Hello, You can try to set with -mcpu a CPU that does not support the SIMD extension at all. There may also be FP instructions that are wrongly (or for simplification purposes) tagged with SIMD opClasses in gem5.
Regards, -- Fernando A. Endo, Post-doc INRIA Rennes-Bretagne Atlantique France 2016-05-02 17:31 GMT+02:00 Ayaz Akram <[email protected]>: > Hi Karthik ! > > Yes, I can see the flag "-mcpu=generic+nosimd " with gcc when benchmark > is compiling ! > > So, it seems that gcc is able to see this flag ! > > Thanks > -Ayaz > > On Sun, May 1, 2016 at 11:14 PM, Karthik Sangaiah <[email protected]> > wrote: > >> Hi Ayaz, >> >> Can you track down any flags used during the benchmark's compilation? >> >> Thanks, >> >> Karthik >> >> On Sun, May 1, 2016 at 12:00 PM, <[email protected]> wrote: >> >>> Send gem5-users mailing list submissions to >>> [email protected] >>> >>> To subscribe or unsubscribe via the World Wide Web, visit >>> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> or, via email, send a message with subject or body 'help' to >>> [email protected] >>> >>> You can reach the person managing the list at >>> [email protected] >>> >>> When replying, please edit your Subject line so it is more specific >>> than "Re: Contents of gem5-users digest..." >>> >>> >>> Today's Topics: >>> >>> 1. SIMD instructions on Arm (Ayaz Akram) >>> 2. Re: SIMD instructions on Arm (Arthur Perais) >>> 3. Re: SIMD instructions on Arm (Ayaz Akram) >>> 4. Re: SIMD instructions on Arm (Arthur Perais) >>> 5. Re: SIMD instructions on Arm (Ayaz Akram) >>> 6. Re: Measuring latency & bandwidth accurately (Abdul Mutaal) >>> >>> >>> ---------------------------------------------------------------------- >>> >>> Message: 1 >>> Date: Sat, 30 Apr 2016 13:58:17 -0400 >>> From: Ayaz Akram <[email protected]> >>> To: gem5 users mailing list <[email protected]> >>> Subject: [gem5-users] SIMD instructions on Arm >>> Message-ID: >>> < >>> cac0+nvblej3941udyf3kfm+pj42ikjmyh90ccsertcmhue_...@mail.gmail.com> >>> Content-Type: text/plain; charset="utf-8" >>> >>> This question is not directly related to gem5, but I am asking here as a >>> lot of community members are using gem5 for Arm simulations. I am trying >>> to compile some SPEC2006 benchmarks for Arm without any simd >>> instructions. >>> I am using -mcpu=generic+nosimd flag but, when I run the benchmarks with >>> gem5, stats still show simd operation types. I wonder if anyone has >>> experienced such problem. >>> >>> Thanks >>> Ayaz >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> URL: < >>> http://m5sim.org/cgi-bin/mailman/private/gem5-users/attachments/20160430/1ac8fa00/attachment-0001.html >>> > >>> >>> ------------------------------ >>> >>> Message: 2 >>> Date: Sat, 30 Apr 2016 20:02:08 +0200 (CEST) >>> From: Arthur Perais <[email protected]> >>> To: gem5 users mailing list <[email protected]> >>> Subject: Re: [gem5-users] SIMD instructions on Arm >>> Message-ID: <[email protected]> >>> Content-Type: text/plain; charset="iso-8859-1" >>> >>> Could the SIMD instructions be in the kernel/libraries rather than in >>> your benchmark (I'm assuming you're using Full System). >>> >>> Best, >>> >>> Arthur. >>> >>> ----- Mail original ----- >>> >>> > De: "Ayaz Akram" <[email protected]> >>> > À: "gem5 users mailing list" <[email protected]> >>> > Envoyé: Samedi 30 Avril 2016 19:58:17 >>> > Objet: [gem5-users] SIMD instructions on Arm >>> >>> > This question is not directly related to gem5, but I am asking here as >>> a lot >>> > of community members are using gem5 for Arm simulations. I am trying to >>> > compile some SPEC2006 benchmarks for Arm without any simd >>> instructions. I am >>> > using -mcpu=generic+nosimd flag but, when I run the benchmarks with >>> gem5, >>> > stats still show simd operation types. I wonder if anyone has >>> experienced >>> > such problem. >>> >>> > Thanks >>> > Ayaz >>> >>> > _______________________________________________ >>> > gem5-users mailing list >>> > [email protected] >>> > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> URL: < >>> http://m5sim.org/cgi-bin/mailman/private/gem5-users/attachments/20160430/6fb033fa/attachment-0001.html >>> > >>> >>> ------------------------------ >>> >>> Message: 3 >>> Date: Sat, 30 Apr 2016 14:05:46 -0400 >>> From: Ayaz Akram <[email protected]> >>> To: gem5 users mailing list <[email protected]> >>> Subject: Re: [gem5-users] SIMD instructions on Arm >>> Message-ID: >>> <CAC0+NVZsgM8_RPK-Qru9timzciAtUv-f2-HuxEK_m= >>> [email protected]> >>> Content-Type: text/plain; charset="utf-8" >>> >>> No, I am using system emulation mode. >>> >>> Thanks >>> >>> On Sat, Apr 30, 2016 at 2:02 PM, Arthur Perais <[email protected]> >>> wrote: >>> >>> > Could the SIMD instructions be in the kernel/libraries rather than in >>> your >>> > benchmark (I'm assuming you're using Full System). >>> > >>> > Best, >>> > >>> > Arthur. >>> > >>> > ------------------------------ >>> > >>> > *De: *"Ayaz Akram" <[email protected]> >>> > *À: *"gem5 users mailing list" <[email protected]> >>> > *Envoyé: *Samedi 30 Avril 2016 19:58:17 >>> > *Objet: *[gem5-users] SIMD instructions on Arm >>> > >>> > This question is not directly related to gem5, but I am asking here as >>> a >>> > lot of community members are using gem5 for Arm simulations. I am >>> trying >>> > to compile some SPEC2006 benchmarks for Arm without any simd >>> instructions. >>> > I am using -mcpu=generic+nosimd flag but, when I run the benchmarks >>> with >>> > gem5, stats still show simd operation types. I wonder if anyone has >>> > experienced such problem. >>> > >>> > Thanks >>> > Ayaz >>> > >>> > _______________________________________________ >>> > gem5-users mailing list >>> > [email protected] >>> > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> > >>> > >>> > >>> > _______________________________________________ >>> > gem5-users mailing list >>> > [email protected] >>> > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> > >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> URL: < >>> http://m5sim.org/cgi-bin/mailman/private/gem5-users/attachments/20160430/e0660b70/attachment-0001.html >>> > >>> >>> ------------------------------ >>> >>> Message: 4 >>> Date: Sat, 30 Apr 2016 20:13:54 +0200 (CEST) >>> From: Arthur Perais <[email protected]> >>> To: gem5 users mailing list <[email protected]> >>> Subject: Re: [gem5-users] SIMD instructions on Arm >>> Message-ID: >>> <[email protected]> >>> Content-Type: text/plain; charset="iso-8859-1" >>> >>> But maybe the library that you are statically linking against when >>> producing your benchmark binary have been compiled using SIMD instructions. >>> Of course if your benchmark does not need any library then this does not >>> apply. >>> >>> Best, >>> >>> Arhtur. >>> >>> ----- Mail original ----- >>> >>> > De: "Ayaz Akram" <[email protected]> >>> > À: "gem5 users mailing list" <[email protected]> >>> > Envoyé: Samedi 30 Avril 2016 20:05:46 >>> > Objet: Re: [gem5-users] SIMD instructions on Arm >>> >>> > No, I am using system emulation mode. >>> >>> > Thanks >>> >>> > On Sat, Apr 30, 2016 at 2:02 PM, Arthur Perais < >>> [email protected] > >>> > wrote: >>> >>> > > Could the SIMD instructions be in the kernel/libraries rather than >>> in your >>> > > benchmark (I'm assuming you're using Full System). >>> > >>> >>> > > Best, >>> > >>> >>> > > Arthur. >>> > >>> >>> > > > De: "Ayaz Akram" < [email protected] > >>> > > >>> > >>> > > > À: "gem5 users mailing list" < [email protected] > >>> > > >>> > >>> > > > Envoyé: Samedi 30 Avril 2016 19:58:17 >>> > > >>> > >>> > > > Objet: [gem5-users] SIMD instructions on Arm >>> > > >>> > >>> >>> > > > This question is not directly related to gem5, but I am asking >>> here as a >>> > > > lot >>> > > > of community members are using gem5 for Arm simulations. I am >>> trying to >>> > > > compile some SPEC2006 benchmarks for Arm without any simd >>> instructions. I >>> > > > am >>> > > > using -mcpu=generic+nosimd flag but, when I run the benchmarks >>> with gem5, >>> > > > stats still show simd operation types. I wonder if anyone has >>> experienced >>> > > > such problem. >>> > > >>> > >>> >>> > > > Thanks >>> > > >>> > >>> > > > Ayaz >>> > > >>> > >>> > > > _______________________________________________ >>> > > >>> > >>> > > > gem5-users mailing list >>> > > >>> > >>> > > > [email protected] >>> > > >>> > >>> > > > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> > > >>> > >>> >>> > > _______________________________________________ >>> > >>> > > gem5-users mailing list >>> > >>> > > [email protected] >>> > >>> > > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> > >>> >>> > _______________________________________________ >>> > gem5-users mailing list >>> > [email protected] >>> > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> URL: < >>> http://m5sim.org/cgi-bin/mailman/private/gem5-users/attachments/20160430/d419edc6/attachment-0001.html >>> > >>> >>> ------------------------------ >>> >>> Message: 5 >>> Date: Sat, 30 Apr 2016 14:23:08 -0400 >>> From: Ayaz Akram <[email protected]> >>> To: gem5 users mailing list <[email protected]> >>> Subject: Re: [gem5-users] SIMD instructions on Arm >>> Message-ID: >>> <CAC0+NVYEO_TYgk=77X9pYenNxmHxC=_ >>> [email protected]> >>> Content-Type: text/plain; charset="utf-8" >>> >>> That might be the case. Actually, I am testing with povray and namd >>> benchmarks from SPEC2006 suite. I am not sure if they are using any such >>> library. >>> >>> Thanks for your help ! >>> >>> >>> On Sat, Apr 30, 2016 at 2:13 PM, Arthur Perais <[email protected]> >>> wrote: >>> >>> > But maybe the library that you are statically linking against when >>> > producing your benchmark binary have been compiled using SIMD >>> instructions. >>> > Of course if your benchmark does not need any library then this does >>> not >>> > apply. >>> > >>> > Best, >>> > >>> > Arhtur. >>> > >>> > ------------------------------ >>> > >>> > *De: *"Ayaz Akram" <[email protected]> >>> > *À: *"gem5 users mailing list" <[email protected]> >>> > *Envoyé: *Samedi 30 Avril 2016 20:05:46 >>> > *Objet: *Re: [gem5-users] SIMD instructions on Arm >>> > >>> > >>> > No, I am using system emulation mode. >>> > >>> > Thanks >>> > >>> > On Sat, Apr 30, 2016 at 2:02 PM, Arthur Perais <[email protected] >>> > >>> > wrote: >>> > >>> >> Could the SIMD instructions be in the kernel/libraries rather than in >>> >> your benchmark (I'm assuming you're using Full System). >>> >> >>> >> Best, >>> >> >>> >> Arthur. >>> >> >>> >> ------------------------------ >>> >> >>> >> *De: *"Ayaz Akram" <[email protected]> >>> >> *À: *"gem5 users mailing list" <[email protected]> >>> >> *Envoyé: *Samedi 30 Avril 2016 19:58:17 >>> >> *Objet: *[gem5-users] SIMD instructions on Arm >>> >> >>> >> This question is not directly related to gem5, but I am asking here >>> as a >>> >> lot of community members are using gem5 for Arm simulations. I am >>> trying >>> >> to compile some SPEC2006 benchmarks for Arm without any simd >>> instructions. >>> >> I am using -mcpu=generic+nosimd flag but, when I run the benchmarks >>> with >>> >> gem5, stats still show simd operation types. I wonder if anyone has >>> >> experienced such problem. >>> >> >>> >> Thanks >>> >> Ayaz >>> >> _______________________________________________ >>> >> gem5-users mailing list >>> >> [email protected] >>> >> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> >> >>> >> >>> >> >>> >> _______________________________________________ >>> >> gem5-users mailing list >>> >> [email protected] >>> >> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> >> >>> > >>> > >>> > _______________________________________________ >>> > gem5-users mailing list >>> > [email protected] >>> > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> > >>> > >>> > >>> > _______________________________________________ >>> > gem5-users mailing list >>> > [email protected] >>> > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> > >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> URL: < >>> http://m5sim.org/cgi-bin/mailman/private/gem5-users/attachments/20160430/01733228/attachment-0001.html >>> > >>> >>> ------------------------------ >>> >>> Message: 6 >>> Date: Sun, 1 May 2016 13:26:43 +0200 >>> From: Abdul Mutaal <[email protected]> >>> To: gem5 users mailing list <[email protected]> >>> Subject: Re: [gem5-users] Measuring latency & bandwidth accurately >>> Message-ID: >>> < >>> calsb4lge6xo0hxcnmo1kpdyad3ygla4l60qor99zhtyc-us...