| 2c92023 | Steven Johnson | 04 October 2018, 17:17:32 UTC | Merge branch 'master' into srj-mak | 04 October 2018, 17:17:32 UTC |
| 558cfad | Andrew Adams | 04 October 2018, 17:06:48 UTC | Merge pull request #3308 from halide/fix_runtime_buffer_performance Add ALWAYS_INLINE qualifiers to fix debug mode performance | 04 October 2018, 17:06:48 UTC |
| 6a3a4bf | Steven Johnson | 04 October 2018, 16:35:00 UTC | Merge pull request #3303 from halide/srj-f16 Add support for Buffer<float16> in Python bindings (Issue #3263) | 04 October 2018, 16:35:00 UTC |
| c07f3e8 | Steven Johnson | 04 October 2018, 01:36:36 UTC | Merge branch 'master' into srj-f16 | 04 October 2018, 01:36:36 UTC |
| 093617a | Steven Johnson | 04 October 2018, 01:35:26 UTC | Expand LLVM_CONFIG and CLANG for test_apps The test_apps target recursively calls sub-makefiles (which can change the WD), but the buildbots define LLVM_CONFIG and CLANG as relative paths, so these can become invalid when building apps. (Why this hasn't caused more failures in the past is a bit of a mystery.) Use $(realpath) to explicitly expand these vars for the sub-makefile calls to address this. | 04 October 2018, 01:35:26 UTC |
| bf7500b | Dillon Sharlet | 03 October 2018, 20:21:29 UTC | Merge pull request #3225 from aankit-ca/hexagon_vgather Hexagon vgather support | 03 October 2018, 20:21:29 UTC |
| 13394eb | Andrew Adams | 03 October 2018, 18:41:42 UTC | Add ALWAYS_INLINE qualifiers to fix debug mode performance | 03 October 2018, 18:41:42 UTC |
| c02750d | Steven Johnson | 03 October 2018, 16:31:59 UTC | Empty | 03 October 2018, 16:31:59 UTC |
| cbf88cb | Steven Johnson | 03 October 2018, 16:13:44 UTC | Merge branch 'master' into srj-f16 | 03 October 2018, 16:13:44 UTC |
| 7c8355c | Ankit Aggarwal | 03 October 2018, 08:46:44 UTC | 1. Added v65 feature check before store_in VTCM. 2. Better failure message for using VTCM without v65. | 03 October 2018, 08:46:44 UTC |
| 5569b3a | Dillon Sharlet | 02 October 2018, 19:38:32 UTC | Merge pull request #3304 from halide/pdb_hvx_profiling_fix Fix profiling when a stage is offloaded to HVX. | 02 October 2018, 19:38:32 UTC |
| c288889 | Steven Johnson | 02 October 2018, 17:59:27 UTC | Merge branch 'master' into srj-f16 | 02 October 2018, 17:59:27 UTC |
| 079c244 | Steven Johnson | 02 October 2018, 01:00:59 UTC | Merge pull request #3302 from halide/srj-quietdiv Add quiet_div, quiet_mod to C backend (Issue #3300) | 02 October 2018, 01:00:59 UTC |
| bb826d3 | Steven Johnson | 02 October 2018, 00:57:55 UTC | Trivial change for buildbot | 02 October 2018, 00:57:55 UTC |
| 682f392 | Steven Johnson | 01 October 2018, 23:27:30 UTC | Back out internal_assert changes for quiet_div/mod | 01 October 2018, 23:27:30 UTC |
| f338362 | Steven Johnson | 01 October 2018, 23:14:20 UTC | assert denom != 0 in quiet_div, quiet_mod for LLVM too | 01 October 2018, 23:14:20 UTC |
| d16a9a9 | Steven Johnson | 01 October 2018, 23:09:49 UTC | change user_error -> internal_assert | 01 October 2018, 23:09:49 UTC |
| de59814 | Steven Johnson | 01 October 2018, 22:13:08 UTC | Add support for Buffer<float16> in Python bindings (Issue #3263) | 01 October 2018, 22:13:08 UTC |
| 3ed06b2 | Steven Johnson | 01 October 2018, 21:00:48 UTC | Add quiet_div, quiet_mod to C backend (Issue #3300) | 01 October 2018, 21:00:48 UTC |
| fd0bef0 | Ankit Aggarwal | 01 October 2018, 18:45:26 UTC | Corrected is_intrinsic(gather) in CodeGen_Hexagon.cpp | 01 October 2018, 18:45:26 UTC |
