add support in archdetect for detecting A64FX #608

boegel · 2024-06-13T19:00:54Z

No description provided.

init/arch_specs/eessi_arch_arm.spec

…io_archdetect-a64fx

…' field is not available

boegel · 2025-01-10T12:32:16Z

bot: build repo:eessi.io-2023.06-software arch:aarch64/neoverse_n1

eessi-bot · 2025-01-10T12:32:19Z

Updates by the bot instance eessi-bot-mc-aws (click for details)

received bot command build repo:eessi.io-2023.06-software arch:aarch64/neoverse_n1 from boegel
- expanded format: build repository:eessi.io-2023.06-software architecture:aarch64/neoverse_n1
handling command build repository:eessi.io-2023.06-software architecture:aarch64/neoverse_n1 resulted in:
- submitted job 39812, for details & status see add support in archdetect for detecting A64FX #608 (comment)

eessi-bot · 2025-01-10T12:32:19Z

Updates by the bot instance eessi-bot-mc-azure (click for details)

received bot command build repo:eessi.io-2023.06-software arch:aarch64/neoverse_n1 from boegel
- expanded format: build repository:eessi.io-2023.06-software architecture:aarch64/neoverse_n1
handling command build repository:eessi.io-2023.06-software architecture:aarch64/neoverse_n1 resulted in:
- no jobs were submitted

eessi-bot · 2025-01-10T12:32:23Z

New job on instance eessi-bot-mc-aws for CPU micro-architecture aarch64-neoverse_n1 for repository eessi.io-2023.06-software in job dir /project/def-users/SHARED/jobs/2025.01/pr_608/39812

date	job status	comment
Jan 10 12:32:22 UTC 2025	submitted	job id `39812` awaits release by job manager
Jan 10 12:32:52 UTC 2025	released	job awaits launch by Slurm scheduler
Jan 10 12:38:03 UTC 2025	running	job `39812` is running
Jan 10 12:47:25 UTC 2025	finished	😁 SUCCESS (click triangle for details) Details ✅ job output file `slurm-39812.out` ✅ no message matching `FATAL:` ✅ no message matching `ERROR:` ✅ no message matching `FAILED:` ✅ no message matching `required modules missing:` ✅ found message(s) matching `No missing installations` ✅ found message matching `.tar.gz created!` Artefacts `eessi-2023.06-software-linux-aarch64-neoverse_n1-1736512762.tar.gz` size: 0 MiB (17750 bytes) entries: 3 modules under 2023.06/software/linux/aarch64/neoverse_n1/modules/all no module files in tarball software under 2023.06/software/linux/aarch64/neoverse_n1/software no software packages in tarball other under 2023.06/software/linux/aarch64/neoverse_n1 `2023.06/init/arch_specs/eessi_arch_arm.spec` `2023.06/init/easybuild/eb_hooks.py` `2023.06/init/eessi_archdetect.sh`
Jan 10 12:47:25 UTC 2025	test result	😁 SUCCESS (click triangle for details) ReFrame Summary [ OK ] ( 1/10) EESSI_LAMMPS_lj %scale=1_node %device_type=cpu %module_name=LAMMPS/29Aug2024-foss-2023b-kokkos /aeb2d9df @BotBuildTests:aarch64-neoverse-n1-node+default P: perf: 663.951 timesteps/s (r:0, l:None, u:None) [ OK ] ( 2/10) EESSI_LAMMPS_lj %scale=1_node %device_type=cpu %module_name=LAMMPS/2Aug2023_update2-foss-2023a-kokkos /04ff9ece @BotBuildTests:aarch64-neoverse-n1-node+default P: perf: 671.001 timesteps/s (r:0, l:None, u:None) [ OK ] ( 3/10) EESSI_OSU_Micro_Benchmarks_coll %benchmark_info=mpi.collective.osu_allreduce %scale=1_node %module_name=OSU-Micro-Benchmarks/7.2-gompi-2023b %device_type=cpu /31ac6ab9 @BotBuildTests:aarch64-neoverse-n1-node+default P: latency: 3.77 us (r:0, l:None, u:None) [ OK ] ( 4/10) EESSI_OSU_Micro_Benchmarks_coll %benchmark_info=mpi.collective.osu_allreduce %scale=1_node %module_name=OSU-Micro-Benchmarks/7.1-1-gompi-2023a %device_type=cpu /f3be40a2 @BotBuildTests:aarch64-neoverse-n1-node+default P: latency: 3.61 us (r:0, l:None, u:None) [ OK ] ( 5/10) EESSI_OSU_Micro_Benchmarks_coll %benchmark_info=mpi.collective.osu_alltoall %scale=1_node %module_name=OSU-Micro-Benchmarks/7.2-gompi-2023b %device_type=cpu /10e66fba @BotBuildTests:aarch64-neoverse-n1-node+default P: latency: 5.67 us (r:0, l:None, u:None) [ OK ] ( 6/10) EESSI_OSU_Micro_Benchmarks_coll %benchmark_info=mpi.collective.osu_alltoall %scale=1_node %module_name=OSU-Micro-Benchmarks/7.1-1-gompi-2023a %device_type=cpu /5be57ae7 @BotBuildTests:aarch64-neoverse-n1-node+default P: latency: 5.48 us (r:0, l:None, u:None) [ OK ] ( 7/10) EESSI_OSU_Micro_Benchmarks_pt2pt %benchmark_info=mpi.pt2pt.osu_latency %scale=1_node %module_name=OSU-Micro-Benchmarks/7.2-gompi-2023b %device_type=cpu /c8c9aff5 @BotBuildTests:aarch64-neoverse-n1-node+default P: latency: 0.46 us (r:0, l:None, u:None) [ OK ] ( 8/10) EESSI_OSU_Micro_Benchmarks_pt2pt %benchmark_info=mpi.pt2pt.osu_latency %scale=1_node %module_name=OSU-Micro-Benchmarks/7.1-1-gompi-2023a %device_type=cpu /9795e491 @BotBuildTests:aarch64-neoverse-n1-node+default P: latency: 0.43 us (r:0, l:None, u:None) [ OK ] ( 9/10) EESSI_OSU_Micro_Benchmarks_pt2pt %benchmark_info=mpi.pt2pt.osu_bw %scale=1_node %module_name=OSU-Micro-Benchmarks/7.2-gompi-2023b %device_type=cpu /48da21c5 @BotBuildTests:aarch64-neoverse-n1-node+default P: bandwidth: 19692.91 MB/s (r:0, l:None, u:None) [ OK ] (10/10) EESSI_OSU_Micro_Benchmarks_pt2pt %benchmark_info=mpi.pt2pt.osu_bw %scale=1_node %module_name=OSU-Micro-Benchmarks/7.1-1-gompi-2023a %device_type=cpu /1b8c1ca2 @BotBuildTests:aarch64-neoverse-n1-node+default P: bandwidth: 19603.35 MB/s (r:0, l:None, u:None) [ PASSED ] Ran 10/10 test case(s) from 10 check(s) (0 failure(s), 0 skipped, 0 aborted) Details ✅ job output file `slurm-39812.out` ✅ no message matching `ERROR:` ✅ no message matching `[\sFAILED\s].Ran . test case`
Jan 10 13:38:51 UTC 2025	uploaded	transfer of `eessi-2023.06-software-linux-aarch64-neoverse_n1-1736512762.tar.gz` to S3 bucket succeeded

ocaisa · 2025-01-10T12:49:57Z

init/arch_specs/eessi_arch_arm.spec

 "aarch64/neoverse_v1"	"ARM"		"asimddp svei8mm"
-"aarch64/neoverse_v1"	""		"asimddp svei8mm"	# AWS Graviton3
+"aarch64/neoverse_v1"	"0x41"		"asimddp svei8mm"	# AWS Graviton3


What happens when there are multiple vendors implementing the instruction set, like with Gravition/Grace for "aarch64/neoverse_v2"

I guess this is only necessary when there is a doubt about what the match is

https://gpages.juszkiewicz.com.pl/arm-socs-table/arm-socs.html only lists Graviton 3 as CPU supporting Neoverse V1, so that makes it very unlikely in practice we'll run into something that's detected as Neoverse V1 but isn't a Graviton 3 (at least today).

For neoverse_v2 with Graviton 4 vs Grace, there's a more complex situation though, because Graviton 4 supports CPU instructions like paca, pacg, rng which Grace doesn't, and the other way around (like sm3, sm4, svesm4).
It gets even more interesting when Google Axion is taken into account, since that doesn't support sbss, which the other two do...

Nevermind, it is right there in the file, just means we need to add each supported CPU explicitly. That seems like a bit of a pity, doesn't it make the archdetect selection very conservative if it encounters a CPU it hasn't seen before? It doesn't matter if the CPU has all the required features

For neoverse_v2 with Graviton 4 vs Grace, there's a more complex situation though, because Graviton 4 supports CPU instructions like paca, pacg, rng which Grace doesn't, and the other way around (like sm3, sm4, svesm4). It gets even more interesting when Google Axion is taken into account, since that doesn't support sbss, which the other two do...

Isn't that quite a big problem? Can we disable some instructions? I guess compilers would be caught by surprise...

This data is probably pretty useful for us: https://github.com/hrw/arm-socs-table/tree/main/cpuinfo-data

I wonder if we shouldn't have a much more considered set of CPU features that reflect the CPU where the software was built for the target and then match that? This would do a much better job of future matching CPUs. Is it really the case that CPU features alone are not enough to distinguish A64FX from Neoverse_N1?

Not going to hold this PR back for that discussion, will open an issue.

I'm also wondering if we can instruct the compiler to not emit particular instructions even if they're supported.
My gut says that should be possible, but I'm not sure it actually is...

Let's follow up on that in a dedicated issue though => #845

For A64FX vs Neoverse N1/V1, see the /proc/cpuinfo dumps we have under tests/archdetect/.

boegel · 2025-01-10T13:02:19Z

eb_hooks.py is in the generated tarball, but untouched in this PR...
Did we overlook a deploy in a previous already merged PR?

ocaisa · 2025-01-10T13:03:01Z

Sure did, #841

bedroge · 2025-01-10T13:03:16Z

eb_hooks.py is in the generated tarball, but untouched in this PR... Did we overlook a deploy in a previous already merged PR?

