b073d759d0
As observed using GCC 10 using -march=native -ftree-vectorize on Knights Landing, it is now smart enough to find clobbers inside non-inlined static functions. In particular, sgemv counted on a kernel to preserve the whole %ymm2 register (since it was not in the clobber list), but the top part was destroyed by vzeroupper. This caused many tests to fail. This patch makes sure all xmm (and ymm/zmm by extension) registers are listed as clobbered to avoid this happening, as most kernels already did correctly in fact.
202 lines
7.6 KiB
C
202 lines
7.6 KiB
C
/***************************************************************************
|
|
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.
|
|
*****************************************************************************/
|
|
|
|
#define HAVE_KERNEL_8 1
|
|
static void caxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y , FLOAT *alpha) __attribute__ ((noinline));
|
|
|
|
static void caxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha)
|
|
{
|
|
|
|
#if !defined(CONJ)
|
|
FLOAT mvec[4] = { -1.0, 1.0, -1.0, 1.0 };
|
|
#else
|
|
FLOAT mvec[4] = { 1.0, -1.0, 1.0, -1.0 };
|
|
#endif
|
|
|
|
BLASLONG register i = 0;
|
|
|
|
if ( n <= 2048 )
|
|
{
|
|
|
|
__asm__ __volatile__
|
|
(
|
|
"vzeroupper \n\t"
|
|
"vbroadcastss (%4), %%xmm0 \n\t" // real part of alpha
|
|
"vbroadcastss 4(%4), %%xmm1 \n\t" // imag part of alpha
|
|
#if !defined(CONJ)
|
|
"vmulps (%5), %%xmm1 , %%xmm1 \n\t"
|
|
#else
|
|
"vmulps (%5), %%xmm0 , %%xmm0 \n\t"
|
|
#endif
|
|
|
|
".align 16 \n\t"
|
|
"1: \n\t"
|
|
|
|
"vmovups (%2,%0,4), %%xmm5 \n\t" // 2 complex values from x
|
|
".align 2 \n\t"
|
|
"vmovups 16(%2,%0,4), %%xmm7 \n\t" // 2 complex values from x
|
|
"vmovups 32(%2,%0,4), %%xmm9 \n\t" // 2 complex values from x
|
|
"vmovups 48(%2,%0,4), %%xmm11 \n\t" // 2 complex values from x
|
|
|
|
"vmovups 64(%2,%0,4), %%xmm12 \n\t" // 2 complex values from x
|
|
"vmovups 80(%2,%0,4), %%xmm13 \n\t" // 2 complex values from x
|
|
"vmovups 96(%2,%0,4), %%xmm14 \n\t" // 2 complex values from x
|
|
"vmovups 112(%2,%0,4), %%xmm15 \n\t" // 2 complex values from x
|
|
|
|
"vpermilps $0xb1 , %%xmm5 , %%xmm4 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm7 , %%xmm6 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm9 , %%xmm8 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm11, %%xmm10 \n\t" // exchange real and imag part
|
|
|
|
"vfmadd213ps (%3,%0,4), %%xmm0 , %%xmm5 \n\t"
|
|
".align 2 \n\t"
|
|
"vfmadd213ps 16(%3,%0,4), %%xmm0 , %%xmm7 \n\t"
|
|
"vfmadd213ps 32(%3,%0,4), %%xmm0 , %%xmm9 \n\t"
|
|
"vfmadd213ps 48(%3,%0,4), %%xmm0 , %%xmm11 \n\t"
|
|
|
|
"vfmadd231ps %%xmm1 , %%xmm4 , %%xmm5 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm6 , %%xmm7 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm8 , %%xmm9 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm10, %%xmm11 \n\t"
|
|
|
|
"vpermilps $0xb1 , %%xmm12, %%xmm4 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm13, %%xmm6 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm14, %%xmm8 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm15, %%xmm10 \n\t" // exchange real and imag part
|
|
|
|
"vfmadd213ps 64(%3,%0,4), %%xmm0 , %%xmm12 \n\t"
|
|
"vfmadd213ps 80(%3,%0,4), %%xmm0 , %%xmm13 \n\t"
|
|
