f2062aa | Shubham Pawar | 05 November 2019, 22:30:58 UTC | Fix offload mode autoscheduler results. Currently, when autoscheduler is invoked in offload mode, the the emit options {featurization and schedule} are not working as expected and bailing out due to auto_scheduler_results not present in the lowered module, resulting in empty feature and schedule files produced. This happens because, when offloaded submodules are recursively resolved, the autoscheduler results are not copied back into the lowered module. This change fixes this issue by setting back the auto_scheduler_results in the ModuleContents of the module that is lowered after resolving submodules. | 05 November 2019, 22:58:31 UTC |
aa33d0a | Steven Johnson | 09 October 2019, 16:58:16 UTC | Merge pull request #4303 from halide/pdb_add_libcpp_makefile_inc Plumb LLVM_CXXFLAGS to apps/support_sa/Makefile.inc via halide_config.make | 09 October 2019, 16:58:16 UTC |
092f528 | Andrew Adams | 09 October 2019, 16:52:36 UTC | Merge pull request #3936 from halide/tzumao-atomics Atomics support | 09 October 2019, 16:52:36 UTC |
67365be | Tzu-Mao Li | 09 October 2019, 02:23:44 UTC | Only process Stores matching producer names | 09 October 2019, 02:23:44 UTC |
79dc0c0 | Pranav Bhandarkar | 08 October 2019, 23:56:21 UTC | Use CXX_FLAGS instead of CXXFLAGS for consistency | 08 October 2019, 23:56:21 UTC |
90701c9 | Pranav Bhandarkar | 08 October 2019, 23:40:48 UTC | Pass LLVM_CXXFLAGS to halide_config.make and use in apps/support_sa/Makefile.inc | 08 October 2019, 23:40:48 UTC |
37559c2 | Pranav Bhandarkar | 08 October 2019, 23:39:32 UTC | Fix untested typo in apps/hexagon_benchmarks/Makefile | 08 October 2019, 23:39:32 UTC |
18bbc91 | Tzu-Mao Li | 08 October 2019, 22:06:22 UTC | lift store indices from mutex & store | 08 October 2019, 22:06:22 UTC |
de55fa5 | Steven Johnson | 08 October 2019, 18:58:56 UTC | Merge pull request #4286 from halide/tzumao-python-funcref-setitem Allowing f[g[1]] = 2 in Python bindings | 08 October 2019, 18:58:56 UTC |
c770075 | Tzu-Mao Li | 07 October 2019, 21:30:55 UTC | tracking github issue for predicated vector store | 07 October 2019, 21:30:55 UTC |
6194540 | Tzu-Mao Li | 07 October 2019, 21:30:32 UTC | typo | 07 October 2019, 21:30:32 UTC |
5b20563 | Tzu-Mao Li | 07 October 2019, 21:29:43 UTC | typo | 07 October 2019, 21:29:43 UTC |
352f95f | Tzu-Mao Li | 07 October 2019, 21:29:02 UTC | github tracking issue for predicated vector atomic store | 07 October 2019, 21:29:02 UTC |
db138f6 | Tzu-Mao Li | 07 October 2019, 21:25:16 UTC | comment on the behavior of fail mutex allocation | 07 October 2019, 21:25:16 UTC |
570590f | Tzu-Mao Li | 07 October 2019, 21:23:29 UTC | typo | 07 October 2019, 21:23:29 UTC |
b8ea4a5 | Tzu-Mao Li | 07 October 2019, 21:21:20 UTC | error for unsupported OPENMP code | 07 October 2019, 21:21:20 UTC |
ec1b457 | Tzu-Mao Li | 07 October 2019, 21:21:10 UTC | grammar/typo | 07 October 2019, 21:21:10 UTC |
