Commit Graph

520 Commits

Author SHA1 Message Date
Jan Vesely 87036d2701 vstore_half: Make sure the helper function is always inline
Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 324370
2018-02-06 18:44:35 +00:00
Jan Vesely 3b8b4eb64d half_powr: Implement using powr
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 323942
2018-02-01 03:00:35 +00:00
Jan Vesely a75677c2b7 math.h: Use logical operations instead of bit operations for readability
Trivial.

Reported-by: Roman Lebedev <lebedev.ri@gmail.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 323920
2018-01-31 21:53:42 +00:00
Jan Vesely 0ecb5e511e math.h: Set HAVE_HW_FMA32 based on compiler provided macro
Fixes sin/cos piglits on non-FMA capable asics.
Bugzilla: https://bugs.llvm.org/show_bug.cgi?id=35983

Reviewer: Tom Stellard
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 323677
2018-01-29 19:05:08 +00:00
Jan Vesely 7013857f95 tanpi: Port from amd_builtins
Passes piglit on turks and carrizo.
Passes CTS on carrizo.

Acked-By: Aaron Watry <awatry@gmail.com>
Tested-By: Aaron Watry <awatry@gmail.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322980
2018-01-19 18:57:22 +00:00
Jan Vesely 03937bdec3 tan: Port from amd_builtins
v2: fixup constant precision
Passes piglit on turks and carrizo.
Passes CTS on carrizo
Fixes half_tan to pass CTS on carrizo

Acked-By: Aaron Watry <awatry@gmail.com>
Tested-By: Aaron Watry <awatry@gmail.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322979
2018-01-19 18:57:19 +00:00
Jan Vesely 44e0522c09 half_divide: Implement using x/y
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322899
2018-01-18 21:12:06 +00:00
Jan Vesely 2813b4f8d9 half_tan: Implement using tan
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322898
2018-01-18 21:12:04 +00:00
Jan Vesely bf38fae8de half_sin: Implement using sin
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322897
2018-01-18 21:12:01 +00:00
Jan Vesely 398108b91e half_recip: Implement using 1/x
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322896
2018-01-18 21:11:58 +00:00
Jan Vesely a1aba44ffa half_log2: Implement using log2
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322895
2018-01-18 21:11:56 +00:00
Jan Vesely b3b72af4b9 half_log10: Implement using log10
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322894
2018-01-18 21:11:53 +00:00
Jan Vesely 6852023802 half_log: Implement using log
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322893
2018-01-18 21:11:50 +00:00
Jan Vesely aa4c3899b5 half_exp10: Implement using exp10
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322892
2018-01-18 21:11:48 +00:00
Jan Vesely 3c0e19b61a half_exp2: Implement using exp2
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322891
2018-01-18 21:11:45 +00:00
Jan Vesely caa9000b1c half_exp: Implement using exp
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322890
2018-01-18 21:11:43 +00:00
Jan Vesely b5d556061d half_cos: Implement using cos
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322889
2018-01-18 21:11:40 +00:00
Jan Vesely e53ae3b596 half_sqrt: Cleanup implementation
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322888
2018-01-18 21:11:38 +00:00
Jan Vesely a95db14461 half_rsqrt: Cleanup implementation
Passes CTS on carrizo
v2: Use full precision implementation

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322887
2018-01-18 21:11:35 +00:00
Jan Vesely fe8e00bc3c rootn: Port from amd_builtins
Passes piglit on turks and carrizo
fp64 passes ctx on carrizo

v2: fix formatting
    check fp32 denormal support at runtime

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322763
2018-01-17 21:22:14 +00:00
Jan Vesely c45ec604f5 powr: Port from amd_builtins
Passes piglit on turks and carrizo
fp64 passes cts on carrizo

v2: fix formatting
    check fp32 denormal support at runtime

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322762
2018-01-17 21:22:06 +00:00
Jan Vesely 5efc8fe321 pown: Port from amd_builtins
Passes piglit on turks and carrizo
fp64 passes CTS on carrizo

v2: fix formatting
    check fp32 denormal support at runtime

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322761
2018-01-17 21:22:03 +00:00
Jan Vesely cc5c65b2c2 pow: Port from amd_builtins
Passes piglit on turks and carrizo
fp64 passes CTS on carrizo

v2: fix formatting
    check fp32 denormal support at runtime

Reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 322760
2018-01-17 21:21:35 +00:00
Vedran Miletic 79b7f4c125 configure.py: Add gfx900 (Vega, Raven)
Sort amdgcn-- and amdgcn--amdhsa in a consistent way.