| 15cef92 | Pranav Bhandarkar | 30 September 2018, 06:37:27 UTC | Fix profiling when a stage is offloaded to HVX. The HVX device side runtime maintains its own profiling state that is sampled by the sampling thread. However, if a stage is offloaded to HVX, the current_func is set on the host, but the HVX side runtime won't know of it until it executes a halide_profiler_set_current_func call on its own. This means all the profiling information up until such a point is lost and counted as "overhead" | 30 September 2018, 06:37:27 UTC |
| 7608320 | Zalman Stern | 29 September 2018, 01:12:51 UTC | Merge pull request #3295 from halide/embedded_buffer_alignment Increase embedded buffer alignment to max vector size of target if | 29 September 2018, 01:12:51 UTC |
| 9f930d5 | Zalman Stern | 28 September 2018, 22:35:25 UTC | Merge pull request #2576 from halide/async Async | 28 September 2018, 22:35:25 UTC |
| de37a31 | Z Stern | 28 September 2018, 19:53:34 UTC | Increase embedded buffer alignment to max vector size of target if that size is greater than 32. We should consider weaker alignment for various items, such as string constants, but I didn't want to decrease alignment without discussion. | 28 September 2018, 19:53:34 UTC |
| d7d53c5 | Zalman Stern | 28 September 2018, 17:37:02 UTC | Merge branch 'master' into async | 28 September 2018, 17:37:02 UTC |
| cc64f51 | Zalman Stern | 28 September 2018, 17:36:31 UTC | Move failing tests into a new directory where they can be tracked with an issue so as to not block merging. | 28 September 2018, 17:36:31 UTC |
| f777721 | Ankit Aggarwal | 27 September 2018, 10:35:46 UTC | Ramp check missing for gathers | 27 September 2018, 10:35:46 UTC |
| 99745c0 | Ankit Aggarwal | 27 September 2018, 06:23:27 UTC | Undo Comment | 27 September 2018, 06:23:27 UTC |
| 977afee | Ankit Aggarwal | 26 September 2018, 16:11:44 UTC | Added qurt_hvx_vtcm in src/CMakelists. | 26 September 2018, 16:11:44 UTC |
| 7274704 | Dillon Sharlet | 25 September 2018, 19:38:00 UTC | Merge pull request #3282 from halide/disable-addusb Temporarily disable paddusb and paddusw due to #3281 | 25 September 2018, 19:38:00 UTC |
| 56bdbdb | Ankit Aggarwal | 24 September 2018, 11:53:00 UTC | Minor Fixes | 24 September 2018, 11:53:00 UTC |
| cfa0905 | Patricia Suriana | 23 September 2018, 22:16:31 UTC | Merge pull request #3280 from halide/perform_inline_with_order Use realization order when inlining to avoid extra works | 23 September 2018, 22:16:31 UTC |
| 8b73456 | Patricia Adriana Suriana | 22 September 2018, 03:09:39 UTC | Merge branch 'master' of https://github.com/halide/Halide into perform_inline_with_order | 22 September 2018, 03:09:39 UTC |
| e5d1fd2 | Dillon Sharlet | 22 September 2018, 00:00:57 UTC | Merge pull request #3234 from aankit-ca/hexagon_wrapper_init Initialize HexagonWrapper with v65 instead of v60. | 22 September 2018, 00:00:57 UTC |
| b3655f4 | Dillon Sharlet | 22 September 2018, 00:00:11 UTC | Merge pull request #3267 from halide/update-hexagon-sdk Update hexagon SDK to 3.3.3 | 22 September 2018, 00:00:11 UTC |
| 382f33a | dsharletg | 21 September 2018, 07:25:35 UTC | Temporarily disable paddusb and paddusw #3281. | 21 September 2018, 07:25:35 UTC |
| 3ce1f20 | Patricia Suriana | 21 September 2018, 01:23:51 UTC | Merge pull request #3279 from halide/compute_with_remove_is_right_level Remove is_the_right_level since compute_at aliasing at fused group is not currently supported | 21 September 2018, 01:23:51 UTC |
| 9c51d7d | Jing Pu | 20 September 2018, 23:59:11 UTC | Merge pull request #3229 from halide/solve_div_master Add a division visitor method to SolveExpression. | 20 September 2018, 23:59:11 UTC |
| c34fe9e | Patricia Adriana Suriana | 20 September 2018, 15:44:00 UTC | Merge branch 'master' of https://github.com/halide/Halide into compute_with_remove_is_right_level | 20 September 2018, 15:44:00 UTC |
