Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | powerpc: Remove power4 mpa optimization | Adhemerval Zanella | 2019-04-29 | 3 | -275/+0 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This patch removes the POWER4 optimized mpa optimization used currently on all powerpc targets. In fact for newer chips, GCC generates *worse* code than generic implementation as below. One possibilty would to add ifunc variants for the mpa routines (as x86_64), but it will add complexity only for older chips (and one would need to check if power5, power5+, and power6 do benefict from this optimization), and only for specific implementation (since most used one such as sin, cos, exp, pow where optimized to avoid calling the slow multiprecision path). * POWER9 patched $ ./testrun.sh benchtests/bench-atan "atan": { "": { "duration": 5.12565e+09, "iterations": 1.552e+08, "max": 100.552, "min": 7.799, "mean": 33.0261 }, "144bits": { "duration": 5.12745e+09, "iterations": 825000, "max": 7517.17, "min": 6186.3, "mean": 6215.09 } } $ ./testrun.sh benchtests/bench-acos "acos": { "": { "duration": 5.21741e+09, "iterations": 1.269e+08, "max": 191.738, "min": 7.931, "mean": 41.1144 }, "slow": { "duration": 5.25999e+09, "iterations": 198000, "max": 26681.7, "min": 26463.6, "mean": 26565.6 } } * POWER9 master $ ./testrun.sh benchtests/bench-atan "atan": { "": { "duration": 5.12815e+09, "iterations": 1.552e+08, "max": 134.788, "min": 7.803, "mean": 33.0422 }, "144bits": { "duration": 5.1209e+09, "iterations": 447000, "max": 11615.8, "min": 11301.8, "mean": 11456.2 } } $ ./testrun.sh benchtests/bench-acos "acos": { "": { "duration": 5.22272e+09, "iterations": 1.269e+08, "max": 115.981, "min": 7.931, "mean": 41.1562 }, "slow": { "duration": 5.28723e+09, "iterations": 96000, "max": 55434.1, "min": 54820.6, "mean": 55075.3 } } * POWER8 patched $ taskset -c 16 ./testrun.sh benchtests/bench-acos "acos": { "": { "duration": 5.16398e+09, "iterations": 9.99e+07, "max": 174.408, "min": 8.645, "mean": 51.6915 }, "slow": { "duration": 5.16982e+09, "iterations": 96000, "max": 54830.5, "min": 53703.8, "mean": 53852.3 } } * POWER8 master $ taskset -c 16 ./testrun.sh benchtests/bench-acos "acos": { "": { "duration": 5.17019e+09, "iterations": 9.99e+07, "max": 186.127, "min": 8.633, "mean": 51.7537 }, "slow": { "duration": 5.34225e+09, "iterations": 90000, "max": 60353.2, "min": 59155.3, "mean": 59358.4 } } * POWER7 patched $ taskset -c 16 benchtests/bench-asin "asin": { "": { "duration": 5.15559e+09, "iterations": 6.5e+07, "max": 193.335, "min": 12.227, "mean": 79.3168 }, "slow": { "duration": 5.20538e+09, "iterations": 80000, "max": 65705.2, "min": 64299.4, "mean": 65067.3 } } * POWER7 master $ taskset -c 16 benchtests/bench-asin "asin": { "": { "duration": 5.15446e+09, "iterations": 6.5e+07, "max": 184.575, "min": 12.226, "mean": 79.2994 }, "slow": { "duration": 5.20616e+09, "iterations": 80000, "max": 65705.1, "min": 64336.6, "mean": 65076.9 } } Checked on powerpc-linux-gnu (built without --with-cpu, with --with-cpu=power4 and with --with-cpu=power5+ and --disable-multi-arch), powerpc64-linux-gnu (built without --with-cp and with --with-cpu=power5+ and --disable-multi-arch). * sysdeps/powerpc/power4/fpu/Makefile: Remove file. * sysdeps/powerpc/power4/fpu/mpa-arch.h: Likewise. * sysdeps/powerpc/power4/fpu/mpa.c: Likewise. Reviewed-by: Gabriel F. T. Gomes <gabriel@inconstante.eti.br> | ||||
* | Update copyright dates with scripts/update-copyrights. | Joseph Myers | 2019-01-01 | 3 | -3/+3 |
| | | | | | | | * All files with FSF copyright notices: Update copyright dates using scripts/update-copyrights. * locale/programs/charmap-kw.h: Regenerated. * locale/programs/locfile-kw.h: Likewise. | ||||
