From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=0.8 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, HTML_MESSAGE,MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,UNPARSEABLE_RELAY, USER_AGENT_SANE_1 autolearn=no autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id CC996C43331 for ; Sun, 29 Mar 2020 04:01:50 +0000 (UTC) Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id 9E0D220659 for ; Sun, 29 Mar 2020 04:01:50 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 9E0D220659 Authentication-Results: mail.kernel.org; dmarc=none (p=none dis=none) header.from=c-sky.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Received: from localhost ([::1]:34070 helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1jIP8v-0002k4-SZ for qemu-devel@archiver.kernel.org; Sun, 29 Mar 2020 00:01:49 -0400 Received: from eggs.gnu.org ([2001:470:142:3::10]:37447) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1jIP7x-0002I7-Lw for qemu-devel@nongnu.org; Sun, 29 Mar 2020 00:00:50 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1jIP7w-000569-8U for qemu-devel@nongnu.org; Sun, 29 Mar 2020 00:00:49 -0400 Received: from smtp2200-217.mail.aliyun.com ([121.197.200.217]:54218) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1jIP7v-0004vn-Py; Sun, 29 Mar 2020 00:00:48 -0400 X-Alimail-AntiSpam: AC=CONTINUE; BC=0.1311125|-1; CH=green; DM=|CONTINUE|false|; DS=CONTINUE|ham_regular_dialog|0.345928-0.000373953-0.653698; FP=0|0|0|0|0|-1|-1|-1; HT=e02c03305; MF=zhiwei_liu@c-sky.com; NM=1; PH=DS; RN=9; RT=9; SR=0; TI=SMTPD_---.H6x3BXQ_1585454437; Received: from 192.168.3.18(mailfrom:zhiwei_liu@c-sky.com fp:SMTPD_---.H6x3BXQ_1585454437) by smtp.aliyun-inc.com(10.147.42.198); Sun, 29 Mar 2020 12:00:39 +0800 Subject: Re: [PATCH v6 18/61] target/riscv: vector single-width integer multiply instructions From: LIU Zhiwei To: Richard Henderson , alistair23@gmail.com, chihmin.chao@sifive.com, palmer@dabbelt.com References: <20200317150653.9008-1-zhiwei_liu@c-sky.com> <20200317150653.9008-19-zhiwei_liu@c-sky.com> <1a5900d9-dc26-04f2-74b2-0bada246a0e7@linaro.org> <8be612c0-4345-c1f2-3b82-73a9a1ea4133@c-sky.com> <53209745-9b54-00b3-2b1f-1271f0843021@linaro.org> <5dabf936-6a4f-5d76-0d16-ff357d1ee376@c-sky.com> Message-ID: <764066b7-c083-1b4e-32c4-616aec3bd9be@c-sky.com> Date: Sun, 29 Mar 2020 12:00:37 +0800 User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:68.0) Gecko/20100101 Thunderbird/68.6.0 MIME-Version: 1.0 In-Reply-To: <5dabf936-6a4f-5d76-0d16-ff357d1ee376@c-sky.com> Content-Type: multipart/alternative; boundary="------------37E6D960588C56DB2585DEDA" Content-Language: en-US X-detected-operating-system: by eggs.gnu.org: GNU/Linux 3.x [generic] [fuzzy] X-Received-From: 121.197.200.217 X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: guoren@linux.alibaba.com, wenmeng_zhang@c-sky.com, qemu-riscv@nongnu.org, qemu-devel@nongnu.org, wxy194768@alibaba-inc.com Errors-To: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Sender: "Qemu-devel" This is a multi-part message in MIME format. --------------37E6D960588C56DB2585DEDA Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit On 2020/3/29 0:13, LIU Zhiwei wrote: > > > On 2020/3/28 23:47, Richard Henderson wrote: >> On 3/28/20 8:17 AM, LIU Zhiwei wrote: >>>> Missed the improvement here.  See tcg_gen_mulsu2_i64. >>> Though I have not gotten the principle, the code in >>> tcg_gen_mulsu2_i64 is much >>> tidier. >> Let A = signed operand, >>      B = unsigned operand >>      P = unsigned product >> >> If the sign bit A is set, then P is too large. >> In that case we subtract 2**64 * B to fix that: >> >>      HI_P -= (A < 0 ? B : 0) >> >> where the conditional is computed as (A >> 63) & B. > > I think I get it. > > LET  A = 2 ** 64  - X > > THEN > > X = 2 ** 64 - A > SIGNED_P = -X * B > > if (A * B == P) then > > (2 ** 64  - X) * B == P > 2 **64 * B - X * B == P > > -X *B == P - 2**64*B > > HI_P -= (A < 0 ? B ：0） > It's confusing here. I paste the clearer code. /* * Let A = signed operand, * B = unsigned operand * P = mulu64(A, B), unsigned product * * LET X = 2 ** 64 - A, 2's complement of A * SP = signed product * THEN * IF A < 0 * SP = -X * B * = -(2 ** 64 - A) * B * = A * B - 2 ** 64 * B * = P - 2 ** 64 * B * ELSE * SP = P * THEN * HI_P -= (A < 0 ? B : 0） */ static int64_t do_mulhsu_d(int64_t s2, uint64_t s1) { uint64_t hi_64, lo_64; mulu64(&lo_64, &hi_64, s2, s1); hi_64 -= s2 < 0 ? s1 : 0; return hi_64; } Zhiwei > Zhiwei >> >> r~ > --------------37E6D960588C56DB2585DEDA Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 8bit

On 2020/3/29 0:13, LIU Zhiwei wrote:


On 2020/3/28 23:47, Richard Henderson wrote:
On 3/28/20 8:17 AM, LIU Zhiwei wrote:
Missed the improvement here.  See tcg_gen_mulsu2_i64.
Though I have not gotten the principle, the code in tcg_gen_mulsu2_i64 is much
tidier.
Let A = signed operand,
     B = unsigned operand
     P = unsigned product

If the sign bit A is set, then P is too large.
In that case we subtract 2**64 * B to fix that:

     HI_P -= (A < 0 ? B : 0)

where the conditional is computed as (A >> 63) & B.

I think I get it.

LET  A = 2 ** 64  - X

THEN

X = 2 ** 64 - A
SIGNED_P = -X * B

if (A * B == P) then

(2 ** 64  - X) * B == P
2 **64 * B - X * B == P

-X *B == P - 2**64*B

HI_P -= (A < 0 ? B ：0）

It's confusing here. I paste the clearer code.
/*
 * Let  A = signed operand,
 *      B = unsigned operand
 *      P = mulu64(A, B), unsigned product
 *
 * LET  X = 2 ** 64  - A, 2's complement of A
 *      SP = signed product
 * THEN
 *      IF A < 0
 *          SP = -X * B
 *             = -(2 ** 64 - A) * B
 *             = A * B - 2 ** 64 * B
 *             = P - 2 ** 64 * B
 *      ELSE
 *          SP = P
 * THEN
 *      HI_P -= (A < 0 ? B : 0）
 */

static int64_t do_mulhsu_d(int64_t s2, uint64_t s1)
{
    uint64_t hi_64, lo_64;

    mulu64(&lo_64, &hi_64, s2, s1);

    hi_64 -= s2 < 0 ? s1 : 0;
    return hi_64;
}
Zhiwei
Zhiwei

r~


--------------37E6D960588C56DB2585DEDA--