Commit Graph

Select branches

Hide Pull Requests

geohot/updates

gh-pages

master

remove_anchors

#1

#2

#3

v0.10.0

v0.10.1

v0.10.2

v0.10.3

v0.11.0

v0.12.0

v0.13.0

v0.4.0

v0.5.0

v0.6.0

v0.7.0

v0.8.0

v0.9.0

v0.9.1

v0.9.2

Mono Color

• 1f9d2442b9 Add Tensor.scatter_reduce (#8947) 3 people 2025-02-13 09:08:54 -05:00
• 2b9ce1235a simple failing case for reorder expand + keep views in tensor_map [pr] (#9057) qazalandGitHub 2025-02-13 12:22:55 +02:00
• 765a936b81 getenv(CC) for clang (#9054) George HotzandGitHub 2025-02-13 15:30:01 +08:00
• 33a1151f2f Revert "match torch rmsnorm implementation (#6799)" (#9052) George HotzandGitHub 2025-02-13 14:42:45 +08:00
• a66b8250e0 match torch rmsnorm implementation (#6799) 3 people 2025-02-13 15:02:51 +10:00
• 19ae829bd1 test float uop in sym_infer (#7456) ggandGitHub 2025-02-12 23:55:28 -05:00
• 095504d094 mulacc_unrolled should happen even with no DEVECTORIZE (#9029) 3 people 2025-02-13 05:54:15 +01:00
• 74742c018f hotfix: setup_mock_nv_osx geohot 2025-02-13 12:25:15 +08:00
• d2ff55e9c6 OSX GPUOcelot (#8209) 4 people 2025-02-13 05:24:29 +01:00
• f4f56d7c15 move time_linearizer to extra.optimization.helpers [pr] (#9048) chenyuandGitHub 2025-02-12 15:49:58 -05:00
• c15486cf39 remove contiguous in test_subbuffer_used [pr] (#9046) chenyuandGitHub 2025-02-12 14:41:16 -05:00
• b3eab03055 Three things to get Windows CI working correctly: (#9047) rmtewandGitHub 2025-02-13 08:41:00 +13:00
• f53b819648 UOps. -> Ops. [pr] (#9044) chenyuandGitHub 2025-02-12 12:53:23 -05:00
• 6811688d29 disallow VIEW(BUFFER) in tensor [pr] (#9041) qazalandGitHub 2025-02-12 18:27:35 +02:00
• 7b5ac2c15e free_intermediates in bert (#9040) chenyuandGitHub 2025-02-12 10:00:39 -05:00
• 916d5e7f08 WebGPU f16 support (f16 bounty part 2) (#8653) Ahmed HarmoucheandGitHub 2025-02-12 12:46:53 +01:00
• aaed315fee add AMX support to LLVM (#8957) 3 people 2025-02-12 05:01:18 -03:00
• 0c97c10814 TestOps: silence pytorch std()/var() degrees of freedom warnings (#9034) Josh MooreandGitHub 2025-02-12 01:49:18 -05:00
• d581afd873 skipdata capstone (#9026) Ignacio SicaandGitHub 2025-02-11 21:11:14 -03:00
• 2845f8797a failed test cases for rsqrt at 0 and similar ones (#9035) chenyuandGitHub 2025-02-11 17:50:16 -05:00
• 101652a55c hcq: thread fence (#8991) nimlgenandGitHub 2025-02-11 18:09:37 +03:00
• 45aae8a6bc hotfix: add External Benchmark Schedule to CI geohot 2025-02-11 22:06:17 +08:00
• 17fa6e7619 disk: better error desc when not opened (#9028) nimlgenandGitHub 2025-02-11 16:31:04 +03:00
• 166670a2f2 nv: fill grid/block sizes (#9025) nimlgenandGitHub 2025-02-11 16:30:30 +03:00
• c80603285e bring back some things from the fix_kernel_ops diff [pr] (#9027) qazalandGitHub 2025-02-11 14:20:31 +01:00
• 9209b85c91 add UOps.CAT (#9022) George HotzandGitHub 2025-02-11 19:50:37 +08:00
• a521260b7a dont reduce the ptr size, sz is base for unaligned [pr] (#9023) George HotzandGitHub 2025-02-11 19:50:23 +08:00
• d0d58a6771 add CUSTOM support to cstyle (#9020) George HotzandGitHub 2025-02-11 18:02:58 +08:00
• fb698920f1 revert scheduler change (#9019) George HotzandGitHub 2025-02-11 11:34:12 +08:00
• 16e9e4db37 make llvm opt the default (#9017) George HotzandGitHub 2025-02-11 10:08:45 +08:00
• bec4f59ce8 workaround f16 cast ambiguity (#8935) divinity76andGitHub 2025-02-11 02:38:56 +01:00
• b741a9aae7 update doc of Tensor.tolist (#9016) chenyuandGitHub 2025-02-10 16:51:23 -05:00
