db88adf | Steven Johnson | 27 August 2019, 18:28:24 UTC | Update DataLayout strings for trunk LLVM (Only tested for linux-x86-64 so far; other targets may also need updating.) | 27 August 2019, 18:28:24 UTC |
a743e63 | Andrew Adams | 27 August 2019, 16:51:11 UTC | Merge pull request #4177 from halide/srj-tidy clang-tidy: prefer empty() to size() == 0 (and similar) | 27 August 2019, 16:51:11 UTC |
8d61fd8 | Steven Johnson | 27 August 2019, 01:00:10 UTC | clang-tidy: prefer empty() to size() == 0 (and similar) | 27 August 2019, 01:00:10 UTC |
65c26cb | Steven Johnson | 26 August 2019, 21:49:31 UTC | Merge pull request #4174 from halide/srj-tidy Remove unused 'using' decls to appease clang-tidy | 26 August 2019, 21:49:31 UTC |
806f467 | Dillon Sharlet | 26 August 2019, 21:18:30 UTC | Merge pull request #4103 from gasparitiago/patch-1 Fix comment on FFT App Generator file | 26 August 2019, 21:18:30 UTC |
feb095d | Steven Johnson | 26 August 2019, 17:47:43 UTC | Remove unused 'using' decls to appease clang-tidy | 26 August 2019, 17:47:43 UTC |
24b37b8 | Andrew Adams | 26 August 2019, 17:07:05 UTC | Merge pull request #3667 from halide/bfloat16 Basic float16 / bfloat16 support | 26 August 2019, 17:07:05 UTC |
5d2b0bf | Steven Johnson | 26 August 2019, 16:26:46 UTC | Merge pull request #4169 from halide/srj-bench Remove min_iter and max_iter from benchmark & RunGen | 26 August 2019, 16:26:46 UTC |
6e309d7 | Andrew Adams | 25 August 2019, 21:22:50 UTC | Merge remote-tracking branch 'origin/master' into bfloat16 | 25 August 2019, 21:22:50 UTC |
0110353 | Zalman Stern | 23 August 2019, 17:15:15 UTC | Merge pull request #4171 from halide/srj-memset Use zero-init instead of memset for PipelineFeatures | 23 August 2019, 17:15:15 UTC |
27bd93c | Steven Johnson | 23 August 2019, 00:55:14 UTC | Use zero-init instead of memset for PipelineFeatures gcc will fail with -Werror=class-memaccess because PipelineFeatures is not TriviallyCopyable | 23 August 2019, 00:55:14 UTC |
5cf2a29 | Steven Johnson | 23 August 2019, 00:15:50 UTC | Remove min_iter and max_iter from benchmark & RunGen These aren't really that useful; min_time is usually the only knob of this sort that is interesting to fiddle with. Remove them entirely to simplify code and reduce the temptation to mess with them. (Some versions of gbenchmark don't support min_iter and max_iter at all, only min_time.) | 23 August 2019, 00:15:50 UTC |
ae7edd3 | Steven Johnson | 22 August 2019, 22:46:45 UTC | Merge pull request #3988 from mgharbi/pytorch-wrapper Pytorch wrapper | 22 August 2019, 22:46:45 UTC |
bdca077 | Steven Johnson | 22 August 2019, 20:43:51 UTC | Merge pull request #4167 from halide/srj-weights-out Remove HL_WEIGHTS_OUT_DIR | 22 August 2019, 20:43:51 UTC |
97f85b8 | Steven Johnson | 22 August 2019, 18:59:11 UTC | Merge pull request #4163 from halide/simplifications_for_unrolling_sliding_window Simplifications for unrolling sliding window | 22 August 2019, 18:59:11 UTC |
8132634 | Steven Johnson | 22 August 2019, 18:58:17 UTC | Merge pull request #4160 from halide/srj-warn Default user errors and user warnings to output on a single line | 22 August 2019, 18:58:17 UTC |
b8d9f89 | Steven Johnson | 22 August 2019, 18:58:04 UTC | Remove HL_WEIGHTS_OUT_DIR It was only referenced in single-shot scheduling, but we shouldn't ever be saving from that path anyway (only from the retraining loop). Also change save_weights() to fail with abort() if you try to save with an empty output path (rather than quietly doing nothing), and upgrade save failures from assert(0) to abort(), because that's not an error we want to risk eliding based on compiler settings. | 22 August 2019, 18:58:04 UTC |
