From 6acbafe45b732b3410e67db5b1a5f05ed5e90f1e Mon Sep 17 00:00:00 2001 From: wernsaar Date: Sun, 20 Jul 2014 14:52:25 +0200 Subject: [PATCH] added sgemv_n microkernel for haswell --- kernel/x86_64/KERNEL.HASWELL | 4 +- kernel/x86_64/sgemv_n_avx.c | 2 + kernel/x86_64/sgemv_n_microk_haswell.c | 461 +++++++++++++++++++++++++++++++++ 3 files changed, 465 insertions(+), 2 deletions(-) create mode 100644 kernel/x86_64/sgemv_n_microk_haswell.c diff --git a/kernel/x86_64/KERNEL.HASWELL b/kernel/x86_64/KERNEL.HASWELL index 288e395..871a7d4 100644 --- a/kernel/x86_64/KERNEL.HASWELL +++ b/kernel/x86_64/KERNEL.HASWELL @@ -1,8 +1,8 @@ ifdef OS_WINDOWS -#SGEMVNKERNEL = ../arm/gemv_n.c +SGEMVNKERNEL = ../arm/gemv_n.c SGEMVTKERNEL = ../arm/gemv_t.c else -#SGEMVNKERNEL = sgemv_n_avx.c +SGEMVNKERNEL = sgemv_n_avx.c SGEMVTKERNEL = sgemv_t_avx.c endif diff --git a/kernel/x86_64/sgemv_n_avx.c b/kernel/x86_64/sgemv_n_avx.c index 96a03ec..57aaad4 100644 --- a/kernel/x86_64/sgemv_n_avx.c +++ b/kernel/x86_64/sgemv_n_avx.c @@ -30,6 +30,8 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. #if defined(BULLDOZER) || defined(PILEDRIVER) #include "sgemv_n_microk_bulldozer.c" +#elif defined(HASWELL) +#include "sgemv_n_microk_haswell.c" #else #include "sgemv_n_microk_sandy.c" #endif diff --git a/kernel/x86_64/sgemv_n_microk_haswell.c b/kernel/x86_64/sgemv_n_microk_haswell.c new file mode 100644 index 0000000..9db3869 --- /dev/null +++ b/kernel/x86_64/sgemv_n_microk_haswell.c @@ -0,0 +1,461 @@ +/*************************************************************************** +Copyright (c) 2014, The OpenBLAS Project +All rights reserved. +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are +met: +1. Redistributions of source code must retain the above copyright +notice, this list of conditions and the following disclaimer. +2. Redistributions in binary form must reproduce the above copyright +notice, this list of conditions and the following disclaimer in +the documentation and/or other materials provided with the +distribution. +3. Neither the name of the OpenBLAS project nor the names of +its contributors may be used to endorse or promote products +derived from this software without specific prior written permission. +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +ARE DISCLAIMED. IN NO EVENT SHALL THE OPENBLAS PROJECT OR CONTRIBUTORS BE +LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE +USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*****************************************************************************/ + +static void sgemv_kernel_64( long n, float alpha, float *a, long lda, float *x, float *y) +{ + + + float *pre = a + lda*2; + + __asm__ __volatile__ + ( + "movq %0, %%rax\n\t" // n -> rax + "vbroadcastss %1, %%ymm1\n\t" // alpha -> ymm1 + "movq %2, %%rsi\n\t" // adress of a -> rsi + "movq %3, %%rcx\n\t" // value of lda > rcx + "movq %4, %%rdi\n\t" // adress of x -> rdi + "movq %5, %%rdx\n\t" // adress of y -> rdx + "movq %6, %%r8\n\t" // address for prefetch + "prefetcht0 (%%r8)\n\t" // Prefetch + "prefetcht0 64(%%r8)\n\t" // Prefetch + + "vxorps %%ymm8 , %%ymm8 , %%ymm8 \n\t" // set to zero + "vxorps %%ymm9 , %%ymm9 , %%ymm9 \n\t" // set to zero + "vxorps %%ymm10, %%ymm10, %%ymm10\n\t" // set to zero + "vxorps %%ymm11, %%ymm11, %%ymm11\n\t" // set to zero + "vxorps %%ymm12, %%ymm12, %%ymm12\n\t" // set to zero + "vxorps %%ymm13, %%ymm13, %%ymm13\n\t" // set to zero + "vxorps %%ymm14, %%ymm14, %%ymm14\n\t" // set to zero + "vxorps %%ymm15, %%ymm15, %%ymm15\n\t" // set to zero + ".align 16 \n\t" + ".L01LOOP%=: \n\t" + "vbroadcastss (%%rdi), %%ymm0 \n\t" // load values of c + "leaq (%%r8 , %%rcx, 4), %%r8 \n\t" // add lda to pointer for prefetch + + "prefetcht0 (%%r8)\n\t" // Prefetch + "vfmadd231ps 0*4(%%rsi), %%ymm0, %%ymm8 \n\t" // multiply a and c and add to temp + "vfmadd231ps 8*4(%%rsi), %%ymm0, %%ymm9 \n\t" // multiply a and c and add to temp + "prefetcht0 64(%%r8)\n\t" // Prefetch + "vfmadd231ps 16*4(%%rsi), %%ymm0, %%ymm10\n\t" // multiply a and c and add to temp + "vfmadd231ps 24*4(%%rsi), %%ymm0, %%ymm11\n\t" // multiply a and c and add to temp + "prefetcht0 128(%%r8)\n\t" // Prefetch + "vfmadd231ps 32*4(%%rsi), %%ymm0, %%ymm12\n\t" // multiply a and c and add to temp + "vfmadd231ps 40*4(%%rsi), %%ymm0, %%ymm13\n\t" // multiply a and c and add to temp + "prefetcht0 192(%%r8)\n\t" // Prefetch + "vfmadd231ps 48*4(%%rsi), %%ymm0, %%ymm14\n\t" // multiply a and c and add to temp + "vfmadd231ps 56*4(%%rsi), %%ymm0, %%ymm15\n\t" // multiply a and c and add to temp + + "addq $4 , %%rdi \n\t" // increment pointer of c + "leaq (%%rsi, %%rcx, 4), %%rsi \n\t" // add lda to pointer of a + + "dec %%rax \n\t" // n = n -1 + "jnz .L01LOOP%= \n\t" + + "vmulps %%ymm8 , %%ymm1, %%ymm8 \n\t" // scale by alpha + "vmulps %%ymm9 , %%ymm1, %%ymm9 \n\t" // scale by alpha + "vmulps %%ymm10, %%ymm1, %%ymm10\n\t" // scale by alpha + "vmulps %%ymm11, %%ymm1, %%ymm11\n\t" // scale by alpha + "vmulps %%ymm12, %%ymm1, %%ymm12\n\t" // scale by alpha + "vmulps %%ymm13, %%ymm1, %%ymm13\n\t" // scale by alpha + "vmulps %%ymm14, %%ymm1, %%ymm14\n\t" // scale by alpha + "vmulps %%ymm15, %%ymm1, %%ymm15\n\t" // scale by alpha + + "vmovups %%ymm8 , (%%rdx) \n\t" // store temp -> y + "vmovups %%ymm9 , 8*4(%%rdx) \n\t" // store temp -> y + "vmovups %%ymm10, 16*4(%%rdx) \n\t" // store temp -> y + "vmovups %%ymm11, 24*4(%%rdx) \n\t" // store temp -> y + "vmovups %%ymm12, 32*4(%%rdx) \n\t" // store temp -> y + "vmovups %%ymm13, 40*4(%%rdx) \n\t" // store temp -> y + "vmovups %%ymm14, 48*4(%%rdx) \n\t" // store temp -> y + "vmovups %%ymm15, 56*4(%%rdx) \n\t" // store temp -> y + + : + : + "m" (n), // 0 + "m" (alpha), // 1 + "m" (a), // 2 + "m" (lda), // 3 + "m" (x), // 4 + "m" (y), // 5 + "m" (pre) // 6 + : "%rax", "%rcx", "%rdx", "%rsi", "%rdi", "%r8", "cc", + "%xmm0", "%xmm1", + "%xmm4", "%xmm5", "%xmm6", "%xmm7", + "%xmm8", "%xmm9", "%xmm10", "%xmm11", + "%xmm12", "%xmm13", "%xmm14", "%xmm15", + "memory" + ); + +} + + + +static void sgemv_kernel_32( long n, float alpha, float *a, long lda, float *x, float *y) +{ + + + float *pre = a + lda*3; + + __asm__ __volatile__ + ( + "movq %0, %%rax\n\t" // n -> rax + "vbroadcastss %1, %%ymm1\n\t" // alpha -> ymm1 + "movq %2, %%rsi\n\t" // adress of a -> rsi + "movq %3, %%rcx\n\t" // value of lda > rcx + "movq %4, %%rdi\n\t" // adress of x -> rdi + "movq %5, %%rdx\n\t" // adress of y -> rdx + "movq %6, %%r8\n\t" // address for prefetch + "prefetcht0 (%%r8)\n\t" // Prefetch + "prefetcht0 64(%%r8)\n\t" // Prefetch + + "vxorps %%ymm8 , %%ymm8 , %%ymm8 \n\t" // set to zero + "vxorps %%ymm9 , %%ymm9 , %%ymm9 \n\t" // set to zero + "vxorps %%ymm10, %%ymm10, %%ymm10\n\t" // set to zero + "vxorps %%ymm11, %%ymm11, %%ymm11\n\t" // set to zero + ".align 16 \n\t" + ".L01LOOP%=: \n\t" + "vbroadcastss (%%rdi), %%ymm0 \n\t" // load values of c + "nop \n\t" + "leaq (%%r8 , %%rcx, 4), %%r8 \n\t" // add lda to pointer for prefetch + + "prefetcht0 (%%r8)\n\t" // Prefetch + "prefetcht0 64(%%r8)\n\t" // Prefetch + + "vmulps 0*4(%%rsi), %%ymm0, %%ymm4 \n\t" // multiply a and c and add to temp + "vmulps 8*4(%%rsi), %%ymm0, %%ymm5 \n\t" // multiply a and c and add to temp + "vmulps 16*4(%%rsi), %%ymm0, %%ymm6 \n\t" // multiply a and c and add to temp + "vmulps 24*4(%%rsi), %%ymm0, %%ymm7 \n\t" // multiply a and c and add to temp + + "vaddps %%ymm8 , %%ymm4, %%ymm8 \n\t" // multiply a and c and add to temp + "vaddps %%ymm9 , %%ymm5, %%ymm9 \n\t" // multiply a and c and add to temp + "vaddps %%ymm10, %%ymm6, %%ymm10\n\t" // multiply a and c and add to temp + "vaddps %%ymm11, %%ymm7, %%ymm11\n\t" // multiply a and c and add to temp + + + + "addq $4 , %%rdi \n\t" // increment pointer of c + "leaq (%%rsi, %%rcx, 4), %%rsi \n\t" // add lda to pointer of a + + "dec %%rax \n\t" // n = n -1 + "jnz .L01LOOP%= \n\t" + + "vmulps %%ymm8 , %%ymm1, %%ymm8 \n\t" // scale by alpha + "vmulps %%ymm9 , %%ymm1, %%ymm9 \n\t" // scale