Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Bug fixed]: fixed a bug in timing/cu/issuearbiter.go; this bug cause… #21

Merged
merged 10 commits into from
Dec 18, 2023
Merged

[Bug fixed]: fixed a bug in timing/cu/issuearbiter.go; this bug cause… #21

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
fd99c4d
[Bug fixed]: fixed a bug in timing/cu/issuearbiter.go; this bug cause…
lchangxii Oct 18, 2023
d4b0438
[Format]: how to replace 4 space with a tab
lchangxii Oct 18, 2023
18782ca
[Issue Arbiter]: fixed a minimal bug; could cause dead loop issue
lchangxii Oct 20, 2023
d9bde6e
Merge branch 'sarchlab:v3' into v3
lchangxii Nov 27, 2023
ae39027
[Bug Fixed]: fixed a bug in vector memory unit
lchangxii Nov 27, 2023
f3b0b8c
Apply suggestions from code review
syifan Nov 27, 2023
ea939c3
update logic
syifan Dec 13, 2023
71393e6
Merge branch 'v3' into v3
syifan Dec 13, 2023
e0717d9
Fix unit test
syifan Dec 18, 2023
57df837
Remove concurrent kernel in test
syifan Dec 18, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
56 changes: 28 additions & 28 deletions tests/acceptance/cases.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -801,32 +801,32 @@ var benchmarks = []benchmark{
{gpus: []int{1, 2, 3, 4}, timing: true, parallel: true, unifiedGPU: true, unifiedMemory: true},
},
},
{
benchmarkPath: "",
executablePath: "../../samples/concurrentkernel",
executable: "concurrentkernel",
sizeArgs: []string{},
cases: []benchmarkCase{
{gpus: []int{1}, timing: false, parallel: false, unifiedGPU: false, unifiedMemory: false},
{gpus: []int{1}, timing: false, parallel: true, unifiedGPU: false, unifiedMemory: false},
{gpus: []int{1}, timing: true, parallel: false, unifiedGPU: false, unifiedMemory: false},
{gpus: []int{1}, timing: true, parallel: true, unifiedGPU: false, unifiedMemory: false},
},
},
{
benchmarkPath: "",
executablePath: "../../samples/concurrentworkload",
executable: "concurrentworkload",
sizeArgs: []string{},
cases: []benchmarkCase{
{gpus: []int{1, 2, 3, 4}, timing: false, parallel: false, unifiedGPU: false, unifiedMemory: false},
{gpus: []int{1, 2, 3, 4}, timing: false, parallel: true, unifiedGPU: false, unifiedMemory: false},
{gpus: []int{1, 2, 3, 4}, timing: true, parallel: false, unifiedGPU: false, unifiedMemory: false},
{gpus: []int{1, 2, 3, 4}, timing: true, parallel: true, unifiedGPU: false, unifiedMemory: false},
// {gpus: []int{1, 2, 3, 4}, timing: false, parallel: false, unifiedGPU: false, unifiedMemory: true},
// {gpus: []int{1, 2, 3, 4}, timing: false, parallel: true, unifiedGPU: false, unifiedMemory: true},
// {gpus: []int{1, 2, 3, 4}, timing: true, parallel: false, unifiedGPU: false, unifiedMemory: true},
// {gpus: []int{1, 2, 3, 4}, timing: true, parallel: true, unifiedGPU: false, unifiedMemory: true},
},
},
// {
// benchmarkPath: "",
// executablePath: "../../samples/concurrentkernel",
// executable: "concurrentkernel",
// sizeArgs: []string{},
// cases: []benchmarkCase{
// {gpus: []int{1}, timing: false, parallel: false, unifiedGPU: false, unifiedMemory: false},
// {gpus: []int{1}, timing: false, parallel: true, unifiedGPU: false, unifiedMemory: false},
// {gpus: []int{1}, timing: true, parallel: false, unifiedGPU: false, unifiedMemory: false},
// {gpus: []int{1}, timing: true, parallel: true, unifiedGPU: false, unifiedMemory: false},
// },
// },
// {
// benchmarkPath: "",
// executablePath: "../../samples/concurrentworkload",
// executable: "concurrentworkload",
// sizeArgs: []string{},
// cases: []benchmarkCase{
// {gpus: []int{1, 2, 3, 4}, timing: false, parallel: false, unifiedGPU: false, unifiedMemory: false},
// {gpus: []int{1, 2, 3, 4}, timing: false, parallel: true, unifiedGPU: false, unifiedMemory: false},
// {gpus: []int{1, 2, 3, 4}, timing: true, parallel: false, unifiedGPU: false, unifiedMemory: false},
// {gpus: []int{1, 2, 3, 4}, timing: true, parallel: true, unifiedGPU: false, unifiedMemory: false},
// {gpus: []int{1, 2, 3, 4}, timing: false, parallel: false, unifiedGPU: false, unifiedMemory: true},
// {gpus: []int{1, 2, 3, 4}, timing: false, parallel: true, unifiedGPU: false, unifiedMemory: true},
// {gpus: []int{1, 2, 3, 4}, timing: true, parallel: false, unifiedGPU: false, unifiedMemory: true},
// {gpus: []int{1, 2, 3, 4}, timing: true, parallel: true, unifiedGPU: false, unifiedMemory: true},
// },
// },
}
35 changes: 20 additions & 15 deletions timing/cu/issuearbiter.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ type IssueArbiter struct {
// NewIssueArbiter returns a newly created IssueArbiter
func NewIssueArbiter() *IssueArbiter {
a := new(IssueArbiter)
a.lastSIMDID = -1
a.lastSIMDID = 0
return a
}

