Benchmark infrastructure - TCWG-360 [5/10]
* More thinking/prototyping sufficiently-secure Jenkins benchmarking
* Converting LAVA end into multinode job
Benchmarking presentation [2/10]
* A couple of helpful discussions
* Read a couple of helpful docs
Misc [3/10]
=Plan=
* Complete multinode job
* Settle on a plan for Jenkins
* Redraft presentation
== This Week ==
* TCWG-777 (4/10)
- Resolved ICE caused by pass during gcc build but hit another ICE:
http://pastebin.com/RUAY6scB
- Current pass state: http://pastebin.com/AGXnSkrZ
- For test-case:
void f(int flags)
{
void foo(void);
if (flags & 1)
foo();
}
- temporaries don't exist for -O1
- for -O2 temps introduced by peephole2 due to define_peephole2
pattern in thumb2.md:1540
http://pastebin.com/3rEF8Te4
So this intentionally transforms rtx from
zeroextractsi_compare0_scratch to rtx from shiftsi3_compare0_scratch.
Why is it beneficial to do this transform ?
- Looking into combine pass
- For above test-case works with -marm for -O2.
* TCWG-830 (3/10)
- trying to understand vect dump
- untested patch: http://pastebin.com/K4UX5iYz
* Misc (2/10)
- Started looking at TCWG-835, loop vectorized on x86 but not arm
- Committed fix to segfault on -dx
- Conference calls
== Next Week ==
- Continue with TCWG-777, TCWG-830, TCWG-835
- Travel to Mumbai on 14th July (Tuesday) for US Visa Interview with
US Consulate.
Hospital and physio (2/10)
== Progress ==
o Upstream GCC (2/10)
* More work on ongoing patches
o Linaro GCC release (3/10)
* Reviewed patches for tcwg-release script
* Looked at validation issues and redo backports for 5.1
o Misc (3/10)
* Various meetings
* Upstream libunwind support
== Plan ==
- Continue ongoing tasks
== Progress ==
LLDB development
-- Debugging problems with process launch and debugserver crash on remote
connection. [TCWG-855] [6/10]
-- Caught the notorious issue mentioned above fix can be found here
http://reviews.llvm.org/D11129.
-- Figured arm lldb and lldbserver host builds on chromebook will put
steps on collaborate LLDB page soon. [1/10]
Miscellaneous [3/10]
-- Travel to Islamabad for Czech Republic visa
-- Meetings, emails, discussions etc.
== Plan ==
LLDB development
-- Follow up review process for process launch bug fix.
-- Run testsuite for armhf and figure out issues to fix.
-- Submit patches to fix build on older version of gcc.
Eid Holidays 17th to 21st July 2015
== This week ==
* TCWG-146 - Detect smin/umin idion (0/10)
- Waiting for upstream approval/review
* TCWG-140 - Transform end of loop conditions to min_expr (1/10)
- Blocked waiting on validation
* TCWG-833 - Exploit Wide Add operations when appropriate (8/10)
- Mere detailed investigation
- Working theory is to develop wide add rtl patterns that
incorporate vec_unpack to widen 16-bit to 32-bit
* Misc (1/10)
- Conference calls
- Conference call with Charles and Prathamesh to discuss
autovectorization progress
== Next week ==
- Validate patch for TCWG-140
- Develop patch for TCWG-833
- Validate TCWG-833 if successful patch is developed
- Investigate Aarch64 implementation
== Progress ==
* Add REG_EQUAL note for arm_emit_movpair (1/10)
- committed patch1 after testing again
* Factor conversion out of COND_EXPR - TCWG-849 (5/10)
- Gone through couple of iterations and committed the patch
- There are still some improvements need as follow up patches
* TACT -TCWG-851 (2/10)
- Started looking into spec2k
* Misc (2/10)
- Looked at interaction between gcc optimization passes
- gcc-patches, gcc-bugs list
- Meetings
== Plan ==
- GCC Bugs
- TACT driven optimization exploration for gcc
- Linaro bug 1318
== Progress ==
* Releases (CARD-1431 1/10)
- Released 3.6.2-final
* Maintenance (CARD-1833 5/10)
- Reducing runtime of some benchmarks in LLVM's
test-suite by getting rid of millions of useless
fprintf calls.
- Working on https://llvm.org/PR20700 some more
* Background (4/10)
- Code review, meetings, discussions, etc.
- Long TargetTuple review (D10969) / discussions
- Replacing broken buildbot USB disks (need to buy more)
- Bisecting self-hosting bot breakage
- Testing patches for ARM
- Jira farming
== Plan ==
* continue PR20700
* continue review/discussion of TargetTuple
* look again at PR20757
* maybe look at PR21000
* Off Monday (2/10)
== Progress ==
* published linaro-4.8 and 4.9 2015.06 releases
* linaro-5.1-2015.07 (1/10)
- backport reviews
- updated my helper script for reviews for cope with the git-only branches
* upstream (1/10)
- started looking at vget_lane Neon intrinsic failure on armeb
* infra/release/backports (2/10)
- reviews
* Misc (4/10)
- meetings, conf-calls, emails
== Next ==
* Off Tuesday
* backports, release, validation: update doc
* backports, reviews
* upstream work
== Later ==
* Off July 18th-Aug 3rd
Hi Linaro Toolchain Group,
I am comparing execution time (run time) of sin() trigonometric function
between following glibc (including libm) libraries for aarch64 (juno cortex
a57) :
Linaro glibc 2.19, Linaro eglibc 2.19, eglibc 2.19 (from
http://www.eglibc.org/) and Linaro glibc 2.21.
My observation for execution time of sin():
with Linaro glibc 2.19 and eglibc 2.19 = 1m24.703s (approx)
whereas,
with Linaro eglibc 2.19 & Linaro glibc 2.21 = 0m25.243s (approx)
Has Linaro optimized the libm functions for aarch64 in Linaro eglibc 2.19 ?
If yes, please point me to relevant reference from where I can find more
information on them.
Since the eglibc development from version 2.19 has stopped, will Linaro
maintain its own development version of glibc ?
I am using below snippet code and linux 'time' command to calculate the
time.
void sin_func(void)
{
double incr = 0.732;
double result, count = 0.0;
printf("%s\n", __func__);
while (count < 105414350.0) {
result = sin(count);
count += incr;
}
}
Thanks.
--
with regards,
Virendra Kumar Pathak
