Anthony Gutierrez has submitted this change. (
https://gem5-review.googlesource.com/c/public/gem5/+/29962 )
Change subject: arch-gcn3: add all s_buffer_load_dword instructions
......................................................................
arch-gcn3: add all s_buffer_load_dword instructions
Adds the other s_buffer_load_dword* instruction implementations to
f134a84.
Change-Id: I8d97527278900dc68c32463ea1824409ccd04e1d
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/29962
Maintainer: Anthony Gutierrez <anthony.gutier...@amd.com>
Tested-by: kokoro <noreply+kok...@google.com>
Reviewed-by: Matt Sinclair <mattdsincl...@gmail.com>
---
M src/arch/gcn3/insts/instructions.cc
1 file changed, 125 insertions(+), 8 deletions(-)
Approvals:
Matt Sinclair: Looks good to me, approved
Anthony Gutierrez: Looks good to me, approved
kokoro: Regressions pass
diff --git a/src/arch/gcn3/insts/instructions.cc
b/src/arch/gcn3/insts/instructions.cc
index 002c4d5..2b992b1 100644
--- a/src/arch/gcn3/insts/instructions.cc
+++ b/src/arch/gcn3/insts/instructions.cc
@@ -4737,17 +4737,46 @@
void
Inst_SMEM__S_BUFFER_LOAD_DWORD::execute(GPUDynInstPtr gpuDynInst)
{
- panicUnimplemented();
- }
+ Wavefront *wf = gpuDynInst->wavefront();
+ gpuDynInst->execUnitId = wf->execUnitId;
+ gpuDynInst->latency.init(gpuDynInst->computeUnit());
+ gpuDynInst->latency.set(gpuDynInst->computeUnit()->clockPeriod());
+ ScalarRegU32 offset(0);
+ ConstScalarOperandU128 rsrcDesc(gpuDynInst, instData.SBASE);
+
+ rsrcDesc.read();
+
+ if (instData.IMM) {
+ offset = extData.OFFSET;
+ } else {
+ ConstScalarOperandU32 off_sgpr(gpuDynInst, extData.OFFSET);
+ off_sgpr.read();
+ offset = off_sgpr.rawData();
+ }
+
+ calcAddr(gpuDynInst, rsrcDesc, offset);
+
+ gpuDynInst->computeUnit()->scalarMemoryPipe
+ .getGMReqFIFO().push(gpuDynInst);
+
+ wf->scalarRdGmReqsInPipe--;
+ wf->scalarOutstandingReqsRdGm++;
+ gpuDynInst->wavefront()->outstandingReqs++;
+ gpuDynInst->wavefront()->validateRequestCounters();
+ } // execute
void
Inst_SMEM__S_BUFFER_LOAD_DWORD::initiateAcc(GPUDynInstPtr gpuDynInst)
{
+ initMemRead<1>(gpuDynInst);
} // initiateAcc
void
Inst_SMEM__S_BUFFER_LOAD_DWORD::completeAcc(GPUDynInstPtr gpuDynInst)
{
+ // 1 request, size 32
+ ScalarOperandU32 sdst(gpuDynInst, instData.SDATA);
+ sdst.write();
} // completeAcc
Inst_SMEM__S_BUFFER_LOAD_DWORDX2::Inst_SMEM__S_BUFFER_LOAD_DWORDX2(
@@ -4767,17 +4796,46 @@
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX2::execute(GPUDynInstPtr gpuDynInst)
{
- panicUnimplemented();
- }
+ Wavefront *wf = gpuDynInst->wavefront();
+ gpuDynInst->execUnitId = wf->execUnitId;
+ gpuDynInst->latency.init(gpuDynInst->computeUnit());
+ gpuDynInst->latency.set(gpuDynInst->computeUnit()->clockPeriod());
+ ScalarRegU32 offset(0);
+ ConstScalarOperandU128 rsrcDesc(gpuDynInst, instData.SBASE);
+
+ rsrcDesc.read();
+
+ if (instData.IMM) {
+ offset = extData.OFFSET;
+ } else {
+ ConstScalarOperandU32 off_sgpr(gpuDynInst, extData.OFFSET);
+ off_sgpr.read();
+ offset = off_sgpr.rawData();
+ }
+
+ calcAddr(gpuDynInst, rsrcDesc, offset);
+
+ gpuDynInst->computeUnit()->scalarMemoryPipe
+ .getGMReqFIFO().push(gpuDynInst);
+
+ wf->scalarRdGmReqsInPipe--;
+ wf->scalarOutstandingReqsRdGm++;
+ gpuDynInst->wavefront()->outstandingReqs++;
+ gpuDynInst->wavefront()->validateRequestCounters();
+ } // execute
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX2::initiateAcc(GPUDynInstPtr gpuDynInst)
{
+ initMemRead<2>(gpuDynInst);
} // initiateAcc
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX2::completeAcc(GPUDynInstPtr gpuDynInst)
{
+ // use U64 because 2 requests, each size 32
+ ScalarOperandU64 sdst(gpuDynInst, instData.SDATA);
+ sdst.write();
} // completeAcc
Inst_SMEM__S_BUFFER_LOAD_DWORDX4::Inst_SMEM__S_BUFFER_LOAD_DWORDX4(
@@ -4797,17 +4855,46 @@
