Illegal instruction #19
Comments
|
Hi @NicolasC0, What is the singularity version you have, i.e., what is the output of: I don't know, but it may help to add the |
|
Thank you for the quick response! python seems to be working and the singularity version is 3.8.7-1.el7. Could this be the issue? I tried the --cleanenv flag but still got the same error |
|
Ok. I don't think the singularity version should be a problem. The container is built using version 3.7.4, and I use it w. version 3.7.3. Version 2.x would cause problems due to changes in the sif format. I have only tested from Linux hosts, however. Does invoking the mixer --help function work? Other basic imports/checks (via bash) you may test: |
|
Thank you for the suggestions. The --help functions does work: When issuing --plink version and simu_linux --elp both work as well: But there's an error when issuing
|
|
Thanks again for testing. You can roll back to the older version of the The file was updated via #2, which may inadvertently have broken things so it would be good to test. |
|
Thank you again for the suggestions, I apologize for the delay in my response! I also get Illegal instruction error when running those commands Should I also try to test the older version of the mixer.sif file? |
|
Hi. It seems like this last command at least exposed which exact file may have caused the issue. I don't have an Intel-based machine at hand, however, and can't reproduce it, but will check with someone else in our group. If you have time, you may of course test the older mixer.sif file -- functionally it should be the same (and may work on your end). Also, please post your operating system / version. |
|
@NicolasC0 Sorry about these troubles!
This is a fairly old cpu, however I'm still surprised about the On the machine were we built mixer.sif container we have Base on https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html I think to try fixing this we need to re-generate mixer.sif containier patching CMakeLists.txt with |
|
Hi @NicolasC0. I've pushed a rebuilt container using the Can you see if it works for you? |
|
Thank you so much for the changes, I believe the "illegal instruction" problem is fixed! Though now it seems like it can't quite identify the sumstats trait file ("ValueError: Input file --trait1_file does not exist: /temporario/10721897/mixer/reference/sumstats/ADHD2022_munge.sumstats.gz") even though the file does exist and is in the correct path. I've tried moving it to different directories, removing and adding it back, tried with different files to no avail. I'm unsure if it's something I'm overlooking when running the script or preparing the files? |
|
Hi again, @NicolasC0. You may also use the |
|
It seems to be working! thank you so much for your great help! I have only one last question, the log files show mixer version 1.2.0 and I was wondering if that was correct? The main file is at around 1.2GB and to pull the repository I used git 2.9.5 and git-lfs 3.4.1. |
|
Excellent. The version 1.2.0 reported sounds correct as this is hardcoded into the Mixer master branch. Better to think of the current master as a v1.3 development version; while the v1.3.0 tag used in this repo is for the codes etc. found here. |
|
Oh I see, so it is a development version. Thank you immensely for your great help! |
|
Hi, I have face the same issue, and try the sif file in tag espenhgn/issue19. I have met the new issue here, please help and check this. Thanks very much. System Info: ISSUE:INFO:root:init(lib_name=/tools/mixer/src/build/lib/libbgmg.so, context_id=0)
Call: |
|
Hi @liyehaha. What is the CPU of the host system you run on? |
|
I'm his friend. Thanks for your quick response. CPU:e5-2650, a fairly old cpu. We have runned this program with a dektop: CPU, Intel(R) Core(TM) i5-7500 CPU @ 3.40GHz. Running slowly, but it runs normally. Additionally, I have a question, when I runnig the program as below: Why does this repetition range from 0 to 19, a total of 20 repetitions, while there are 22 reference genome files (22 chromosomes)? |
|
Thanks for the update. I think the e5-2650 CPU is now so outdated that we will not support it (officially); supporting only AVX. The i5-7500 CPU supports Intel® SSE4.1, Intel® SSE4.2, Intel® AVX2 and is nearly current. Similar to what I did earlier, it may be possible to use the For your latter question, the repetitions are over random seeds in each run, not chromosomes. It's not super clear from the MiXeR documentation, but in the code, the random seed is set as |
|
Thanks for your answer. I have completed this run, but I encountered another issue. It seems there's an issue with power analysis. Can you help me address this problem? |
|
This seems normal, so what is the exact issue? Please also have a look at this section |
|
Sorry for late response. when I run these two lines commands, I reveived a message "--json argument is a wild-card (contains *), skip figures generation", and then I hadn't obtained these two svg format figures. what's the meanning of "--json argument is a wild-card (contains *)"? |
|
Hi. These messages you are receiving are for information, and this discussion belongs in another issue, not with the original illegal instruction problem reported here. The arguments to each script are explained by invoking the help option (output truncated): Wildcards are commonly used in the shell to refer to e.g., files matching a pattern. |
|
As for the "Illegal instruction" issue, also occurs with the Docker image now pushed to ghcr.io:` with Mac hosts (M1 chip): Only a subset of instructions are supported: The fix appears to be to build the image locally. |
|
Thanks for your help. I understand it.
李存
***@***.***
…------------------ 原始邮件 ------------------
发件人: "Espen ***@***.***>;
发送时间: 2024年5月13日(星期一) 晚上6:48
收件人: ***@***.***>;
抄送: ***@***.***>; ***@***.***>;
主题: Re: [comorment/mixer] Illegal instruction (Issue #19)
As for the "Illegal instruction" issue, also occurs with the Docker image now pushed to ghcr.io:` with Mac hosts (M1 chip):
$ export ISHELL="docker run --platform=linux/amd64 --rm -it -v ${MIXER}/usecases:/work --entrypoint=bash ghcr.io/comorment/mixer:espenhgn-issue20" $ISHELL ***@***.***:/tools# python -c "import ctypes; ctypes.CDLL('/tools/mixer/src/build/lib/libbgmg.so')" Illegal instruction
Only a subset of instructions are supported:
***@***.***:/tools# lscpu | grep Flags Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 asimddp sha512 asimdfhm dit uscat ilrcpc flagm ssbs sb paca pacg dcpodp flagm2 frint
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: ***@***.***>
|
Thank you so much for developing mixer!
I believe I've installed and set up the singularity container for mixer correctly, however, when I try to run the job I get "Illegal instruction" error such as this:
There is no problem with the singulrity execution:
neither with using python
The error appears when trying to pass any arguments to mixer.py:
I believe the server achitecture is compatible (Architecture: x86_64, CPU op-mode(s): 32-bit, 64-bit, Model name: Intel(R) Xeon(R) CPU E7- 2870 @ 2.40GHz), but I'm very new to this level of programming and have never used or installed a container/singularity before.
Has anyone had this problem?
The text was updated successfully, but these errors were encountered: