Re: [v8-users] what does simd256 in V8 used for

35 vues
Passer au premier message non lu

ji qiu

non lue,
16 févr. 2023, 04 h 22 min 58 s2023-02-16
à v8-u...@googlegroups.com
By searching in the Chromium gerrit for the git commit history of the addition of the ymm register, it's found that Intel had been working for 256bit wide vector instructions since 2021. (https://chromium-review.googlesource.com/q/owner:jie...@intel.com) .

From the CL https://chromium-review.googlesource.com/c/v8/v8/+/3411357 you can see re-vectorization phase is added to the wasm pipeline so that for Intel platforms 256bit-wide vector can be used to squeeze hardware computation power.


'Qiaowen Yang' via v8-users <v8-u...@googlegroups.com> 于2023年2月15日周三 17:36写道:
I found that x86 assembler includes ymm registers, which are 256-bits in length. I am wondering what this will be used in V8. 
V8 does not have auto-vectorization, and the simd instructions in V8 all come from webassembly, which only has a `v128` data type. So, how will ymm registers be used in V8?

--
--
v8-users mailing list
v8-u...@googlegroups.com
http://groups.google.com/group/v8-users
---
You received this message because you are subscribed to the Google Groups "v8-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to v8-users+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/v8-users/2f54d449-d6e6-4184-9465-b90c189a629cn%40googlegroups.com.

Qiaowen Yang

non lue,
21 févr. 2023, 06 h 16 min 20 s2023-02-21
à v8-users
Thank you. This is very helpful.
Répondre à tous
Répondre à l'auteur
Transférer
0 nouveau message