From patchwork Thu Jul 29 14:28:04 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Prathamesh Kulkarni X-Patchwork-Id: 1511310 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (sender SPF authorized) smtp.mailfrom=gcc.gnu.org (client-ip=8.43.85.97; helo=sourceware.org; envelope-from=gcc-patches-bounces+incoming=patchwork.ozlabs.org@gcc.gnu.org; receiver=) Authentication-Results: ozlabs.org; dkim=pass (1024-bit key; unprotected) header.d=gcc.gnu.org header.i=@gcc.gnu.org header.a=rsa-sha256 header.s=default header.b=nk3rjEY/; dkim-atps=neutral Received: from sourceware.org (server2.sourceware.org [8.43.85.97]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 4GbCbG61Lnz9sRK for ; Fri, 30 Jul 2021 00:29:29 +1000 (AEST) Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id 442D839B2861 for ; Thu, 29 Jul 2021 14:29:26 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 442D839B2861 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org; s=default; t=1627568966; bh=+QVAqIjnj4/+U3DkcFvy6Y/gFY7ND6PQQ9zgcxEyR7U=; h=Date:Subject:To:List-Id:List-Unsubscribe:List-Archive:List-Post: List-Help:List-Subscribe:From:Reply-To:From; b=nk3rjEY/5C3sMjhZH8kimdjEjx9okU5wFBXE9OBPyVXUior0QFoCue5XcBm2OWMV2 ROb3b8pT6zQycrf51OA9Dz5VxLdTx3PG0upfzBdzl9PS1j8QbiJ4RfbQ3IMV3uRcI0 hbUvHbHSpYdUs57bNDmMY7Ztj2O+uCJR71FcUZfk= X-Original-To: gcc-patches@gcc.gnu.org Delivered-To: gcc-patches@gcc.gnu.org Received: from mail-ej1-x632.google.com (mail-ej1-x632.google.com [IPv6:2a00:1450:4864:20::632]) by sourceware.org (Postfix) with ESMTPS id E13613864827 for ; Thu, 29 Jul 2021 14:28:41 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org E13613864827 Received: by mail-ej1-x632.google.com with SMTP id gs8so10991602ejc.13 for ; Thu, 29 Jul 2021 07:28:41 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:from:date:message-id:subject:to; bh=+QVAqIjnj4/+U3DkcFvy6Y/gFY7ND6PQQ9zgcxEyR7U=; b=qhi4PhKta03lgXXUNKKB+rbU7M1DC9uu4GrbKLbZGDu1HW2HD4RvPiBa+J/E41DwUA 7zt8IyGtWzZVYbe9Fv5Dw22SmWkmGktyAw6VfAxy3yb281SLz0nMsHRsU7Xt3o6+sRb+ FH1v1Ehjy+1RfuRw6RWD5dGK+iZHN1EZNEyIBDQQrJS/llFddb+NGApb75u2JAThqMkr G/vCF6NUiD6BYXO5KxbEStgJU1GhhYpYEDOmfq4SQt3UHTi6h9W+foWghtIBH1Jnw4oa FZCIIySTebuUSzdO5HsYwsgOnkZVj/SvPrFkYmihX380fb9+iczAQT96tiKg9HO9sjTD 7dOw== X-Gm-Message-State: AOAM532V42Glk5jCxRYAhdKCVw3QgYSFAO7g/A+3jjxZm67gA33Vq5ni +WvuLfu90ue0bXP2n/69olzijMT9eDrPbT6Z75HPA++hWmc= X-Google-Smtp-Source: ABdhPJzitojJhAf5UNdfTE8Z6jrP0oit8hHsd4O+EGeGsTJlyxCd3T+f36+8LR6kH8YFC902TkrFR2ARANlmQxM7wYo= X-Received: by 2002:a17:906:180a:: with SMTP id v10mr4786754eje.112.1627568920858; Thu, 29 Jul 2021 07:28:40 -0700 (PDT) MIME-Version: 1.0 Date: Thu, 29 Jul 2021 19:58:04 +0530 Message-ID: Subject: [ARM] PR66791: Replace builtin in vld1_dup intrinsics To: gcc Patches , Kyrill Tkachov , Richard Earnshaw X-Spam-Status: No, score=-9.2 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.4 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on server2.sourceware.org X-BeenThere: gcc-patches@gcc.gnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Gcc-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-Patchwork-Original-From: Prathamesh Kulkarni via Gcc-patches From: Prathamesh Kulkarni Reply-To: Prathamesh Kulkarni Errors-To: gcc-patches-bounces+incoming=patchwork.ozlabs.org@gcc.gnu.org Sender: "Gcc-patches" Hi, The attached patch replaces builtins in vld1_dup intrinsics with call to corresponding vdup_n intrinsic and removes entry for vld1_dup from arm_neon_builtins.def. Bootstrapped+tested on arm-linux-gnueabihf. OK to commit ? Thanks, Prathamesh gcc/ChangeLog: PR target/66791 * config/arm/arm_neon.h (vld1_dup_s8): Replace builtin with call to corresponding vdup_n