derek/gem5
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

arch-gcn3,arch-vega,gpu-compute: Move request counters

Browse Source 
When the Vega ISA got committed, it lacked the request counter
tracking for memory requests that existed in the GCN3 code.

Instead of copying over the same lines from the GCN3 code to the Vega
code, this commit makes the various memory pipelines handle updating the
request counter information instead, as every memory instruction calls a
memory pipeline.

This commit also adds an issueRequest in scalar_memory_pipeline, as
previously, the gpuDynInsts were explicitly placed in the queue of
issuedRequests.

Change-Id: I5140d3b2f12be582f2ae9ff7c433167aeec5b68e
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/45347
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
This commit is contained in:
Kyle Roarty
2021-05-07 22:12:45 -05:00
committed by Matt Sinclair
parent 3f9b03522c
commit 02dd6b77ff
6 changed files with 82 additions and 408 deletions
Download Patch File Download Diff File Expand all files Collapse all files

408
src/arch/amdgpu/gcn3/insts/instructions.cc
View File

@@ -4497,12 +4497,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
.issueRequest(gpuDynInst);
}
void
@@ -4556,12 +4551,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
issueRequest(gpuDynInst);
}
void
@@ -4613,12 +4603,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
issueRequest(gpuDynInst);
}
void
@@ -4670,12 +4655,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
issueRequest(gpuDynInst);
}
void
@@ -4727,12 +4707,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
issueRequest(gpuDynInst);
}
void
@@ -4785,12 +4760,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
.issueRequest(gpuDynInst);
} // execute
void
@@ -4844,12 +4814,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
.issueRequest(gpuDynInst);
} // execute
void
@@ -4903,12 +4868,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
.issueRequest(gpuDynInst);
} // execute
void
@@ -4962,12 +4922,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
.issueRequest(gpuDynInst);
} // execute
void
@@ -5021,12 +4976,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
.issueRequest(gpuDynInst);
} // execute
void
@@ -5087,12 +5037,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
wf->scalarWrGmReqsInPipe--;
wf->scalarOutstandingReqsWrGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
issueRequest(gpuDynInst);
}
void
@@ -5147,12 +5092,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
wf->scalarWrGmReqsInPipe--;
wf->scalarOutstandingReqsWrGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
issueRequest(gpuDynInst);
}
void
@@ -5207,12 +5147,7 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
wf->scalarWrGmReqsInPipe--;
wf->scalarOutstandingReqsWrGm++;
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
issueRequest(gpuDynInst);
}
void
@@ -31262,11 +31197,6 @@ namespace Gcn3ISA
}
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -31326,11 +31256,6 @@ namespace Gcn3ISA
}
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -31389,11 +31314,6 @@ namespace Gcn3ISA
}
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -31564,11 +31484,6 @@ namespace Gcn3ISA
}
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -31623,11 +31538,6 @@ namespace Gcn3ISA
}
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -32078,11 +31988,6 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr);
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -32139,11 +32044,6 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr);
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -32203,11 +32103,6 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr);
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -32286,11 +32181,6 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr);
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -32366,11 +32256,6 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr);
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
Inst_DS__DS_READ_U16::initiateAcc(GPUDynInstPtr gpuDynInst)
@@ -32931,11 +32816,6 @@ namespace Gcn3ISA
}
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -32995,11 +32875,6 @@ namespace Gcn3ISA
}
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -33517,11 +33392,6 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr);
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -33578,11 +33448,6 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr);
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -33642,11 +33507,6 @@ namespace Gcn3ISA
calcAddr(gpuDynInst, addr);
