diff options
Diffstat (limited to 'sysdeps/x86_64')
-rw-r--r-- | sysdeps/x86_64/fpu/multiarch/svml_d_asin2_core_sse4.S | 461 |
1 files changed, 230 insertions, 231 deletions
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_asin2_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_d_asin2_core_sse4.S index 00723482a2..e37f36560b 100644 --- a/sysdeps/x86_64/fpu/multiarch/svml_d_asin2_core_sse4.S +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_asin2_core_sse4.S @@ -27,262 +27,261 @@ /* Offsets for data table __svml_dasin_data_internal */ -#define AbsMask 0 -#define OneHalf 16 -#define SmallNorm 32 -#define One 48 -#define Two 64 -#define sqrt_coeff 80 -#define poly_coeff 144 -#define Pi2H 336 +#define AbsMask 0 +#define OneHalf 16 +#define SmallNorm 32 +#define One 48 +#define Two 64 +#define sqrt_coeff 80 +#define poly_coeff 144 +#define Pi2H 336 #include <sysdep.h> - .text - .section .text.sse4,"ax",@progbits + .section .text.sse4, "ax", @progbits ENTRY(_ZGVbN2v_asin_sse4) - subq $72, %rsp - cfi_def_cfa_offset(80) - movaps %xmm0, %xmm5 - movups __svml_dasin_data_internal(%rip), %xmm3 - movups OneHalf+__svml_dasin_data_internal(%rip), %xmm8 - -/* x = |arg| */ - movaps %xmm3, %xmm4 - andps %xmm5, %xmm4 - -/* Y = 0.5 - 0.5*x */ - movaps %xmm8, %xmm6 - mulpd %xmm4, %xmm6 - movaps %xmm8, %xmm14 - -/* x^2 */ - movaps %xmm4, %xmm2 - subpd %xmm6, %xmm14 - mulpd %xmm4, %xmm2 - -/* S ~ -2*sqrt(Y) */ - cvtpd2ps %xmm14, %xmm9 - minpd %xmm14, %xmm2 - movlhps %xmm9, %xmm9 - movaps %xmm14, %xmm15 - rsqrtps %xmm9, %xmm10 - cmpltpd SmallNorm+__svml_dasin_data_internal(%rip), %xmm15 - addpd %xmm14, %xmm14 - cvtps2pd %xmm10, %xmm11 - andnps %xmm11, %xmm15 - movaps %xmm4, %xmm1 - movaps %xmm15, %xmm12 - andnps %xmm5, %xmm3 - mulpd %xmm15, %xmm12 - mulpd %xmm14, %xmm15 - mulpd %xmm12, %xmm14 - cmpnltpd %xmm8, %xmm1 - subpd Two+__svml_dasin_data_internal(%rip), %xmm14 - -/* polynomial */ - movups poly_coeff+__svml_dasin_data_internal(%rip), %xmm6 - movaps %xmm2, %xmm12 - mulpd %xmm2, %xmm6 - mulpd %xmm2, %xmm12 - addpd poly_coeff+16+__svml_dasin_data_internal(%rip), %xmm6 - movups One+__svml_dasin_data_internal(%rip), %xmm7 - movaps %xmm12, %xmm8 - cmpltpd %xmm4, %xmm7 - mulpd %xmm12, %xmm6 - movmskpd %xmm7, %edx - movups poly_coeff+32+__svml_dasin_data_internal(%rip), %xmm9 - movaps %xmm14, %xmm0 - movups poly_coeff+64+__svml_dasin_data_internal(%rip), %xmm7 - mulpd %xmm2, %xmm9 - mulpd %xmm2, %xmm7 - addpd poly_coeff+48+__svml_dasin_data_internal(%rip), %xmm9 - addpd poly_coeff+80+__svml_dasin_data_internal(%rip), %xmm7 - mulpd %xmm12, %xmm8 - mulpd %xmm12, %xmm7 - addpd %xmm6, %xmm9 - mulpd %xmm15, %xmm0 - mulpd %xmm8, %xmm9 - movups poly_coeff+96+__svml_dasin_data_internal(%rip), %xmm10 - mulpd %xmm2, %xmm10 - movups sqrt_coeff+__svml_dasin_data_internal(%rip), %xmm13 - mulpd %xmm14, %xmm13 - addpd