[gem5-dev] Re: Modeling Golden Cove in Gem5

I am not an expert at the CPU model, but a couple things I can point you two: 1. The X86 O3CPU model is based on the (very old) Alpha 21264 CPU ( https://ieeexplore.ieee.org/document/755465), with some bells and whistles that have partially updated it to try and represent more modern CPUs. For

[gem5-dev] Re: Build failed in Jenkins: nightly #659

Thanks Bobby, appreciate it. Matt On Fri, Jul 7, 2023 at 11:04 AM Bobby Bruce wrote: > This appears to be a Docker failure of some kind. Both the compiler and > nightly tests dropped at the same time. I think the Docker daemon crashed > for some reason > > No tests failed here. The patches

[gem5-dev] Re: Build failed in Jenkins: nightly #659

Hi Bobby, How do I read this output? I wanted to check if Square was still passing, because of the patch Matt P pushed yesterday, but I don't seem to see it. I realize we are replacing this setup shortly, but nonetheless want to make sure bugs aren't slipping in. Thanks, Matt On Fri, Jul 7,

[gem5-dev] Re: Build failed in Jenkins: nightly #620

Agreed, thanks Matt P! Matt On Wed, May 31, 2023 at 3:43 PM Bobby Bruce wrote: > Ah! Thanks Matt P. I didn’t realize this was the fix! Thanks! > > -- > Dr. Bobby R. Bruce > Room 3050, > Kemper Hall, UC Davis > Davis, > CA, 95616 > > web: https://www.bobbybruce.net > > On May 31, 2023, at 1:28

[gem5-dev] Re: Build failed in Jenkins: nightly #620

I’m confused, there weren’t any recent GPU commits, right? So someone other than Matt P or us must have broken things? Matt On Wed, May 31, 2023 at 3:19 PM Bobby Bruce via gem5-dev wrote: > Hey Matt, > > At least one of the reason’s the Nightly tests are failing is a GPU test > failure. You

[gem5-dev] Re: GEM5-GCN-DNNMark get Invalid filter channel number when running: dnnmark_test_VGG, dnnmark_test_alexnet, dnnmark_test_fwd_conv

Thanks, this is helpful. Looking through those old email chains, I don't see any specific resolution to them, unfortunately. I do not have a ton of time to dig into this (end of the semester is keeping me busy) but if you can keep digging I may be able to provide some ideas. First, are you

[gem5-dev] Re: GEM5-GCN-DNNMark get Invalid filter channel number when running: dnnmark_test_VGG, dnnmark_test_alexnet, dnnmark_test_fwd_conv

I have not tried running these specific benchmarks in gem5 personally, so I cannot say for certain what the error is or even if they are expected to run to completion in gem5. But, normally the error you're seeing happens because you have not created the appropriate "cache" files for the GPU

[gem5-dev] Re: Problem on simulating GCN3 GPU: Running DNNMark too slow.

I have not personally tried to run FWD_FC lately -- we do test some of the DNNMark layers in the weekly test, but testing all of them is not practical. Looking at the error though, it appears that the simulated gem5 memory system is not big enough. I would try increasing --mem-size on the

[gem5-dev] Re: Problem on simulating GCN3 GPU: Running DNNMark too slow.

Hi, Trying to answer your various questions: 1. Similar to #2 below, I am unclear what "blocked" means. It sounds like the program is just running, but is slower than you were hoping it would be? If so, unfortunately, this is a well known problem with detailed simulators like gem5 -- they can

[gem5-dev] Re: 回复：[gem5-users] Re: Hi, i'm new to gem5. is there a way to make simulation for a CPU-GPU heterogeneous computing？

Right, you need to specify the values you want for them or don't include them. Matt On Fri, Apr 28, 2023 at 9:45 AM 429442672 <429442...@qq.com> wrote: > Oh， sorry, i got the point. that's the optional args. > > > -- 原始邮件 -- > *发件人:* "The gem5 Developer List"

[gem5-dev] Re: Hi, i'm new to gem5. is there a way to make simulation for a CPU-GPU heterogeneous computing？

1. I am not personally aware of any CPU-FPGA support, but maybe others on this mailing list can chime in if they are aware of it. 2. gem5 already has support for some small ML workloads. For example DNNMark (https://resources.gem5.org/resources/dnn-mark) is already integrated into

