From 9cdd929342ef078fd2599f7a25fab59706125a1f Mon Sep 17 00:00:00 2001
From: Rakesh Roy <rakesh.roy@amd.com>
Date: Mon, 5 Sep 2022 23:17:30 +0530
Subject: [PATCH] SWDEV-353941 - Fix hipMemset latency issue for
 hipMallocManaged

- In case of HMM, use blit kernel instead of CPU memcpy for hipMemset

Change-Id: I89bfc96ff01a2375ed8df1b1c6bc05357dea84f7


[ROCm/clr commit: f097cda948d9c1627c6f3e84473749d8668f38fd]
---
 projects/clr/rocclr/device/pal/palvirtual.cpp  | 12 +++++++++++-
 projects/clr/rocclr/device/rocm/rocvirtual.cpp | 12 +++++++++++-
 2 files changed, 22 insertions(+), 2 deletions(-)

diff --git a/projects/clr/rocclr/device/pal/palvirtual.cpp b/projects/clr/rocclr/device/pal/palvirtual.cpp
index 0a5e1484c5..dfbd198b4b 100644
--- a/projects/clr/rocclr/device/pal/palvirtual.cpp
+++ b/projects/clr/rocclr/device/pal/palvirtual.cpp
@@ -1872,11 +1872,21 @@ void VirtualGPU::submitFillMemory(amd::FillMemoryCommand& cmd) {
     amd::BufferRect rect;
     rect.create(static_cast<size_t*>(origin), static_cast<size_t*>(region),
         pitch, 0);
+
+    bool force_blit = false;
+    if (amd::IS_HIP) {
+      constexpr uint32_t kManagedAlloc = (CL_MEM_SVM_FINE_GRAIN_BUFFER | CL_MEM_ALLOC_HOST_PTR);
+      // In case of HMM, use blit kernel instead of CPU memcpy
+      if ((cmd.memory().getMemFlags() & kManagedAlloc) == kManagedAlloc) {
+        force_blit = true;
+      }
+    }
+
     for (size_t slice = 0; slice < depth; slice++) {
       for (size_t row = 0; row < height; row++) {
         const size_t rowOffset = rect.offset(0, row, slice);
         if (!fillMemory(cmd.type(), &cmd.memory(), cmd.pattern(), cmd.patternSize(),
-            amd::Coord3D{rowOffset, 0, 0}, amd::Coord3D{width, 1, 1})) {
+            amd::Coord3D{rowOffset, 0, 0}, amd::Coord3D{width, 1, 1}, force_blit)) {
           cmd.setStatus(CL_INVALID_OPERATION);
         }
       }
diff --git a/projects/clr/rocclr/device/rocm/rocvirtual.cpp b/projects/clr/rocclr/device/rocm/rocvirtual.cpp
index 61daa52240..bba29a253b 100644
--- a/projects/clr/rocclr/device/rocm/rocvirtual.cpp
+++ b/projects/clr/rocclr/device/rocm/rocvirtual.cpp
@@ -2296,8 +2296,18 @@ void VirtualGPU::submitFillMemory(amd::FillMemoryCommand& cmd) {
   amd::ScopedLock lock(execution());
 
   profilingBegin(cmd);
+
+  bool force_blit = false;
+  if (amd::IS_HIP) {
+    constexpr uint32_t kManagedAlloc = (CL_MEM_SVM_FINE_GRAIN_BUFFER | CL_MEM_ALLOC_HOST_PTR);
+    // In case of HMM, use blit kernel instead of CPU memcpy
+    if ((cmd.memory().getMemFlags() & kManagedAlloc) == kManagedAlloc) {
+      force_blit = true;
+    }
+  }
+
   if (!fillMemory(cmd.type(), &cmd.memory(), cmd.pattern(), cmd.patternSize(),
-    cmd.surface(), cmd.origin(), cmd.size())) {
+    cmd.surface(), cmd.origin(), cmd.size(), force_blit)) {
     cmd.setStatus(CL_INVALID_OPERATION);
   }
   profilingEnd(cmd);