P4 to Git Change 1288106 by jsjodin@jsjodin-git2p4-llvm on 2016/07/06 18:17:28

SWDEV-3 - AMDGPU: Improve load/store of illegal types.

	There was a combine before to handle the simple copy case.
	Split this into handling loads and stores separately.

	We might want to change how this handles some of the vector
	extloads, since this can result in large code size increases.

	git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@274394 91177308-0d34-0410-b5e6-96231b3b80d8

	GitHash: df587174ebb04a30c8b55aef601121c25af11986

Affected files ...

... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/lib/Target/AMDGPU/AMDGPUISelLowering.cpp#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/lib/Target/AMDGPU/AMDGPUISelLowering.h#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/lib/Target/AMDGPU/SIISelLowering.cpp#6 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/add.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/copy-illegal-type.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/ctpop64.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/cvt_f32_ubyte.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/half.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/insert_vector_elt.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-constant-i16.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-constant-i32.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-constant-i64.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-constant-i8.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-global-f64.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-global-i16.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-global-i32.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-global-i64.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-global-i8.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-local-f32.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-local-i16.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-local-i8.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/load-weird-sizes.ll#1 add
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/scalar_to_vector.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/sext-in-reg.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/shift-and-i128-ubfe.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/store-barrier.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/store.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/sub.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/trunc-bitcast-vector.ll#2 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm.git/test/CodeGen/AMDGPU/trunc-store.ll#2 edit

rocclr/runtime/device/pal/paldevice.cpp

@@ -706,7 +706,6 @@ Device::create(Pal::IDevice* device)
     // palSettings ...
     palSettings->textureOptLevel = Pal::TextureFilterOptimizationsDisabled;
     palSettings->forceHighClocks = appProfile_.enableHighPerformanceState();
-    palSettings->cmdBufBatchedSubmitChainLimit = 0;
 
     // Commit the new settings for the device
     result = iDev()->CommitSettingsAndInit();