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

• 44a67bf783 constant folding (#3675) George HotzandGitHub 2024-03-10 14:47:24 -07:00
• 25aede6fd9 truncate for exec_alu (#3674) George HotzandGitHub 2024-03-10 14:19:04 -07:00
• 957ae9b594 Fix Tensor's __repr__ for printing out grad (#3673) Francis LataandGitHub 2024-03-10 17:04:29 -04:00
• 0f16729023 RDNA3: restore launch bounds (#3672) George HotzandGitHub 2024-03-10 10:27:52 -07:00
• d7452c2a20 clean up llvmir builder (#3671) chenyuandGitHub 2024-03-09 21:19:36 -05:00
• 1143c62519 tensor.py touchups (#3667) George HotzandGitHub 2024-03-09 16:12:20 -08:00
• 69ca7f7bf9 changes for teenygrad (#3665) George HotzandGitHub 2024-03-09 15:30:34 -08:00
• 89b8b5d549 Fix missing import. (#3666) Quentin WachandGitHub 2024-03-09 23:55:23 +01:00
• 8ae85b2cf5 add inference_mode context manager with decorator support (#3621) Maximilian WolfandGitHub 2024-03-09 17:38:26 +01:00
• b5cbf1792a Fix Tensor.cumsum when axis of length 0 is selected (#3473) Obada KhaliliandGitHub 2024-03-09 18:26:41 +02:00
• 915f98791c use custom KernelOptError in kernel opt (#3661) chenyuandGitHub 2024-03-08 15:36:16 -05:00
• ac02e7347d ptx timing vs cuda timing (#3659) George HotzandGitHub 2024-03-08 10:17:49 -08:00
• daa4034e80 No more metal flakiness (#3643) uuuvnandGitHub 2024-03-08 18:54:44 +02:00
• e25879d50e don't get new var_val for the same ast in fuzz_linearizer (#3657) chenyuandGitHub 2024-03-08 09:49:24 -05:00
• 1130c73844 add FUZZ_NTH to fuzz_linearizer (#3656) chenyuandGitHub 2024-03-08 09:16:49 -05:00
• 9f66dcf718 PolynomialDecayWithWarmup + tests (#3649) David HouandGitHub 2024-03-07 15:53:36 -08:00
• 57df8e8d82 update fuzz_linearizer (#3648) chenyuandGitHub 2024-03-07 18:41:22 -05:00
• b282a45e39 fix direct store float4 with same vin (#3652) chenyuandGitHub 2024-03-07 18:11:50 -05:00
• a66ffec6d3 update kernel dataset to exclude the disktensor ones (#3651) chenyuandGitHub 2024-03-07 17:35:19 -05:00
• fcf4a5ccf2 fix example that calls Tensor.__bool__ (#3650) chenyuandGitHub 2024-03-07 16:59:26 -05:00
• 6e50582e62 working to improve ptx (#3647) George HotzandGitHub 2024-03-07 12:39:31 -08:00
• 1853ec9a02 add tests for bfloat16 on HIP (#3638) ZafferandGitHub 2024-03-07 13:45:36 -05:00
• 0cef284aac fix typing FlopCounter.flops can be sint (#3646) chenyuandGitHub 2024-03-07 12:49:17 -05:00
• 906cc3a69b cleanup tests Device[Device.DEFAULT] is always Compiled (#3645) chenyuandGitHub 2024-03-07 11:15:42 -05:00
• bdd62c7fd8 make the bf16 include dynamic (#3642) qazalandGitHub 2024-03-07 17:31:35 +02:00
• 4552248c84 fix Tensor.to preserves grad.data (#3636) chenyuandGitHub 2024-03-06 21:44:49 -05:00
• d33311ebe0 remove parens of ALU if it has associative property (#3635) chenyuandGitHub 2024-03-06 21:12:11 -05:00
• fe6b6e38c1 remove parentheses of GEP if it's from SSA (#3634) chenyuandGitHub 2024-03-06 20:22:46 -05:00
• 0afaf70d57 lars optimizer + tests (#3631) 3 people 2024-03-06 15:11:01 -08:00
• b2e92d44fa skip METAL sin test in test_dtype_alu (#3633) chenyuandGitHub 2024-03-06 17:29:19 -05:00
• 8f10bfa2ff ban __bool__ on Tensor (#3632) chenyuandGitHub 2024-03-06 17:12:35 -05:00
• 81baf3eed3 bring ptx back (#3623) George HotzandGitHub 2024-03-06 13:34:21 -08:00
• c270d54c32 update test_dtype_alu for METAL (#3629) chenyuandGitHub 2024-03-06 14:55:19 -05:00
• abc5f3a6a0 hip bf16 hotfix (#3630) qazalandGitHub 2024-03-06 21:42:30 +02:00