9deb78d | Tzu-Mao Li | 07 October 2019, 21:18:35 UTC | EOL at the end of file | 07 October 2019, 21:18:35 UTC |
3e6b00f | Andrew Adams | 04 October 2019, 20:05:39 UTC | Fix for mingw | 04 October 2019, 20:05:39 UTC |
dc35013 | Andrew Adams | 04 October 2019, 18:11:19 UTC | Speed up atomics test | 04 October 2019, 18:11:19 UTC |
abef53d | Andrew Adams | 04 October 2019, 17:24:15 UTC | clang-format | 04 October 2019, 17:24:15 UTC |
fc9cb6b | Tzu-Mao Li | 04 October 2019, 14:36:42 UTC | class style consistency | 04 October 2019, 14:36:42 UTC |
074315a | Tzu-Mao Li | 04 October 2019, 02:24:50 UTC | fix memory bug | 04 October 2019, 02:24:50 UTC |
a6da1d1 | Tzu-Mao Li | 04 October 2019, 02:15:13 UTC | Add CUDA vectorize test | 04 October 2019, 02:15:13 UTC |
da1664b | Tzu-Mao Li | 04 October 2019, 00:27:29 UTC | more comment style fix | 04 October 2019, 00:27:29 UTC |
38ccad5 | Tzu-Mao Li | 04 October 2019, 00:20:22 UTC | fix comment style | 04 October 2019, 00:20:22 UTC |
4b1104b | Tzu-Mao Li | 03 October 2019, 23:47:06 UTC | abstract halide mutex array | 03 October 2019, 23:47:06 UTC |
2750a83 | Tzu-Mao Li | 03 October 2019, 23:22:40 UTC | simplify simplify_stmts | 03 October 2019, 23:22:40 UTC |
b349bd4 | Tzu-Mao Li | 03 October 2019, 23:20:40 UTC | forbid self reference indexing on atomics | 03 October 2019, 23:20:40 UTC |
6125bbf | Tzu-Mao Li | 03 October 2019, 23:02:16 UTC | extract let binding of Store index | 03 October 2019, 23:02:16 UTC |
14d3819 | Tzu-Mao Li | 03 October 2019, 22:55:03 UTC | relax assertion criteria | 03 October 2019, 22:55:03 UTC |
ee17e9b | Tzu-Mao Li | 03 October 2019, 21:43:40 UTC | use halide_malloc instead of halide_default_malloc. Also fix a memory leaking bug. | 03 October 2019, 21:43:40 UTC |
d1dc0e6 | Tzu-Mao Li | 03 October 2019, 21:41:12 UTC | commments on mutex allocation optimization | 03 October 2019, 21:41:12 UTC |
d13015b | Tzu-Mao Li | 03 October 2019, 21:38:39 UTC | explain why we don't do anything in a fake thread pool | 03 October 2019, 21:38:39 UTC |
ce992d7 | Tzu-Mao Li | 03 October 2019, 21:37:37 UTC | simplify atomic irmutator | 03 October 2019, 21:37:37 UTC |
6ccfdb6 | Tzu-Mao Li | 03 October 2019, 21:36:35 UTC | Atomic must have body defined | 03 October 2019, 21:36:35 UTC |
8cfacc8 | Tzu-Mao Li | 03 October 2019, 21:33:51 UTC | use OutputImageParam to infer output buffer size | 03 October 2019, 21:33:51 UTC |
31b72d7 | Tzu-Mao Li | 03 October 2019, 21:28:09 UTC | better comment | 03 October 2019, 21:28:09 UTC |
32a490d | Tzu-Mao Li | 03 October 2019, 21:26:26 UTC | use string() instead of "" | 03 October 2019, 21:26:26 UTC |
d04a934 | Tzu-Mao Li | 03 October 2019, 21:25:32 UTC | Fix let lifting checking for mutex lock | 03 October 2019, 21:25:32 UTC |
6bfbf74 | Tzu-Mao Li | 03 October 2019, 19:24:48 UTC | class style & add comments & avoid copy | 03 October 2019, 19:24:48 UTC |
e12a198 | Tzu-Mao Li | 03 October 2019, 17:35:15 UTC | get rid of some name matching code | 03 October 2019, 17:35:15 UTC |