c581b46 | Andrew Adams | 22 August 2019, 18:15:26 UTC | Add Halide tracking issue for llvm bug | 22 August 2019, 18:15:26 UTC |
52e3c3a | Andrew Adams | 22 August 2019, 18:13:29 UTC | Pass Type by const ref | 22 August 2019, 18:13:29 UTC |
2b9e18e | Andrew Adams | 22 August 2019, 17:42:14 UTC | More simplifier rules | 22 August 2019, 17:42:14 UTC |
11c3415 | Steven Johnson | 22 August 2019, 17:41:03 UTC | Update Error.cpp | 22 August 2019, 17:41:03 UTC |
d1f470b | Steven Johnson | 22 August 2019, 17:36:19 UTC | Merge branch 'master' into srj-warn | 22 August 2019, 17:36:19 UTC |
26d0a35 | Steven Johnson | 22 August 2019, 17:26:00 UTC | Merge pull request #4159 from halide/pdb_fix_clang_build Fix build when LLVM is built with libc++ | 22 August 2019, 17:26:00 UTC |
167f9b1 | Steven Johnson | 22 August 2019, 17:21:46 UTC | Merge pull request #4164 from halide/srj-sched_h Generate legal .schedule.h file for non-auto-schedule code | 22 August 2019, 17:21:46 UTC |
130610d | Steven Johnson | 22 August 2019, 01:14:55 UTC | Generate legal .schedule.h file for non-auto-schedule code Previous codegen was bogus and assert-fail-laden; it's not a useful case, but it shouldn't crater. | 22 August 2019, 01:14:55 UTC |
0ba002f | Steven Johnson | 22 August 2019, 01:06:12 UTC | Merge pull request #4162 from halide/srj-sched_h Improve assertion in FunctionDAG | 22 August 2019, 01:06:12 UTC |
b269166 | Andrew Adams | 21 August 2019, 23:58:32 UTC | Fix new rules. Add test | 21 August 2019, 23:58:32 UTC |
f99977e | Steven Johnson | 21 August 2019, 23:55:53 UTC | Improve assertion in FunctionDAG The failure mode for autoscheduling-a-func-with-a-schedule-on-it is currently inscrutable; change from an internal_assert to a user_assert with a more-helpful message. | 21 August 2019, 23:55:53 UTC |
366af58 | Andrew Adams | 21 August 2019, 23:37:11 UTC | New select rewrites for simplifying unrolled slid loops Unrolling happens after sliding but before partitioning, so the extents of the unrolled loop should have an upper bound, but it can be hard for the compiler to see this through all the selects. This PR makes things way more aggressive in the about-to-fail case in unrolling. It now exhaustively substitutes in everything for maximally aggressive simplification, and there are a bunch of new simplifier rules to help cancel things through select nodes. This exhaustive substitution is somewhat dangerous, but this is a case where the compiler is about to throw an error if it doesn't simplify down to almost a constant. | 21 August 2019, 23:37:11 UTC |
af485ac | Steven Johnson | 21 August 2019, 23:34:27 UTC | Merge pull request #4161 from halide/srj-sched_h Tweak .schedule.h output to be appease lint | 21 August 2019, 23:34:27 UTC |
07c1875 | Steven Johnson | 21 August 2019, 22:52:25 UTC | Tweak .schedule.h output to be appease lint | 21 August 2019, 22:52:25 UTC |
e32e6c6 | Pranav Bhandarkar | 21 August 2019, 21:45:12 UTC | Revert "temp patch to fix broken build using clang" This reverts commit b99972475aa14d61671481b534157d9e0a1f90e0. | 21 August 2019, 21:45:12 UTC |
34d0eb4 | Steven Johnson | 21 August 2019, 21:38:53 UTC | Default user errors and user warnings to output on a single line Tweak the formatting of ErrorReport to use only a single line of output (aside from newlines inserted by the user <<), to make it more tractable to filter warnings/errors via grep. (I'm not 100% sold on this change; my motivation is to be able to silence spammy warnings during autoscheduler, but there may be a better / cleaner way.) | 21 August 2019, 21:38:53 UTC |