| 97573bc | Patricia Adriana Suriana | 20 September 2018, 15:36:48 UTC | Use realization order when inlining to avoid extra works | 20 September 2018, 15:36:48 UTC |
| c646ef2 | Jing Pu | 20 September 2018, 00:39:00 UTC | fix comment. | 20 September 2018, 00:39:00 UTC |
| f53d7d4 | Zalman Stern | 20 September 2018, 00:15:40 UTC | Add a test to cover a couple of cases of storage folding interacting with async. Both are currently failing. | 20 September 2018, 00:15:40 UTC |
| 0e39b46 | Dillon Sharlet | 19 September 2018, 23:10:43 UTC | Merge pull request #3266 from halide/pdb_add_precompiled_sim_qurt Add binary libsim_qurt.a to hexagon_remote | 19 September 2018, 23:10:43 UTC |
| caf9f8d | Patricia Suriana | 19 September 2018, 21:42:39 UTC | Update comments for compute_with | 19 September 2018, 21:42:43 UTC |
| 96195f7 | Patricia Suriana | 19 September 2018, 21:35:37 UTC | Remove is_the_right_level since compute_at aliasing at fused group is not currently supported | 19 September 2018, 21:35:37 UTC |
| 8eb5c62 | Jing Pu | 19 September 2018, 18:48:40 UTC | Merge branch 'master' into solve_div_master | 19 September 2018, 18:48:40 UTC |
| 2de2f25 | Andrew Adams | 19 September 2018, 03:25:07 UTC | Merge pull request #3278 from halide/fix_warp_shuffles_2 Add new simplifier rules to handle warp shuffle breakage | 19 September 2018, 03:25:07 UTC |
| 31f6dce | Zalman Stern | 18 September 2018, 23:52:37 UTC | Merge branch 'master' into async | 18 September 2018, 23:52:37 UTC |
| 45ae466 | Andrew Adams | 18 September 2018, 18:58:13 UTC | Add new simplifier rules to handle warp shuffle breakage | 18 September 2018, 18:58:13 UTC |
| e852bb0 | Jing Pu | 15 September 2018, 02:41:27 UTC | fixed a overflow issue. | 15 September 2018, 02:41:27 UTC |
| 1aec3c8 | Jing Pu | 15 September 2018, 02:36:45 UTC | Merge branch 'master' into solve_div_master | 15 September 2018, 02:36:45 UTC |
| e48e6e2 | Andrew Adams | 14 September 2018, 23:08:50 UTC | Merge pull request #2845 from halide/metaprogrammed_simplifier_rules Metaprogrammed simplifier rules | 14 September 2018, 23:08:50 UTC |
| a5117ef | Andrew Adams | 14 September 2018, 16:28:15 UTC | Merge pull request #3264 from halide/use_host_cuda_capability_v3 Use host cuda capability v3 | 14 September 2018, 16:28:15 UTC |
| 44bddf8 | Andrew Adams | 14 September 2018, 16:15:15 UTC | Merge pull request #3265 from halide/fix_3061_v2 Fix 3061 v2 | 14 September 2018, 16:15:15 UTC |
| a9276a6 | Andrew Adams | 14 September 2018, 15:38:25 UTC | Merge branch 'metaprogrammed_simplifier_rules' of https://github.com/halide/Halide into metaprogrammed_simplifier_rules | 14 September 2018, 15:38:25 UTC |
| 7dafba1 | Andrew Adams | 14 September 2018, 15:38:17 UTC | Make failed proof probing conditional on debug level | 14 September 2018, 15:38:17 UTC |
| 9f8da5e | Zalman Stern | 14 September 2018, 07:35:29 UTC | Renable full runtime folding validity check for dynamic footprint case. The above could fail due to a race condition in reseting the boundaries of fold (head/tail) at the end of a loop. This was fixed by adding "_next" versions of the head and tail counters which allow setting the value to update the corresponding counter to for the next loop *after* its semaphore has beein acquired. Fix a bug where the dynamic footprint name could get an extra ".head" appended as that should only happen if the Func is async. | 14 September 2018, 07:35:29 UTC |
| c072dcc | Andrew Adams | 14 September 2018, 04:04:59 UTC | Switch to unordered_map in scope. It's slightly faster | 14 September 2018, 04:04:59 UTC |
| 2a39b9f | Andrew Adams | 13 September 2018, 17:20:36 UTC | Add some missing library loads | 13 September 2018, 17:20:36 UTC |