intrinsic. (vld1_dup_s16): Likewise. (vld1_dup_s32): Likewise. (vld1_dup_f32): Likewise. (vld1_dup_u8): Likewise. (vld1_dup_u16): Likewise. (vld1_dup_u32): Likewise. (vld1_dup_p8): Likewise. (vld1_dup_p16): Likewise. (vld1_dup_p64): Likewise. (vld1_dup_s64): Likewise. (vld1_dup_u64): Likewise. (vld1q_dup_s8): Likewise. (vld1q_dup_s16): Likewise. (vld1q_dup_s32): Likewise. (vld1q_dup_f32): Likewise. (vld1q_dup_u8): Likewise. (vld1q_dup_u16): Likewise. (vld1q_dup_u32): Likewise. (vld1q_dup_p8): Likewise. (vld1q_dup_p16): Likewise. (vld1q_dup_p64): Likewise. (vld1q_dup_s64): Likewise. (vld1q_dup_u64): Likewise. * config/arm/arm_neon_builtins.def (vld1_dup): Remove entry. diff --git a/gcc/config/arm/arm_neon.h b/gcc/config/arm/arm_neon.h index 41b596b5fc6..bc55dacffd3 100644 --- a/gcc/config/arm/arm_neon.h +++ b/gcc/config/arm/arm_neon.h @@ -10683,21 +10683,21 @@ __extension__ extern __inline int8x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_s8 (const int8_t * __a) { - return (int8x8_t)__builtin_neon_vld1_dupv8qi ((const __builtin_neon_qi *) __a); + return vdup_n_s8 (*__a); } __extension__ extern __inline int16x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_s16 (const int16_t * __a) { - return (int16x4_t)__builtin_neon_vld1_dupv4hi ((const __builtin_neon_hi *) __a); + return vdup_n_s16 (*__a); } __extension__ extern __inline int32x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_s32 (const int32_t * __a) { - return (int32x2_t)__builtin_neon_vld1_dupv2si ((const __builtin_neon_si *) __a); + return vdup_n_s32 (*__a); } #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE) @@ -10714,42 +10714,42 @@ __extension__ extern __inline float32x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_f32 (const float32_t * __a) { - return (float32x2_t)__builtin_neon_vld1_dupv2sf ((const __builtin_neon_sf *) __a); + return vdup_n_f32 (*__a); } __extension__ extern __inline uint8x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_u8 (const uint8_t * __a) { - return (uint8x8_t)__builtin_neon_vld1_dupv8qi ((const __builtin_neon_qi *) __a); + return vdup_n_u8 (*__a); } __extension__ extern __inline uint16x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_u16 (const uint16_t * __a) { - return (uint16x4_t)__builtin_neon_vld1_dupv4hi ((const __builtin_neon_hi *) __a); + return vdup_n_u16 (*__a); } __extension__ extern __inline uint32x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_u32 (const uint32_t * __a) { - return (uint32x2_t)__builtin_neon_vld1_dupv2si ((const __builtin_neon_si *) __a); + return vdup_n_u32 (*__a); } __extension__ extern __inline poly8x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_p8 (const poly8_t * __a) { - return (poly8x8_t)__builtin_neon_vld1_dupv8qi ((const __builtin_neon_qi *) __a); + return vdup_n_p8 (*__a); } __extension__ extern __inline poly16x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_p16 (const poly16_t * __a) { - return (poly16x4_t)__builtin_neon_vld1_dupv4hi ((const __builtin_neon_hi *) __a); + return vdup_n_p16 (*__a); } #pragma GCC push_options @@ -10758,7 +10758,7 @@ __extension__ extern __inline poly64x1_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_p64 (const poly64_t * __a) { - return (poly64x1_t)__builtin_neon_vld1_dupdi ((const __builtin_neon_di *) __a); + return vdup_n_p64 (*__a); } #pragma GCC pop_options @@ -10766,35 +10766,35 @@ __extension__ extern __inline int64x1_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_s64 (const int64_t * __a) { - return (int64x1_t)__builtin_neon_vld1_dupdi ((const __builtin_neon_di *) __a); + return vdup_n_s64 (*__a); } __extension__ extern __inline uint64x1_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1_dup_u64 (const uint64_t * __a) { - return (uint64x1_t)__builtin_neon_vld1_dupdi ((const __builtin_neon_di *) __a); + return vdup_n_u64 (*__a); } __extension__ extern __inline int8x16_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_s8 (const int8_t * __a) { - return (int8x16_t)__builtin_neon_vld1_dupv16qi ((const __builtin_neon_qi *) __a); + return vdupq_n_s8 (*__a); } __extension__ extern __inline int16x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_s16 (const int16_t * __a) { - return (int16x8_t)__builtin_neon_vld1_dupv8hi ((const __builtin_neon_hi *) __a); + return vdupq_n_s16 (*__a); } __extension__ extern __inline int32x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_s32 (const int32_t * __a) { - return (int32x4_t)__builtin_neon_vld1_dupv4si ((const __builtin_neon_si *) __a); + return vdupq_n_s32 (*__a); } #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE) @@ -10811,42 +10811,42 @@ __extension__ extern __inline float32x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_f32 (const float32_t * __a) { - return (float32x4_t)__builtin_neon_vld1_dupv4sf ((const __builtin_neon_sf *) __a); + return vdupq_n_f32 (*__a); } __extension__ extern __inline uint8x16_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_u8 (const uint8_t * __a) { - return (uint8x16_t)__builtin_neon_vld1_dupv16qi ((const __builtin_neon_qi *) __a); + return vdupq_n_u8 (*__a); } __extension__ extern __inline uint16x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_u16 (const uint16_t * __a) { - return (uint16x8_t)__builtin_neon_vld1_dupv8hi ((const __builtin_neon_hi *) __a); + return vdupq_n_u16 (*__a); } __extension__ extern __inline uint32x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_u32 (const uint32_t * __a) { - return (uint32x4_t)__builtin_neon_vld1_dupv4si ((const __builtin_neon_si *) __a); + return vdupq_n_u32 (*__a); } __extension__ extern __inline poly8x16_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_p8 (const poly8_t * __a) { - return (poly8x16_t)__builtin_neon_vld1_dupv16qi ((const __builtin_neon_qi *) __a); + return vdupq_n_p8 (*__a); } __extension__ extern __inline poly16x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_p16 (const poly16_t * __a) { - return (poly16x8_t)__builtin_neon_vld1_dupv8hi ((const __builtin_neon_hi *) __a); + return vdupq_n_p16 (*__a); } #pragma GCC push_options @@ -10855,7 +10855,7 @@ __extension__ extern __inline poly64x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_p64 (const poly64_t * __a) { - return (poly64x2_t)__builtin_neon_vld1_dupv2di ((const __builtin_neon_di *) __a); + return vdupq_n_p64 (*__a); } #pragma GCC pop_options @@ -10863,14 +10863,14 @@ __extension__ extern __inline int64x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_s64 (const int64_t * __a) { - return (int64x2_t)__builtin_neon_vld1_dupv2di ((const __builtin_neon_di *) __a); + return vdupq_n_s64 (*__a); } __extension__ extern __inline uint64x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vld1q_dup_u64 (const uint64_t * __a) { - return (uint64x2_t)__builtin_neon_vld1_dupv2di ((const __builtin_neon_di *) __a); + return vdupq_n_u64 (*__a); } #pragma GCC push_options diff --git a/gcc/config/arm/arm_neon_builtins.def b/gcc/config/arm/arm_neon_builtins.def index 70438ac1848..a2aeb21f350 100644 --- a/gcc/config/arm/arm_neon_builtins.def +++ b/gcc/config/arm/arm_neon_builtins.def @@ -307,8 +307,6 @@ VAR14 (LOAD1, vld1, v4bf, v8bf) VAR12 (LOAD1LANE, vld1_lane, v8qi, v4hi, v2si, v2sf, di, v16qi, v8hi, v4si, v4sf, v2di, v4bf, v8bf) -VAR10 (LOAD1, vld1_dup, - v8qi, v4hi, v2si, v2sf, di, v16qi, v8hi, v4si, v4sf, v2di) VAR14 (STORE1, vst1, v8qi, v4hi, v4hf, v2si, v2sf, di, v16qi, v8hi, v8hf, v4si, v4sf, v2di, v4bf, v8bf)