* | Remove slow paths from exp | Szabolcs Nagy | 2018-02-12 | 1 | -1/+0 |
| | | | | | | | | | | | | | | | | | | | | | | | | | Remove the __slowexp code, so exp is no longer correctly rounded. The result is computed to about 70 bits precision so the worst case ulp error is about 0.500007 in nearest rounding mode. * manual/probes.texi: Remove slowexp probes. * math/Makefile: Remove slowexp. * sysdeps/generic/math_private.h (__slowexp): Remove. * sysdeps/ieee754/dbl-64/e_exp.c (__ieee754_exp): Remove __slowexp and document error bounds. * sysdeps/i386/fpu/slowexp.c: Remove. * sysdeps/ia64/fpu/slowexp.c: Remove. * sysdeps/ieee754/dbl-64/slowexp.c: Remove. * sysdeps/ieee754/dbl-64/uexp.h (err_0): Remove. * sysdeps/m68k/m680x0/fpu/slowexp.c: Remove. * sysdeps/powerpc/power4/fpu/Makefile (CPPFLAGS-slowexp.c): Remove. * sysdeps/x86_64/fpu/multiarch/Makefile: Remove slowexp-fma. * sysdeps/x86_64/fpu/multiarch/e_exp-avx.c (__slowexp): Remove. * sysdeps/x86_64/fpu/multiarch/e_exp-fma.c (__slowexp): Remove. * sysdeps/x86_64/fpu/multiarch/e_exp-fma4.c (__slowexp): Remove. * sysdeps/x86_64/fpu/multiarch/slowexp-avx.c: Remove. * sysdeps/x86_64/fpu/multiarch/slowexp-fma.c: Remove. * sysdeps/x86_64/fpu/multiarch/slowexp-fma4.c: Remove. | ||||
* | Remove slow paths from pow | Wilco Dijkstra | 2018-02-12 | 1 | -1/+0 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Remove the slow paths from pow. Like several other double precision math functions, pow is exactly rounded. This is not required from math functions and causes major overheads as it requires multiple fallbacks using higher precision arithmetic if a result is close to 0.5ULP. Ridiculous slowdowns of up to 100000x have been reported when the highest precision path triggers. All GLIBC math tests pass on AArch64 and x64 (with ULP of pow set to 1). The worst case error is ~0.506ULP. A simple test over a few hundred million values shows pow is 10% faster on average. This fixes BZ #13932. [BZ #13932] * sysdeps/ieee754/dbl-64/uexp.h (err_1): Remove. * benchtests/pow-inputs: Update comment for slow path cases. * manual/probes.texi (slowpow_p10): Delete removed probe. (slowpow_p10): Likewise. * math/Makefile: Remove halfulp.c and slowpow.c. * sysdeps/aarch64/libm-test-ulps: Set ULP of pow to 1. * sysdeps/generic/math_private.h (__exp1): Remove error argument. (__halfulp): Remove. (__slowpow): Remove. * sysdeps/i386/fpu/halfulp.c: Delete file. * sysdeps/i386/fpu/slowpow.c: Likewise. * sysdeps/ia64/fpu/halfulp.c: Likewise. * sysdeps/ia64/fpu/slowpow.c: Likewise. * sysdeps/ieee754/dbl-64/e_exp.c (__exp1): Remove error argument, improve comments and add error analysis. * sysdeps/ieee754/dbl-64/e_pow.c (__ieee754_pow): Add error analysis. (power1): Remove function: (log1): Remove error argument, add error analysis. (my_log2): Remove function. * sysdeps/ieee754/dbl-64/halfulp.c: Delete file. * sysdeps/ieee754/dbl-64/slowpow.c: Likewise. * sysdeps/m68k/m680x0/fpu/halfulp.c: Likewise. * sysdeps/m68k/m680x0/fpu/slowpow.c: Likewise. * sysdeps/powerpc/power4/fpu/Makefile: Remove CPPFLAGS-slowpow.c. * sysdeps/x86_64/fpu/libm-test-ulps: Set ULP of pow to 1. * sysdeps/x86_64/fpu/multiarch/Makefile: Remove slowpow-fma.c, slowpow-fma4.c, halfulp-fma.c, halfulp-fma4.c. * sysdeps/x86_64/fpu/multiarch/e_pow-fma.c (__slowpow): Remove define. * sysdeps/x86_64/fpu/multiarch/e_pow-fma4.c (__slowpow): Likewise. * sysdeps/x86_64/fpu/multiarch/halfulp-fma.c: Delete file. * sysdeps/x86_64/fpu/multiarch/halfulp-fma4.c: Likewise. * sysdeps/x86_64/fpu/multiarch/slowpow-fma.c: Likewise. * sysdeps/x86_64/fpu/multiarch/slowpow-fma4.c: Likewise. | ||||
* | Update copyright dates with scripts/update-copyrights. | Joseph Myers | 2018-01-01 | 3 | -3/+3 |
