM)))));
// for(int iM=0; iM<128; iM++){
// printf("%04x ", *(reinterpret_cast<uint16_t*>(&(q_host(0, 0, iK, iM)))));
// if(iM%16==15){
// printf("|");
// }
// }
// printf("\n");
// }