Expand description
gpu dialect.
Structs
- An
all_reduceoperation. Reduce values among workgroup.. - A builder for an
all_reduceoperation. - An
allocoperation. GPU memory allocation operation.. - A builder for an
allocoperation. - A
barrieroperation. Synchronizes all work items of a workgroup.. - A builder for a
barrieroperation. - A
block_dimoperation. - A builder for a
block_dimoperation. - A
block_idoperation. - A builder for a
block_idoperation. - A
create_2to4_spmatoperation. Create sparse matrix with 2:4 sparsity operation. - A builder for a
create_2to4_spmatoperation. - A
create_coo_aosoperation. Create sparse matrix in COO format operation (AoS). - A builder for a
create_coo_aosoperation. - A
create_coooperation. Create sparse matrix in COO format operation. - A builder for a
create_coooperation. - A
create_csroperation. Create sparse matrix in CSR format operation. - A builder for a
create_csroperation. - A
create_dn_tensoroperation. Create dense tensor operation. - A builder for a
create_dn_tensoroperation. - A
deallocoperation. GPU memory deallocation operation. - A builder for a
deallocoperation. - A
destroy_dn_tensoroperation. Destroy dense tensor operation. - A builder for a
destroy_dn_tensoroperation. - A
destroy_sp_matoperation. Destroy sparse matrix operation. - A builder for a
destroy_sp_matoperation. - A
funcoperation. Function executable on a GPU. - A builder for a
funcoperation. - A
moduleoperation. A top level compilation unit containing code to be run on a GPU.. - A builder for a
moduleoperation. - A
global_idoperation. - A builder for a
global_idoperation. - A
grid_dimoperation. - A builder for a
grid_dimoperation. - A
host_registeroperation. Registers a memref for access from device.. - A builder for a
host_registeroperation. - A
host_unregisteroperation. Unregisters a memref for access from device.. - A builder for a
host_unregisteroperation. - A
lane_idoperation. - A builder for a
lane_idoperation. - A
launch_funcoperation. Launches a function as a GPU kernel. - A builder for a
launch_funcoperation. - A
launchoperation. GPU kernel launch operation. - A builder for a
launchoperation. - A
memcpyoperation. GPU memcpy operation. - A builder for a
memcpyoperation. - A
memsetoperation. GPU memset operation. - A builder for a
memsetoperation. - A
module_endoperation. A pseudo op that marks the end of a gpu.module.. - A builder for a
module_endoperation. - A
num_subgroupsoperation. - A builder for a
num_subgroupsoperation. - A
printfoperation. Device-side printf, as in CUDA or OpenCL, for debugging. - A builder for a
printfoperation. - A
returnoperation. Terminator for GPU functions.. - A builder for a
returnoperation. - A
sddmm_buffer_sizeoperation. Precompute buffersize for SDDMM operation. - A builder for a
sddmm_buffer_sizeoperation. - A
sddmmoperation. SDDMM operation. - A builder for a
sddmmoperation. - A
set_default_deviceoperation. Set default GPU for operations after this by index. - A builder for a
set_default_deviceoperation. - A
shuffleoperation. Shuffles values within a subgroup.. - A builder for a
shuffleoperation. - A
spmm_buffer_sizeoperation. Precompute buffersize for SpMM operation. - A builder for a
spmm_buffer_sizeoperation. - A
spmmoperation. SpMM operation. - A builder for a
spmmoperation. - A
spmv_buffer_sizeoperation. Precompute buffersize for SpMV operation. - A builder for a
spmv_buffer_sizeoperation. - A
spmvoperation. SpMV operation. - A builder for a
spmvoperation. - A
subgroup_idoperation. - A builder for a
subgroup_idoperation. - A
subgroup_mma_computeoperation. GPU warp synchronous matrix multiply accumulate. - A builder for a
subgroup_mma_computeoperation. - A
subgroup_mma_constant_matrixoperation. GPU warp synchronous constant matrix. - A builder for a
subgroup_mma_constant_matrixoperation. - A
subgroup_mma_elementwiseoperation. GPU warp elementwise operation on a matrix. - A builder for a
subgroup_mma_elementwiseoperation. - A
subgroup_mma_load_matrixoperation. GPU warp synchronous matrix load. - A builder for a
subgroup_mma_load_matrixoperation. - A
subgroup_mma_store_matrixoperation. GPU warp synchronous matrix store. - A builder for a
subgroup_mma_store_matrixoperation. - A
subgroup_reduceoperation. Reduce values among subgroup.. - A builder for a
subgroup_reduceoperation. - A
subgroup_sizeoperation. - A builder for a
subgroup_sizeoperation. - A
terminatoroperation. Terminator for GPU launch regions.. - A builder for a
terminatoroperation. - A
thread_idoperation. - A builder for a
thread_idoperation. - A
waitoperation. Wait for async gpu ops to complete.. - A builder for a
waitoperation. - A
yieldoperation. GPU yield operation. - A builder for a
yieldoperation.
Functions
- Creates an
all_reduceoperation. - Creates an
allocoperation. - Creates a
barrieroperation. - Creates a
block_dimoperation. - Creates a
block_idoperation. - Creates a
create_2to4_spmatoperation. - Creates a
create_coooperation. - Creates a
create_coo_aosoperation. - Creates a
create_csroperation. - Creates a
create_dn_tensoroperation. - Creates a
deallocoperation. - Creates a
destroy_dn_tensoroperation. - Creates a
destroy_sp_matoperation. - Creates a
funcoperation. - Creates a
global_idoperation. - Creates a
grid_dimoperation. - Creates a
host_registeroperation. - Creates a
host_unregisteroperation. - Creates a
lane_idoperation. - Creates a
launchoperation. - Creates a
launch_funcoperation. - Creates a
memcpyoperation. - Creates a
memsetoperation. - Creates a
moduleoperation. - Creates a
module_endoperation. - Creates a
num_subgroupsoperation. - Creates a
printfoperation. - Creates a
returnoperation. - Creates a
sddmmoperation. - Creates a
sddmm_buffer_sizeoperation. - Creates a
set_default_deviceoperation. - Creates a
shuffleoperation. - Creates a
spmmoperation. - Creates a
spmm_buffer_sizeoperation. - Creates a
spmvoperation. - Creates a
spmv_buffer_sizeoperation. - Creates a
subgroup_idoperation. - Creates a
subgroup_mma_computeoperation. - Creates a
subgroup_mma_constant_matrixoperation. - Creates a
subgroup_mma_elementwiseoperation. - Creates a
subgroup_mma_load_matrixoperation. - Creates a
subgroup_mma_store_matrixoperation. - Creates a
subgroup_reduceoperation. - Creates a
subgroup_sizeoperation. - Creates a
terminatoroperation. - Creates a
thread_idoperation. - Creates a
waitoperation. - Creates a
yieldoperation.