llvm-svn: 319017
2017-11-27 11:14:06 +00:00
Jan Vesely fe7c045753 math: Implement minmag
Reviewer: Aaron Watry
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 318265
2017-11-15 04:10:39 +00:00
Jan Vesely 7ba243cc3d math: Implement maxmag
Reviewer: Aaron Watry
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 318264
2017-11-15 04:10:37 +00:00
Jan Vesely 383fbd050c native_powr: Switch implementation to native_exp2 and native_log2
v2: don't use assume
    check only for x<0, the other conditions are handled transparently
v3: don't check inputs at all, nan propagation works as expected

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 318204
2017-11-14 21:55:41 +00:00
Jan Vesely f38b40daf7 native_divide: provide function implementation instead of macro
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 318067
2017-11-13 18:28:56 +00:00
Jan Vesely 1b9825f982 native_recip: provide function implementation instead of macro
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 318066
2017-11-13 18:28:53 +00:00
Jan Vesely a6758c94ef native_rsqrt: Switch implementation to 1 / native_sqrt
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 318065
2017-11-13 18:28:51 +00:00
Jan Vesely 541a3f0758 native_tan: Switch implementation to use native_sin/native_cos
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 318064
2017-11-13 18:28:48 +00:00
Jan Vesely 79b7566210 math: Use precomputed constant for log2(10.0)
exp10 CTS fails with or without this change

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 318063
2017-11-13 18:28:45 +00:00
Jan Vesely 6b4a625438 native_exp10: Switch implementation to llvm intrinsic
v2: Use native_log2 instead of wrong constant

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317941
2017-11-10 22:16:41 +00:00
Jan Vesely 4301e6d0c9 native_sqrt: Switch implementation to llvm intrinsic
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317940
2017-11-10 22:16:39 +00:00
Jan Vesely 1f34c851e0 native_sin: Switch implementation to llvm intrinsic
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317939
2017-11-10 22:16:36 +00:00
Jan Vesely 0750b7df51 native_cos: Switch implementation to llvm intrinsic
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317938
2017-11-10 22:16:33 +00:00
Jan Vesely edbde58de0 native_exp2: Switch implementation to llvm intrinsic
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317937
2017-11-10 22:16:31 +00:00
Jan Vesely 504f85c551 native_exp: Switch implementation to llvm intrinsic
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317936
2017-11-10 22:16:28 +00:00
Jan Vesely 8dc6e98d47 amdgpu: Add workaround for unimplemented llvm.exp intrinsic
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317935
2017-11-10 22:16:25 +00:00
Jan Vesely adc1eaedf8 native_log10: Switch to generic native intrinsic inc file
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317934
2017-11-10 22:16:22 +00:00
Jan Vesely 086e796053 native_log: Switch to generic native intrinsic inc file
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317933
2017-11-10 22:16:20 +00:00
Jan Vesely f58dee9f3a native_log2: Switch to generic native intrinsic inc file
v2: Add __CLC_XCONCAT instead of function name redirection
    Use __CLC_XCONCAT for intrinsic functions as well