[gem5-dev] Re: Hi, i'm new to gem5. is there a way to make simulation for a CPU-GPU heterogeneous computing？

Yes, all of the “GPU” examples posted on gem5-resources do this, for both GCN3 and Vega models. For example, I usually recommend people start with square: https://resources.gem5.org/resources/square You can find many more examples of this on the homepage of gem5 resources too:

[gem5-dev] Re: Build failed in Jenkins: nightly #566

Sounds good, thanks for letting me know. Matt On Wed, Apr 5, 2023 at 12:13 PM Bobby Bruce wrote: > I don’t know the exact cause of this problem but it’s a Docker failure of > some kind and likely not due to any actual test failure. These just happen > very so often on our Jenkin’s setup and

[gem5-dev] Re: Build failed in Jenkins: nightly #566

Hi Bobby, I'm trying to interpret the console output here, as it seems like the replacement policy tests are failing: ... *03:41:45* Test: test-replacement-policy-traces/tree_plru_test1_st-NULL-x86_64-opt-MI_example-MatchStdoutNoPerf Passed*04:02:55* time="2023-04-05T04:02:55-05:00" level=error

[gem5-dev] Re: Build failed in Jenkins: nightly #482

No problem, great. Matt On Fri, Jan 13, 2023 at 8:18 AM Bobby Bruce wrote: > Sorry for the delay. Yesterday I cleared up some space on the server. We > shouldn't see space issue again. > > -- > Dr. Bobby R. Bruce > Room 3050, > Kemper Hall, UC Davis > Davis, > CA, 95616 > > web:

[gem5-dev] Re: Build failed in Jenkins: nightly #482

Hi Bobby, Just checking on this again. Thanks, Matt On Mon, Jan 9, 2023 at 11:21 PM Matt Sinclair wrote: > Bobby do you need me to clean up the machine or open a ticket with IT at > UW: > > *19:22:13* {standard input}: Fatal error: can't write 97 bytes to section > .text of

[gem5-dev] Re: Build failed in Jenkins: nightly #482

Bobby do you need me to clean up the machine or open a ticket with IT at UW: *19:22:13* {standard input}: Fatal error: can't write 97 bytes to section .text of build/ARM/python/_m5/param_TickedObject.o: 'No space left on device' Matt On Mon, Jan 9, 2023 at 7:23 PM jenkins-no-reply--- via

[gem5-dev] Re: Build failed in Jenkins: weekly #99

8462%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C638088767854623756%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C=Wxw5ByN%2BTeiOQC4FczRdTg0RA28U5gsznTB9Ax3VofY%3D=0> > > > > > > On Sun, Jan 8, 2023 at 12:48 AM Matt Sinclair via gem5-dev < >

[gem5-dev] Re: Build failed in Jenkins: weekly #99

: > Thanks for quickly digging into this, Matt! > > On Sat, Jan 7, 2023 at 1:41 PM Matt Sinclair via gem5-dev < > gem5-dev@gem5.org> wrote: > >> I have confirmed that the Pannotia benchmarks (to my surprise) are using >> AMD's cache bypassing flags for some memory access

[gem5-dev] Re: Build failed in Jenkins: weekly #99

I have confirmed that the Pannotia benchmarks (to my surprise) are using AMD's cache bypassing flags for some memory accesses, which Vishnu added support for this week. Good thing the support is added now! But that is why they are failing here -- they hit a corner case Vishnu and I had

[gem5-dev] Re: Build failed in Jenkins: weekly #99

Hi Matt P & Vishnu, It appears something with the GPU support must have broken between your changes this week -- as far as I can tell all of the nightly tests passed when you checked in your commits, but something in the more complex benchmarks (BC in this case) is breaking: gem5.opt:

[gem5-dev] Re: Build failed in Jenkins: weekly #7

This should fix weekly error: https://gem5-review.googlesource.com/c/public/gem5/+/51907 Matt On Fri, Oct 22, 2021 at 11:23 AM jenkins-no-reply--- via gem5-dev < gem5-dev@gem5.org> wrote: > See > > Changes: > > [Bobby R.