dcb040b | Steven Johnson | 21 August 2019, 17:51:57 UTC | Merge pull request #4156 from halide/srj-tidy Remove unused 'using' | 21 August 2019, 17:51:57 UTC |
9cca54c | Steven Johnson | 21 August 2019, 16:59:29 UTC | Merge pull request #4157 from benoitsteiner/master Made model.cpp compile again | 21 August 2019, 16:59:29 UTC |
6325da5 | Benoit Steiner | 21 August 2019, 00:48:36 UTC | Updated the test filters: Filter test that cover ops recently added that we don't support yet (e.g. quantization) Stop filtering tests that we are now able to run properly | 21 August 2019, 00:48:36 UTC |
9bab4b8 | Benoit Steiner | 20 August 2019, 23:24:44 UTC | Reverted incorrect API change | 20 August 2019, 23:24:44 UTC |
e0141bc | Benoit Steiner | 20 August 2019, 23:22:49 UTC | Fixed Makefile on non OS-X platforms | 20 August 2019, 23:22:49 UTC |
a0654dc | Benoit Steiner | 20 August 2019, 22:47:35 UTC | Fixed model test | 20 August 2019, 22:47:35 UTC |
13f56cc | Benoit Steiner | 20 August 2019, 21:53:55 UTC | Made model.cpp compile again | 20 August 2019, 21:53:55 UTC |
0e655da | Steven Johnson | 20 August 2019, 21:02:26 UTC | Remove unused 'using' | 20 August 2019, 21:02:26 UTC |
c9d5ffd | Steven Johnson | 20 August 2019, 20:31:56 UTC | Merge pull request #4145 from halide/srj-train Improve (re)train_cost_model tool | 20 August 2019, 20:31:56 UTC |
765ddba | Andrew Adams | 20 August 2019, 20:01:37 UTC | Merge branch 'bfloat16' of https://github.com/halide/Halide into bfloat16 | 20 August 2019, 20:01:37 UTC |
380aa63 | Andrew Adams | 20 August 2019, 20:01:28 UTC | Add llvm bug link | 20 August 2019, 20:01:28 UTC |
affe012 | Steven Johnson | 20 August 2019, 17:03:48 UTC | Merge branch 'master' into srj-train | 20 August 2019, 17:03:48 UTC |
40931b1 | Steven Johnson | 20 August 2019, 17:02:42 UTC | Merge pull request #4151 from halide/srj-likely-extent Smarten UnrollLoops; remove HL_PERMIT_FAILED_UNROLL=1 defaults | 20 August 2019, 17:02:42 UTC |
61def08 | Steven Johnson | 20 August 2019, 17:02:18 UTC | Merge pull request #4153 from halide/srj-auto_schedule_mat_mul auto_schedule_mat_mul: increase thresh from 3.5 -> 5.0 (Issue #4152) | 20 August 2019, 17:02:18 UTC |
03e8fc9 | Pranav Bhandarkar | 19 August 2019, 23:15:51 UTC | If the host compiler is clang, then the tests may need-stdlib=libc++ | 19 August 2019, 23:15:51 UTC |
f5cacaa | Andrew Adams | 19 August 2019, 23:08:09 UTC | Merge remote-tracking branch 'origin/master' into bfloat16 | 19 August 2019, 23:08:09 UTC |
2005576 | Andrew Adams | 19 August 2019, 23:07:52 UTC | Delete unused intrinsics | 19 August 2019, 23:07:52 UTC |
be4ee9a | Andrew Adams | 19 August 2019, 23:05:03 UTC | Move comment to less confusing place | 19 August 2019, 23:05:03 UTC |
5ef1dbb | Steven Johnson | 19 August 2019, 22:10:44 UTC | auto_schedule_mat_mul: increase thresh from 3.5 -> 5.0 (Issue #4152) | 19 August 2019, 22:10:44 UTC |
792de10 | Steven Johnson | 19 August 2019, 21:52:15 UTC | Update IROperator.cpp | 19 August 2019, 21:52:15 UTC |
5f5aaf4 | Steven Johnson | 19 August 2019, 21:50:22 UTC | More remove_likelies() use | 19 August 2019, 21:50:22 UTC |
9b8b1a5 | Steven Johnson | 19 August 2019, 21:45:58 UTC | consolidate remove_likelies() | 19 August 2019, 21:45:58 UTC |
0aa3766 | Steven Johnson | 19 August 2019, 21:36:33 UTC | remove HL_PERMIT_FAILED_UNROLL | 19 August 2019, 21:36:33 UTC |
e110b79 | Steven Johnson | 19 August 2019, 21:34:07 UTC | UnrollLoops: remove likely() (etc) for the one-last-chance case So that extents of the form `(a - likely(a))` can correctly be deduced as constant. | 19 August 2019, 21:34:07 UTC |