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 317932
2017-11-10 22:16:15 +00:00
Jan Vesely 39ef293533 tgamma: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317260
2017-11-02 19:49:00 +00:00
Jan Vesely 91d7b92d8a tanh: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317259
2017-11-02 19:48:58 +00:00
Jan Vesely e4d5d10076 tan: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317258
2017-11-02 19:48:57 +00:00
Jan Vesely b0fab2696a sqrt: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317257
2017-11-02 19:48:55 +00:00
Jan Vesely e3802356e2 sinpi: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317256
2017-11-02 19:48:53 +00:00
Jan Vesely 4708b10878 sinh: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317255
2017-11-02 19:48:51 +00:00
Jan Vesely a3febe3fa9 sin: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317254
2017-11-02 19:48:50 +00:00
Jan Vesely 25671b40d7 native_log: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317253
2017-11-02 19:48:48 +00:00
Jan Vesely fd13434d83 native_log2: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317252
2017-11-02 19:48:46 +00:00
Jan Vesely d6ad07687d native_log10: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317251
2017-11-02 19:48:44 +00:00
Jan Vesely 139185dfc7 log: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317250
2017-11-02 19:48:43 +00:00
Jan Vesely 27dffff6e8 logb: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317249
2017-11-02 19:48:41 +00:00
Jan Vesely 7fc23fbdcb log2: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317248
2017-11-02 19:48:39 +00:00
Jan Vesely a9132ce347 log1p: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317247
2017-11-02 19:48:37 +00:00
Jan Vesely 4e062cb74e lgamma: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317246
2017-11-02 19:48:35 +00:00
Jan Vesely 4cb612e140 exp2: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317245
2017-11-02 19:48:33 +00:00
Jan Vesely e99ba9a23d cospi: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317244
2017-11-02 19:48:31 +00:00
Jan Vesely c708278f13 cosh: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317243
2017-11-02 19:48:30 +00:00
Jan Vesely f76371d948 cos: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317242
2017-11-02 19:48:27 +00:00
Jan Vesely 50a3cccdbe cbrt: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317241
2017-11-02 19:48:25 +00:00
Jan Vesely a4df39bcad atanpi: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317240
2017-11-02 19:48:23 +00:00
Jan Vesely 1bd2ac257a atanh: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317239
2017-11-02 19:48:22 +00:00
Jan Vesely 0942b5e1bf atan: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317238
2017-11-02 19:48:20 +00:00
Jan Vesely d3d5e322e3 asinpi: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317237
2017-11-02 19:48:18 +00:00
Jan Vesely 48bda32986 asinh: Use unary_dec instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317236
2017-11-02 19:48:16 +00:00
Jan Vesely ba4b98c691 asin: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317235
2017-11-02 19:48:15 +00:00
Jan Vesely 61171847b7 acospi: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317234
2017-11-02 19:48:13 +00:00
Jan Vesely 720783d9f5 acosh: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317233
2017-11-02 19:48:11 +00:00
Jan Vesely caca914218 acos: Use unary_decl instead of custom inc file
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 317232
2017-11-02 19:48:06 +00:00
Jan Vesely 47e093da9b math: Implement native_log10
Use llvm instrinsic by default
Provide amdgpu workaround

v2: drop old amd copyrights

Reviewer: Aaron Watry
Reviewed-by: Vedran Miletić <vedran@miletic.net>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 316588
2017-10-25 16:49:22 +00:00
Jan Vesely 9fedbb9d8e amdgpu/math: Don't use llvm instrinsic for native_log
AMDGPU targets don't have insturction for it,
so it'll be expanded to C * log2 anyway.

v2: use native_log2 instead of the more precise sw implementation
v3: move to amdgpu
v4: drop old AMD copyright

Reviewer: Aaron Watry
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 316587
2017-10-25 16:49:17 +00:00
Jan Vesely 7ab2d0bdcd shared: Implement aligned vector stores (vstorea_half)
Float version passes newly posted piglit tests on turks, float and double pass on carrizo.
v2: scalar vstorea_half
v3: fix typo

Reviewer: Aaron Watry
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 316291
2017-10-22 14:21:59 +00:00
Jan Vesely 12061c7125 shared: Implement aligned vector loads (vloada_half)
Passes newly posted piglits on turks and carrizo
v2: add scalar vloada_half
v3: fix typo

Reviewer: Aaron Watry
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 316290
2017-10-22 14:21:56 +00:00
Jan Vesely c420b61b26 amdgcn: Add missing datalayout info to .ll files
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Acked-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 316239
2017-10-20 21:10:18 +00:00
Jan Vesely 66b32ad9ad r600: Add missing datalayout to .ll files
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Acked-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 316238
2017-10-20 21:00:31 +00:00
Jan Vesely 577c52b9c7 travis: enable checks of nvptx libraries
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315343
2017-10-10 18:10:25 +00:00
Jan Vesely 2601429bac travis: Enable external function call checks on llvm-{4,5}
Reviewer: Aaron Watry
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315342
2017-10-10 18:10:24 +00:00
Jan Vesely 3d349ea98e Make image builtins r600/llvm-3.9 only
The implementation uses r600 sepcific intrinsics
LLVM-4 switched to _ro_t and _rw_t image types
Portions of the code can be moved back as more targets/llvm versions add image support

Reviewer: Aaron Watry
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315341
2017-10-10 18:10:21 +00:00
Jeroen Ketema 1364d268a4 Implement mem_fence on ptx
PTX does not differentiate between read and write fences. Hence, these a
lowered to a mem_fence call. The mem_fence function compiles to the
“member.cta” instruction, which commits all outstanding reads and writes
of a thread such that these become visible to all other threads in the same
CTA (i.e., work-group). The instruction does not differentiate between
global and local memory. Hence, the flags parameter is ignored, except
for deciding whether a “member.cta” instruction should be issued at all.

Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315235
2017-10-09 19:43:04 +00:00
Jeroen Ketema 4f5a3d5d6f Make ptx barrier work irrespective of the cl_mem_fence_flags
This generates a "bar.sync 0” instruction, which not only causes the
threads to wait, but does acts as a memory fence, as required by
OpenCL. The fence does not differentiate between local and global
memory. Unfortunately, there is no similar instruction which does
not include a memory fence. Hence, we cannot optimize the case
where neither CLK_LOCAL_MEM_FENCE nor CLK_GLOBAL_MEM_FENCE is
passed.

llvm-svn: 315228
2017-10-09 18:36:48 +00:00
Jan Vesely 3c51ae5bd9 travis: Make sure we report failure even if only earlier checked files fail
for loop would only report status of the last command
v2: return '1'
    call test instead of '['

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315193
2017-10-08 20:07:58 +00:00
Jan Vesely 136381dc38 check_external_calls.sh: Print number of calls in tested file.
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315192
2017-10-08 20:07:56 +00:00
Jan Vesely 80bb52ae75 ptx: Use __clc_nextafter to implement nextafter
using clang builtin results in external library call

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315191
2017-10-08 19:34:00 +00:00
Jan Vesely 1de1444d62 Do not include clc_nextafter header globally
Drop unused clc/math/clc_nextafter.h header

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315190
2017-10-08 19:33:58 +00:00
Jan Vesely 6a5c8ddb3a math/nextafter: Use custom declaration inc file
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315189
2017-10-08 19:33:55 +00:00
Jan Vesely 72be1cc0be math/binary_decl.inc: Do not declare mixed float/double functions
fmin/fmax only need vector/scalar mix

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315188
2017-10-08 19:33:53 +00:00
Jan Vesely beb6591753 ldexp: Fix double precision function return type
Fixes ~1200 external calls from nvtpx library.

Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315170
2017-10-08 06:56:14 +00:00
Jan Vesely 391305638c configure: Fix handling of directories with compats only source lists
Reviewer: Jeroen Ketema
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 315018
2017-10-05 20:16:28 +00:00
Jeroen Ketema 957151bd86 Add vload_half helpers for ptx
The removes the vload_half unresolved calls from the nvptx libraries.

Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 314998
2017-10-05 18:17:40 +00:00
Jeroen Ketema feefb0870f Add vstore_half helpers for ptx
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 314925
2017-10-04 19:07:48 +00:00
Jan Vesely a02d0e2c50 integer/sub_sat: Use clang builtin instead of llvm asm
reviewer: Tom Stellard

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 314703
2017-10-02 18:39:03 +00:00
Jan Vesely 1964df8fad integer/add_sat: Use clang builtin instead of llvm asm
reviewer: Tom Stellard

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 314702
2017-10-02 18:39:00 +00:00
Jan Vesely 943057a288 integer/clz: Use clang builtin instead of llvm asm
The generated llvm IR mostly identical. char/uchar case is a bit worse.

reviewer: Tom Stellard

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 314701
2017-10-02 18:38:57 +00:00
Jeroen Ketema fe9fa89854 Let get_work_dim take exactly 0 arguments
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 314634
2017-10-01 20:11:46 +00:00
Jeroen Ketema 17fdf263c5 Do no circularly define NULL
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 314633
2017-10-01 20:10:14 +00:00
Jan Vesely 2b7fa1c6f6 Fix amdgcn-amdhsa on llvm-3.9
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Acked-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 314548
2017-09-29 19:06:52 +00:00
Jan Vesely aee030f284 travis: Check built libraries on llvm-3.9
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Acked-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 314547
2017-09-29 19:06:50 +00:00
Jan Vesely 8c8c287adf Add script to check for unresolved function calls
v2: add shell shebang
    improve error checks and reporting
v3: fix typo

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 314546
2017-09-29 19:06:48 +00:00