[gem5-dev] Re: Build failed in Jenkins: nightly #10

t; Dr. Bobby R. Bruce >> Room 3050, >> Kemper Hall, UC Davis >> Davis, >> CA, 95616 >> >> web: https://www.bobbybruce.net >> >> >> On Wed, Oct 13, 2021 at 10:36 AM Matt Sinclair via gem5-dev < >> gem5-dev@gem5.org> wrote: >> >>

[gem5-dev] Re: Build failed in Jenkins: nightly #10

(Resending as my prior email bounced) Matt On Wed, Oct 13, 2021 at 10:21 AM Matt Sinclair wrote: > I believe the weekly tests will fail until this is merged in: > https://gem5-review.googlesource.com/c/public/gem5/+/51207 > > Jason, I will follow-up with Kyle today about the additional change

[gem5-dev] Re: Build failed in Jenkins: Nightly #205

(Reposting since prior message bounced) As noted by Gabe last night on the review board, this commit: https://gem5-review.googlesource.com/c/public/gem5/+/40216, should fix the failure. Matt On Sun, Jan 31, 2021 at 3:08 AM jenkins-no-reply--- via gem5-dev < gem5-dev@gem5.org> wrote: > See < >

[gem5-dev] Re: Testing fail on GCN3_X86 on Arm machine

Hi Nathanael, I have personally never tested GCN3 on an ARM machine, so I can't say definitively if it will work there or not. But if you are not, I strongly recommend testing anything you need to test with GCN3 inside the docker we created: -

[gem5-dev] Re: Build failed in Jenkins: Nightly #162

Sounds good, thanks Bobby! Matt On Fri, Dec 18, 2020 at 11:02 AM Bobby Bruce wrote: > I think the Jenkins might have an older Docker image. I'll fix this. > > -- > Dr. Bobby R. Bruce > Room 2235, > Kemper Hall, UC Davis > Davis, > CA, 95616 > > web: https://www.bobbybruce.net > > Please

[gem5-dev] Re: Build failed in Jenkins: Nightly #162

Hi Bobby, Looks like this failed because the GCN3 stuff requires the older version of Python. I know we've discussed this before and I believe Kyle/Matt P. had pushed a patch to deal with this, so I'm wondering if perhaps the way you added the GCN3 stuff to the regression is missing that patch?

[gem5-dev] Re: Build failed in Jenkins: Compiler-Checks #33

(Resending so message doesn't bounce on mailing list) Hi Bobby, Can you point us to what you would want in testlibs? Matt On Mon, Nov 30, 2020 at 2:38 PM Bobby Bruce via gem5-dev wrote: > @Kyle: GCN3 is only built weekly via the compiler tests. Compiling every > gem5 variant each time we run

[gem5-dev] Re: AMD GCN3 - HSA Memory Mapping

Hi Sampad, I don't know the answer to this question directly, but I'm CC'ing the AMD folks who hopefully can provide the answer. I mentioned this to Matt P earlier and he thinks it may be a bug. Thanks, Matt On Fri, Jul 31, 2020 at 10:47 AM Sampad Mohapatra via gem5-dev < gem5-dev@gem5.org>

[gem5-dev] Re: AMD GCN3 on Research Cluster. Toolchain Compilation Help.

The AMD staging branch is in the process of being merged into develop. The two patches that update that file in particular are the next ones to be merged in and are currently going through review: - https://gem5-review.googlesource.com/c/public/gem5/+/30274 -

[gem5-dev] Re: AMD GCN3 on Research Cluster. Toolchain Compilation Help.

I recommend starting here: http://www.gem5.org/2020/05/27/modern-gpu-applications.html and here: http://www.gem5.org/documentation/general_docs/gpu_models/GCN3 (old version may also be useful: http://old.gem5.org/GPU_Models.html). The above includes information about the required driver versions

[gem5-dev] Re: AMD GPU cache stats correctness

I don't know much about the HSAIL model, and it was deprecated a while ago in favor of the more up-to-date GCN3 model (which is in the process of being merged into the develop branch, and is also fully available on its own staging branch). If HSAIL is the same as GCN3 though, there weren't any

[gem5-dev] Re: GCN3 Integration

To add onto what Matt said, Kyle Roarty and I will also be giving a talk in the gem5 workshop about the docker support we've added for the GPU model. We are also working on merging that into develop currently. Matt S. On Mon, May 18, 2020 at 1:57 PM Poremba, Matthew via gem5-dev <