Michael Matloob (Gerrit)

unread,

Apr 30, 2026, 3:30:30 PM (11 days ago) Apr 30

to goph...@pubsubhelper.golang.org, Michael Pratt, golang...@luci-project-accounts.iam.gserviceaccount.com, golang-co...@googlegroups.com

Attention needed from Michael Pratt

New activity on the change

Open in Gerrit

Related details

Attention is currently required from:

Michael Pratt

Submit Requirements:

Code-Review
No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

satisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

Apr 30, 2026, 4:53:11 PM (11 days ago) Apr 30

to Michael Matloob, goph...@pubsubhelper.golang.org, Michael Pratt, golang...@luci-project-accounts.iam.gserviceaccount.com, golang-co...@googlegroups.com

Attention needed from Michael Matloob and Michael Pratt

Michael Matloob voted Code-Review+1

Code-Review

+1

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob
Michael Pratt

Submit Requirements:

Code-Review
No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

satisfied_requirement

open

diffy

Michael Pratt (Gerrit)

unread,

May 6, 2026, 3:07:36 PM (5 days ago) May 6

to Michael Matloob, goph...@pubsubhelper.golang.org, golang...@luci-project-accounts.iam.gserviceaccount.com, Michael Matloob, Michael Pratt, golang-co...@googlegroups.com

Attention needed from Michael Matloob

Michael Pratt added 6 comments

Commit Message

Line 16, Patchset 6 (Latest):function was slower.

Michael Pratt . unresolved

trailing whitespace

Line 18, Patchset 6 (Latest):Look at the change description of CL 766980 for the results of those

Michael Pratt . unresolved

IIUC, these results compare tiny allocations with specialized malloc today vs consolidated specialized malloc (this CL).

How do they compare against no specialized malloc (GOEXPERIMENT=nosizespecializedmalloc)? I assume that they are still faster?

File src/runtime/_mkmalloc/mkmalloc.go

Line 634, Patchset 6 (Latest): if i < 16 {

Michael Pratt . unresolved

See comment in malloc_table_generated.go. I think we still want to emit panic for size 0?

File src/runtime/malloc_generated.go

Line 4381, Patchset 6 (Latest):func mallocgcTinySC2(size uintptr, typ *_type, needzero bool) unsafe.Pointer {

Michael Pratt . resolved

It's nice that we can easily see the actual diff here 😄

Line 4417, Patchset 6 (Latest):

	if size&7 == 0 {
		off = alignUp(off, 8)
	} else if goarch.PtrSize == 4 && size == 12 {

		off = alignUp(off, 8)
	} else if size&3 == 0 {
		off = alignUp(off, 4)
	} else if size&1 == 0 {
		off = alignUp(off, 2)
	}

Michael Pratt . resolved

I'd assume that the only real difference in microbenchmarks is that previously these branches would be eliminated, leaving exactly the right alignUp call.

The other uses below go from add/compare of register and immediate to register and register, which I'd guess is negligible.

File src/runtime/malloc_tables_generated.go

Line 525, Patchset 6 (Parent): mallocPanic,

Michael Pratt . unresolved

Why did this go away?

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob

Submit Requirements:

Code-Review

No-Unresolved-Comments

Review-Enforcement

TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

May 6, 2026, 5:54:42 PM (5 days ago) May 6

to goph...@pubsubhelper.golang.org, golang...@luci-project-accounts.iam.gserviceaccount.com, Michael Matloob, Michael Pratt, golang-co...@googlegroups.com

Attention needed from Michael Matloob and Michael Pratt

Michael Matloob added 3 comments

Commit Message

Line 16, Patchset 6:function was slower.

Michael Pratt . resolved

trailing whitespace

Michael Matloob

Done

File src/runtime/_mkmalloc/mkmalloc.go

Line 634, Patchset 6: if i < 16 {

Michael Pratt . resolved

See comment in malloc_table_generated.go. I think we still want to emit panic for size 0?

Michael Matloob

Done

File src/runtime/malloc_tables_generated.go

Line 525, Patchset 6 (Parent): mallocPanic,

Michael Pratt . resolved

Why did this go away?

Michael Matloob

Done

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob
Michael Pratt

Submit Requirements:

Code-Review
No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

May 6, 2026, 5:54:42 PM (5 days ago) May 6

to goph...@pubsubhelper.golang.org, golang-co...@googlegroups.com

Attention needed from Michael Matloob and Michael Pratt

Michael Matloob uploaded new patchset

Michael Matloob uploaded patch set #7 to this change.