@mail.gmail.com> >>> Content-Type: text/plain; charset="utf-8" >>> >>> Hi Radhika, >>> >>> Thanks for your answer. >>> >>> My system consists of Traffic generator replaying some recorded traces. I >>> would like to measure the latency & bandwidth of my system attached with >>> traffic generator. >>> >>> Just a side question - After running the trace I see that there is stat >>> called 'numRetries' and 'retryTicks' , which are the ticks wasted during >>> back pressure. So my question is, does CommMonitor has the ability to >>> count >>> latency of packet if the packet experiences a back-pressure and retry is >>> being done from traffic generator. >>> >>> Thanks >>> >>> 2016-04-29 9:42 GMT+02:00 Radhika Jagtap <[email protected]>: >>> >>> > Hi Abdul, >>> > >>> > >>> > >>> > What do you mean by certain architecture? As you may know, CommMonitor >>> is >>> > connected between two ports and so it (reliably and accurately) >>> measures >>> > the latency and bandwidth statistics of requests between those two >>> ports. I >>> > you want to measure these as seen by the application, you could >>> connect the >>> > CommMonitor between the core and L1 data cache ports. If you are >>> looking >>> > for characterizing, that is things like maximum bandwidth and static >>> > latency (ie round trip latency per single request is in flight), then >>> > running lmbench benchmarks could be useful . >>> > >>> > >>> > >>> > Radhika >>> > >>> > >>> > >>> > *From:* gem5-users [mailto:[email protected]] *On Behalf Of >>> *Abdul >>> > Mutaal >>> > *Sent:* 25 April 2016 22:20 >>> > *To:* gem5 users mailing list >>> > *Subject:* [gem5-users] Measuring latency & bandwidth accurately >>> > >>> > >>> > >>> > Hi, >>> > >>> > >>> > >>> > How can I accurately measure bandwidth & latency of certain >>> architecture ? >>> > I am currently using CommMonitor. Is that reliable or is there any >>> other >>> > method to measure accurately in gem5. >>> > >>> > >>> > >>> > >>> > >>> > >>> > >>> > Thanks >>> > >>> > >>> > >>> > -- >>> > >>> > >>> > >>> > Regards, >>> > >>> > Abdul Mutaal >>> > >>> > >>> > >>> > >>> > IMPORTANT NOTICE: The contents of this email and any attachments are >>> > confidential and may also be privileged. If you are not the intended >>> > recipient, please notify the sender immediately and do not disclose the >>> > contents to any other person, use it for any purpose, or store or copy >>> the >>> > information in any medium. Thank you. >>> > >>> > _______________________________________________ >>> > gem5-users mailing list >>> > [email protected] >>> > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> > >>> >>> >>> >>> -- >>> >>> Regards, >>> Abdul Mutaal >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> URL: < >>> http://m5sim.org/cgi-bin/mailman/private/gem5-users/attachments/20160501/1ecc97cb/attachment-0001.html >>> > >>> >>> ------------------------------ >>> >>> Subject: Digest Footer >>> >>> _______________________________________________ >>> gem5-users mailing list >>> [email protected] >>> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >>> >>> ------------------------------ >>> >>> End of gem5-users Digest, Vol 118, Issue 1 >>> ****************************************** >>> >> >> >> _______________________________________________ >> gem5-users mailing list >> [email protected] >> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >> > > > _______________________________________________ > gem5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users >
_______________________________________________ gem5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