58f0028 | Tzu-Mao Li | 03 October 2019, 16:38:41 UTC | remove outdated comments | 03 October 2019, 16:38:41 UTC |
14fb1a6 | Tzu-Mao Li | 03 October 2019, 08:36:51 UTC | simplify atomic store codegen | 03 October 2019, 08:36:51 UTC |
ec2b544 | Tzu-Mao Li | 03 October 2019, 08:18:28 UTC | build realize group outside atomic when possible | 03 October 2019, 08:18:28 UTC |
b3abc8d | Tzu-Mao Li | 03 October 2019, 02:47:43 UTC | fix test | 03 October 2019, 02:47:43 UTC |
359fe49 | Tzu-Mao Li | 03 October 2019, 02:31:30 UTC | fix CodeGen_C & CodeGen_OpenCL | 03 October 2019, 02:31:30 UTC |
18f96f7 | Tzu-Mao Li | 03 October 2019, 00:39:49 UTC | add nested atomic test | 03 October 2019, 00:39:49 UTC |
110cb61 | Tzu-Mao Li | 03 October 2019, 00:32:28 UTC | Matching Atomic's IR semantics better by only issuing atomic stores for corresponding producers | 03 October 2019, 00:32:28 UTC |
fd89966 | Tzu-Mao Li | 02 October 2019, 23:45:16 UTC | allocate mutex buffers at Allocate nodes except for output buffers | 02 October 2019, 23:45:16 UTC |
27f9651 | Tzu-Mao Li | 02 October 2019, 21:36:05 UTC | remove unused let bindings | 02 October 2019, 21:36:05 UTC |
b52e285 | Tzu-Mao Li | 02 October 2019, 21:32:36 UTC | remove commented out code | 02 October 2019, 21:32:36 UTC |
19f7f50 | Tzu-Mao Li | 02 October 2019, 21:20:09 UTC | supports atomic add virtual method | 02 October 2019, 21:20:09 UTC |
60db39e | Tzu-Mao Li | 02 October 2019, 21:00:57 UTC | remove mutex_indices field in Atomic | 02 October 2019, 21:00:57 UTC |
76c288f | Tzu-Mao Li | 02 October 2019, 19:17:45 UTC | use string::empty() | 02 October 2019, 19:17:45 UTC |
ef65127 | Tzu-Mao Li | 02 October 2019, 19:17:11 UTC | use string::empty() | 02 October 2019, 19:17:11 UTC |
71d29f8 | Tzu-Mao Li | 02 October 2019, 19:16:28 UTC | typo in error message | 02 October 2019, 19:16:28 UTC |
af09b3a | Tzu-Mao Li | 02 October 2019, 19:15:57 UTC | use ScopedBindings | 02 October 2019, 19:15:57 UTC |
3f20c9a | Tzu-Mao Li | 02 October 2019, 19:13:01 UTC | check nested atomic mutex locks | 02 October 2019, 19:13:01 UTC |
19a6bd6 | Tzu-Mao Li | 02 October 2019, 19:01:46 UTC | comments about exception handling inside mutex locks | 02 October 2019, 19:01:46 UTC |
7bf1f5c | Tzu-Mao Li | 02 October 2019, 18:55:05 UTC | Let halide_mutex_array_create returns a single pointer | 02 October 2019, 18:55:05 UTC |
7ed763b | Tzu-Mao Li | 02 October 2019, 18:34:21 UTC | fix compile | 02 October 2019, 18:34:21 UTC |
f1ef78f | Tzu-Mao Li | 02 October 2019, 18:20:26 UTC | minor style change | 02 October 2019, 18:20:26 UTC |
cb0063b | Tzu-Mao Li | 02 October 2019, 18:18:32 UTC | minor style change | 02 October 2019, 18:18:32 UTC |
1d5a6dd | Tzu-Mao Li | 02 October 2019, 18:13:53 UTC | fix atomic test | 02 October 2019, 18:13:53 UTC |
a281bcc | Tzu-Mao Li | 02 October 2019, 18:12:40 UTC | cleanup | 02 October 2019, 18:12:40 UTC |
3039790 | Tzu-Mao Li | 02 October 2019, 18:08:58 UTC | Merge branch 'master' of https://github.com/halide/Halide into tzumao-atomics | 02 October 2019, 18:08:58 UTC |
