[vectorIntrinsics] RFR: 8266775: Add VectorLoadConst node implementation for Arm SVE [v3]
Ningsheng Jian
njian at openjdk.java.net
Wed May 12 06:05:26 UTC 2021
On Tue, 11 May 2021 03:13:35 GMT, Wang Huang <whuang at openjdk.org> wrote:
>> * add VectorLoadConst node implementation for Arm SVE by
>> * ~~add new data in `generate_iota_indices`~~
>> * ~~`sve_ld1b` these data~~
>> * `sve_index`
>>
>> Thank you for your review. Any suggestion is welcome.
>
> Wang Huang has updated the pull request incrementally with one additional commit since the last revision:
>
> use sve_index
src/hotspot/cpu/aarch64/aarch64_sve_ad.m4 line 2496:
> 2494:
> 2495: // ------------------------------ Vector Load Const -------------------------------
> 2496: instruct loadconB(vReg dst, immI0 src)
Add a blank line between them?
src/hotspot/cpu/aarch64/aarch64_sve_ad.m4 line 2499:
> 2497: %{
> 2498: predicate(UseSVE > 0 &&
> 2499: n->bottom_type()->is_vect()->element_basic_type() == T_BYTE);
When the size is 16, the neon pattern will also be matched. Can you please add "UseSVE == 0" for NEON VectorLoadConst patterns?
-------------
PR: https://git.openjdk.java.net/panama-vector/pull/79
More information about the panama-dev
mailing list