• bc2a13a5f7 test case to show clang and python doing math in double (#3628) chenyuandGitHub 2024-03-06 13:49:03 -05:00
• 568353fa84 hotfix: bump line count to 6500 geohot 2024-03-06 07:52:18 -08:00
• a1507c7fd4 Fix Tensor.dropout() with multigpu (#3619) Elias WahlandGitHub 2024-03-06 00:26:21 +01:00
• e5ee6bb2bd fix outdated url in showcase doc (#3624) Jungwan WooandGitHub 2024-03-06 07:44:40 +09:00
• 8500265561 this mem fault still happening (#3620) George HotzandGitHub 2024-03-05 10:39:32 -08:00
• 3c3f846c45 tinybox benchmark with HSA (#3603) chenyuandGitHub 2024-03-05 11:03:52 -05:00
• f500be1313 out of bounds access caused by launch bounds (#3615) George HotzandGitHub 2024-03-05 06:34:00 -08:00
• eb83e2d3a0 decouple buffer mutability from cstyle (#3617) qazalandGitHub 2024-03-05 16:20:59 +02:00
• 3275260c98 Revert "test: add failing bfloat16 test case for metal backend (#3481)" (#3618) chenyuandGitHub 2024-03-05 09:08:42 -05:00
• 1e12a2ae80 test: add failing bfloat16 test case for metal backend (#3481) SkoshandGitHub 2024-03-05 15:44:54 +02:00
• 957e9800f1 llama + beam to mac benchmark, full cifar to nvidia benchmark (#3612) chenyuandGitHub 2024-03-04 21:35:57 -05:00
• 282bbd5acb check the input length into argfix (#3610) chenyuandGitHub 2024-03-04 19:50:17 -05:00
• 7db6dd725d multilazybuffer fix (#3609) Elias WahlandGitHub 2024-03-04 23:36:23 +01:00
• c3b8d285aa cleanup uops (#3605) chenyuandGitHub 2024-03-04 11:03:14 -05:00
• 94679322a3 simpler float4 direct store and locals support (#3592) qazalandGitHub 2024-03-04 16:28:28 +02:00
• 3db826e195 hsa in lin opts (#3602) nimlgenandGitHub 2024-03-04 17:17:32 +03:00
• 7c90005c65 search: hotfix to make sure TC behavior is all in applied_opts (#3598) Francis LamandGitHub 2024-03-03 18:44:38 -08:00
• 8e5d60a322 add more gpt2 variant in mac/nvidia benchmark (#3599) chenyuandGitHub 2024-03-03 17:55:30 -05:00
• 968d109453 apply more create_lt_node (#3597) chenyuandGitHub 2024-03-03 16:12:39 -05:00
• bc562c4747 Python div alu behavior differs slightly from others (#3596) 3 people 2024-03-03 13:48:25 -05:00
• 56d21d77b3 Fix two bugs concerning Tensor.to. (#3593) Marcin SłowikandGitHub 2024-03-03 17:48:56 +01:00
• 0082300a59 Fix symbolic negative floordiv (#3594) 3 people 2024-03-03 11:40:52 -05:00
• e09619ab6c explicitly create_lt_node when used in shapetracker _expr_view (#3561) chenyuandGitHub 2024-03-03 10:08:21 -05:00
• 640dc0fc51 hsa flush hdp (#3591) nimlgenandGitHub 2024-03-03 15:55:07 +03:00
• 660df3cff1 Add test for .softmax.argmax (#3559) 4 people 2024-03-02 23:51:52 -05:00
• ee41fafdab use operator instead of lambda in python_alu (#3590) chenyuandGitHub 2024-03-02 19:33:21 -05:00
• a89afd4ffa Directly store float4 nodes (#3564) 3 people 2024-03-03 01:58:20 +02:00
• 770707b376 hotfix: gpuocelot no rebuild geohot 2024-03-02 15:57:38 -08:00
• 74c9acddb0 simple python ALU (#3589) George HotzandGitHub 2024-03-02 15:50:58 -08:00
• 162dfb07d9 fuzz_linearizer: fix uops and add to test.yml (#3588) Francis LamandGitHub 2024-03-02 15:03:42 -08:00
• 8978488565 add sanity tests for bufs_from_lin (#3586) Jovan SardinhaandGitHub 2024-03-02 14:17:43 -08:00
• aa9b013d79 add constant folding for WHERE in uops (#3584) George HotzandGitHub 2024-03-02 10:37:14 -08:00
• 3b7e3fa2e4 fix sync in hsa graph (#3582) nimlgenandGitHub 2024-03-02 18:37:51 +03:00
• 6c36264790 Improve type hints for optimizer (#3583) Szymon OżógandGitHub 2024-03-02 16:35:44 +01:00