| | | | | | | | * All files with FSF copyright notices: Update copyright dates using scripts/update-copyrights. * locale/programs/charmap-kw.h: Regenerated. * locale/programs/locfile-kw.h: Likewise. | ||||
* | Revert exp reimplementation (causes test failures). | Joseph Myers | 2017-12-19 | 1 | -0/+1 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Revert: 2017-12-19 Joseph Myers <joseph@codesourcery.com> * sysdeps/x86_64/fpu/libm-test-ulps: Update. 2017-12-19 Patrick McGehearty <patrick.mcgehearty@oracle.com> * sysdeps/ieee754/dbl-64/e_exp.c: Include <math-svid-compat.h> and <errno.h>. Include "eexp.tbl". (half): New constant. (one): Likewise. (__ieee754_exp): Rewrite. (__slowexp): Remove prototype. * sysdeps/ieee754/dbl-64/eexp.tbl: New file. * sysdeps/ieee754/dbl-64/slowexp.c: Remove file. * sysdeps/i386/fpu/slowexp.c: Likewise. * sysdeps/ia64/fpu/slowexp.c: Likewise. * sysdeps/m68k/m680x0/fpu/slowexp.c: Likewise. * sysdeps/x86_64/fpu/multiarch/slowexp-avx.c: Likewise. * sysdeps/x86_64/fpu/multiarch/slowexp-fma.c: Likewise. * sysdeps/x86_64/fpu/multiarch/slowexp-fma4.c: Likewise. * sysdeps/generic/math_private.h (__slowexp): Remove prototype. * sysdeps/ieee754/dbl-64/e_pow.c: Remove mention of slowexp.c in comment. * sysdeps/powerpc/power4/fpu/Makefile [$(subdir) = math] (CPPFLAGS-slowexp.c): Remove variable. * sysdeps/x86_64/fpu/multiarch/Makefile (libm-sysdep_routines): Remove slowexp-fma, slowexp-fma4 and slowexp-avx. (CFLAGS-slowexp-fma.c): Remove variable. (CFLAGS-slowexp-fma4.c): Likewise. (CFLAGS-slowexp-avx.c): Likewise. * sysdeps/x86_64/fpu/multiarch/e_exp-avx.c (__slowexp): Do not define as macro. * sysdeps/x86_64/fpu/multiarch/e_exp-fma.c (__slowexp): Likewise. * sysdeps/x86_64/fpu/multiarch/e_exp-fma4.c (__slowexp): Likewise. * math/Makefile (type-double-routines): Remove slowexp. * manual/probes.texi (slowexp_p6): Remove. (slowexp_p32): Likewise. | ||||
* | Improve __ieee754_exp() performance by greater than 5x on sparc/x86. | Patrick McGehearty | 2017-12-19 | 1 | -1/+0 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | These changes will be active for all platforms that don't provide their own exp() routines. They will also be active for ieee754 versions of ccos, ccosh, cosh, csin, csinh, sinh, exp10, gamma, and erf. Typical performance gains is typically around 5x when measured on Sparc s7 for common values between exp(1) and exp(40). Using the glibc perf tests on sparc, sparc (nsec) x86 (nsec) old new old new max 17629 395 5173 144 min 399 54 15 13 mean 5317 200 1349 23 The extreme max times for the old (ieee754) exp are due to the multiprecision computation in the old algorithm when the true value is very near 0.5 ulp away from an value representable in double precision. The new algorithm does not take special measures for those cases. The current glibc exp perf tests overrepresent those values. Informal testing suggests approximately one in 200 cases might invoke the high cost computation. The performance advantage of the new algorithm for other values is still large but not as large as indicated by the chart above. Glibc correctness tests for exp() and expf() were run. Within the test suite 3 input values were found to cause 1 bit differences (ulp) when "FE_TONEAREST" rounding mode is set. No differences in exp() were seen for the tested values for the other rounding modes. Typical example: exp(-0x1.760cd2p+0) (-1.46113312244415283203125) new code: 2.31973271630014299393707e-01 0x1.db14cd799387ap-3 old code: 2.31973271630014271638132e-01 0x1.db14cd7993879p-3 exp = 2.31973271630014285508337 (high precision) Old delta: off by 0.49 ulp New delta: off by 0.51 ulp In addition, because ieee754_exp() is used by other routines, cexp() showed test results with very small imaginary input values where the imaginary portion