Following approvals got outdated and were removed:

Code-Review: +1 by Michael Matloob
TryBots-Pass: LUCI-TryBot-Result-1 by golang...@luci-project-accounts.iam.gserviceaccount.com

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob
Michael Pratt

Submit Requirements:

Code-Review
No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

May 6, 2026, 6:00:26 PM (5 days ago) May 6

to goph...@pubsubhelper.golang.org, golang...@luci-project-accounts.iam.gserviceaccount.com, Michael Matloob, Michael Pratt, golang-co...@googlegroups.com

Attention needed from Michael Matloob and Michael Pratt

Michael Matloob added 1 comment

Commit Message

Line 18, Patchset 6:Look at the change description of CL 766980 for the results of those

Michael Pratt . unresolved

IIUC, these results compare tiny allocations with specialized malloc today vs consolidated specialized malloc (this CL).
How do they compare against no specialized malloc (GOEXPERIMENT=nosizespecializedmalloc)? I assume that they are still faster?

Michael Matloob

I need to look at this more. They are faster when invoked directly, but not anymore through the jump table. I'm going to see if I can skip the jump table in the tiny case.

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob
Michael Pratt

Submit Requirements:

Code-Review
No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

May 7, 2026, 11:35:24 AM (4 days ago) May 7

to goph...@pubsubhelper.golang.org, golang-co...@googlegroups.com

Attention needed from Michael Matloob and Michael Pratt

Michael Matloob uploaded new patchset

Michael Matloob uploaded patch set #8 to this change.

Following approvals got outdated and were removed:

TryBots-Pass: LUCI-TryBot-Result+1 by golang...@luci-project-accounts.iam.gserviceaccount.com

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob
Michael Pratt

Submit Requirements:

Code-Review
No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

May 7, 2026, 12:03:12 PM (4 days ago) May 7

to goph...@pubsubhelper.golang.org, golang-co...@googlegroups.com

Attention needed from Michael Matloob and Michael Pratt

Michael Matloob uploaded new patchset

Michael Matloob uploaded patch set #9 to this change.

Following approvals got outdated and were removed:

TryBots-Pass: LUCI-TryBot-Result+1 by golang...@luci-project-accounts.iam.gserviceaccount.com

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob
Michael Pratt

Submit Requirements:

Code-Review
No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

May 7, 2026, 2:10:36 PM (4 days ago) May 7

to goph...@pubsubhelper.golang.org, golang...@luci-project-accounts.iam.gserviceaccount.com, Michael Matloob, Michael Pratt, golang-co...@googlegroups.com

Attention needed from Michael Matloob and Michael Pratt

Michael Matloob added 1 comment

Commit Message

Line 18, Patchset 6:Look at the change description of CL 766980 for the results of those

Michael Pratt . resolved

IIUC, these results compare tiny allocations with specialized malloc today vs consolidated specialized malloc (this CL).
How do they compare against no specialized malloc (GOEXPERIMENT=nosizespecializedmalloc)? I assume that they are still faster?

Michael Matloob

I need to look at this more. They are faster when invoked directly, but not anymore through the jump table. I'm going to see if I can skip the jump table in the tiny case.

Michael Matloob

Okay, I was able to make a change to them so that they are now faster:

```
goos: linux
goarch: amd64
pkg: runtime
cpu: AMD Ryzen 9 9950X3D 16-Core Processor          
        │    base     │        sizespecializedmalloc        │
        │   sec/op    │   sec/op     vs base                │
1         4.029n ± 2%   3.736n ± 1%   -7.27% (p=0.000 n=10)
2         4.048n ± 1%   3.595n ± 1%  -11.17% (p=0.000 n=10)
3         4.352n ± 1%   4.062n ± 1%   -6.64% (p=0.000 n=10)
4         4.485n ± 1%   4.277n ± 1%   -4.64% (p=0.000 n=10)
5         4.969n ± 1%   4.715n ± 1%   -5.10% (p=0.000 n=10)
6         5.800n ± 1%   5.234n ± 0%   -9.77% (p=0.000 n=10)
7         5.729n ± 1%   5.328n ± 0%   -7.00% (p=0.000 n=10)
8         5.390n ± 2%   5.173n ± 1%   -4.02% (p=0.000 n=10)
9         7.328n ± 1%   7.258n ± 1%   -0.95% (p=0.015 n=10)
10        7.362n ± 2%   7.125n ± 0%   -3.22% (p=0.000 n=10)
11        7.335n ± 1%   7.223n ± 1%   -1.51% (p=0.001 n=10)
12        7.068n ± 1%   7.072n ± 1%        ~ (p=0.529 n=10)
13        7.363n ± 2%   7.218n ± 1%   -1.96% (p=0.001 n=10)
14        7.347n ± 2%   7.082n ± 1%   -3.61% (p=0.000 n=10)
15        7.365n ± 1%   7.199n ± 1%   -2.26% (p=0.002 n=10)
geomean   5.843n        5.571n        -4.66%
```