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX4::execute(GPUDynInstPtr gpuDynInst)
{
- panicUnimplemented();
- }
+ Wavefront *wf = gpuDynInst->wavefront();
+ gpuDynInst->execUnitId = wf->execUnitId;
+ gpuDynInst->latency.init(gpuDynInst->computeUnit());
+ gpuDynInst->latency.set(gpuDynInst->computeUnit()->clockPeriod());
+ ScalarRegU32 offset(0);
+ ConstScalarOperandU128 rsrcDesc(gpuDynInst, instData.SBASE);
+
+ rsrcDesc.read();
+
+ if (instData.IMM) {
+ offset = extData.OFFSET;
+ } else {
+ ConstScalarOperandU32 off_sgpr(gpuDynInst, extData.OFFSET);
+ off_sgpr.read();
+ offset = off_sgpr.rawData();
+ }
+
+ calcAddr(gpuDynInst, rsrcDesc, offset);
+
+ gpuDynInst->computeUnit()->scalarMemoryPipe
+ .getGMReqFIFO().push(gpuDynInst);
+
+ wf->scalarRdGmReqsInPipe--;
+ wf->scalarOutstandingReqsRdGm++;
+ gpuDynInst->wavefront()->outstandingReqs++;
+ gpuDynInst->wavefront()->validateRequestCounters();
+ } // execute
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX4::initiateAcc(GPUDynInstPtr gpuDynInst)
{
+ initMemRead<4>(gpuDynInst);
} // initiateAcc
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX4::completeAcc(GPUDynInstPtr gpuDynInst)
{
+ // 4 requests, each size 32
+ ScalarOperandU128 sdst(gpuDynInst, instData.SDATA);
+ sdst.write();
} // completeAcc
Inst_SMEM__S_BUFFER_LOAD_DWORDX8::Inst_SMEM__S_BUFFER_LOAD_DWORDX8(
@@ -4827,17 +4914,46 @@
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX8::execute(GPUDynInstPtr gpuDynInst)
{
- panicUnimplemented();
- }
+ Wavefront *wf = gpuDynInst->wavefront();
+ gpuDynInst->execUnitId = wf->execUnitId;
+ gpuDynInst->latency.init(gpuDynInst->computeUnit());
+ gpuDynInst->latency.set(gpuDynInst->computeUnit()->clockPeriod());
+ ScalarRegU32 offset(0);
+ ConstScalarOperandU128 rsrcDesc(gpuDynInst, instData.SBASE);
+
+ rsrcDesc.read();
+
+ if (instData.IMM) {
+ offset = extData.OFFSET;
+ } else {
+ ConstScalarOperandU32 off_sgpr(gpuDynInst, extData.OFFSET);
+ off_sgpr.read();
+ offset = off_sgpr.rawData();
+ }
+
+ calcAddr(gpuDynInst, rsrcDesc, offset);
+
+ gpuDynInst->computeUnit()->scalarMemoryPipe
+ .getGMReqFIFO().push(gpuDynInst);
+
+ wf->scalarRdGmReqsInPipe--;
+ wf->scalarOutstandingReqsRdGm++;
+ gpuDynInst->wavefront()->outstandingReqs++;
+ gpuDynInst->wavefront()->validateRequestCounters();
+ } // execute
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX8::initiateAcc(GPUDynInstPtr gpuDynInst)
{
+ initMemRead<8>(gpuDynInst);
} // initiateAcc
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX8::completeAcc(GPUDynInstPtr gpuDynInst)
{
+ // 8 requests, each size 32
+ ScalarOperandU256 sdst(gpuDynInst, instData.SDATA);
+ sdst.write();
} // completeAcc
Inst_SMEM__S_BUFFER_LOAD_DWORDX16::Inst_SMEM__S_BUFFER_LOAD_DWORDX16(
@@ -4894,6 +5010,7 @@
void
Inst_SMEM__S_BUFFER_LOAD_DWORDX16::completeAcc(GPUDynInstPtr
gpuDynInst)
{
+ // 16 requests, each size 32
ScalarOperandU512 sdst(gpuDynInst, instData.SDATA);
sdst.write();
} // completeAcc
--
To view, visit https://gem5-review.googlesource.com/c/public/gem5/+/29962
To unsubscribe, or for help writing mail filters, visit
https://gem5-review.googlesource.com/settings
Gerrit-Project: public/gem5
Gerrit-Branch: develop
Gerrit-Change-Id: I8d97527278900dc68c32463ea1824409ccd04e1d
Gerrit-Change-Number: 29962
Gerrit-PatchSet: 8
Gerrit-Owner: Anthony Gutierrez <anthony.gutier...@amd.com>
Gerrit-Reviewer: Anthony Gutierrez <anthony.gutier...@amd.com>
Gerrit-Reviewer: Matt Sinclair <mattdsincl...@gmail.com>
Gerrit-Reviewer: Tony Gutierrez <anthony.gutier...@amd.com>
Gerrit-Reviewer: Tuan Ta <q...@cornell.edu>
Gerrit-Reviewer: kokoro <noreply+kok...@google.com>
Gerrit-MessageType: merged
_______________________________________________
gem5-dev mailing list -- gem5-dev@gem5.org
To unsubscribe send an email to gem5-dev-le...@gem5.org
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s