gpuDynInst->computeUnit()->localMemoryPipe.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -35070,17 +34930,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe.
issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -35200,17 +35053,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -35330,17 +35176,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -35429,17 +35268,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -35533,17 +35365,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -35642,17 +35467,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -35765,17 +35583,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -35859,17 +35670,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -35953,17 +35757,10 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -36042,13 +35839,9 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
}
for (int lane = 0; lane < NumVecElemPerVecReg; ++lane) {
@@ -36059,9 +35852,6 @@ namespace Gcn3ISA
= data1[lane];
}
}
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -36142,13 +35932,9 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
}
for (int lane = 0; lane < NumVecElemPerVecReg; ++lane) {
@@ -36161,9 +35947,6 @@ namespace Gcn3ISA
= data2[lane];
}
}
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -36246,13 +36029,9 @@ namespace Gcn3ISA
if (isLocalMem()) {
gpuDynInst->computeUnit()->localMemoryPipe
.issueRequest(gpuDynInst);
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
} else {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
}
for (int lane = 0; lane < NumVecElemPerVecReg; ++lane) {
@@ -36267,9 +36046,6 @@ namespace Gcn3ISA
= data3[lane];
}
}
wf->outstandingReqs++;
wf->validateRequestCounters();
} // execute
void
@@ -36328,17 +36104,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->rdGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -36385,17 +36153,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->rdGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
Inst_MUBUF__BUFFER_WBINVL1_VOL::initiateAcc(GPUDynInstPtr gpuDynInst)
@@ -39415,14 +39175,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
} // execute
void
@@ -39484,14 +39239,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -39552,14 +39302,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -39649,14 +39394,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -39718,14 +39458,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -39787,14 +39522,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -39865,14 +39595,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -39955,14 +39680,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
} // execute
void
@@ -40023,14 +39743,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -40091,14 +39806,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -40160,14 +39870,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
wf->outstandingReqs++;
wf->validateRequestCounters();
}
void
@@ -40237,14 +39942,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -40318,14 +40018,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe
.issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -40390,17 +40085,10 @@ namespace Gcn3ISA
"Flats to private aperture not tested yet\n");
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
ConstVecOperandU32 data(gpuDynInst, extData.DATA);
data.read();
@@ -40511,16 +40199,9 @@ namespace Gcn3ISA
"Flats to private aperture not tested yet\n");
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -40602,16 +40283,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -40693,16 +40367,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
Inst_FLAT__FLAT_ATOMIC_SUB::initiateAcc(GPUDynInstPtr gpuDynInst)
@@ -40958,16 +40625,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -41049,16 +40709,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -41182,16 +40835,9 @@ namespace Gcn3ISA
"Flats to private aperture not tested yet\n");
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -41274,16 +40920,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -41367,16 +41006,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -41642,16 +41274,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void
@@ -41736,16 +41361,9 @@ namespace Gcn3ISA
if (gpuDynInst->executedAs() == enums::SC_GLOBAL) {
gpuDynInst->computeUnit()->globalMemoryPipe.
issueRequest(gpuDynInst);
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else {
fatal("Non global flat instructions not implemented yet.\n");
}
gpuDynInst->wavefront()->outstandingReqs++;
gpuDynInst->wavefront()->validateRequestCounters();
}
void

