AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

SWDEV-439628 - hipGraphExecKernelNodeSetParams to update graph kernel node params with graph performance optimizations.

Browse Source 
During hipGraphExecKernelNodeSetParams kernel function can also be updated.
Hence size required for kernel parameters differs from what is allocated during graphInstantiation.
So, create new 128KB kernel pool and allocate kernel args from the pool.
If the pool is full create new 128KB pool. Release kernel pools when graph exec object is destroyed.

Change-Id: I9567946d63400c79cbfd4c5439c654c92557ceae


[ROCm/clr commit: e9957151f3]
This commit is contained in:
Anusha GodavarthySurya
2024-01-08 17:42:55 +00:00
committed by Maneesh Gupta
parent 401a2635ba
commit 3dc533e1e1
6 changed files with 90 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/clr/hipamd/src/hip_device.cpp
+5 -3
View File
@@ -30,7 +30,7 @@
namespace hip {
// ================================================================================================
hip::Stream* Device::NullStream() {
hip::Stream* Device::NullStream(bool wait) {
if (null_stream_ == nullptr) {
null_stream_ = new Stream(this, Stream::Priority::Normal, 0, true);
}
@@ -38,8 +38,10 @@ hip::Stream* Device::NullStream() {
if (null_stream_ == nullptr) {
return nullptr;
}
// Wait for all active streams before executing commands on the default
iHipWaitActiveStreams(null_stream_);
if (wait == true) {
// Wait for all active streams before executing commands on the default
iHipWaitActiveStreams(null_stream_);
}
return null_stream_;
}