d7a9e96 | Steven Johnson | 19 August 2019, 18:04:53 UTC | Use host Target for training, with smarts to avoid AVX512 | 19 August 2019, 18:04:53 UTC |
56e2041 | Steven Johnson | 19 August 2019, 17:24:53 UTC | Merge pull request #4146 from halide/srj-tsan-features Make can_use_target TSAN-safe (Issue #4140) | 19 August 2019, 17:24:53 UTC |
c8af918 | Steven Johnson | 17 August 2019, 00:44:04 UTC | Only test autoscheduler on Linux | 17 August 2019, 00:44:04 UTC |
34364d7 | Steven Johnson | 16 August 2019, 23:57:47 UTC | Avoid unused-var warnings, grr | 16 August 2019, 23:57:47 UTC |
3cc46cb | Andrew Adams | 16 August 2019, 23:42:11 UTC | Merge remote-tracking branch 'origin/master' into bfloat16 | 16 August 2019, 23:42:11 UTC |
8c7338a | Steven Johnson | 16 August 2019, 23:22:24 UTC | Make can_use_target TSAN-safe (Issue #4140) | 16 August 2019, 23:22:24 UTC |
37edd63 | Steven Johnson | 16 August 2019, 22:48:44 UTC | Update autotune_loop.sh | 16 August 2019, 22:48:44 UTC |
7871e70 | Steven Johnson | 16 August 2019, 22:15:04 UTC | Update retrain_cost_model.cpp | 16 August 2019, 22:15:04 UTC |
b9eaa49 | Steven Johnson | 16 August 2019, 21:28:51 UTC | Update retrain_cost_model.cpp | 16 August 2019, 21:28:51 UTC |
9d36f67 | Steven Johnson | 16 August 2019, 21:24:03 UTC | --weights_out -> required; --weights -> --initial_weights (optional) | 16 August 2019, 21:24:03 UTC |
dc07855 | Steven Johnson | 16 August 2019, 21:01:11 UTC | Update autotune_loop.sh | 16 August 2019, 21:01:11 UTC |
692be4f | Steven Johnson | 16 August 2019, 19:56:44 UTC | Rename | 16 August 2019, 19:56:44 UTC |
e400274 | Steven Johnson | 16 August 2019, 19:55:14 UTC | Improve train_cost_model tool - Rename to retrain_cost_model, since (1) that's what it does, and (2) this avoids confusion with the Generator-produces filter of the same name - Remove all use of env vars for inputs; explicit commandline flags are now required instead. Update autotune_loop.sh accordingly. - add an option to copy the best .schedule.h file somewhere at the end (note that the previous HL_BEST_SCHEDULE_FILE env var actually output benchmark results, not the schedule file) - Check the return value of calls to train_cost_model() and cost_model(), just in case someone uses a nonstandard override of halide_error. - Tighten some verbosity of output (display just filenames instead of pathnames in some cases) - Some minor drive-by style cleanups. | 16 August 2019, 19:55:14 UTC |
872c56d | Michael Gharbi | 16 August 2019, 19:22:57 UTC | fixes Make dependencies | 16 August 2019, 19:22:57 UTC |
bfc184d | Michael Gharbi | 16 August 2019, 19:20:36 UTC | fixes python indent | 16 August 2019, 19:20:36 UTC |
e206cc8 | Michael Gharbi | 16 August 2019, 19:16:11 UTC | removes WEAK qualifier for cuda hooks | 16 August 2019, 19:16:11 UTC |
cb23358 | Michael Gharbi | 16 August 2019, 19:04:33 UTC | fixes cpp indent | 16 August 2019, 19:04:33 UTC |
b39793d | Michael Gharbi | 16 August 2019, 18:25:43 UTC | fixes typos in readme | 16 August 2019, 18:25:43 UTC |
60f1c5c | Steven Johnson | 16 August 2019, 17:28:17 UTC | Merge pull request #4143 from halide/fix_simplifier_bounds_overflow Avoid 64-bit overflow when tracking constant bounds in simplifier | 16 August 2019, 17:28:17 UTC |
83ab2c9 | Michael Gharbi | 16 August 2019, 01:21:28 UTC | rewrite adjoint using Derivative.cpp | 16 August 2019, 01:21:28 UTC |
27e9440 | Michael Gharbi | 16 August 2019, 00:53:08 UTC | adds a gradient check | 16 August 2019, 00:53:08 UTC |
538a804 | Michael Gharbi | 16 August 2019, 00:47:31 UTC | adds note on test | 16 August 2019, 00:47:31 UTC |
