Chandler Carruth via llvm-dev
2017-May-30 07:58 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev < llvm-dev at lists.llvm.org> wrote:> I will run it on Cyclone/AArch64 next week. >FYI, we're still waiting on these Adam... -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170530/7cb390ca/attachment.html>
Flamedoge via llvm-dev
2017-May-30 08:14 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
Thank you for running these. May I suggest testing on AVX2 capable hardware? That would be Intel Haswell, AMD Carrizo and up. I'm not sure what "vectorizer-maximize-bandwidth" implies, but doubling the vector lanes may help light up parallel regions. Kevin On Tue, May 30, 2017 at 12:58 AM, Chandler Carruth via llvm-dev < llvm-dev at lists.llvm.org> wrote:> On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev < > llvm-dev at lists.llvm.org> wrote: > >> I will run it on Cyclone/AArch64 next week. >> > > FYI, we're still waiting on these Adam... > > _______________________________________________ > LLVM Developers mailing list > llvm-dev at lists.llvm.org > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev > >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170530/829591a3/attachment.html>
Chandler Carruth via llvm-dev
2017-May-30 08:27 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
If you care about such hardware, please run benchmarks with the flag? Dehao has made this flag available. It is important that those who care about particular hardware provid ebenchmark results. Not everyone in the community will have access to particular hardware variants. On Tue, May 30, 2017 at 1:14 AM Flamedoge <code.kchoi at gmail.com> wrote:> Thank you for running these. > > May I suggest testing on AVX2 capable hardware? That would be Intel > Haswell, AMD Carrizo and up. > I'm not sure what "vectorizer-maximize-bandwidth" implies, but doubling > the vector lanes may help light up parallel regions. > > Kevin > > On Tue, May 30, 2017 at 12:58 AM, Chandler Carruth via llvm-dev < > llvm-dev at lists.llvm.org> wrote: > >> On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev < >> llvm-dev at lists.llvm.org> wrote: >> >>> I will run it on Cyclone/AArch64 next week. >>> >> >> FYI, we're still waiting on these Adam... >> >> _______________________________________________ >> LLVM Developers mailing list >> llvm-dev at lists.llvm.org >> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev >> >>-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170530/531b6e18/attachment.html>
Agabaria, Mohammed via llvm-dev
2017-May-30 08:39 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
We’re seeing nice improvements but also significant degradations on IA, which we would like to investigate before the patch is committed. Major degradations we see: networking ip_pktcheckb1m -6.80 % ip_pktcheckb2m -6.74 % ip_pktcheckb4m -7.57 % ip_pktcheckb512k -6.58 % Telecom autcor00data_1 -78.02 % autcor00data_2 -76.80 % autcor00data_3 -77.00 % (on Atom) We still working on creating reproducers. In general we support this patch, just want to have a chance to investigate the issues. We need a few days for that. From: llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org] On Behalf Of Chandler Carruth via llvm-dev Sent: Tuesday, May 30, 2017 11:27 To: Flamedoge <code.kchoi at gmail.com<mailto:code.kchoi at gmail.com>>; Dehao Chen <dehao at google.com<mailto:dehao at google.com>> Cc: llvm-dev <llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org>> Subject: Re: [llvm-dev] Enable vectorizer-maximize-bandwidth by default? If you care about such hardware, please run benchmarks with the flag? Dehao has made this flag available. It is important that those who care about particular hardware provid ebenchmark results. Not everyone in the community will have access to particular hardware variants. On Tue, May 30, 2017 at 1:14 AM Flamedoge <code.kchoi at gmail.com<mailto:code.kchoi at gmail.com>> wrote: Thank you for running these. May I suggest testing on AVX2 capable hardware? That would be Intel Haswell, AMD Carrizo and up. I'm not sure what "vectorizer-maximize-bandwidth" implies, but doubling the vector lanes may help light up parallel regions. Kevin On Tue, May 30, 2017 at 12:58 AM, Chandler Carruth via llvm-dev <llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org>> wrote: On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev <llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org>> wrote: I will run it on Cyclone/AArch64 next week. FYI, we're still waiting on these Adam... _______________________________________________ LLVM Developers mailing list llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev --------------------------------------------------------------------- Intel Israel (74) Limited This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170530/163d3d2d/attachment.html>
Agabaria, Mohammed via llvm-dev
2017-May-30 09:00 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