poly_coeff+112+__svml_dasin_data_internal(%rip), %xmm10 - addpd sqrt_coeff+16+__svml_dasin_data_internal(%rip), %xmm13 - addpd %xmm7, %xmm10 - mulpd %xmm14, %xmm13 - addpd %xmm9, %xmm10 - addpd sqrt_coeff+32+__svml_dasin_data_internal(%rip), %xmm13 - mulpd %xmm12, %xmm10 - mulpd %xmm13, %xmm14 - movups poly_coeff+128+__svml_dasin_data_internal(%rip), %xmm11 - mulpd %xmm2, %xmm11 - addpd sqrt_coeff+48+__svml_dasin_data_internal(%rip), %xmm14 - addpd poly_coeff+144+__svml_dasin_data_internal(%rip), %xmm11 - mulpd %xmm14, %xmm0 - addpd %xmm10, %xmm11 - subpd %xmm15, %xmm0 - mulpd %xmm11, %xmm12 - movups poly_coeff+160+__svml_dasin_data_internal(%rip), %xmm13 - movaps %xmm1, %xmm14 - mulpd %xmm2, %xmm13 - addpd poly_coeff+176+__svml_dasin_data_internal(%rip), %xmm13 - addpd %xmm12, %xmm13 - mulpd %xmm13, %xmm2 - andnps %xmm4, %xmm14 - andps %xmm1, %xmm0 - orps %xmm0, %xmm14 - mulpd %xmm14, %xmm2 - addpd %xmm2, %xmm14 - movups Pi2H+__svml_dasin_data_internal(%rip), %xmm0 - andps %xmm1, %xmm0 - addpd %xmm14, %xmm0 - pxor %xmm3, %xmm0 - testl %edx, %edx - -/* Go to special inputs processing branch */ - jne L(SPECIAL_VALUES_BRANCH) - # LOE rbx rbp r12 r13 r14 r15 edx xmm0 xmm5 - -/* Restore registers - * and exit the function - */ + subq $72, %rsp + cfi_def_cfa_offset(80) + movaps %xmm0, %xmm5 + movups __svml_dasin_data_internal(%rip), %xmm3 + movups OneHalf+__svml_dasin_data_internal(%rip), %xmm8 + + /* x = |arg| */ + movaps %xmm3, %xmm4 + andps %xmm5, %xmm4 + + /* Y = 0.5 - 0.5*x */ + movaps %xmm8, %xmm6 + mulpd %xmm4, %xmm6 + movaps %xmm8, %xmm14 + + /* x^2 */ + movaps %xmm4, %xmm2 + subpd %xmm6, %xmm14 + mulpd %xmm4, %xmm2 + + /* S ~ -2*sqrt(Y) */ + cvtpd2ps %xmm14, %xmm9 + minpd %xmm14, %xmm2 + movlhps %xmm9, %xmm9 + movaps %xmm14, %xmm15 + rsqrtps %xmm9, %xmm10 + cmpltpd SmallNorm+__svml_dasin_data_internal(%rip), %xmm15 + addpd %xmm14, %xmm14 + cvtps2pd %xmm10, %xmm11 + andnps %xmm11, %xmm15 + movaps %xmm4, %xmm1 + movaps %xmm15, %xmm12 + andnps %xmm5, %xmm3 + mulpd %xmm15, %xmm12 + mulpd %xmm14, %xmm15 + mulpd %xmm12, %xmm14 + cmpnltpd %xmm8, %xmm1 + subpd Two+__svml_dasin_data_internal(%rip), %xmm14 + + /* polynomial */ + movups poly_coeff+__svml_dasin_data_internal(%rip), %xmm6 + movaps %xmm2, %xmm12 + mulpd %xmm2, %xmm6 + mulpd %xmm2, %xmm12 + addpd poly_coeff+16+__svml_dasin_data_internal(%rip), %xmm6 + movups One+__svml_dasin_data_internal(%rip), %xmm7 + movaps %xmm12, %xmm8 + cmpltpd %xmm4, %xmm7 + mulpd %xmm12, %xmm6 + movmskpd %xmm7, %edx + movups poly_coeff+32+__svml_dasin_data_internal(%rip), %xmm9 + movaps %xmm14, %xmm0 + movups poly_coeff+64+__svml_dasin_data_internal(%rip), %xmm7 + mulpd %xmm2, %xmm9 + mulpd %xmm2, %xmm7 + addpd poly_coeff+48+__svml_dasin_data_internal(%rip), %xmm9 + addpd poly_coeff+80+__svml_dasin_data_internal(%rip), %xmm7 + mulpd %xmm12, %xmm8 + mulpd %xmm12, %xmm7 + addpd %xmm6, %xmm9 + mulpd %xmm15, %xmm0 + mulpd %xmm8, %xmm9 + movups poly_coeff+96+__svml_dasin_data_internal(%rip), %xmm10 + mulpd %xmm2, %xmm10 + movups sqrt_coeff+__svml_dasin_data_internal(%rip), %xmm13 + mulpd %xmm14, %xmm13 + addpd poly_coeff+112+__svml_dasin_data_internal(%rip), %xmm10 + addpd sqrt_coeff+16+__svml_dasin_data_internal(%rip), %xmm13 + addpd %xmm7, %xmm10 + mulpd %xmm14, %xmm13 + addpd %xmm9, %xmm10 + addpd sqrt_coeff+32+__svml_dasin_data_internal(%rip), %xmm13 + mulpd %xmm12, %xmm10 + mulpd %xmm13, %xmm14 + movups poly_coeff+128+__svml_dasin_data_internal(%rip), %xmm11 + mulpd %xmm2, %xmm11 + addpd sqrt_coeff+48+__svml_dasin_data_internal(%rip), %xmm14 + addpd poly_coeff+144+__svml_dasin_data_internal(%rip), %xmm11 + mulpd %xmm14, %xmm0 + addpd %xmm10, %xmm11 + subpd %xmm15, %xmm0 + mulpd %xmm11, %xmm12 + movups poly_coeff+160+__svml_dasin_data_internal(%rip), %xmm13 + movaps %xmm1, %xmm14 + mulpd %xmm2, %xmm13 + addpd poly_coeff+176+__svml_dasin_data_internal(%rip), %xmm13 + addpd %xmm12, %xmm13 + mulpd %xmm13, %xmm2 + andnps %xmm4, %xmm14 + andps %xmm1, %xmm0 + orps %xmm0, %xmm14 + mulpd %xmm14, %xmm2 + addpd %xmm2, %xmm14 + movups Pi2H+__svml_dasin_data_internal(%rip), %xmm0 + andps %xmm1, %xmm0 + addpd %xmm14, %xmm0 + pxor %xmm3, %xmm0 + testl %edx, %edx + + /* Go to special inputs processing branch */ + jne L(SPECIAL_VALUES_BRANCH) + # LOE rbx rbp r12 r13 r14 r15 edx xmm0 xmm5 + + /* Restore registers + * and exit the function + */ L(EXIT): - addq $72, %rsp - cfi_def_cfa_offset(8) - ret - cfi_def_cfa_offset(80) + addq $72, %rsp + cfi_def_cfa_offset(8) + ret + cfi_def_cfa_offset(80) -/* Branch to process - * special inputs - */ + /* Branch to process + * special inputs + */ L(SPECIAL_VALUES_BRANCH): - movups %xmm5, 32(%rsp) - movups %xmm0, 48(%rsp) - # LOE rbx rbp r12 r13 r14 r15 edx - - xorl %eax, %eax - movq %r12, 16(%rsp) - cfi_offset(12, -64) - movl %eax, %r12d - movq %r13, 8(%rsp) - cfi_offset(13, -72) - movl %edx, %r13d - movq %r14, (%rsp) - cfi_offset(14, -80) - # LOE rbx rbp r15 r12d r13d - -/* Range mask - * bits check - */ + movups %xmm5, 32(%rsp) + movups %xmm0, 48(%rsp) + # LOE rbx rbp r12 r13 r14 r15 edx + + xorl %eax, %eax + movq %r12, 16(%rsp) + cfi_offset(12, -64) + movl %eax, %r12d + movq %r13, 8(%rsp) + cfi_offset(13, -72) + movl %edx, %r13d + movq %r14, (%rsp) + cfi_offset(14, -80) + # LOE rbx rbp r15 r12d r13d + + /* Range mask + * bits check + */ L(RANGEMASK_CHECK): - btl %r12d, %r13d + btl %r12d, %r13d -/* Call scalar math function */ - jc L(SCALAR_MATH_CALL) - # LOE rbx rbp r15 r12d r13d + /* Call scalar math function */ + jc L(SCALAR_MATH_CALL) + # LOE rbx rbp r15 r12d r13d -/* Special inputs - * processing loop - */ + /* Special inputs + * processing loop + */ L(SPECIAL_VALUES_LOOP): - incl %r12d - cmpl $2, %r12d - -/* Check bits in range mask */ - jl