Hello.
I've got a patch candidate that leverages partial linking for a couple of
selected object files.
I'm sending make all-host- jX results for my machine:
before: 3m18s (user 32m52s)
https://gist.githubusercontent.com/marxin/223890df4d8d8e490b6b2918b77dacad/raw/1dd5eae5001295ba0230a689f7edc
On Thu, May 20, 2021 at 2:34 PM Martin Liška wrote:
>
> Hello.
>
> I've got a patch candidate that leverages partial linking for a couple of
> selected object files.
>
> I'm sending make all-host- jX results for my machine:
>
> before: 3m18s (user 32m52s)
> https://gist.githubusercontent.com/marx
On 5/20/21 2:54 PM, Richard Biener wrote:
So why did you go from applying this per-file to multiple files?
When I did per-file for {gimple,generic}-match.c I hit the following issue with
lto.priv symbols:
/usr/lib64/gcc/x86_64-suse-linux/10/../../../../x86_64-suse-linux/bin/ld:
error: libbac
On Thu, May 20, 2021 at 3:06 PM Martin Liška wrote:
>
> On 5/20/21 2:54 PM, Richard Biener wrote:
> > So why did you go from applying this per-file to multiple files?
>
> When I did per-file for {gimple,generic}-match.c I hit the following issue
> with lto.priv symbols:
>
> /usr/lib64/gcc/x86_64-
On Thu, May 20, 2021 at 3:16 PM Richard Biener
wrote:
>
> On Thu, May 20, 2021 at 3:06 PM Martin Liška wrote:
> >
> > On 5/20/21 2:54 PM, Richard Biener wrote:
> > > So why did you go from applying this per-file to multiple files?
> >
> > When I did per-file for {gimple,generic}-match.c I hit the
> On Thu, May 20, 2021 at 3:16 PM Richard Biener
> wrote:
> >
> > On Thu, May 20, 2021 at 3:06 PM Martin Liška wrote:
> > >
> > > On 5/20/21 2:54 PM, Richard Biener wrote:
> > > > So why did you go from applying this per-file to multiple files?
> > >
> > > When I did per-file for {gimple,generic}
On 5/20/21 5:55 PM, Jan Hubicka wrote:
Quick solution is to also modify partitioner to use the local symbol
names when doing incremental linking (those mixing in source code and
random seeds) to avoid clashes.
Good hint. I added hash based on object file name (I don't want to handle
proper stri
On 5/20/21 2:54 PM, Richard Biener wrote:
On Thu, May 20, 2021 at 2:34 PM Martin Liška wrote:
Hello.
I've got a patch candidate that leverages partial linking for a couple of
selected object files.
I'm sending make all-host- jX results for my machine:
before: 3m18s (user 32m52s)
https://gi
On Fri, May 21, 2021 at 5:25 AM Martin Liška wrote:
>
> On 5/20/21 2:54 PM, Richard Biener wrote:
> > On Thu, May 20, 2021 at 2:34 PM Martin Liška wrote:
> >>
> >> Hello.
> >>
> >> I've got a patch candidate that leverages partial linking for a couple of
> >> selected object files.
> >>
> >> I'm
On 5/21/21 2:35 PM, David Edelsohn wrote:
Please remember that not all targets support LTO so a fallback to a
non-partial-LTO build needs to be provided and automatically invoked
for those targets.
Sure, for now it's definitely going to be a opt-in, enabled by something like:
make PARTIAL_LTO=1
PING^1
On 5/21/21 10:29 AM, Martin Liška wrote:
On 5/20/21 5:55 PM, Jan Hubicka wrote:
Quick solution is to also modify partitioner to use the local symbol
names when doing incremental linking (those mixing in source code and
random seeds) to avoid clashes.
Good hint. I added hash based on ob
@Richi: Can you please reply to this email?
On 5/21/21 10:43 AM, Martin Liška wrote:
On 5/20/21 2:54 PM, Richard Biener wrote:
On Thu, May 20, 2021 at 2:34 PM Martin Liška wrote:
Hello.
I've got a patch candidate that leverages partial linking for a couple of
selected object files.
I'm se
On Tue, Jun 1, 2021 at 9:33 AM Martin Liška wrote:
>
> @Richi: Can you please reply to this email?
Not sure what I should add here? Honza suggested to mangle the
promoted symbol names. I don't
really like the idea to compile multiple TUs into one object. Also
+LTO_LINKER_FLAGS = -flto=auto --
On Tue, Jun 1, 2021 at 1:25 PM Martin Liška wrote:
>
> On 6/1/21 9:42 AM, Richard Biener wrote:
> > On Tue, Jun 1, 2021 at 9:33 AM Martin Liška wrote:
> >>
> >> @Richi: Can you please reply to this email?
> >
> > Not sure what I should add here? Honza suggested to mangle the
> > promoted symbol
On 6/1/21 9:42 AM, Richard Biener wrote:
On Tue, Jun 1, 2021 at 9:33 AM Martin Liška wrote:
@Richi: Can you please reply to this email?
Not sure what I should add here? Honza suggested to mangle the
promoted symbol names.
Sure and I sent a patch for that.
I don't
really like the idea to
Hi, all.
Please CC me when I am mentioned into a mail.
On Thu, 2021-05-20 at 15:16 +0200, Richard Biener via Gcc wrote:
> On Thu, May 20, 2021 at 3:06 PM Martin Liška wrote:
> >
> > On 5/20/21 2:54 PM, Richard Biener wrote:
> > > So why did you go from applying this per-file to multiple files?
On 5/21/21 10:29 AM, Martin Liška wrote:
On 5/20/21 5:55 PM, Jan Hubicka wrote:
Quick solution is to also modify partitioner to use the local symbol
names when doing incremental linking (those mixing in source code and
random seeds) to avoid clashes.
Good hint. I added hash based on object fil
PING^2
@Honza: Can you please review the change?
Martin
On 6/23/21 3:53 PM, Martin Liška wrote:
On 5/21/21 10:29 AM, Martin Liška wrote:
On 5/20/21 5:55 PM, Jan Hubicka wrote:
Quick solution is to also modify partitioner to use the local symbol
names when doing incremental linking (those mix
On 8/16/21 3:58 PM, Martin Liška wrote:
PING^2
@Honza: Can you please review the change?
I've tested the patch and apparently it's not enough for
{gimple,generic}-match.o not clashing
in symbol names. Apparently there are more IPA clones that collide.
Leaving that for now.
Martin
Martin
On 6/1/21 3:19 PM, Richard Biener wrote:
On Tue, Jun 1, 2021 at 1:25 PM Martin Liška wrote:
On 6/1/21 9:42 AM, Richard Biener wrote:
On Tue, Jun 1, 2021 at 9:33 AM Martin Liška wrote:
@Richi: Can you please reply to this email?
Not sure what I should add here? Honza suggested to mangle
> Good hint. I added hash based on object file name (I don't want to handle
> proper string escaping) and -frandom-seed.
>
> What do you think about the patch?
Sorry for taking so long - I remember I was sending reply earlier but it
seems I only wrote it and never sent.
> Thanks,
> Martin
> From
21 matches
Mail list logo