4d6340d | Tzu-Mao Li | 02 October 2019, 17:54:38 UTC | remove debug prints & add some comments | 02 October 2019, 17:54:38 UTC |
6c60cf6 | Tzu-Mao Li | 02 October 2019, 17:11:27 UTC | fix cmake compile | 02 October 2019, 17:11:27 UTC |
48aad76 | Tzu-Mao Li | 02 October 2019, 16:59:23 UTC | separate pass for atomic mutex allocation & lock/unlock | 02 October 2019, 16:59:23 UTC |
2d02dfd | Tzu-Mao Li | 01 October 2019, 21:56:47 UTC | fix halide runtime compile | 01 October 2019, 21:56:47 UTC |
7763a6c | Tzu-Mao Li | 01 October 2019, 21:22:12 UTC | fix compile | 01 October 2019, 21:22:12 UTC |
b994431 | Tzu-Mao Li | 01 October 2019, 21:20:26 UTC | add mutex array runtime | 01 October 2019, 21:20:26 UTC |
1609d79 | Tzu-Mao Li | 01 October 2019, 20:31:56 UTC | better detection for atomic mutex lock removal | 01 October 2019, 20:31:56 UTC |
4a89a8f | Steven Johnson | 01 October 2019, 20:31:06 UTC | Merge pull request #4287 from LebedevRI/shared-llvm CMake: add optional LLVM_USE_SHARED_LLVM_LIBRARY option. | 01 October 2019, 20:31:06 UTC |
aff87e6 | Tzu-Mao Li | 01 October 2019, 19:38:13 UTC | fix comment style | 01 October 2019, 19:38:13 UTC |
94da5ab | Tzu-Mao Li | 01 October 2019, 19:37:31 UTC | fix comment typo | 01 October 2019, 19:37:31 UTC |
dffb5f2 | Tzu-Mao Li | 01 October 2019, 17:07:38 UTC | Merge branch 'master' of https://github.com/halide/Halide into tzumao-atomics | 01 October 2019, 17:07:38 UTC |
fc6d889 | Tzu-Mao Li | 01 October 2019, 17:06:04 UTC | Merge branch 'master' of https://github.com/halide/Halide into tzumao-python-funcref-setitem | 01 October 2019, 17:06:04 UTC |
eb63cfd | Tzu-Mao Li | 01 October 2019, 17:03:57 UTC | print double-to-float conversion loss and remove float overload and actually test it | 01 October 2019, 17:03:57 UTC |
50908b2 | Tzu-Mao Li | 01 October 2019, 16:49:30 UTC | address review comments | 01 October 2019, 16:49:30 UTC |
283b4a4 | Steven Johnson | 01 October 2019, 16:34:06 UTC | Merge pull request #4272 from halide/srj-deriv3 Clean up Derivative API | 01 October 2019, 16:34:06 UTC |
ba064cf | Steven Johnson | 01 October 2019, 16:33:29 UTC | Merge pull request #4290 from halide/srj-llvmfix Fix for trunk LLVM changes | 01 October 2019, 16:33:29 UTC |
fcbdd78 | Steven Johnson | 01 October 2019, 16:33:00 UTC | Merge pull request #4292 from ghost-luck/master Fix of a typo in a Makefile. | 01 October 2019, 16:33:00 UTC |
2b3e7ff | Tzu-Mao Li | 01 October 2019, 16:13:52 UTC | Merge branch 'master' of https://github.com/halide/Halide into tzumao-python-funcref-setitem | 01 October 2019, 16:13:52 UTC |
c3d7f14 | Nikita Glukhov | 01 October 2019, 06:30:33 UTC | Fix of a typo in a Makefile. | 01 October 2019, 06:30:33 UTC |
00d7205 | Steven Johnson | 30 September 2019, 19:57:45 UTC | Merge branch 'master' into srj-deriv3 | 30 September 2019, 19:57:45 UTC |
fb0d1e5 | Steven Johnson | 30 September 2019, 18:55:27 UTC | Fix for trunk LLVM changes | 30 September 2019, 18:55:27 UTC |