| 5c517cf | Andrew Adams | 13 September 2018, 17:09:20 UTC | Merge remote-tracking branch 'origin/master' into fix_3061_v2 | 13 September 2018, 17:09:20 UTC |
| 676d8bf | Steven Johnson | 12 September 2018, 18:42:44 UTC | Merge pull request #3270 from halide/srj-bug Add correct LLVM bug to simd_op_check comment | 12 September 2018, 18:42:44 UTC |
| c43fcea | Steven Johnson | 12 September 2018, 18:03:10 UTC | Add correct LLVM bug to simd_op_check comment | 12 September 2018, 18:03:10 UTC |
| a39b712 | Andrew Adams | 12 September 2018, 15:48:53 UTC | Merge remote-tracking branch 'origin/master' into fix_3061_v2 | 12 September 2018, 15:48:53 UTC |
| 7ba586a | Andrew Adams | 12 September 2018, 15:39:24 UTC | Fix generator_aot_acquire_release for cuda | 12 September 2018, 15:39:24 UTC |
| bf93ae1 | Steven Johnson | 11 September 2018, 20:10:28 UTC | Merge pull request #3268 from halide/srj-addus Re-enable addus/subus in simd_op_check | 11 September 2018, 20:10:28 UTC |
| 804a203 | Zalman Stern | 11 September 2018, 18:51:39 UTC | Ensure consistent semaphore behavior, i.e. not depending on semaphore value, if requested to acquire or release 0 count. This makes zero count requests into reliable nops. (Before the return value of the try_acquire call and whether wakeups happened or not depended on the semaphore value.) | 11 September 2018, 18:51:39 UTC |
| 04c4440 | Andrew Adams | 11 September 2018, 17:30:38 UTC | Wrap long line | 11 September 2018, 17:30:38 UTC |
| 47cdabc | Andrew Adams | 11 September 2018, 17:29:48 UTC | Add Halide namespace to test | 11 September 2018, 17:29:48 UTC |
| 8264d19 | Steven Johnson | 11 September 2018, 17:04:47 UTC | Re-enable addus/subus in simd_op_check Underlying bug in LLVM fixed in trunk. Note that pcmp is still disabled due to related bug which is not yet fixed. | 11 September 2018, 17:04:47 UTC |
| 91f19f9 | Ankit Aggarwal | 11 September 2018, 09:55:20 UTC | Minor change in CodeGen_Hexagon | 11 September 2018, 09:55:20 UTC |
| b4953ba | dsharletg | 11 September 2018, 00:57:39 UTC | Rebuild hexagon_sim_remote prebuilt to revert back to v60. | 11 September 2018, 00:57:39 UTC |
| 3579c61 | dsharletg | 11 September 2018, 00:55:38 UTC | Merge branch 'master' of https://github.com/halide/Halide into update-hexagon-sdk | 11 September 2018, 00:55:38 UTC |
| c24009b | dsharletg | 11 September 2018, 00:54:59 UTC | Update Makefile and fix a few breakages for Hexagon SDK 3.3.3. | 11 September 2018, 00:54:59 UTC |
| 723a1c3 | Pranav Bhandarkar | 10 September 2018, 22:06:47 UTC | A previous PR had moved libsim_qurt.a to from Halide/tools to Halide/src/runtime/hexagon_remote. However, we had forgotten to include a library binary in the PR. This made the PR inconsistent with our previous approach of leaving the binary in place in the repo. This PR fixes that. | 10 September 2018, 22:06:47 UTC |
| 74f1678 | Andrew Adams | 10 September 2018, 21:15:22 UTC | Merge remote-tracking branch 'origin/master' into fix_3061 | 10 September 2018, 21:15:22 UTC |
| bf444ae | Ankit Aggarwal | 10 September 2018, 09:13:59 UTC | Minor change in gather.cpp | 10 September 2018, 09:13:59 UTC |
| 1198ce2 | Ankit Aggarwal | 10 September 2018, 08:51:02 UTC | Added Padding and minor correction in test/correctness/gather.cpp | 10 September 2018, 08:51:02 UTC |
| b54a9ec | Ankit Aggarwal | 08 September 2018, 21:40:26 UTC | Remove sim_hap_vtcm.cpp. Was renamed to sim_qurt_vtcm.cpp | 08 September 2018, 21:40:26 UTC |
| 227bac4 | Ankit Aggarwal | 05 September 2018, 09:14:34 UTC | Modified/Added comments, moved malloc-free logic for VTCM to CodeGen_Hexagon Modified gather correctness test | 08 September 2018, 17:17:55 UTC |
| 9dd2c68 | Zalman Stern | 07 September 2018, 19:38:00 UTC | Merge branch 'async' of https://github.com/halide/Halide into async | 07 September 2018, 19:38:00 UTC |
