[AMD Official Use Only - Internal Distribution Only]
Hi Daniel,
We're in the process of merging it into develop. It should be in gem5 20.1.
There are a few CLs that need reviews first. Tony and I will be giving a talk
during the gem5 workshop in 2 weeks.
-Matt
-Original Message-
[AMD Public Use]
Hi all,
I would also be interested if anyone has found a solution to this. I currently
cannot test any of the patches I was hoping to finish this week due to this
issue. I am not very familiar with docker, and after about 30 minutes of going
through documentation wasn't
[AMD Public Use]
Hi Bobby,
What is/was the fix for this issue? I also cannot build even though I have
python 3.6 installed (via apt-get on Ubuntu 20.04). I am manually passing
python3-config to scons but I'm getting a syntax error when it sees the
metaclass keyword.
-Matt
From: Bobby
[AMD Public Use]
Hi Jason & Tuan,
Once I have checked in the current tester, I will post a follow up patch to add
DMA threads. I also still need to test the current (GPU) script to make sure it
works with other protocols. At some point this will also include atomics to do
fence tests as
[AMD Public Use]
Hi Evey,
I've seen similar issues in the past caused by protobuf. I was able to
workaround this by passing a non-existent command to the PROTOC variable to
scons, e.g.:
scons build/X86/gem5.opt -j 9 PROTOC=asdf
Due to the way gem5 stores the build variables, I never had to
[AMD Public Use]
The line that has the error should be deleted to fix the MIPS build. I missed
that in the review. I can post a fix if you want but I thought I'd give Gabe a
chance to take a look.
-Matt
From: Jason Lowe-Power via gem5-dev
Sent: Thursday, August 6, 2020 7:36 AM
To: gem5
[AMD Official Use Only - Approved for External Use]
Hi Sampad,
I'm fairly certain the map call is not needed. The scripts I use do not have
this, for example. My guess is this was intended to be
"map(hsapp_gpu_map_vaddr, hsapp_gpu_map_paddr, hsapp_gpu_map_size)". If that
is the case the
[AMD Public Use]
Microops are proprietary. However, the *number* of microops for a given
architecture can be determined with a benchmark executing the instruction
billions/trillions of times and comparing the ratio of microops to instructions
using performance counters. People have done this
[AMD Public Use]
I could try and help, but the compiler-test.sh script does not work for me
locally and I always get the following error (on multiple machines):
Error: No non-leaf 'build' dir found on target path. /gem5
That said, probably deleting the whole if block in
[AMD Public Use]
The defines causing an error seem to be set in the HSA header file itself but
only for x86 type hosts. I'm hesitant to change this since it is pulled
directly, unmodified, from the AMD HSA runtime repo including license, etc.
However, we might be able to do some scons magic to
[AMD Public Use]
This should fix the immediate error (there may be more):
https://gem5-review.googlesource.com/c/public/gem5/+/44785
-Matt
-Original Message-
From: jenkins-no-reply=gem5@mg.gem5.org
On Behalf Of jenkins-no-reply--- via
gem5-dev
Sent: Friday, April 23, 2021 2:57
[AMD Public Use]
Hi Paul,
The trick is to checkpoint "before the GPU starts doing work." That's a bit
vague because I am not sure where that point is on the stable release,
unfortunately. In the new upcoming release for the GPU model it is clear this
point is when the first dispatch packet
[AMD Public Use]
Hi Paul,
Checkpoints (and similarly switchcpu) in GCN3 do not work once the GPU has
launched a kernel because the drain functions are not implemented in the GPU
model. In general, there is a lot of state to be drained for the GPU. If you
drain/serialize between kernels it
[AMD Public Use]
Hi Gabe,
We are seeing similar errors, so +1 priority on this. Some of our changes that
where ready to submit got rerun (for some reason?) and then failed with the
same error. My (weak) theory is that maybe it has something to do with how it
is checking for timeouts and
[AMD Official Use Only]
Hi folks,
I am curious if anyone has had success booting more recent kernels in X86 in
gem5? I am running into a very early CPU soft lock on boot up with Linux 5.11
in X86. I was wondering if anyone has had any luck getting around that either
via gem5 changes,
own%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000=s1PKc18gM1G%2BswH2VIb5nAHooBEQzfJMX5bOj1c5YsM%3D=0>
On Tue, Feb 22, 2022 at 8:20 AM Poremba, Matthew via gem5-dev
mailto:gem5-dev@gem5.org>> wrote:
[AMD Official Use Only]
Yes, I will take a look soon.
-Matt
From: Jason Lowe-Pow
[AMD Official Use Only]
Yes, I will take a look soon.
-Matt
From: Jason Lowe-Power
Sent: Sunday, February 20, 2022 9:18 AM
To: gem5 Developer List ; Poremba, Matthew
Cc: jenkins-no-re...@gem5.org
Subject: Re: [gem5-dev] Build failed in Jenkins: nightly #135
[CAUTION: External Email]
Hey
[AMD Official Use Only - General]
I don't think I have tried pennant on Vega, but the current status (as of
several months ago) is:
Two weekly tests have issues with Vega right now: (1) heterosync with
sleepMutex and (2) lulesh
Two weekly tests have issues running in full-system: (1) sssp and
[AMD Official Use Only - General]
Thanks all, it looks like everything was taken care of over the weekend.
To answer MattS' question: I've tested weekly on most of my changes and haven't
found any issues. I've been primarily making arch-vega changes and the weekly
tester only tests gcn3 by
[AMD Official Use Only - General]
Hi Bobby,
I have seen this issue with do_boot_cpu as well. The short story is I had to
use multiple event queues and set the "sim_quantum" of the Root object to 1e8:
https://gem5-review.googlesource.com/c/public/gem5/+/65131. Note that 1e8 is
different
[AMD Official Use Only - General]
This should fix the build:
https://gem5-review.googlesource.com/c/public/gem5/+/67977
-Matt
-Original Message-
From: jenkins-no-reply=gem5@mg.gem5.org
On Behalf Of jenkins-no-reply--- via
gem5-dev
Sent: Wednesday, February 15, 2023 12:20 AM
To:
[AMD Official Use Only - General]
https://gem5-review.googlesource.com/c/public/gem5/+/71078
-Matt
From: Matt Sinclair
Sent: Wednesday, May 31, 2023 1:22 PM
To: The gem5 Developer List
Cc: Bobby Bruce ; Poremba, Matthew
Subject: Re: [gem5-dev] Re: Build failed in Jenkins: nightly #620
[AMD Official Use Only - General]
Hi Hugo,
You would have to do at least the step you mentioned. As a first step that
should be enough to "see" the device (with lspci for example). If you want to
actually *use* the device though, I suspect you will have to build the ROCk
kernel driver and
23 matches
Mail list logo