Yeah, I also just noticed that. I guess it's because of #841 (comment).

bedroge · 2025-01-10T14:14:39Z

The tarball has been approved and ingested.

eessi-bot · 2025-01-10T14:14:47Z

PR merged! Moved ['/project/def-users/SHARED/jobs/2025.01/pr_608/39812'] to /project/def-users/SHARED/trash_bin/EESSI/software-layer/2025.01.10

eessi-bot · 2025-01-10T14:14:48Z

PR merged! Moved [] to /project/def-users/SHARED/trash_bin/EESSI/software-layer/2025.01.10

boegel added aarch64 related to Arm 64-bit targets (aarch64) 2023.06-software.eessi.io 2023.06 version of software.eessi.io labels Jun 13, 2024

This comment was marked as outdated.

Sign in to view

boegel force-pushed the 2023.06-software.eessi.io_archdetect-a64fx branch from d426d99 to e3225a8 Compare June 13, 2024 19:04

ocaisa reviewed Jun 13, 2024

View reviewed changes

init/arch_specs/eessi_arch_arm.spec Outdated Show resolved Hide resolved

add support for detecting A64FX to archdetect

cb1672b

boegel force-pushed the 2023.06-software.eessi.io_archdetect-a64fx branch from 688d36e to cb1672b Compare June 13, 2024 19:10

boegel commented Jun 13, 2024

View reviewed changes

init/arch_specs/eessi_arch_arm.spec Outdated Show resolved Hide resolved

boegel added the a64fx label Jul 6, 2024

boegel added 6 commits January 10, 2025 10:31

Merge branch '2023.06-software.eessi.io' into 2023.06-software.eessi.…

f376b9a

…io_archdetect-a64fx

use 0x46 as 'CPU implementer' field for A64FX

5bad2d9

fall back to 'CPU implementer' field from /proc/cpuinfo if 'Vendor ID…

e53ac16

…' field is not available

add Deucalion-Rocky85.all.output for archdetect tests

f112406

set correct 'CPU implementer' for graviton2/graviton3

d82a188

fix logic for determining field to use to determine CPU flags

562fd4d

boegel marked this pull request as ready for review January 10, 2025 12:30

boegel changed the title ~~add support for detecting A64FX to archdetect~~ add support in archdetect for detecting A64FX Jan 10, 2025

ocaisa reviewed Jan 10, 2025

View reviewed changes

boegel mentioned this pull request Jan 10, 2025

Print LmodError when loading GCCcore-12.2.0-based modules on zen4 #841

Merged

1 task

bedroge approved these changes Jan 10, 2025

View reviewed changes

bedroge added the bot:deploy Ask bot to deploy missing software installations to EESSI label Jan 10, 2025

bedroge merged commit 3af9cf6 into EESSI:2023.06-software.eessi.io Jan 10, 2025
51 checks passed

boegel deleted the 2023.06-software.eessi.io_archdetect-a64fx branch January 12, 2025 16:30

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

add support in archdetect for detecting A64FX #608

add support in archdetect for detecting A64FX #608

boegel commented Jun 13, 2024

This comment was marked as outdated.

This comment was marked as outdated.

boegel commented Jan 10, 2025

eessi-bot bot commented Jan 10, 2025 •

edited

Loading

eessi-bot bot commented Jan 10, 2025 •

edited

Loading

eessi-bot bot commented Jan 10, 2025 •

edited

Loading

ocaisa Jan 10, 2025

ocaisa Jan 10, 2025

boegel Jan 10, 2025

ocaisa Jan 10, 2025

ocaisa Jan 10, 2025

ocaisa Jan 10, 2025

ocaisa Jan 10, 2025

boegel Jan 10, 2025

boegel Jan 10, 2025

boegel commented Jan 10, 2025

ocaisa commented Jan 10, 2025

bedroge commented Jan 10, 2025

bedroge commented Jan 10, 2025

eessi-bot bot commented Jan 10, 2025

eessi-bot bot commented Jan 10, 2025

add support in archdetect for detecting A64FX #608

add support in archdetect for detecting A64FX #608

Conversation

boegel commented Jun 13, 2024

This comment was marked as outdated.

This comment was marked as outdated.

boegel commented Jan 10, 2025

eessi-bot bot commented Jan 10, 2025 • edited Loading

eessi-bot bot commented Jan 10, 2025 • edited Loading

eessi-bot bot commented Jan 10, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

boegel commented Jan 10, 2025

ocaisa commented Jan 10, 2025

bedroge commented Jan 10, 2025

bedroge commented Jan 10, 2025

eessi-bot bot commented Jan 10, 2025

eessi-bot bot commented Jan 10, 2025

eessi-bot bot commented Jan 10, 2025 •

edited

Loading

eessi-bot bot commented Jan 10, 2025 •

edited

Loading

eessi-bot bot commented Jan 10, 2025 •

edited

Loading