• 83530a585f add quick external data select test geohot 2024-03-02 05:38:32 -08:00
• 9a37273d36 consts don't have nodes in the graph (#3579) George HotzandGitHub 2024-03-02 04:19:11 -08:00
• 41f0a25b53 lazy.py: cache consts (#3577) George HotzandGitHub 2024-03-02 03:50:05 -08:00
• fb8acd1851 Don't touch UOps.DEFINE_GLOBAL (#3575) uuuvnandGitHub 2024-03-02 13:30:05 +02:00
• 50e1445e60 Revert "allow overriding weight init for Linear (#3569)" (#3576) George HotzandGitHub 2024-03-02 03:17:13 -08:00
• 2d0973a852 allow overriding weight init for Linear (#3569) David HouandGitHub 2024-03-02 03:16:04 -08:00
• 9642a8f547 search: add BEAM UPCAST/LOCAL params and loosen TC criteria during BEAM (#3563) Francis LamandGitHub 2024-03-02 03:11:25 -08:00
• ba6c041eab fix SCE ignore_index with label_smoothing (#3574) David HouandGitHub 2024-03-01 19:19:45 -08:00
• e17f1821a7 wmma: add CUDA tensor core and fix test_speed_v_torch failure (#3544) Francis LamandGitHub 2024-03-01 17:51:02 -08:00
• b3cdc11a58 label_smoothing in sparse_cat_crossentropy (#3568) David HouandGitHub 2024-03-01 17:02:46 -08:00
• 6b29c70b3d Refactor to UOpGraph class (#3566) George HotzandGitHub 2024-03-01 15:14:48 -08:00
• b7e555f6c0 run test_linearizer_failures on PYTHON backend (#3565) chenyuandGitHub 2024-03-01 17:00:18 -05:00
• 48d22067ca clean up test_linearizer_failures (#3562) chenyuandGitHub 2024-03-01 15:57:17 -05:00
• d16aa89561 don't allow MLB assigns with different axes (#3557) David HouandGitHub 2024-03-01 04:59:06 -08:00
• cfd23f398d Revert "don't allow MLB assigns with different axes (#3483)" (#3554) chenyuandGitHub 2024-02-29 23:13:07 -05:00
• f19d8bb7b4 don't allow MLB assigns with different axes (#3483) David HouandGitHub 2024-02-29 20:04:12 -08:00
• 35d998efa8 disable flaky test_conv_beam in CI (#3553) chenyuandGitHub 2024-02-29 22:59:41 -05:00
• e5385eecfc UnsyncedBatchNorm with synced trainable weights for hlb cifar (#3472) 3 people 2024-02-29 19:52:07 -08:00
• 5a6e151844 no barrier side effect (#3550) George HotzandGitHub 2024-02-29 18:10:04 -08:00
• bd9c2ced07 define var can be removed from vars to keep (#3549) George HotzandGitHub 2024-02-29 17:44:19 -08:00
• 2c19ab6561 define var (#3548) George HotzandGitHub 2024-02-29 16:43:27 -08:00
• 83cdc85790 add index to DEFINE_GLOBAL (#3542) George HotzandGitHub 2024-02-29 15:22:26 -08:00
• 978a997d1f print nvidia-smi in CI benchmark (#3546) chenyuandGitHub 2024-02-29 17:31:37 -05:00
• 5d434801fa search: add tensor core to beam search space (#3275) Francis LamandGitHub 2024-02-29 13:05:10 -08:00
• f90caa4b92 Escape table name in diskcache queries. (#3543) Marcin SłowikandGitHub 2024-02-29 22:04:21 +01:00
• 0afde98ba5 scan all gpu agents at launch (#3535) nimlgenandGitHub 2024-02-29 20:37:37 +03:00
• 2e82c5b7a4 README: ops_cpu and ops_torch have been removed (#3539) Mark McLoughlinandGitHub 2024-02-29 15:22:11 +00:00
• b05776ef3e fix addresses of dispatch packets (#3534) nimlgenandGitHub 2024-02-29 16:43:55 +03:00
• 9268a8b154 remove MULACC (#3459) geohotstanandGitHub 2024-02-29 20:40:40 +08:00
• 94fc0fd546 uop the float4 acc upcast in group_for_reduce kernels (#3466) qazalandGitHub 2024-02-29 03:33:47 +02:00
• 48918fa75a fix disktensor offset issue (#3532) George HotzandGitHub 2024-02-28 17:22:17 -08:00
• 0b1fc5888a fix 'Import Error: cannot import name compile_cuda from tinygrad.runtime.ops_cuda' error in extra/gemm/cuda_matmul.py (#3531) Caleb BunchandGitHub 2024-02-28 20:15:32 -05:00