Discussion:
[ACTIVITY] 25-29 Jan 2016
Prathamesh Kulkarni
2016-01-31 21:57:29 UTC
Permalink
== This Week ==

* PR43721 (1/10)
- Submitted upstream.
- GIMPLE parts approved by Richard, wants someone else to approve
rtl-expansion bits.

* benchmarking (1/10)
- benchmarking tcwg-310 on cortex-a53, cortex-a57, r1-a12, cortex-a17
- submitted job for INT benchmarks for SPEC2006 for aarch64

* backports (2/10)
- r232688 automated backport build failed for aarch64-linux-gnu on
tcwg-backport-gerrit:
Unrecognized macro 'GERRIT_PATCHSET_REVISION' in '#664-
${GERRIT_PATCHSET_REVISION,length=8}'
Build successful on tcwg-backport-maxim

- r232737, r232631 have at least 4 dependencies.
- wrote bare bones script to find dependencies for revision

* ccp (2/10)
- enhancement to hoist stmt thru phi with all constant args
- patch: http://people.linaro.org/~prathamesh.kulkarni/ccp-mov.diff
- rejected by Richard, since PRE handles this case when it's
trivially profitable and phiprop does similar transform. In general,
it's not profitable to have this transform included.

* Public Holiday (2/10)

* Misc (2/10)
- Meetings
- Slides: short & long term activities

== Next Week ==
- PR43721: look at using define_expand instead of hook, add test-cases
and few nits
- tcwg-319 - analyse with perf
- LTO INT benchmarking for arm, aarch64 and x86
- LTO Slides
Prathamesh Kulkarni
2016-01-31 21:58:09 UTC
Permalink
On 1 February 2016 at 03:27, Prathamesh Kulkarni
Post by Prathamesh Kulkarni
== This Week ==
* PR43721 (1/10)
- Submitted upstream.
- GIMPLE parts approved by Richard, wants someone else to approve
rtl-expansion bits.
* benchmarking (1/10)
- benchmarking tcwg-310 on cortex-a53, cortex-a57, r1-a12, cortex-a17
- submitted job for INT benchmarks for SPEC2006 for aarch64
* backports (2/10)
- r232688 automated backport build failed for aarch64-linux-gnu on
Unrecognized macro 'GERRIT_PATCHSET_REVISION' in '#664-
${GERRIT_PATCHSET_REVISION,length=8}'
Build successful on tcwg-backport-maxim
- r232737, r232631 have at least 4 dependencies.
- wrote bare bones script to find dependencies for revision
* ccp (2/10)
- enhancement to hoist stmt thru phi with all constant args
- patch: http://people.linaro.org/~prathamesh.kulkarni/ccp-mov.diff
- rejected by Richard, since PRE handles this case when it's
trivially profitable and phiprop does similar transform. In general,
it's not profitable to have this transform included.
* Public Holiday (2/10)
* Misc (2/10)
- Meetings
- Slides: short & long term activities
== Next Week ==
- PR43721: look at using define_expand instead of hook, add test-cases
and few nits
- tcwg-319 - analyse with perf
tcwg-310
Post by Prathamesh Kulkarni
- LTO INT benchmarking for arm, aarch64 and x86
- LTO Slides
Pinski, Andrew
2016-01-31 22:59:14 UTC
Permalink
-----Original Message-----
From: linaro-toolchain [mailto:linaro-toolchain-***@lists.linaro.org] On Behalf Of Prathamesh Kulkarni
Sent: Sunday, January 31, 2016 1:58 PM
To: Linaro Toolchain Mailman List <linaro-***@lists.linaro.org>
Subject: Re: [ACTIVITY] 25-29 Jan 2016
Post by Prathamesh Kulkarni
== This Week ==
* PR43721 (1/10)
- Submitted upstream.
- GIMPLE parts approved by Richard, wants someone else to approve
rtl-expansion bits.
* benchmarking (1/10)
- benchmarking tcwg-310 on cortex-a53, cortex-a57, r1-a12, cortex-a17
- submitted job for INT benchmarks for SPEC2006 for aarch64
* backports (2/10)
- r232688 automated backport build failed for aarch64-linux-gnu on
Unrecognized macro 'GERRIT_PATCHSET_REVISION' in '#664-
${GERRIT_PATCHSET_REVISION,length=8}'
Build successful on tcwg-backport-maxim
- r232737, r232631 have at least 4 dependencies.
- wrote bare bones script to find dependencies for revision
* ccp (2/10)
- enhancement to hoist stmt thru phi with all constant args
- patch: http://people.linaro.org/~prathamesh.kulkarni/ccp-mov.diff
- rejected by Richard, since PRE handles this case when it's
trivially profitable and phiprop does similar transform. In general,
it's not profitable to have this transform included.
<Pinski> I don't know if this is a good optimization at all. Because even the RTL level will try to revert it back to the one single call. If it was back propagating simplified statements that is one thing but calls are not the ones which are simplifiable.
Post by Prathamesh Kulkarni
* Public Holiday (2/10)
* Misc (2/10)
- Meetings
- Slides: short & long term activities
== Next Week ==
- PR43721: look at using define_expand instead of hook, add test-cases
and few nits
- tcwg-319 - analyse with perf
tcwg-310
Post by Prathamesh Kulkarni
- LTO INT benchmarking for arm, aarch64 and x86
- LTO Slides
_______________________________________________
linaro-toolchain mailing list
linaro-***@lists.linaro.org
https://lists.linaro.org/mailman/listinfo/linaro-toolchain

Loading...