• 04e64765c4 Minor typo in ReadMe (#9015) JoelandGitHub 2025-02-10 21:30:20 +01:00
• 6c39aa4a6b adjust cuda ci test targets (#9014) chenyuandGitHub 2025-02-10 15:29:59 -05:00
• dfc9d6827f am_smi: print power state (#9013) nimlgenandGitHub 2025-02-10 23:07:39 +03:00
• bf0bcb2d5a cleanup ast rewriter [pr] (#9012) qazalandGitHub 2025-02-10 19:07:59 +01:00
• 586e48d696 a few more backward tests now pass (#9010) chenyuandGitHub 2025-02-10 12:46:21 -05:00
• f9898f7554 update gpuocelot commit (#9011) chenyuandGitHub 2025-02-10 12:18:44 -05:00
• c52cd2b437 kernel op cleanups + use ScheduleItem [pr] (#9009) qazalandGitHub 2025-02-10 17:54:30 +01:00
• 25fa5e4d5f enable backward tests in test_std_one_in_axis [pr] (#9007) 3 people 2025-02-10 10:44:05 -05:00
• d426f1ad6e don't open devices in lowering (#9008) qazalandGitHub 2025-02-10 15:28:51 +01:00
• cfd3db7862 construct the schedule sink 2 (#8925) qazalandGitHub 2025-02-10 15:23:02 +01:00
• 3e005ca0c2 am: resize bar0 to max supported (#9006) nimlgenandGitHub 2025-02-10 16:48:44 +03:00
• 07cb7e701c am: fix gfx usage at 100% (#9003) nimlgenandGitHub 2025-02-10 16:48:23 +03:00
• f91409f038 am: fix proclogs (#9004) nimlgenandGitHub 2025-02-10 16:38:58 +03:00
• cd77e51810 fix tensor realization bug in #8975 (#8984) qazalandGitHub 2025-02-10 13:51:30 +01:00
• b17ec42b56 remove const_arg (#9002) qazalandGitHub 2025-02-10 12:45:11 +01:00
• 0568720a68 delete revectorize (#9000) George HotzandGitHub 2025-02-10 18:32:35 +08:00
• fd9f9ec772 realized base tensors become RESHAPE(BUFFER) [pr] (#8994) qazalandGitHub 2025-02-10 10:17:54 +01:00
• 910ae260cd dsp float4 fold + revectorize [pr] (#8995) George HotzandGitHub 2025-02-10 12:14:32 +08:00
• e618efce22 COMMUTATIVE flipping is only for ints (#8996) George HotzandGitHub 2025-02-10 12:01:28 +08:00
• 2983285315 use HEX_REG_QEMU_INSN_CNT from qemu as a DSP timer [pr] (#8993) George HotzandGitHub 2025-02-10 11:07:35 +08:00
• 9119716761 update Tensor.maximum (#8992) chenyuandGitHub 2025-02-09 21:26:27 -05:00
• 88add71c25 amd: increase sdma copy size (#8989) nimlgenandGitHub 2025-02-09 20:53:35 +03:00
• 7eba5fb413 Tensor.empty is RESHAPE(BUFFER) (#8987) qazalandGitHub 2025-02-09 18:42:51 +01:00
• 44479f8ad6 raise ValueError in view reshape for negative dims [pr] (#8988) qazalandGitHub 2025-02-09 17:27:15 +01:00
• c6c2373bc0 replace libpciaccess autogen with just pci regs (#8983) nimlgenandGitHub 2025-02-09 18:40:45 +03:00
• 55351ebb31 minimal failing test for #8975 [pr] (#8982) qazalandGitHub 2025-02-09 14:10:37 +01:00
• e5a3f60fc2 am: remove libpciaccess dep (#8980) nimlgenandGitHub 2025-02-09 16:06:55 +03:00
• 52a69dd5e9 Revert "use am in training benchmarks (#8965)" (#8981) nimlgenandGitHub 2025-02-09 15:43:45 +03:00
• 0b26cee2f1 fix some slow tests [pr] (#8979) George HotzandGitHub 2025-02-09 15:57:04 +08:00
• 208097d488 try reducing testing deps [pr] (#8976) George HotzandGitHub 2025-02-09 15:22:32 +08:00
• 6ffee2fca9 reduce speed example [pr] (#8978) George HotzandGitHub 2025-02-09 14:13:59 +08:00
• ac3765c043 use getpass instead of os.getlogin() (#8972) Samuel AyalaandGitHub 2025-02-08 15:29:26 -05:00
• 308516e439 fix viz paginate + cleanups [pr] (#8973) qazalandGitHub 2025-02-08 20:26:57 +01:00
• 107e616857 use am in training benchmarks (#8965) nimlgenandGitHub 2025-02-08 20:20:47 +03:00
• 79de980565 am: do not fork pci bars (#8969) nimlgenandGitHub 2025-02-08 19:03:17 +03:00