i also checked the larger sizes to make sure that they weren't slower either since we made changes before the jump table

```
goos: linux
goarch: amd64
pkg: runtime
cpu: AMD Ryzen 9 9950X3D 16-Core Processor          
        │    base     │        sizespecializedmalloc        │
        │   sec/op    │   sec/op     vs base                │
1         4.029n ± 2%   3.736n ± 1%   -7.27% (p=0.000 n=10)
2         4.048n ± 1%   3.595n ± 1%  -11.17% (p=0.000 n=10)
3         4.352n ± 1%   4.062n ± 1%   -6.64% (p=0.000 n=10)
4         4.485n ± 1%   4.277n ± 1%   -4.64% (p=0.000 n=10)
5         4.969n ± 1%   4.715n ± 1%   -5.10% (p=0.000 n=10)
6         5.800n ± 1%   5.234n ± 0%   -9.77% (p=0.000 n=10)
7         5.729n ± 1%   5.328n ± 0%   -7.00% (p=0.000 n=10)
8         5.390n ± 2%   5.173n ± 1%   -4.02% (p=0.000 n=10)
9         7.328n ± 1%   7.258n ± 1%   -0.95% (p=0.015 n=10)
10        7.362n ± 2%   7.125n ± 0%   -3.22% (p=0.000 n=10)
11        7.335n ± 1%   7.223n ± 1%   -1.51% (p=0.001 n=10)
12        7.068n ± 1%   7.072n ± 1%        ~ (p=0.529 n=10)
13        7.363n ± 2%   7.218n ± 1%   -1.96% (p=0.001 n=10)
14        7.347n ± 2%   7.082n ± 1%   -3.61% (p=0.000 n=10)
15        7.365n ± 1%   7.199n ± 1%   -2.26% (p=0.002 n=10)
geomean   5.843n        5.571n        -4.66%
```

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob
Michael Pratt

Submit Requirements:

Code-Review

No-Unresolved-Comments

Review-Enforcement

TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

satisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

May 8, 2026, 4:28:47 PM (3 days ago) May 8

to goph...@pubsubhelper.golang.org, golang-co...@googlegroups.com

Attention needed from Michael Pratt

Michael Matloob uploaded new patchset

Michael Matloob uploaded patch set #10 to this change.

Open in Gerrit

Related details

Attention is currently required from:

Michael Pratt

Submit Requirements:

Code-Review
No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

unsatisfied_requirement

satisfied_requirement

open

diffy

Michael Pratt (Gerrit)

unread,

5:06 PM (6 hours ago) 5:06 PM

to Michael Matloob, goph...@pubsubhelper.golang.org, Michael Pratt, golang...@luci-project-accounts.iam.gserviceaccount.com, Michael Matloob, golang-co...@googlegroups.com

Attention needed from Michael Matloob

Michael Pratt voted Code-Review+2

Code-Review

+2

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob

Submit Requirements:

Code-Review

No-Unresolved-Comments
Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

satisfied_requirement

unsatisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

5:06 PM (6 hours ago) 5:06 PM

to Michael Matloob, goph...@pubsubhelper.golang.org, Michael Pratt, golang...@luci-project-accounts.iam.gserviceaccount.com, golang-co...@googlegroups.com

Attention needed from Michael Matloob

Michael Matloob voted Code-Review+1

Code-Review

+1

Open in Gerrit

Related details

Attention is currently required from:

Michael Matloob

Submit Requirements:

Code-Review
No-Unresolved-Comments

Review-Enforcement
TryBots-Pass

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

satisfied_requirement

open

diffy

Michael Matloob (Gerrit)

unread,

5:06 PM (6 hours ago) 5:06 PM

to Michael Matloob, goph...@pubsubhelper.golang.org, golang-...@googlegroups.com, Michael Pratt, golang...@luci-project-accounts.iam.gserviceaccount.com, Michael Knyszek, golang-co...@googlegroups.com

Michael Matloob submitted the change

Change information

Commit message:

runtime: consolidate tiny sizespecializedmalloc functions

In the sizespecializedmalloc goexperiment, we specialized the tiny
function per tiny size, so there was a different allocation function per
size from 1-15. This created a lot of functions for a code path that was
not executed that often. From the microbenchmarks, comparing the
consolidated tiny function in this cl with the per-size functions, the
specialized functions could be up to 20% faster, but for 8 byte
allocations, which are almost certainly the most common, the per-size
function was slower.



Look at the change description of CL 766980 for the results of those

microbenchmarks. The CL also contains the code used to run the
benchmark.

Since we've noticed significant icache pressure from all the functions,
and, the tiny functions aren't used as much as the other ones, and the
benefits seem to be mixed, consolidate the 15 functions into a single
function.

This cuts the size of the mallocgc* functions by about 20%.

For #79286

Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64_c2s16-perf_vs_parent-sizespecializedmalloc,gotip-linux-amd64_c3h88-perf_vs_parent-sizespecializedmalloc,gotip-linux-arm64_c4ah72-perf_vs_parent-sizespecializedmalloc,gotip-linux-arm64_c4as16-perf_vs_parent-sizespecializedmalloc,gotip-linux-arm64_c4as16-perf_vs_parent,gotip-linux-arm64_c4ah72-perf_vs_parent,gotip-linux-amd64_c3h88-perf_vs_parent,gotip-linux-amd64_c2s16-perf_vs_parent

Change-Id: I824f65727a858158c14d2edd6fea1e846a6a6964

Reviewed-on: https://go-review.googlesource.com/c/go/+/772540

LUCI-TryBot-Result: golang...@luci-project-accounts.iam.gserviceaccount.com <golang...@luci-project-accounts.iam.gserviceaccount.com>

Reviewed-by: Michael Matloob <mat...@google.com>

Reviewed-by: Michael Pratt <mpr...@google.com>

Files:

M src/cmd/compile/internal/ir/symtab.go
M src/cmd/compile/internal/ssa/rewrite.go
M src/cmd/compile/internal/ssagen/ssa.go
M src/runtime/_mkmalloc/mkmalloc.go
M src/runtime/malloc.go
M src/runtime/malloc_generated.go
M src/runtime/malloc_stubs.go
M src/runtime/malloc_tables_generated.go
M test/codegen/strings.go
M test/live.go
M test/live_regabi.go

Change size: XL

Delta: 11 files changed, 101 insertions(+), 2277 deletions(-)

Branch: refs/heads/master

Submit Requirements:

Code-Review: +1 by Michael Matloob, +2 by Michael Pratt
TryBots-Pass: LUCI-TryBot-Result+1 by golang...@luci-project-accounts.iam.gserviceaccount.com

Open in Gerrit

Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings.

Gerrit

open

diffy

satisfied_requirement

Reply all

Reply to author

Forward

[go] runtime: consolidate tiny sizespecializedmalloc functions

Michael Matloob (Gerrit)

New activity on the change

Related details

Michael Matloob (Gerrit)

Michael Matloob voted Code-Review+1

Related details

Michael Pratt (Gerrit)

Michael Pratt added 6 comments

Related details

Michael Matloob (Gerrit)

Michael Matloob added 3 comments

Related details

Michael Matloob (Gerrit)

Michael Matloob uploaded new patchset

Related details

Michael Matloob (Gerrit)

Michael Matloob added 1 comment

Related details

Michael Matloob (Gerrit)

Michael Matloob uploaded new patchset

Related details

Michael Matloob (Gerrit)

Michael Matloob uploaded new patchset

Related details

Michael Matloob (Gerrit)

Michael Matloob added 1 comment

Related details

Michael Matloob (Gerrit)

Michael Matloob uploaded new patchset

Related details

Michael Pratt (Gerrit)

Michael Pratt voted Code-Review+2

Related details

Michael Matloob (Gerrit)

Michael Matloob voted Code-Review+1

Related details

Michael Matloob (Gerrit)

Michael Matloob submitted the change

Change information