by alpha + "vmulps %%ymm10, %%ymm1, %%ymm10\n\t" // scale by alpha + "vmulps %%ymm11, %%ymm1, %%ymm11\n\t" // scale by alpha + + "vmovups %%ymm8 , (%%rdx) \n\t" // store temp -> y + "vmovups %%ymm9 , 8*4(%%rdx) \n\t" // store temp -> y + "vmovups %%ymm10, 16*4(%%rdx) \n\t" // store temp -> y + "vmovups %%ymm11, 24*4(%%rdx) \n\t" // store temp -> y + + : + : + "m" (n), // 0 + "m" (alpha), // 1 + "m" (a), // 2 + "m" (lda), // 3 + "m" (x), // 4 + "m" (y), // 5 + "m" (pre) // 6 + : "%rax", "%rcx", "%rdx", "%rsi", "%rdi", "%r8", "cc", + "%xmm0", "%xmm1", + "%xmm4", "%xmm5", "%xmm6", "%xmm7", + "%xmm8", "%xmm9", "%xmm10", "%xmm11", + "memory" + ); + + + +} + +static void sgemv_kernel_16( long n, float alpha, float *a, long lda, float *x, float *y) +{ + + float *pre = a + lda*3; + + __asm__ __volatile__ + ( + "movq %0, %%rax\n\t" // n -> rax + "vbroadcastss %1, %%ymm1\n\t" // alpha -> ymm1 + "movq %2, %%rsi\n\t" // adress of a -> rsi + "movq %3, %%rcx\n\t" // value of lda > rcx + "movq %4, %%rdi\n\t" // adress of x -> rdi + "movq %5, %%rdx\n\t" // adress of y -> rdx + "movq %6, %%r8\n\t" // address for prefetch + "prefetcht0 (%%r8)\n\t" // Prefetch + "prefetcht0 64(%%r8)\n\t" // Prefetch + + "vxorps %%ymm8 , %%ymm8 , %%ymm8 \n\t" // set to zero + "vxorps %%ymm9 , %%ymm9 , %%ymm9 \n\t" // set to zero + ".align 16 \n\t" + ".L01LOOP%=: \n\t" + "vbroadcastss (%%rdi), %%ymm0 \n\t" // load values of c + "nop \n\t" + "leaq (%%r8 , %%rcx, 4), %%r8 \n\t" // add lda to pointer for prefetch + + "prefetcht0 (%%r8)\n\t" // Prefetch + + "vmulps 0*4(%%rsi), %%ymm0, %%ymm4 \n\t" // multiply a and c and add to temp + "vmulps 8*4(%%rsi), %%ymm0, %%ymm5 \n\t" // multiply a and c and add to temp + + "vaddps %%ymm8 , %%ymm4, %%ymm8 \n\t" // multiply a and c and add to temp + "vaddps %%ymm9 , %%ymm5, %%ymm9 \n\t" // multiply a and c and add to temp + + "addq $4 , %%rdi \n\t" // increment pointer of c + "leaq (%%rsi, %%rcx, 4), %%rsi \n\t" // add lda to pointer of a + + "dec %%rax \n\t" // n = n -1 + "jnz .L01LOOP%= \n\t" + + "vmulps %%ymm8 , %%ymm1, %%ymm8 \n\t" // scale by alpha + "vmulps %%ymm9 , %%ymm1, %%ymm9 \n\t" // scale by alpha + + "vmovups %%ymm8 , (%%rdx) \n\t" // store temp -> y + "vmovups %%ymm9 , 8*4(%%rdx) \n\t" // store temp -> y + + : + : + "m" (n), // 0 + "m" (alpha), // 1 + "m" (a), // 2 + "m" (lda), // 3 + "m" (x), // 4 + "m" (y), // 5 + "m" (pre) // 6 + : "%rax", "%rcx", "%rdx", "%rsi", "%rdi", "%r8", "cc", + "%xmm0", "%xmm1", + "%xmm4", "%xmm5", "%xmm6", "%xmm7", + "%xmm8", "%xmm9", "%xmm10", "%xmm11", + "memory" + ); + + +} + + +static void sgemv_kernel_8( long n, float alpha, float *a, long lda, float *x, float *y) +{ + + __asm__ __volatile__ + ( + "movq %0, %%rax\n\t" // n -> rax + "vbroadcastss %1, %%ymm1\n\t" // alpha -> ymm1 + "movq %2, %%rsi\n\t" // adress of a -> rsi + "movq %3, %%rcx\n\t" // value of lda > rcx + "movq %4, %%rdi\n\t" // adress of x -> rdi + "movq %5, %%rdx\n\t" // adress of y -> rdx + + "vxorps %%ymm8 , %%ymm8 , %%ymm8 \n\t" // set to zero + ".align 16 \n\t" + ".L01LOOP%=: \n\t" + "vbroadcastss (%%rdi), %%ymm0 \n\t" // load values of c + + "vmulps 0*4(%%rsi), %%ymm0, %%ymm4 \n\t" // multiply a and c and add to temp + "vaddps %%ymm8 , %%ymm4, %%ymm8 \n\t" // multiply a and c and add to temp + + "addq $4 , %%rdi \n\t" // increment pointer of c + "leaq (%%rsi, %%rcx, 4), %%rsi \n\t" // add lda to pointer of a + + "dec %%rax \n\t" // n = n -1 + "jnz .L01LOOP%= \n\t" + + "vmulps %%ymm8 , %%ymm1, %%ymm8 \n\t" // scale by alpha + "vmovups %%ymm8 , (%%rdx) \n\t" // store temp -> y + + : + : + "m" (n), // 0 + "m" (alpha), // 1 + "m" (a), // 2 + "m" (lda), // 3 + "m" (x), // 4 + "m" (y) // 5 + : "%rax", "%rcx", "%rdx", "%rsi", "%rdi", "%r8", "cc", + "%xmm0", "%xmm1", + "%xmm4", "%xmm5", "%xmm6", "%xmm7", + "%xmm8", "%xmm9", "%xmm10", "%xmm11", + "memory" + ); + + +} + + +static void sgemv_kernel_4( long n, float alpha, float *a, long lda, float *x, float *y) +{ + + + __asm__ __volatile__ + ( + "movq %0, %%rax\n\t" // n -> rax + "vbroadcastss %1, %%xmm1\n\t" // alpha -> xmm1 + "movq %2, %%rsi\n\t" // adress of a -> rsi + "movq %3, %%rcx\n\t" // value of lda > rcx + "movq %4, %%rdi\n\t" // adress of x -> rdi + "movq %5, %%rdx\n\t" // adress of y -> rdx + + "vxorps %%xmm12, %%xmm12, %%xmm12\n\t" // set to zero + + ".L01LOOP%=: \n\t" + "vbroadcastss (%%rdi), %%xmm0 \n\t" // load values of c + + "vmulps 0*4(%%rsi), %%xmm0, %%xmm4 \n\t" // multiply a and c and add to temp + "vaddps %%xmm12, %%xmm4, %%xmm12 \n\t" // multiply a and c and add to temp + + "addq $4 , %%rdi \n\t" // increment pointer of c + "leaq (%%rsi, %%rcx, 4), %%rsi \n\t" // add lda to pointer of a + + "dec %%rax \n\t" // n = n -1 + "jnz .L01LOOP%= \n\t" + + "vmulps %%xmm12, %%xmm1, %%xmm12\n\t" // scale by alpha + + "vmovups %%xmm12, (%%rdx) \n\t" // store temp -> y + + : + : + "m" (n), // 0 + "m" (alpha), // 1 + "m" (a), // 2 + "m" (lda), // 3 + "m" (x), // 4 + "m" (y) // 5 + : "%rax", "%rcx", "%rdx", "%rsi", "%rdi", "%r8", + "%xmm0", "%xmm1", + "%xmm8", "%xmm9", "%xmm10", "%xmm11", + "%xmm12", "%xmm13", "%xmm14", "%xmm15", + "memory" + ); + +} + +static void sgemv_kernel_2( long n, float alpha, float *a, long lda, float *x, float *y) +{ + + + __asm__ __volatile__ + ( + "movq %0, %%rax\n\t" // n -> rax + "vmovss %1, %%xmm1\n\t" // alpha -> xmm1 + "movq %2, %%rsi\n\t" // adress of a -> rsi + "movq %3, %%rcx\n\t" // value of lda > rcx + "movq %4, %%rdi\n\t" // adress of x -> rdi + "movq %5, %%rdx\n\t" // adress of y -> rdx + + "vxorps %%xmm12, %%xmm12, %%xmm12\n\t" // set to zero + "vxorps %%xmm13, %%xmm13, %%xmm13\n\t" // set to zero + + ".L01LOOP%=: \n\t" + "vmovss (%%rdi), %%xmm0 \n\t" // load values of c + + "vmulps 0*4(%%rsi), %%xmm0, %%xmm4 \n\t" // multiply a and c and add to temp + "vmulps 1*4(%%rsi), %%xmm0, %%xmm5 \n\t" // multiply a and c and add to temp + + "vaddps %%xmm12, %%xmm4, %%xmm12 \n\t" // multiply a and c and add to temp + "vaddps %%xmm13, %%xmm5, %%xmm13 \n\t" // multiply a and c and add to temp + + "addq $4 , %%rdi \n\t" // increment pointer of c + "leaq (%%rsi, %%rcx, 4), %%rsi \n\t" // add lda to pointer of a + + "dec %%rax \n\t" // n = n -1 + "jnz .L01LOOP%= \n\t" + + "vmulss %%xmm12, %%xmm1, %%xmm12\n\t" // scale by alpha + "vmulss %%xmm13, %%xmm1, %%xmm13\n\t" // scale by alpha + + "vmovss %%xmm12, (%%rdx) \n\t" // store temp -> y + "vmovss %%xmm13, 4(%%rdx) \n\t" // store temp -> y + + : + : + "m" (n), // 0 + "m" (alpha), // 1 + "m" (a), // 2 + "m" (lda), // 3 + "m" (x), // 4 + "m" (y) // 5 + : "%rax", "%rcx", "%rdx", "%rsi", "%rdi", "%r8", + "%xmm0", "%xmm1", + "%xmm8", "%xmm9", "%xmm10", "%xmm11", + "%xmm12", "%xmm13", "%xmm14", "%xmm15", + "memory" + ); + +} + + + +static void sgemv_kernel_1( long n, float alpha, float *a, long lda, float *x, float *y) +{ + + + __asm__ __volatile__ + ( + "movq %0, %%rax\n\t" // n -> rax + "vmovss %1, %%xmm1\n\t" // alpha -> xmm1 + "movq %2, %%rsi\n\t" // adress of a -> rsi + "movq %3, %%rcx\n\t" // value of lda > rcx + "movq %4, %%rdi\n\t" // adress of x -> rdi + "movq %5, %%rdx\n\t" // adress of y -> rdx + + "vxorps %%xmm12, %%xmm12, %%xmm12\n\t" // set to zero + + ".L01LOOP%=: \n\t" + "vmovss (%%rdi), %%xmm0 \n\t" // load values of c + "addq $4 , %%rdi \n\t" // increment pointer of c + + "vmulss 0*4(%%rsi), %%xmm0, %%xmm4 \n\t" // multiply a and c and add to temp + "vaddss %%xmm12, %%xmm4, %%xmm12 \n\t" // multiply a and c and add to temp + + "leaq (%%rsi, %%rcx, 4), %%rsi \n\t" // add lda to pointer of a + + "dec %%rax \n\t" // n = n -1 + "jnz .L01LOOP%= \n\t" + + "vmulss %%xmm12, %%xmm1, %%xmm12\n\t" // scale by alpha + + "vmovss %%xmm12, (%%rdx) \n\t" // store temp -> y + + : + : + "m" (n), // 0 + "m" (alpha), // 1 + "m" (a), // 2 + "m" (lda), // 3 + "m" (x), // 4 + "m" (y) // 5 + : "%rax", "%rcx", "%rdx", "%rsi", "%rdi", "%r8", + "%xmm0", "%xmm1", + "%xmm8", "%xmm9", "%xmm10", "%xmm11", + "%xmm12", "%xmm13", "%xmm14", "%xmm15", + "memory" + ); + +} + + -- 2.7.4