• 0cac941af1 move xpow to sym instead of late_rewrite (#8968) chenyuandGitHub 2025-02-08 10:09:24 -05:00
• e7182bbb2c fix "fatal bad object" log in process replay [pr] (#8966) qazalandGitHub 2025-02-08 11:57:38 +01:00
• 9b9c1e14da Late MTLCompiler load (#8963) uuuvnandGitHub 2025-02-08 14:29:23 +05:00
• a3c78d47b3 speed docs + upgrades [pr] (#8964) George HotzandGitHub 2025-02-08 17:28:52 +08:00
• 5bdd6a1cc4 increase CI speed with more runners [pr] (#8961) George HotzandGitHub 2025-02-08 09:04:36 +08:00
• 11d50324d8 am: tiny cleanups (#8958) nimlgenandGitHub 2025-02-07 23:44:43 +03:00
• cfd28517df move pow folding tests to test_schedule [pr] (#8955) chenyuandGitHub 2025-02-07 12:51:43 -05:00
• c2b4c43edb handle stride 0 reduce (#8068) 3 people 2025-02-07 22:40:58 +08:00
• cf21e27d78 little better VIEW simplifier pattern [pr] (#8954) qazalandGitHub 2025-02-07 12:55:54 +01:00
• 329013f577 fix UOp.metadata on KERNEL op [pr] (#8953) qazalandGitHub 2025-02-07 12:40:11 +01:00
• 4de084a835 cleanup ci, split docs/autogen, testing_minimal, LLVM Speed [pr] (#8952) George HotzandGitHub 2025-02-07 19:01:59 +08:00
• 6090cbe3be Try to open llvm first when opening metal (#8949) uuuvnandGitHub 2025-02-07 15:58:37 +05:00
• 67b70e4f6c Fix incorrect __del__ (#8950) uuuvnandGitHub 2025-02-07 15:21:44 +05:00
• 9ed2d0dfa2 refactor into subactions (#8946) George HotzandGitHub 2025-02-07 18:06:44 +08:00
• 133cacadde Autogen webgpu dawn, removing wgpu-py dependency (f16 support part 1) (#8646) 3 people 2025-02-07 08:16:59 +01:00
• dbda72f91d hotfix: raise line limit to 11200 for new webgpu backend geohot 2025-02-07 14:29:20 +08:00
• b1e1319972 ci speed on the enterprise plan [pr] (#8942) George HotzandGitHub 2025-02-07 11:18:12 +08:00
• 3b67712892 [bounty] Fix LLVM=1 NO_DEVECTORIZE=1 python3 test/test_ops.py TestOps.test_strided_conv2d_simple (#8937) 3 people 2025-02-06 18:07:54 -08:00
• f54242849d failing test for the devectorize [pr] (#8940) George HotzandGitHub 2025-02-07 09:44:54 +08:00
• ee1a0fb8ec am_smi: print device name (#8939) nimlgenandGitHub 2025-02-07 03:01:25 +03:00
• a092b6395d Tuple -> tuple, List -> list [pr] (#8936) chenyuandGitHub 2025-02-06 14:21:19 -05:00
• d5183e1584 remove unneeded annotation import (#8934) chenyuandGitHub 2025-02-06 13:12:35 -05:00
• 00d72a5144 setitem isinstance cleanup [pr] (#8932) chenyuandGitHub 2025-02-06 11:44:57 -05:00
• 81e241150a hotfix: save 1 line (#8931) qazalandGitHub 2025-02-06 16:26:05 +01:00
• eb1144be8b hotfix: only check current graph when excluding nodes in viz (#8930) qazalandGitHub 2025-02-06 15:58:53 +01:00
• 3cc05081f4 llvm no devectorize, the right way (#8901) George HotzandGitHub 2025-02-06 22:53:49 +08:00
• 8b16c65bca add compile3 benchmark [pr] (#8929) George HotzandGitHub 2025-02-06 22:49:31 +08:00
• 79fb5c6470 hotfix: test_shard_no_recompile shouldn't rely on schedule order [pr] (#8928) qazalandGitHub 2025-02-06 15:27:59 +01:00
• 1249e8dd3b objc fast msg, try 2 [pr] (#8927) George HotzandGitHub 2025-02-06 19:06:21 +08:00
• 86feb98dcd am: add support for 7600 (#8910) nimlgenandGitHub 2025-02-06 14:04:07 +03:00
• ae45826758 hotfix: GRAPH_ONE_KERNEL + fix timing geohot 2025-02-06 17:52:20 +08:00
• 1c53e8bf27 Revert "objc fast msg (#8922)" (#8926) George HotzandGitHub 2025-02-06 17:50:49 +08:00
• c3f99a727e objc fast msg (#8922) George HotzandGitHub 2025-02-06 17:49:06 +08:00