26
src/arch/amdgpu/vega/insts/instructions.cc
View File

@@ -4972,7 +4972,7 @@ namespace VegaISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
.issueRequest(gpuDynInst);
} // execute
void
@@ -5026,7 +5026,7 @@ namespace VegaISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
issueRequest(gpuDynInst);
} // execute
void
@@ -5079,7 +5079,7 @@ namespace VegaISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
issueRequest(gpuDynInst);
} // execute
void
@@ -5132,7 +5132,7 @@ namespace VegaISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
issueRequest(gpuDynInst);
} // execute
void
@@ -5185,7 +5185,7 @@ namespace VegaISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
issueRequest(gpuDynInst);
} // execute
void
@@ -5239,7 +5239,7 @@ namespace VegaISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
.issueRequest(gpuDynInst);
} // execute
void
@@ -5294,7 +5294,7 @@ namespace VegaISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
.issueRequest(gpuDynInst);
} // execute
void
@@ -5349,7 +5349,7 @@ namespace VegaISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
.issueRequest(gpuDynInst);
} // execute
void
@@ -5404,7 +5404,7 @@ namespace VegaISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
.issueRequest(gpuDynInst);
} // execute
void
@@ -5459,7 +5459,7 @@ namespace VegaISA
calcAddr(gpuDynInst, rsrcDesc, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe
.getGMReqFIFO().push(gpuDynInst);
.issueRequest(gpuDynInst);
} // execute
void
@@ -5521,7 +5521,7 @@ namespace VegaISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
issueRequest(gpuDynInst);
} // execute
void
@@ -5577,7 +5577,7 @@ namespace VegaISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
issueRequest(gpuDynInst);
} // execute
void
@@ -5633,7 +5633,7 @@ namespace VegaISA
calcAddr(gpuDynInst, addr, offset);
gpuDynInst->computeUnit()->scalarMemoryPipe.
getGMReqFIFO().push(gpuDynInst);
issueRequest(gpuDynInst);
} // execute
void

18
src/gpu-compute/global_memory_pipeline.cc
View File

@@ -276,6 +276,24 @@ GlobalMemPipeline::completeRequest(GPUDynInstPtr gpuDynInst)
void
GlobalMemPipeline::issueRequest(GPUDynInstPtr gpuDynInst)
{
Wavefront *wf = gpuDynInst->wavefront();
if (gpuDynInst->isLoad()) {
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
} else if (gpuDynInst->isStore()) {
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
} else {
// Atomic, both read and write
wf->rdGmReqsInPipe--;
wf->outstandingReqsRdGm++;
wf->wrGmReqsInPipe--;
wf->outstandingReqsWrGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
gpuDynInst->setAccessTime(curTick());
gpuDynInst->profileRoundTripTime(curTick(), InstMemoryHop::Initiate);
gmIssuedRequests.push(gpuDynInst);

18
src/gpu-compute/local_memory_pipeline.cc
View File

@@ -123,6 +123,24 @@ LocalMemPipeline::exec()
void
LocalMemPipeline::issueRequest(GPUDynInstPtr gpuDynInst)
{
Wavefront *wf = gpuDynInst->wavefront();
if (gpuDynInst->isLoad()) {
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
} else if (gpuDynInst->isStore()) {
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
} else {
// Atomic, both read and write
wf->rdLmReqsInPipe--;
wf->outstandingReqsRdLm++;
wf->wrLmReqsInPipe--;
wf->outstandingReqsWrLm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
gpuDynInst->setAccessTime(curTick());
lmIssuedRequests.push(gpuDynInst);
}

18
src/gpu-compute/scalar_memory_pipeline.cc
View File

@@ -144,4 +144,22 @@ ScalarMemPipeline::exec()
}
}
void
ScalarMemPipeline::issueRequest(GPUDynInstPtr gpuDynInst)
{
Wavefront *wf = gpuDynInst->wavefront();
if (gpuDynInst->isLoad()) {
wf->scalarRdGmReqsInPipe--;
wf->scalarOutstandingReqsRdGm++;
} else if (gpuDynInst->isStore()) {
wf->scalarWrGmReqsInPipe--;
wf->scalarOutstandingReqsWrGm++;
}
wf->outstandingReqs++;
wf->validateRequestCounters();
issuedRequests.push(gpuDynInst);
}
} // namespace gem5

2
src/gpu-compute/scalar_memory_pipeline.hh
View File

@@ -67,6 +67,8 @@ class ScalarMemPipeline
std::queue<GPUDynInstPtr> &getGMStRespFIFO() { return returnedStores; }
std::queue<GPUDynInstPtr> &getGMLdRespFIFO() { return returnedLoads; }
void issueRequest(GPUDynInstPtr gpuDynInst);
bool
isGMLdRespFIFOWrRdy() const
{