BTW, we also tested on AVX2, and saw a few smaller degradations there: denbench cjpegv2data1 -5.02 % cjpegv2data2 -4.29 % cjpegv2data3 -4.89 % cjpegv2data4 -5.44 % cjpegv2data5 -4.80 % cjpegv2data6 -3.98 % cjpegv2data7 -5.48 % coremark-pro cjpeg-rose7-preset -4.91 % core -6.04 % telecom autcor00data_1 -2.00 % From: Agabaria, Mohammed Sent: Tuesday, May 30, 2017 11:40 To: 'Flamedoge' <code.kchoi at gmail.com>; 'Dehao Chen' <dehao at google.com> Cc: 'llvm-dev' <llvm-dev at lists.llvm.org> Subject: RE: [llvm-dev] Enable vectorizer-maximize-bandwidth by default? We’re seeing nice improvements but also significant degradations on IA, which we would like to investigate before the patch is committed. Major degradations we see: networking ip_pktcheckb1m -6.80 % ip_pktcheckb2m -6.74 % ip_pktcheckb4m -7.57 % ip_pktcheckb512k -6.58 % Telecom autcor00data_1 -78.02 % autcor00data_2 -76.80 % autcor00data_3 -77.00 % (on Atom) We still working on creating reproducers. In general we support this patch, just want to have a chance to investigate the issues. We need a few days for that. From: llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org] On Behalf Of Chandler Carruth via llvm-dev Sent: Tuesday, May 30, 2017 11:27 To: Flamedoge <code.kchoi at gmail.com<mailto:code.kchoi at gmail.com>>; Dehao Chen <dehao at google.com<mailto:dehao at google.com>> Cc: llvm-dev <llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org>> Subject: Re: [llvm-dev] Enable vectorizer-maximize-bandwidth by default? If you care about such hardware, please run benchmarks with the flag? Dehao has made this flag available. It is important that those who care about particular hardware provid ebenchmark results. Not everyone in the community will have access to particular hardware variants. On Tue, May 30, 2017 at 1:14 AM Flamedoge <code.kchoi at gmail.com<mailto:code.kchoi at gmail.com>> wrote: Thank you for running these. May I suggest testing on AVX2 capable hardware? That would be Intel Haswell, AMD Carrizo and up. I'm not sure what "vectorizer-maximize-bandwidth" implies, but doubling the vector lanes may help light up parallel regions. Kevin On Tue, May 30, 2017 at 12:58 AM, Chandler Carruth via llvm-dev <llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org>> wrote: On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev <llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org>> wrote: I will run it on Cyclone/AArch64 next week. FYI, we're still waiting on these Adam... _______________________________________________ LLVM Developers mailing list llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev --------------------------------------------------------------------- Intel Israel (74) Limited This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170530/9ff19e77/attachment.html>
Adam Nemet via llvm-dev
2017-May-30 16:16 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
> On May 30, 2017, at 12:58 AM, Chandler Carruth <chandlerc at gmail.com> wrote: > > On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev <llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>> wrote: > I will run it on Cyclone/AArch64 next week. > > FYI, we're still waiting on these Adam…I was going to test spec but this does not seem to trigger on spec according to Dehao, so there is really no reason for me to test this. We have some spec perf bots that test trunk; if there is some unexpected regression we should pick it up. Sorry for not being explicit about this. Adam -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170530/e197d7ef/attachment.html>
Chandler Carruth via llvm-dev
2017-May-30 20:04 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
On Tue, May 30, 2017 at 9:17 AM Adam Nemet <anemet at apple.com> wrote:> On May 30, 2017, at 12:58 AM, Chandler Carruth <chandlerc at gmail.com> > wrote: > > On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev < > llvm-dev at lists.llvm.org> wrote: > >> I will run it on Cyclone/AArch64 next week. >> > > FYI, we're still waiting on these Adam… > > > I was going to test spec but this does not seem to trigger on spec > according to Dehao, so there is really no reason for me to test this. We > have some spec perf bots that test trunk; if there is some unexpected > regression we should pick it up. Sorry for not being explicit about this. >Ok, if you're fine with that, cool. Only reason I asked was because I wasn't sure that it fundamentally didn't fire on SPEC or just didn't with the particular cost model and might fire/behave differently with a different cost model. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170530/98a4ea1b/attachment.html>
Chandler Carruth via llvm-dev
2017-May-30 20:07 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
