[gem5-users] X86 Elastic Trace Replay
Micron Confidential I am trying to replay an elastic trace taken with an X86 system (built similarly as in fs.py - using common.FSConfig.makeLinuxX86System). However, when I try to replay the trace, using the same system, but switching to the TraceCPU, I get the following assertion error: gem5.opt: build/X86/sim/system.hh:133: const System::Threads::Thread& System::Threads::thread(ContextID) const: Assertion `id < size()' failed. In looking at the example etrace_replay.py, I see that, by contrast, the system is constructed much more simply - using the System() constructor directly. However, I was wondering if it is possible to make use of the common.FSConfig.makeLinuxX86System function (which uses makeX86System) as it does a lot with regard to setting up the bus hierarchy correctly. Thanks, Patrick Micron Confidential ___ gem5-users mailing list -- gem5-users@gem5.org To unsubscribe send an email to gem5-users-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
[gem5-users] gem5 v20.1.0.2 Hotfix Release
Dear all, We've fixed two bugs as part of a hotfix release for gem5 v20.1. A "ValueError: invalid literal for int() with base..." error was being thrown in certain circumstances due to a non-integer being passed to "MemorySize" via a division. This has been fixed. An assertion in Stats was being triggered due to a name collision between two ThreadStateStats Objects. This has also been fixed. As always, we advise users to pull the latest release of gem5. Kind regards, Bobby -- Dr. Bobby R. Bruce Room 2235, Kemper Hall, UC Davis Davis, CA, 95616 web: https://www.bobbybruce.net ___ gem5-users mailing list -- gem5-users@gem5.org To unsubscribe send an email to gem5-users-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
[gem5-users] Re: How to use custom gpu models with gem5 ?
Hi, maybe you can check this work: https://github.com/ntampouratzis/ACSIM . I believe that you may contact the author directly as I know it is still being developed (although not publicly available at the moment). Best regards, Andreas On Mon, Nov 16, 2020 at 11:45 AM hgedek--- via gem5-users < gem5-users@gem5.org> wrote: > Hi again, > > Thanks for replies and I learned so much about gem5. I was trying gem5 for > a while to learn it. I have one question: is se mode or fs mode more > suitable to test tlm & gpus & etc... ? Our own testing environment based on > qemu and it runs ubuntu. So we can install gpu driver and test our hw model > using tlm & systemc. I tested gem5 with ubuntu in fs mode but booting it > took a huge time ( x86 ). I tried multiple times but I couldn't succeed. So > I decided to test arm platform this week because it seems you have > binaries. But before all which mode do you advice ? > > At the end we are looking for running an opengl code on gem5 based system > and see triangles ! thats all. > > Thanks > ___ > gem5-users mailing list -- gem5-users@gem5.org > To unsubscribe send an email to gem5-users-le...@gem5.org > %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s > ___ gem5-users mailing list -- gem5-users@gem5.org To unsubscribe send an email to gem5-users-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
[gem5-users] CleanInvalidate speculative loaded cache blocks.
Hi all, I am trying to cleanInvalidate the cache entries that are loaded speculatively. When a branch miss prediction is detected, the load queue will squash its entries corresponding to miss predicted instructions. For each load queue entry, I am trying to create a new request with the same address as the one in the load queue and then create a new packet in the lsq unit by using: > new Packet(req, MemCmd::CleanInvalidReq) And send this packet to the cache. However, I am encountering error that says: > panic: Tried to execute unmapped address 0x6ec0. > Memory Usage: 695344 KBytes > Program aborted at tick 109387000 > or for a different executable binary: > panic: Tried to read unmapped address 0xff800db8. > PC: 0x780023c1, Instr: MOV_R_M : ld rdx, DS:[rcx] > Memory Usage: 684344 KBytes > Program aborted at tick 50074000 > Does anyone have an idea what could be causing this? Any help would be appreciated. Regards, Zelong ___ gem5-users mailing list -- gem5-users@gem5.org To unsubscribe send an email to gem5-users-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
[gem5-users] Re: How to use custom gpu models with gem5 ?
Hi again, Thanks for replies and I learned so much about gem5. I was trying gem5 for a while to learn it. I have one question: is se mode or fs mode more suitable to test tlm & gpus & etc... ? Our own testing environment based on qemu and it runs ubuntu. So we can install gpu driver and test our hw model using tlm & systemc. I tested gem5 with ubuntu in fs mode but booting it took a huge time ( x86 ). I tried multiple times but I couldn't succeed. So I decided to test arm platform this week because it seems you have binaries. But before all which mode do you advice ? At the end we are looking for running an opengl code on gem5 based system and see triangles ! thats all. Thanks ___ gem5-users mailing list -- gem5-users@gem5.org To unsubscribe send an email to gem5-users-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s