mindspore-Ecosystem
/
mindspore
21
0
Fork 27
Code Issues Pull Requests 6 Projects Releases Wiki Activity

!39282 WeNet: Replace DynamicAtomicCleanAddr with Memset 

Merge pull request !39282 from zhengzuohe/memset_master
This commit is contained in:
i-robot 2022-08-01 04:34:09 +00:00 committed by Gitee
parent 93c01050c8 586a14f49e
commit a5fe1f0b3b
No known key found for this signature in database
GPG Key ID: 173E9B9CA92EEF8F
2 changed files with 33 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
mindspore/ccsrc/plugin/device/ascend/kernel/tbe/dynamic_tbe_kernel_mod.cc
View File

@ -38,6 +38,7 @@
#include "include/common/utils/utils.h"
#include "register/op_tiling.h"
#include "nlohmann/json.hpp"
#include "runtime/device/memory_manager.h"
namespace mindspore {
namespace kernel {
@ -238,9 +239,14 @@ bool DynamicTbeKernelMod::Launch(const std::vector<AddressPtr> &inputs, const st
MS_EXCEPTION_IF_NULL(kernel_mod);
device::KernelRuntime::GenLaunchArgs(*kernel_mod, atomic_clean_node.lock(), &kernel_launch_info);
auto atomic_inputs = kernel_launch_info.inputs_;
std::vector<AddressPtr> atomic_outputs;
std::vector<AddressPtr> atomic_workspace;
kernel_mod->Launch(atomic_inputs, atomic_workspace, atomic_outputs, stream_ptr);
// Temporary scheme to avoid SyncStream error in dynamic-shaped WeNet network
for (auto input : atomic_inputs) {
auto align_size = device::MemoryManager::GetCommonAlignSize(input->size);
auto ret = aclrtMemsetAsync(input->addr, align_size, 0, align_size, stream_ptr);
if (ret != RT_ERROR_NONE) {
MS_LOG(EXCEPTION) << "AclrtMemset failed for " << cnode->fullname_with_scope();
}
}
}
}

25
tests/ut/cpp/stub/runtime/acl_rt.cc
View File

@ -33,7 +33,6 @@ ACL_FUNC_VISIBILITY aclError aclrtMemcpy(void *dst, size_t destMax, const void *
return ACL_ERROR_NONE;
}
/**
* @ingroup AscendCL
* @brief Asynchronous memory replication between Host and Device
@ -76,3 +75,27 @@ ACL_FUNC_VISIBILITY aclError aclrtMemcpyAsync(void *dst, size_t destMax, const v
ACL_FUNC_VISIBILITY aclError aclrtSetDevice(int32_t deviceId) {
return ACL_ERROR_NONE;
}
/**
* @ingroup AscendCL
* @brief asynchronously initialize memory and set contents to specified value
*
* @par Function
* The memory to be initialized is on the Host or Device side
* and the system uses address to recognize that
*
* @param dst [IN] destination address pointer
* @param destMax [IN] max length of the destination address memory
* @param value [IN] set value
* @param count [IN] the number of byte to set
* @param stream [IN] asynchronized task stream
*
* @retval ACL_SUCCESS The function is successfully executed.
* @retval OtherValues Failure
*
* @see aclrtSynchronizeStream
*/
ACL_FUNC_VISIBILITY aclError aclrtMemsetAsync(void *dst, size_t destMax, int32_t value, size_t count,
aclrtStream stream) {
return ACL_ERROR_NONE;
}