cbad285 | Roman Lebedev | 30 September 2019, 16:26:14 UTC | Address review notes. | 30 September 2019, 17:18:18 UTC |
d8126f0 | Steven Johnson | 30 September 2019, 15:57:08 UTC | Merge pull request #4288 from LebedevRI/clang10 Autoscheduler: Weights::randomize(): silence clang's -Wimplicit-int-float-conversion | 30 September 2019, 15:57:08 UTC |
4c90964 | Roman Lebedev | 30 September 2019, 15:09:29 UTC | Autoscheduler: Weights::randomize(): silence clang's -Wimplicit-int-float-conversion ``` Halide/apps/autoscheduler/Weights.cpp:22:34: error: implicit conversion from 'std::mersenne_twister_engine<unsigned long, 32, 624, 397, 31, 2567483615, 11, 4294967295, 7, 2636928640, 15, 4022730752, 18, 1812433253>::result_type' (aka 'unsigned long') to 'float' changes value from 4294967295 to 4294967296 [-Werror,-Wimplicit-int-float-conversion] f = ((float)rng()) / rng.max() - 0.5f; ~ ^~~~~~~~~ ``` I think we should just make the cast explicit here? | 30 September 2019, 15:09:29 UTC |
d45ff96 | Roman Lebedev | 28 September 2019, 09:16:42 UTC | CMake: add optional HALIDE_SHARED_LLVM_LIBRARY option. As discussed in https://github.com/halide/Halide/issues/4284 one important thing is to be able to link to shared versions of LLVM libraries. This is off-by-default. The important [existing] caveat to note is that i don't see where those llvm libraries-to-be-linked are specified anywhere in the halide_config.cmake, so as long as you are a. building static halide and b. not building halide as a cmake sub-project (via add_subdirectory) i don't see how the linking can succeed, both either with old static llvm libraries, and now with shared llvm library. The best i can tell `llvm-config --system-libs` is only needed when linking to static libraries, to propagate their deps, so let's not do that for shared llvm libs. And the switch to `llvm_config()` is the main thing here, it internally handles `USE_SHARED`, and does `target_link_libraries()`. Since this is opt-in, i don't believe this should have any impact on existing users. | 30 September 2019, 14:41:01 UTC |
333fbc0 | Tzu-Mao Li | 29 September 2019, 20:09:37 UTC | fix test | 29 September 2019, 20:09:37 UTC |
37dffe5 | Tzu-Mao Li | 29 September 2019, 20:00:06 UTC | Allows f[g[1]] = 2 in Python bindings | 29 September 2019, 20:00:06 UTC |
e49f560 | Tzu-Mao Li | 29 September 2019, 03:57:24 UTC | minor comment change | 29 September 2019, 03:57:24 UTC |
1449d5b | Tzu-Mao Li | 29 September 2019, 02:56:21 UTC | python bindings | 29 September 2019, 02:56:21 UTC |
2bdc230 | Tzu-Mao Li | 29 September 2019, 00:02:26 UTC | fix error tests | 29 September 2019, 00:02:26 UTC |
b6c6018 | Tzu-Mao Li | 28 September 2019, 23:55:25 UTC | cleanup & errors & revive tests & add failing test | 28 September 2019, 23:55:25 UTC |
99aad46 | Tzu-Mao Li | 28 September 2019, 23:03:53 UTC | uncomment commented out code | 28 September 2019, 23:03:53 UTC |
402f5b4 | Tzu-Mao Li | 28 September 2019, 23:02:31 UTC | fix async & improve the test | 28 September 2019, 23:02:31 UTC |