Skip to content

float double and all complex using Tensile on 950 #1657

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: develop
Choose a base branch
from
+17 −0
Open

float double and all complex using Tensile on 950 #1657

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions library/src/include/utility.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -789,6 +789,9 @@ constexpr double rocblas_internal_value_category(const T& beta)
return beta == T(0) ? 0.0 : beta == T(1) ? 1.0 : beta == T(-1) ? -1.0 : 2.0;
}

// Internal use
int rocblas_internal_get_arch(rocblas_handle handle);

// Internal use, whether Tensile supports ldc != ldd
// We assume true if the value is greater than or equal to 906
bool rocblas_internal_tensile_supports_ldc_ne_ldd(rocblas_handle handle);
Expand Down
5 changes: 5 additions & 0 deletions library/src/rocblas_auxiliary.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -897,6 +897,11 @@ struct XnackMode<PROP, void_t<decltype(PROP::gcnArchName)>>
}
};

int rocblas_internal_get_arch(rocblas_handle handle)
{
return handle->getArch();
}

bool rocblas_internal_tensile_supports_ldc_ne_ldd(rocblas_handle handle)
{
return handle->getArch() >= 906;
Expand Down
9 changes: 9 additions & 0 deletions library/src/tensile_host.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1081,6 +1081,15 @@ template <typename Ti, typename To, typename Tc>
bool useHipBLASLt(const RocblasContractionProblem<Ti, To, Tc>& prob)
{
#ifdef BUILD_WITH_HIPBLASLT
if constexpr(sizeof(Ti) >= 4)
{
// TODO remove after tuning
if(rocblas_internal_get_arch(prob.handle) == 950)
{
return false;
}
}

bool batched = prob.batch_A != nullptr;
return prob.handle->tryHipBLASLt(batched);
#else
Expand Down
Loading