of the result was off by 3 ulp when in upward rounding mode, but not in the other rounding modes. For x86, tgamma showed a few values where the ulp increased to 6 (max ulp for tgamma is 5). Sparc tgamma did not show these failures. I presume the tgamma differences are due to compiler optimization differences within the gamma function.The gamma function is known to be difficult to compute accurately. * sysdeps/ieee754/dbl-64/e_exp.c: Include <math-svid-compat.h> and <errno.h>. Include "eexp.tbl". (half): New constant. (one): Likewise. (__ieee754_exp): Rewrite. (__slowexp): Remove prototype. * sysdeps/ieee754/dbl-64/eexp.tbl: New file. * sysdeps/ieee754/dbl-64/slowexp.c: Remove file. * sysdeps/i386/fpu/slowexp.c: Likewise. * sysdeps/ia64/fpu/slowexp.c: Likewise. * sysdeps/m68k/m680x0/fpu/slowexp.c: Likewise. * sysdeps/x86_64/fpu/multiarch/slowexp-avx.c: Likewise. * sysdeps/x86_64/fpu/multiarch/slowexp-fma.c: Likewise. * sysdeps/x86_64/fpu/multiarch/slowexp-fma4.c: Likewise. * sysdeps/generic/math_private.h (__slowexp): Remove prototype. * sysdeps/ieee754/dbl-64/e_pow.c: Remove mention of slowexp.c in comment. * sysdeps/powerpc/power4/fpu/Makefile [$(subdir) = math] (CPPFLAGS-slowexp.c): Remove variable. * sysdeps/x86_64/fpu/multiarch/Makefile (libm-sysdep_routines): Remove slowexp-fma, slowexp-fma4 and slowexp-avx. (CFLAGS-slowexp-fma.c): Remove variable. (CFLAGS-slowexp-fma4.c): Likewise. (CFLAGS-slowexp-avx.c): Likewise. * sysdeps/x86_64/fpu/multiarch/e_exp-avx.c (__slowexp): Do not define as macro. * sysdeps/x86_64/fpu/multiarch/e_exp-fma.c (__slowexp): Likewise. * sysdeps/x86_64/fpu/multiarch/e_exp-fma4.c (__slowexp): Likewise. * math/Makefile (type-double-routines): Remove slowexp. * manual/probes.texi (slowexp_p6): Remove. (slowexp_p32): Likewise. | ||||
* | Update copyright dates with scripts/update-copyrights. | Joseph Myers | 2017-01-01 | 3 | -3/+3 |
| | |||||
* | Update copyright dates with scripts/update-copyrights. | Joseph Myers | 2016-01-04 | 3 | -3/+3 |
| | |||||
* | Update copyright dates with scripts/update-copyrights. | Joseph Myers | 2015-01-02 | 3 | -3/+3 |
| | |||||
* | Update copyright notices with scripts/update-copyrights | Allan McRae | 2014-01-01 | 3 | -3/+3 |
| | |||||
* | PowerPC: multiarch wordcopy routines for PowerPC32 | Adhemerval Zanella | 2013-12-06 | 1 | -0/+212 |
| | |||||
* | Remove usage of ONE macro | Siddhesh Poyarekar | 2013-03-29 | 1 | -1/+1 |
| | | | | I missed this instance, which caused a build failure in powerpc. | ||||
* | Remove ZERO and MZERO macros | Siddhesh Poyarekar | 2013-03-29 | 1 | -10/+10 |
| | |||||
* | Make mantissa type of mp_no configurable | Siddhesh Poyarekar | 2013-03-26 | 1 | -0/+56 |
| | | | | | | | | | | The mantissa of mp_no is intended to take only integral values. This is a relatively good choice for powerpc due to its 4 fpus, but not for other architectures, which suffer due to this choice. This change makes the default mantissa a long integer and allows powerpc to override it. Additionally, some operations have been optimized for integer manipulation, resulting in a significant improvement in performance. | ||||
* | Revert configurable mantissa patch | Siddhesh Poyarekar | 2013-03-15 | 1 | -56/+0 |
| | | | | | | Reverts d22ca8cdfb98001d03772ef264b244930d439b3f since it is severely broken on 32-bit. | ||||
* | Make mantissa type configurable | Siddhesh Poyarekar | 2013-03-15 | 1 | -0/+56 |
| | | | | | This allows the default mantissa to be integral, with powerpc overriding it to take advantage of its FPUs. | ||||
* | Consolidate copies of mp code in powerpc | Siddhesh Poyarekar | 2013-03-08 | 2 | -0/+221 |
Retain a single copy of the mp code in power4 instead of the two identical copies in powerpc32 and powerpc64. |