"vfmadd213ps 96(%3,%0,4), %%xmm0 , %%xmm14 \n\t"
|
|
"vfmadd213ps 112(%3,%0,4), %%xmm0 , %%xmm15 \n\t"
|
|
|
|
"vfmadd231ps %%xmm1 , %%xmm4 , %%xmm12 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm6 , %%xmm13 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm8 , %%xmm14 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm10, %%xmm15 \n\t"
|
|
|
|
"vmovups %%xmm5 , (%3,%0,4) \n\t"
|
|
".align 2 \n\t"
|
|
"vmovups %%xmm7 , 16(%3,%0,4) \n\t"
|
|
"vmovups %%xmm9 , 32(%3,%0,4) \n\t"
|
|
"vmovups %%xmm11, 48(%3,%0,4) \n\t"
|
|
"vmovups %%xmm12, 64(%3,%0,4) \n\t"
|
|
"vmovups %%xmm13, 80(%3,%0,4) \n\t"
|
|
"vmovups %%xmm14, 96(%3,%0,4) \n\t"
|
|
"vmovups %%xmm15,112(%3,%0,4) \n\t"
|
|
|
|
"addq $32, %0 \n\t"
|
|
"subq $16, %1 \n\t"
|
|
"jnz 1b \n\t"
|
|
"vzeroupper \n\t"
|
|
|
|
:
|
|
"+r" (i), // 0
|
|
"+r" (n) // 1
|
|
:
|
|
"r" (x), // 2
|
|
"r" (y), // 3
|
|
"r" (alpha), // 4
|
|
"r" (mvec) // 5
|
|
: "cc",
|
|
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
|
|
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
|
|
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
|
|
"%xmm12", "%xmm13", "%xmm14", "%xmm15",
|
|
"memory"
|
|
);
|
|
return;
|
|
}
|
|
|
|
__asm__ __volatile__
|
|
(
|
|
"vzeroupper \n\t"
|
|
"vbroadcastss (%4), %%xmm0 \n\t" // real part of alpha
|
|
"vbroadcastss 4(%4), %%xmm1 \n\t" // imag part of alpha
|
|
#if !defined(CONJ)
|
|
"vmulps (%5), %%xmm1 , %%xmm1 \n\t"
|
|
#else
|
|
"vmulps (%5), %%xmm0 , %%xmm0 \n\t"
|
|
#endif
|
|
|
|
".align 16 \n\t"
|
|
"1: \n\t"
|
|
|
|
"prefetcht0 512(%2,%0,4) \n\t"
|
|
"vmovups (%2,%0,4), %%xmm5 \n\t" // 2 complex values from x
|
|
".align 2 \n\t"
|
|
"vmovups 16(%2,%0,4), %%xmm7 \n\t" // 2 complex values from x
|
|
"vmovups 32(%2,%0,4), %%xmm9 \n\t" // 2 complex values from x
|
|
"vmovups 48(%2,%0,4), %%xmm11 \n\t" // 2 complex values from x
|
|
|
|
"vpermilps $0xb1 , %%xmm5 , %%xmm4 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm7 , %%xmm6 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm9 , %%xmm8 \n\t" // exchange real and imag part
|
|
"vpermilps $0xb1 , %%xmm11, %%xmm10 \n\t" // exchange real and imag part
|
|
|
|
"prefetcht0 512(%3,%0,4) \n\t"
|
|
"vfmadd213ps (%3,%0,4), %%xmm0 , %%xmm5 \n\t"
|
|
".align 2 \n\t"
|
|
"vfmadd213ps 16(%3,%0,4), %%xmm0 , %%xmm7 \n\t"
|
|
"vfmadd213ps 32(%3,%0,4), %%xmm0 , %%xmm9 \n\t"
|
|
"vfmadd213ps 48(%3,%0,4), %%xmm0 , %%xmm11 \n\t"
|
|
|
|
"vfmadd231ps %%xmm1 , %%xmm4 , %%xmm5 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm6 , %%xmm7 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm8 , %%xmm9 \n\t"
|
|
"vfmadd231ps %%xmm1 , %%xmm10, %%xmm11 \n\t"
|
|
|
|
"vmovups %%xmm5 , (%3,%0,4) \n\t"
|
|
".align 2 \n\t"
|
|
"vmovups %%xmm7 , 16(%3,%0,4) \n\t"
|
|
"vmovups %%xmm9 , 32(%3,%0,4) \n\t"
|
|
"vmovups %%xmm11, 48(%3,%0,4) \n\t"
|
|
|
|
"addq $16, %0 \n\t"
|
|
"subq $8, %1 \n\t"
|
|
"jnz 1b \n\t"
|
|
"vzeroupper \n\t"
|
|
|
|
:
|
|
"+r" (i), // 0
|
|
"+r" (n) // 1
|
|
:
|
|
"r" (x), // 2
|
|
"r" (y), // 3
|
|
"r" (alpha), // 4
|
|
"r" (mvec) // 5
|
|
: "cc",
|
|
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
|
|
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
|
|
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
|
|
"%xmm12", "%xmm13", "%xmm14", "%xmm15",
|
|
"memory"
|
|
);
|
|
|
|
|
|
}
|
|
|