L(RANGEMASK_CHECK) - # LOE rbx rbp r15 r12d r13d - - movq 16(%rsp), %r12 - cfi_restore(12) - movq 8(%rsp), %r13 - cfi_restore(13) - movq (%rsp), %r14 - cfi_restore(14) - movups 48(%rsp), %xmm0 - -/* Go to exit */ - jmp L(EXIT) - cfi_offset(12, -64) - cfi_offset(13, -72) - cfi_offset(14, -80) - # LOE rbx rbp r12 r13 r14 r15 xmm0 - -/* Scalar math fucntion call - * to process special input - */ + incl %r12d + cmpl $2, %r12d + + /* Check bits in range mask */ + jl L(RANGEMASK_CHECK) + # LOE rbx rbp r15 r12d r13d + + movq 16(%rsp), %r12 + cfi_restore(12) + movq 8(%rsp), %r13 + cfi_restore(13) + movq (%rsp), %r14 + cfi_restore(14) + movups 48(%rsp), %xmm0 + + /* Go to exit */ + jmp L(EXIT) + cfi_offset(12, -64) + cfi_offset(13, -72) + cfi_offset(14, -80) + # LOE rbx rbp r12 r13 r14 r15 xmm0 + + /* Scalar math fucntion call + * to process special input + */ L(SCALAR_MATH_CALL): - movl %r12d, %r14d - movsd 32(%rsp,%r14,8), %xmm0 - call asin@PLT - # LOE rbx rbp r14 r15 r12d r13d xmm0 + movl %r12d, %r14d + movsd 32(%rsp, %r14, 8), %xmm0 + call asin@PLT + # LOE rbx rbp r14 r15 r12d r13d xmm0 - movsd %xmm0, 48(%rsp,%r14,8) + movsd %xmm0, 48(%rsp, %r14, 8) -/* Process special inputs in loop */ - jmp L(SPECIAL_VALUES_LOOP) - # LOE rbx rbp r15 r12d r13d + /* Process special inputs in loop */ + jmp L(SPECIAL_VALUES_LOOP) + # LOE rbx rbp r15 r12d r13d END(_ZGVbN2v_asin_sse4) - .section .rodata, "a" - .align 16 + .section .rodata, "a" + .align 16 #ifdef __svml_dasin_data_internal_typedef typedef unsigned int VUINT32; typedef struct { - __declspec(align(16)) VUINT32 AbsMask[2][2]; - __declspec(align(16)) VUINT32 OneHalf[2][2]; - __declspec(align(16)) VUINT32 SmallNorm[2][2]; - __declspec(align(16)) VUINT32 One[2][2]; - __declspec(align(16)) VUINT32 Two[2][2]; - __declspec(align(16)) VUINT32 sqrt_coeff[4][2][2]; - __declspec(align(16)) VUINT32 poly_coeff[12][2][2]; - __declspec(align(16)) VUINT32 Pi2H[2][2]; + __declspec(align(16)) VUINT32 AbsMask[2][2]; + __declspec(align(16)) VUINT32 OneHalf[2][2]; + __declspec(align(16)) VUINT32 SmallNorm[2][2]; + __declspec(align(16)) VUINT32 One[2][2]; + __declspec(align(16)) VUINT32 Two[2][2]; + __declspec(align(16)) VUINT32 sqrt_coeff[4][2][2]; + __declspec(align(16)) VUINT32 poly_coeff[12][2][2]; + __declspec(align(16)) VUINT32 Pi2H[2][2]; } __svml_dasin_data_internal; #endif __svml_dasin_data_internal: - /*== AbsMask ==*/ - .quad 0x7fffffffffffffff, 0x7fffffffffffffff - /*== OneHalf ==*/ - .align 16 - .quad 0x3fe0000000000000, 0x3fe0000000000000 - /*== SmallNorm ==*/ - .align 16 - .quad 0x3000000000000000, 0x3000000000000000 - /*== One ==*/ - .align 16 - .quad 0x3ff0000000000000, 0x3ff0000000000000 - /*== Two ==*/ - .align 16 - .quad 0x4000000000000000, 0x4000000000000000 - /*== sqrt_coeff[4] ==*/ - .align 16 - .quad 0xbf918000993B24C3, 0xbf918000993B24C3 /* sqrt_coeff4 */ - .quad 0x3fa400006F70D42D, 