On Tue, May 30, 2017 at 1:40 AM Agabaria, Mohammed via llvm-dev < llvm-dev at lists.llvm.org> wrote:> We’re seeing nice improvements but also significant degradations on IA, > which we would like to investigate before the patch is committed. > > > > Major degradations we see: > > > > networking > > ip_pktcheckb1m -6.80 % > > ip_pktcheckb2m -6.74 % > > ip_pktcheckb4m -7.57 % > > ip_pktcheckb512k -6.58 % > > Telecom > > autcor00data_1 -78.02 % > > autcor00data_2 -76.80 % > > autcor00data_3 -77.00 % > > (on Atom) > > > > We still working on creating reproducers. > > In general we support this patch, just want to have a chance to > investigate the issues. We need a few days for that. >I mean, OK... but keep in mind that Dehao's original email went out over a week ago, so this patch has already been held up a while. As these benchmarks aren't readily available, we also can't do anything to help until a test case is posted. Have you considered contributing these benchmarks to the LLVM test suite?> > > > > > > *From:* llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org > <llvm-dev-bounces at lists.llvm.org>] *On Behalf Of *Chandler Carruth via > llvm-dev > *Sent:* Tuesday, May 30, 2017 11:27 > *To:* Flamedoge <code.kchoi at gmail.com>; Dehao Chen <dehao at google.com> > *Cc:* llvm-dev <llvm-dev at lists.llvm.org> > *Subject:* Re: [llvm-dev] Enable vectorizer-maximize-bandwidth by default? > > > > If you care about such hardware, please run benchmarks with the flag? > > > > Dehao has made this flag available. It is important that those who care > about particular hardware provid ebenchmark results. Not everyone in the > community will have access to particular hardware variants. > > > > On Tue, May 30, 2017 at 1:14 AM Flamedoge <code.kchoi at gmail.com> wrote: > > Thank you for running these. > > > > May I suggest testing on AVX2 capable hardware? That would be Intel > Haswell, AMD Carrizo and up. > > I'm not sure what "vectorizer-maximize-bandwidth" implies, but doubling > the vector lanes may help light up parallel regions. > > > > Kevin > > > > On Tue, May 30, 2017 at 12:58 AM, Chandler Carruth via llvm-dev < > llvm-dev at lists.llvm.org> wrote: > > On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev < > llvm-dev at lists.llvm.org> wrote: > > I will run it on Cyclone/AArch64 next week. > > > > FYI, we're still waiting on these Adam... > > > > _______________________________________________ > LLVM Developers mailing list > llvm-dev at lists.llvm.org > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev > > --------------------------------------------------------------------- > Intel Israel (74) Limited > > This e-mail and any attachments may contain confidential material for > the sole use of the intended recipient(s). Any review or distribution > by others is strictly prohibited. If you are not the intended > recipient, please contact the sender and delete all copies. > _______________________________________________ > LLVM Developers mailing list > llvm-dev at lists.llvm.org > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170530/fa12519f/attachment.html>
Nema, Ashutosh via llvm-dev
2017-Jun-01 11:47 UTC
[llvm-dev] Enable vectorizer-maximize-bandwidth by default?
Hi, We enabled “vectorizer-maximize-bandwidth” and ran SPEC CPU2006 (base,rate) on Ryzen 8 core, 16 copies with below config: Base: -m64 -O3 -march=znver1 -mavx2 Base + VMB: -m64 -O3 -march=znver1 -mavx2 -mllvm -vectorizer-maximize-bandwidth There’s a small uplift for gcc and some small regression for sjeng. Others are within noise levels. CPU2006 Results: Benchmark Base/(Base + VMB) 400.perlbench 1 401.bzip2 1 403.gcc 1.01517 429.mcf 1.00222 445.gobmk 1 456.hmmer 1 458.sjeng 0.98641 462.libquantum 1 464.h264ref 1.01005 471.omnetpp 1.00187 473.astar 1 483.xalancbmk 1.00149 433.milc 1 444.namd 1 447.dealII 1 450.soplex 1 453.povray 1.00515 470.lbm 1 482.sphinx3 1 * Ratio more than 1 indicates vectorizer-maximize-bandwidth is improving the performance * Ratio less than 1 indicates vectorizer-maximize-bandwidth is regressing the performance. * Ratio 1 indicates no change. Regards, Ashutosh From: llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org] On Behalf Of Adam Nemet via llvm-dev Sent: Tuesday, May 30, 2017 9:47 PM To: Chandler Carruth <chandlerc at gmail.com> Cc: llvm-dev <llvm-dev at lists.llvm.org> Subject: Re: [llvm-dev] Enable vectorizer-maximize-bandwidth by default? On May 30, 2017, at 12:58 AM, Chandler Carruth <chandlerc at gmail.com<mailto:chandlerc at gmail.com>> wrote: On Fri, May 19, 2017 at 4:01 PM Adam Nemet via llvm-dev <llvm-dev at lists.llvm.org<mailto:llvm-dev at lists.llvm.org>> wrote: I will run it on Cyclone/AArch64 next week. FYI, we're still waiting on these Adam… I was going to test spec but this does not seem to trigger on spec according to Dehao, so there is really no reason for me to test this. We have some spec perf bots that test trunk; if there is some unexpected regression we should pick it up. Sorry for not being explicit about this. Adam -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170601/aa985876/attachment.html>