Skip to content

[Feature] Allow using multiple nodes to caculate on DSP hardware #6708

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

Merged
merged 1 commit into from
Nov 17, 2025
Merged

[Feature] Allow using multiple nodes to caculate on DSP hardware #6708

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
2 changes: 1 addition & 1 deletion source/source_base/module_device/memory_op.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -452,7 +452,7 @@ struct resize_memory_op_mt<FPTYPE, base_device::DEVICE_CPU>
{
mtfunc::free_ht(arr);
}
arr = (FPTYPE*)mtfunc::malloc_ht(sizeof(FPTYPE) * size, GlobalV::MY_RANK);
arr = (FPTYPE*)mtfunc::malloc_ht(sizeof(FPTYPE) * size, GlobalV::MY_RANK % 4);
std::string record_string;
if (record_in != nullptr)
{
Expand Down
16 changes: 8 additions & 8 deletions source/source_base/module_external/blas_connector_matrix.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@ void BlasConnector::gemm(const char transa, const char transb, const int m, cons
else if (device_type == base_device::AbacusDevice_t::DspDevice){
mtfunc::sgemm_mth_(&transb, &transa, &n, &m, &k,
&alpha, b, &ldb, a, &lda,
&beta, c, &ldc, GlobalV::MY_RANK);
&beta, c, &ldc, GlobalV::MY_RANK % 4);
}
#endif
#ifdef __CUDA
Expand Down Expand Up @@ -67,7 +67,7 @@ void BlasConnector::gemm(const char transa,
#ifdef __DSP
else if (device_type == base_device::AbacusDevice_t::DspDevice)
{
mtfunc::dgemm_mth_(&transb, &transa, &n, &m, &k, &alpha, b, &ldb, a, &lda, &beta, c, &ldc, GlobalV::MY_RANK);
mtfunc::dgemm_mth_(&transb, &transa, &n, &m, &k, &alpha, b, &ldb, a, &lda, &beta, c, &ldc, GlobalV::MY_RANK % 4);
}
#endif
else if (device_type == base_device::AbacusDevice_t::GpuDevice)
Expand Down Expand Up @@ -106,7 +106,7 @@ void BlasConnector::gemm(const char transa,
#ifdef __DSP
else if (device_type == base_device::AbacusDevice_t::DspDevice)
{
mtfunc::cgemm_mth_(&transb, &transa, &n, &m, &k, &alpha, b, &ldb, a, &lda, &beta, c, &ldc, GlobalV::MY_RANK);
mtfunc::cgemm_mth_(&transb, &transa, &n, &m, &k, &alpha, b, &ldb, a, &lda, &beta, c, &ldc, GlobalV::MY_RANK % 4);
}
#endif
else if (device_type == base_device::AbacusDevice_t::GpuDevice)
Expand Down Expand Up @@ -157,7 +157,7 @@ void BlasConnector::gemm(const char transa,
#ifdef __DSP
else if (device_type == base_device::AbacusDevice_t::DspDevice)
{
mtfunc::zgemm_mth_(&transb, &transa, &n, &m, &k, &alpha, b, &ldb, a, &lda, &beta, c, &ldc, GlobalV::MY_RANK);
mtfunc::zgemm_mth_(&transb, &transa, &n, &m, &k, &alpha, b, &ldb, a, &lda, &beta, c, &ldc, GlobalV::MY_RANK % 4);
}
#endif
else if (device_type == base_device::AbacusDevice_t::GpuDevice)
Expand Down Expand Up @@ -200,7 +200,7 @@ void BlasConnector::gemm_cm(const char transa, const char transb, const int m, c
else if (device_type == base_device::AbacusDevice_t::DspDevice){
mtfunc::sgemm_mth_(&transb, &transa, &m, &n, &k,
&alpha, a, &lda, b, &ldb,
&beta, c, &ldc, GlobalV::MY_RANK);
&beta, c, &ldc, GlobalV::MY_RANK % 4);
}
#endif
#ifdef __CUDA
Expand Down Expand Up @@ -237,7 +237,7 @@ void BlasConnector::gemm_cm(const char transa,
#ifdef __DSP
else if (device_type == base_device::AbacusDevice_t::DspDevice)
{
mtfunc::dgemm_mth_(&transa, &transb, &m, &n, &k, &alpha, a, &lda, b, &ldb, &beta, c, &ldc, GlobalV::MY_RANK);
mtfunc::dgemm_mth_(&transa, &transb, &m, &n, &k, &alpha, a, &lda, b, &ldb, &beta, c, &ldc, GlobalV::MY_RANK % 4);
}
#endif
#ifdef __CUDA
Expand Down Expand Up @@ -276,7 +276,7 @@ void BlasConnector::gemm_cm(const char transa,
#ifdef __DSP
else if (device_type == base_device::AbacusDevice_t::DspDevice)
{
mtfunc::cgemm_mth_(&transa, &transb, &m, &n, &k, &alpha, a, &lda, b, &ldb, &beta, c, &ldc, GlobalV::MY_RANK);
mtfunc::cgemm_mth_(&transa, &transb, &m, &n, &k, &alpha, a, &lda, b, &ldb, &beta, c, &ldc, GlobalV::MY_RANK % 4);
}
#endif
#ifdef __CUDA
Expand Down Expand Up @@ -327,7 +327,7 @@ void BlasConnector::gemm_cm(const char transa,
#ifdef __DSP
else if (device_type == base_device::AbacusDevice_t::DspDevice)
{
mtfunc::zgemm_mth_(&transa, &transb, &m, &n, &k, &alpha, a, &lda, b, &ldb, &beta, c, &ldc, GlobalV::MY_RANK);
mtfunc::zgemm_mth_(&transa, &transb, &m, &n, &k, &alpha, a, &lda, b, &ldb, &beta, c, &ldc, GlobalV::MY_RANK % 4);
}
#endif
#ifdef __CUDA
Expand Down
2 changes: 1 addition & 1 deletion source/source_base/module_fft/fft_dsp.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@ void FFT_DSP<double>::initfft(int nx_in, int ny_in, int nz_in)
this->nx = nx_in;
this->ny = ny_in;
this->nz = nz_in;
cluster_id = GlobalV::MY_RANK;
cluster_id = GlobalV::MY_RANK % 4;
nxyz = this->nx * this->ny * this->nz;
}
template <>
Expand Down
2 changes: 1 addition & 1 deletion source/source_main/driver_run.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -129,7 +129,7 @@ void Driver::init_hardware()

#ifdef __DSP
std::cout << " ** Initializing DSP Hardware..." << std::endl;
mtfunc::dspInitHandle(GlobalV::MY_RANK);
mtfunc::dspInitHandle(GlobalV::MY_RANK % 4);
#endif
}

Expand Down
Loading