a7a9574 | Michael Gharbi | 16 August 2019, 00:32:20 UTC | adds readme | 16 August 2019, 00:32:20 UTC |
c93ea38 | Michael Gharbi | 16 August 2019, 00:32:11 UTC | fix output | 16 August 2019, 00:32:11 UTC |
7ab366c | Michael Gharbi | 16 August 2019, 00:06:32 UTC | merge | 16 August 2019, 00:14:21 UTC |
7fa1028 | Andrew Adams | 15 August 2019, 20:38:15 UTC | Avoid 64-bit overflow when tracking constant bounds in simplifier These are no-overflow types, so treat such cases not as things that overflow (in which an overflow in the upper bound would affect the lower bound), but rather as things which still have upper/lower bounds that we can no longer represent. | 15 August 2019, 20:38:15 UTC |
4628a24 | Steven Johnson | 15 August 2019, 18:58:21 UTC | Merge pull request #4137 from halide/srj-has-feat Allow Target::has_feature() to operator on halide_target_feature_t | 15 August 2019, 18:58:21 UTC |
1d49dbd | Steven Johnson | 15 August 2019, 00:36:51 UTC | Allow Target::has_feature() to operator on halide_target_feature_t This is really a workaround to allow generated .schedule.h files to compile happily in all cases. A better long-term answer would be for those files to use the Target::Feature names (rather than the halide_target_feature_t names), but doing so will require an additional value->name map (which must be maintained and could get out of sync), so I'm reluctant to do that without a little more thought. | 15 August 2019, 17:38:11 UTC |
07c3c27 | Steven Johnson | 15 August 2019, 17:07:02 UTC | Merge pull request #4138 from halide/srj-llvm-fix Fix for https://reviews.llvm.org/rL368647, from alinas@ | 15 August 2019, 17:07:02 UTC |
70485a7 | Steven Johnson | 15 August 2019, 17:03:16 UTC | Update CodeGen_LLVM.cpp | 15 August 2019, 17:03:16 UTC |
3affe21 | Andrew Adams | 15 August 2019, 16:38:35 UTC | Merge remote-tracking branch 'origin/master' into bfloat16 | 15 August 2019, 16:38:35 UTC |
98991c0 | Steven Johnson | 15 August 2019, 01:16:00 UTC | Fix for https://reviews.llvm.org/rL368647, from alinas@ | 15 August 2019, 01:16:00 UTC |
7eb5b24 | Steven Johnson | 14 August 2019, 22:35:38 UTC | Merge pull request #4131 from halide/srj-tic Change halide_toc_impl to use debug(1) instead of debug(0) | 14 August 2019, 22:35:38 UTC |
3d29b66 | Andrew Adams | 14 August 2019, 22:16:24 UTC | Avoid half types in the x86 backend | 14 August 2019, 22:16:24 UTC |
8766bc4 | Steven Johnson | 14 August 2019, 21:13:43 UTC | Merge pull request #4133 from halide/srj-aslog2 Clean up default autotuner output | 14 August 2019, 21:13:43 UTC |
5afe043 | Steven Johnson | 14 August 2019, 18:45:51 UTC | Merge pull request #4129 from benoitsteiner/master Improved onnx support | 14 August 2019, 18:45:51 UTC |
2576f3d | Steven Johnson | 14 August 2019, 18:09:11 UTC | Merge pull request #4134 from halide/srj-llvm-fix Fix for Trunk LLVM | 14 August 2019, 18:09:11 UTC |
19964e9 | Steven Johnson | 14 August 2019, 16:58:31 UTC | Fix for Trunk LLVM | 14 August 2019, 17:03:19 UTC |
7c78788 | Steven Johnson | 14 August 2019, 16:44:04 UTC | Update DefaultCostModel.cpp | 14 August 2019, 16:44:04 UTC |
cdec241 | Steven Johnson | 13 August 2019, 23:48:05 UTC | Back to dump | 13 August 2019, 23:48:05 UTC |
5f56d69 | Steven Johnson | 13 August 2019, 23:37:29 UTC | Update AutoSchedule.cpp | 13 August 2019, 23:37:29 UTC |
a6599e7 | Steven Johnson | 13 August 2019, 23:20:43 UTC | Update AutoSchedule.cpp | 13 August 2019, 23:20:43 UTC |
d99d640 | Steven Johnson | 13 August 2019, 23:19:40 UTC | Also emit the target being used | 13 August 2019, 23:19:40 UTC |