Expand All @@ -23,25 +23,30 @@ func (a *IssueArbiter) Arbitrate(
return []*wavefront.Wavefront{}
}

a.moveToNextSIMD(wfPools)
for len(wfPools[a.lastSIMDID].wfs) == 0 {
a.moveToNextSIMD(wfPools)
}
wfToIssue := make([]*wavefront.Wavefront, 0)
for i := 0; i < len(wfPools); i++ {
simdID := (a.lastSIMDID + i) % len(wfPools)

typeMask := make([]bool, 7)
wfPool := wfPools[simdID]
for _, wf := range wfPool.wfs {
if wf.State != wavefront.WfReady || wf.InstToIssue == nil {
continue
}

typeMask := make([]bool, 7)
wfPool := wfPools[a.lastSIMDID]
list := make([]*wavefront.Wavefront, 0)
for _, wf := range wfPool.wfs {
if wf.State != wavefront.WfReady || wf.InstToIssue == nil {
continue
if typeMask[wf.InstToIssue.ExeUnit] == false {
wfToIssue = append(wfToIssue, wf)
typeMask[wf.InstToIssue.ExeUnit] = true
}
}

if typeMask[wf.InstToIssue.ExeUnit] == false {
list = append(list, wf)
typeMask[wf.InstToIssue.ExeUnit] = true
if len(wfToIssue) != 0 {
a.lastSIMDID = simdID
break
}
}
return list

return wfToIssue
}

func (a *IssueArbiter) moveToNextSIMD(wfPools []*WavefrontPool) {
Expand Down
3 changes: 2 additions & 1 deletion timing/cu/vectormemoryunit.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -109,7 +109,7 @@ func (u *VectorMemoryUnit) insertTransactionToPipeline(
}

func (u *VectorMemoryUnit) execute(now sim.VTimeInSec) (madeProgress bool) {
item := u.postInstructionPipelineBuffer.Pop()
item := u.postInstructionPipelineBuffer.Peek()
if item == nil {
return false
}
Expand All @@ -126,6 +126,7 @@ func (u *VectorMemoryUnit) execute(now sim.VTimeInSec) (madeProgress bool) {
log.Panicf("running inst %s in vector memory unit is not supported", inst.String(nil))
}

u.postInstructionPipelineBuffer.Pop()
u.cu.UpdatePCAndSetReady(wave)
u.numInstInFlight--

Expand Down
2 changes: 2 additions & 0 deletions timing/cu/vectormemoryunit_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,6 +91,7 @@ var _ = Describe("Vector Memory Unit", func() {
transactions[i].Read = read
}
coalescer.EXPECT().generateMemTransactions(wave).Return(transactions)
instBuffer.EXPECT().Peek().Return(vectorMemInst{wavefront: wave})
instBuffer.EXPECT().Pop().Return(vectorMemInst{wavefront: wave})

madeProgress := vecMemUnit.instToTransaction(10)
Expand Down Expand Up @@ -122,6 +123,7 @@ var _ = Describe("Vector Memory Unit", func() {
transactions[i].Write = write
}
coalescer.EXPECT().generateMemTransactions(wave).Return(transactions)
instBuffer.EXPECT().Peek().Return(vectorMemInst{wavefront: wave})
instBuffer.EXPECT().Pop().Return(vectorMemInst{wavefront: wave})

madeProgress := vecMemUnit.instToTransaction(10)
Expand Down
Loading