| 5704eaa | Zalman Stern | 07 September 2018, 19:37:58 UTC | Merge branch 'master' into async | 07 September 2018, 19:37:58 UTC |
| 29e1173 | Zalman Stern | 07 September 2018, 19:36:32 UTC | Make thread pool propagate task errors downward to children tasks. Otherwise error case can deadlock. Add TODO to figure out 0 count semaphore calls situation. (Related to dynamic footprint issue which needs to be resolved before this can land.) | 07 September 2018, 19:36:32 UTC |
| 285da63 | Zalman Stern | 07 September 2018, 18:23:15 UTC | Remove extra blank line per review feedback. | 07 September 2018, 18:23:15 UTC |
| 66a9c86 | Zalman Stern | 07 September 2018, 18:22:34 UTC | Remove extra debug print. | 07 September 2018, 18:22:34 UTC |
| dabd88c | Andrew Adams | 07 September 2018, 16:59:08 UTC | Merge remote-tracking branch 'origin/master' into use_host_cuda_capability_v2 | 07 September 2018, 16:59:08 UTC |
| b4ea2b0 | Andrew Adams | 07 September 2018, 16:57:48 UTC | Merge branch 'master' into metaprogrammed_simplifier_rules | 07 September 2018, 16:57:48 UTC |
| 65f22a4 | Ankit Aggarwal | 20 August 2018, 20:00:33 UTC | Cleaned gather correctness test | 07 September 2018, 08:34:15 UTC |
| f53795a | Ankit Aggarwal | 20 August 2018, 19:57:06 UTC | Newline at file end in mini_qurt_vtcm.h | 07 September 2018, 08:34:15 UTC |
| e18da49 | Ankit Aggarwal | 20 August 2018, 09:40:59 UTC | Adding asserts, changing to Scatter-Gather pass to IRMutator2, additional comments, minor changes | 07 September 2018, 08:34:15 UTC |
| 75c0315 | Ankit Aggarwal | 13 August 2018, 17:56:08 UTC | Hexagon: Added support for generating vgather instruction on versions>=v65 | 07 September 2018, 08:34:15 UTC |
| 09dc5d5 | Steven Johnson | 07 September 2018, 01:45:09 UTC | Merge pull request #3260 from halide/srj-fill Buffer::fill() and similar should return ref-to-self | 07 September 2018, 01:45:09 UTC |
| ffa7e0e | Steven Johnson | 06 September 2018, 23:17:45 UTC | Merge pull request #3261 from halide/srj-asconst Add Buffer::as_const() | 06 September 2018, 23:17:45 UTC |
| a98714f | Steven Johnson | 06 September 2018, 23:17:26 UTC | Merge pull request #3258 from halide/srj-alias Add Buffer<>::alias() | 06 September 2018, 23:17:26 UTC |
| 37581d3 | Pranav Bhandarkar | 06 September 2018, 20:18:34 UTC | Merge pull request #3259 from aankit-ca/sim_qurt_cleanup Changed the directory structure for tools/sim_qurt | 06 September 2018, 20:18:34 UTC |
| 6b2d2de | Ankit Aggarwal | 06 September 2018, 19:18:59 UTC | Moved libsim_qurt.a from bin/v60/sim_qurt to bin/v60 | 06 September 2018, 19:18:59 UTC |
| 0bf8b5c | Ankit Aggarwal | 06 September 2018, 18:44:48 UTC | Minor change | 06 September 2018, 18:44:48 UTC |
| 7622ba7 | Steven Johnson | 06 September 2018, 18:38:57 UTC | Address review comments | 06 September 2018, 18:38:57 UTC |
| 6e298e9 | Steven Johnson | 06 September 2018, 18:20:06 UTC | Add Buffer::as_const() This is just syntactic sugar for `as<const T>()`, for the case of having a Buffer<T>& (or Buffer<T>*) and needing the const variation; it avoid needing to recapitulate the type. (Looks likely to be used commonly enough in my use case to be worthwhile, but if there is pushback as too-much-sugar, LMK.) Also, drive-by addition of HALIDE_ALWAYS_INLINE. | 06 September 2018, 18:20:06 UTC |
| b83ead4 | Steven Johnson | 05 September 2018, 23:28:29 UTC | Add proper return-type forwarding for Halide::Buffer as well | 05 September 2018, 23:28:29 UTC |
| 5d05655 | Steven Johnson | 05 September 2018, 23:05:57 UTC | Merge pull request #3257 from halide/srj-copyfrom Buffer<const T>::copy_from() should be illegal | 05 September 2018, 23:05:57 UTC |
