summaryrefslogtreecommitdiffstats
path: root/llvm/lib/Target/AMDGPU
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/lib/Target/AMDGPU')
-rw-r--r--llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp14
-rw-r--r--llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h3
2 files changed, 11 insertions, 6 deletions
diff --git a/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp b/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
index 74dac6561cf..3896bcc0b0e 100644
--- a/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
@@ -358,9 +358,11 @@ unsigned getMinNumSGPRs(const FeatureBitset &Features, unsigned WavesPerEU) {
if (WavesPerEU >= getMaxWavesPerEU(Features))
return 0;
- unsigned MinNumSGPRs =
- alignDown(getTotalNumSGPRs(Features) / (WavesPerEU + 1),
- getSGPRAllocGranule(Features)) + 1;
+
+ unsigned MinNumSGPRs = getTotalNumSGPRs(Features) / (WavesPerEU + 1);
+ if (Features.test(FeatureTrapHandler))
+ MinNumSGPRs -= std::min(MinNumSGPRs, (unsigned)TRAP_NUM_SGPRS);
+ MinNumSGPRs = alignDown(MinNumSGPRs, getSGPRAllocGranule(Features)) + 1;
return std::min(MinNumSGPRs, getAddressableNumSGPRs(Features));
}
@@ -369,11 +371,13 @@ unsigned getMaxNumSGPRs(const FeatureBitset &Features, unsigned WavesPerEU,
assert(WavesPerEU != 0);
IsaVersion Version = getIsaVersion(Features);
- unsigned MaxNumSGPRs = alignDown(getTotalNumSGPRs(Features) / WavesPerEU,
- getSGPRAllocGranule(Features));
unsigned AddressableNumSGPRs = getAddressableNumSGPRs(Features);
if (Version.Major >= 8 && !Addressable)
AddressableNumSGPRs = 112;
+ unsigned MaxNumSGPRs = getTotalNumSGPRs(Features) / WavesPerEU;
+ if (Features.test(FeatureTrapHandler))
+ MaxNumSGPRs -= std::min(MaxNumSGPRs, (unsigned)TRAP_NUM_SGPRS);
+ MaxNumSGPRs = alignDown(MaxNumSGPRs, getSGPRAllocGranule(Features));
return std::min(MaxNumSGPRs, AddressableNumSGPRs);
}
diff --git a/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h b/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h
index 5459ddfc7ef..54932d2bd3e 100644
--- a/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h
+++ b/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h
@@ -42,7 +42,8 @@ namespace IsaInfo {
enum {
// The closed Vulkan driver sets 96, which limits the wave count to 8 but
// doesn't spill SGPRs as much as when 80 is set.
- FIXED_NUM_SGPRS_FOR_INIT_BUG = 96
+ FIXED_NUM_SGPRS_FOR_INIT_BUG = 96,
+ TRAP_NUM_SGPRS = 16
};
/// Instruction set architecture version.
OpenPOWER on IntegriCloud