From: Jiajie Chen <c@jia.je>
To: Richard Henderson <richard.henderson@linaro.org>, qemu-devel@nongnu.org
Cc: gaosong@loongson.cn, git@xen0n.name
Subject: Re: [PATCH v2 03/14] tcg/loongarch64: Lower cmp_vec to vseq/vsle/vslt
Date: Sat, 2 Sep 2023 01:28:20 +0800 [thread overview]
Message-ID: <3ddba7dd-db34-c017-8dfa-fa805145e2f2@jia.je> (raw)
In-Reply-To: <1f963ece-d5b6-50c7-5e40-bc04867494f7@linaro.org>
On 2023/9/2 01:24, Richard Henderson wrote:
> On 9/1/23 02:30, Jiajie Chen wrote:
>> Signed-off-by: Jiajie Chen <c@jia.je>
>> ---
>> tcg/loongarch64/tcg-target-con-set.h | 1 +
>> tcg/loongarch64/tcg-target.c.inc | 60 ++++++++++++++++++++++++++++
>> 2 files changed, 61 insertions(+)
>
> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
>
>
>>
>> diff --git a/tcg/loongarch64/tcg-target-con-set.h
>> b/tcg/loongarch64/tcg-target-con-set.h
>> index 37b3f80bf9..d04916db25 100644
>> --- a/tcg/loongarch64/tcg-target-con-set.h
>> +++ b/tcg/loongarch64/tcg-target-con-set.h
>> @@ -31,4 +31,5 @@ C_O1_I2(r, 0, rZ)
>> C_O1_I2(r, rZ, ri)
>> C_O1_I2(r, rZ, rJ)
>> C_O1_I2(r, rZ, rZ)
>> +C_O1_I2(w, w, wJ)
>
> Notes for improvement: 'J' is a signed 32-bit immediate.
I was wondering about the behavior of 'J' on i128 types: in
tcg_target_const_match(), the argument type is int, so will the higher
bits be truncated?
Besides, tcg_target_const_match() does not know the vector element width.
>
>> + if (const_args[2]) {
>> + /*
>> + * cmp_vec dest, src, value
>> + * Try vseqi/vslei/vslti
>> + */
>> + int64_t value = sextract64(a2, 0, 8 << vece);
>> + if ((cond == TCG_COND_EQ || cond == TCG_COND_LE || \
>> + cond == TCG_COND_LT) && (-0x10 <= value && value <=
>> 0x0f)) {
>> + tcg_out32(s,
>> encode_vdvjsk5_insn(cmp_vec_imm_insn[cond][vece], \
>> + a0, a1, value));
>> + break;
>> + } else if ((cond == TCG_COND_LEU || cond ==
>> TCG_COND_LTU) &&
>> + (0x00 <= value && value <= 0x1f)) {
>> + tcg_out32(s,
>> encode_vdvjuk5_insn(cmp_vec_imm_insn[cond][vece], \
>> + a0, a1, value));
>
> Better would be a new constraint that only matches
>
> -0x10 <= x <= 0x1f
>
> If the sign is wrong for the comparison, it can *always* be loaded
> with just vldi.
>
> Whereas at present, using J,
>
>> + tcg_out_dupi_vec(s, type, vece, temp_vec, a2);
>> + a2 = temp_vec;
>
> this may require 3 instructions (lu12i.w + ori + vreplgr2vr).
>
> By constraining the constants allowed, you allow the register
> allocator to see that a register is required, which may be reused for
> another instruction.
>
>
> r~
next prev parent reply other threads:[~2023-09-01 17:28 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-09-01 9:30 [PATCH v2 00/14] Lower TCG vector ops to LSX Jiajie Chen
2023-09-01 9:30 ` [PATCH v2 01/14] tcg/loongarch64: Import LSX instructions Jiajie Chen
2023-09-01 17:06 ` Richard Henderson
2023-09-01 9:30 ` [PATCH v2 02/14] tcg/loongarch64: Lower basic tcg vec ops to LSX Jiajie Chen
2023-09-01 17:05 ` Richard Henderson
2023-09-01 9:30 ` [PATCH v2 03/14] tcg/loongarch64: Lower cmp_vec to vseq/vsle/vslt Jiajie Chen
2023-09-01 17:24 ` Richard Henderson
2023-09-01 17:28 ` Jiajie Chen [this message]
2023-09-01 17:48 ` Richard Henderson
2023-09-02 1:06 ` Jiajie Chen
2023-09-01 9:30 ` [PATCH v2 04/14] tcg/loongarch64: Lower add/sub_vec to vadd/vsub Jiajie Chen
2023-09-01 17:58 ` Richard Henderson
2023-09-01 9:30 ` [PATCH v2 05/14] tcg/loongarch64: Lower vector bitwise operations Jiajie Chen
2023-09-01 17:59 ` Richard Henderson
2023-09-01 9:30 ` [PATCH v2 06/14] tcg/loongarch64: Lower neg_vec to vneg Jiajie Chen
2023-09-01 9:31 ` [PATCH v2 07/14] tcg/loongarch64: Lower mul_vec to vmul Jiajie Chen
2023-09-01 9:31 ` [PATCH v2 08/14] tcg/loongarch64: Lower vector min max ops Jiajie Chen
2023-09-01 9:31 ` [PATCH v2 09/14] tcg/loongarch64: Lower vector saturated ops Jiajie Chen
2023-09-01 9:31 ` [PATCH v2 10/14] tcg/loongarch64: Lower vector shift vector ops Jiajie Chen
2023-09-01 9:31 ` [PATCH v2 11/14] tcg/loongarch64: Lower bitsel_vec to vbitsel Jiajie Chen
2023-09-01 9:31 ` [PATCH v2 12/14] tcg/loongarch64: Lower vector shift integer ops Jiajie Chen
2023-09-01 18:01 ` Richard Henderson
2023-09-01 9:31 ` [PATCH v2 13/14] tcg/loongarch64: Lower rotv_vec ops to LSX Jiajie Chen
2023-09-01 18:02 ` Richard Henderson
2023-09-01 9:31 ` [PATCH v2 14/14] tcg/loongarch64: Lower rotli_vec to vrotri Jiajie Chen
2023-09-01 18:06 ` Richard Henderson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=3ddba7dd-db34-c017-8dfa-fa805145e2f2@jia.je \
--to=c@jia.je \
--cc=gaosong@loongson.cn \
--cc=git@xen0n.name \
--cc=qemu-devel@nongnu.org \
--cc=richard.henderson@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).