0x3fa400006F70D42D /* sqrt_coeff3 */ - .quad 0xbfb7FFFFFFFFFE97, 0xbfb7FFFFFFFFFE97 /* sqrt_coeff2 */ - .quad 0x3fcFFFFFFFFFFF9D, 0x3fcFFFFFFFFFFF9D /* sqrt_coeff1 */ - /*== poly_coeff[12] ==*/ - .align 16 - .quad 0x3fa07520C70EB909, 0x3fa07520C70EB909 /* poly_coeff12 */ - .quad 0xbf90FB17F7DBB0ED, 0xbf90FB17F7DBB0ED /* poly_coeff11 */ - .quad 0x3f943F44BFBC3BAE, 0x3f943F44BFBC3BAE /* poly_coeff10 */ - .quad 0x3f7A583395D45ED5, 0x3f7A583395D45ED5 /* poly_coeff9 */ - .quad 0x3f88F8DC2AFCCAD6, 0x3f88F8DC2AFCCAD6 /* poly_coeff8 */ - .quad 0x3f8C6DBBCB88BD57, 0x3f8C6DBBCB88BD57 /* poly_coeff7 */ - .quad 0x3f91C6DCF538AD2E, 0x3f91C6DCF538AD2E /* poly_coeff6 */ - .quad 0x3f96E89CEBDEFadd, 0x3f96E89CEBDEFadd /* poly_coeff5 */ - .quad 0x3f9F1C72E13AD8BE, 0x3f9F1C72E13AD8BE /* poly_coeff4 */ - .quad 0x3fa6DB6DB3B445F8, 0x3fa6DB6DB3B445F8 /* poly_coeff3 */ - .quad 0x3fb333333337E0DE, 0x3fb333333337E0DE /* poly_coeff2 */ - .quad 0x3fc555555555529C, 0x3fc555555555529C /* poly_coeff1 */ - /*== Pi2H ==*/ - .align 16 - .quad 0x3ff921fb54442d18, 0x3ff921fb54442d18 - .align 16 - .type __svml_dasin_data_internal,@object - .size __svml_dasin_data_internal,.-__svml_dasin_data_internal + /* AbsMask */ + .quad 0x7fffffffffffffff, 0x7fffffffffffffff + /* OneHalf */ + .align 16 + .quad 0x3fe0000000000000, 0x3fe0000000000000 + /* SmallNorm */ + .align 16 + .quad 0x3000000000000000, 0x3000000000000000 + /* One */ + .align 16 + .quad 0x3ff0000000000000, 0x3ff0000000000000 + /* Two */ + .align 16 + .quad 0x4000000000000000, 0x4000000000000000 + /* sqrt_coeff[4] */ + .align 16 + .quad 0xbf918000993B24C3, 0xbf918000993B24C3 /* sqrt_coeff4 */ + .quad 0x3fa400006F70D42D, 0x3fa400006F70D42D /* sqrt_coeff3 */ + .quad 0xbfb7FFFFFFFFFE97, 0xbfb7FFFFFFFFFE97 /* sqrt_coeff2 */ + .quad 0x3fcFFFFFFFFFFF9D, 0x3fcFFFFFFFFFFF9D /* sqrt_coeff1 */ + /* poly_coeff[12] */ + .align 16 + .quad 0x3fa07520C70EB909, 0x3fa07520C70EB909 /* poly_coeff12 */ + .quad 0xbf90FB17F7DBB0ED, 0xbf90FB17F7DBB0ED /* poly_coeff11 */ + .quad 0x3f943F44BFBC3BAE, 0x3f943F44BFBC3BAE /* poly_coeff10 */ + .quad 0x3f7A583395D45ED5, 0x3f7A583395D45ED5 /* poly_coeff9 */ + .quad 0x3f88F8DC2AFCCAD6, 0x3f88F8DC2AFCCAD6 /* poly_coeff8 */ + .quad 0x3f8C6DBBCB88BD57, 0x3f8C6DBBCB88BD57 /* poly_coeff7 */ + .quad 0x3f91C6DCF538AD2E, 0x3f91C6DCF538AD2E /* poly_coeff6 */ + .quad 0x3f96E89CEBDEFadd, 0x3f96E89CEBDEFadd /* poly_coeff5 */ + .quad 0x3f9F1C72E13AD8BE, 0x3f9F1C72E13AD8BE /* poly_coeff4 */ + .quad 0x3fa6DB6DB3B445F8, 0x3fa6DB6DB3B445F8 /* poly_coeff3 */ + .quad 0x3fb333333337E0DE, 0x3fb333333337E0DE /* poly_coeff2 */ + .quad 0x3fc555555555529C, 0x3fc555555555529C /* poly_coeff1 */ + /* Pi2H */ + .align 16 + .quad 0x3ff921fb54442d18, 0x3ff921fb54442d18 + .align 16 + .type __svml_dasin_data_internal, @object + .size __svml_dasin_data_internal, .-__svml_dasin_data_internal |