sirhcm/tinygrad
1
0
Fork 0
forked from tinygrad/tinygrad
Code Pull Requests 2 Actions Activity
11,971 Commits 336 Branches 15 Tags
pyremu_try
Commit Graph
11971 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
George HotzandGitHub ef3c614b4d Merge branch 'master' into pyremu_try 2026-01-31 11:26:41 +08:00
George HotzandGitHub c9a3ddb341 benchmark llama walltime script (#14454) 
* benchmark llama walltime script

* adj layers
 2026-01-31 10:21:54 +08:00
George HotzandGitHub f5346d6a1a fix USE_ATOMICS for non float dtypes and make it the default (#14444) 
* embedded multistep test

* complex test

* with jit

* fix dtypes and reenable USE_ATOMICS

* that test didn't catch anything
 2026-01-31 09:44:16 +08:00
sirhcmandGitHub e575dd8275 prevent UB in long decomp and more emulated tests (#14447) 2026-01-30 19:38:41 -05:00
chenyuandGitHub 3204f94454 correct var_vals schedule filter (#14451) 
complete_create_schedule_with_vars returns var_vals that's used in schedule
 2026-01-30 17:10:07 -05:00
chenyuandGitHub cfcd1debb5 test schedule with multiple AFTER (#14449) 2026-01-30 15:59:00 -05:00
nimlgenandGitHub 486d53d646 device: call free for external_ptr (#14448) 
* device: call free for external_ptr

* lin
 2026-01-30 23:53:17 +03:00
nimlgenandGitHub e0978498dc amd: read_ptr/write_ptr/doorbells are not lists (#14445) 2026-01-30 23:11:57 +03:00
sirhcmandGitHub 1803ee939d EMULATED_DTYPES=long works with CPU_LLVM (#14446) 2026-01-30 13:54:43 -05:00
chenyuandGitHub 03613e83ad update TestTensorMetadata (#14443) 
run with SCACHE=0 some more TODOs
 2026-01-30 12:39:01 -05:00
geohot cbb1eed57b hotfix: partial revert of 9eb449f88, caused llama NaN 2026-01-30 17:19:27 +00:00
chenyuandGitHub 26f5c00265 move TestTensorMetadata to unit (#14442) 2026-01-30 12:14:21 -05:00
chenyuandGitHub c05a0b85ae flip unique const src order [pr] (#14441) 
* flip unique const src order [pr]

matches buffer, simplifies replace_input_buffer

* combine rules
 2026-01-30 11:44:18 -05:00
geohot ee2c78709d mlperf/llama: disable USE_ATOMICS for now 2026-01-31 00:42:08 +08:00
chenyuandGitHub beecac4d85 expand ranges -> unroll outer ranges [pr] (#14440) 2026-01-30 11:26:05 -05:00
chenyuandGitHub 9eb449f882 clean up toposort sched_sink [pr] (#14439) 2026-01-30 10:18:28 -05:00
George HotzandGitHub 838cd078bc use atomics for embedding backward (#14400) 
* embedding is slow

* failing

* float is fine

* null

* it fails

* simplify embedding with broadcasting

* ATOMIC_ADD incoming

* min change

* simpler test

* better test

* fix test

* real test

* simpler

* cleanups

* types and names

* _zero_kernel

* grad multi

* hack

* none

* multi unshard

* more for call

* don't tag in call

* good

* call_multi

* call_multi wow claude is useless

* embedding backward mutli test

* test passes

* fix as_param

* shape_to_shape_arg

* add clip

* before cast

* fix spec=2, use atomics
 2026-01-30 18:10:59 +08:00
nimlgenandGitHub 1998e0bb28 nv: add prof props to dev (#14437) 2026-01-30 12:51:43 +03:00
George HotzandGitHub 7a9dee4e50 add call/param UOps (#14433) 
* add call/param UOps

* resolve call

* skip that for now

* grad on call

* fix tests
 2026-01-30 14:51:45 +08:00
qazalandGitHub 66d6a68016 viz: sqtt work from cdna gemm (#14434) 
* it's the tag

* initialize rows based on the disasm

* test_cfg with Ops.BINARY

* pyremu wants s_code_end?

* test_diamond

* diff cleanup
 2026-01-30 14:00:56 +09:00
sirhcmandGitHub 88caf57ef4 ci: unify python versions (#14430) 2026-01-29 21:42:03 -05:00
chenyuandGitHub 86a204d22a allow Tensor setitem input to be list/tuple (#14432) 
matches assign, and generally matches numpy
 2026-01-29 21:26:58 -05:00
chenyuandGitHub 4a80319093 clean up split_store final logic [pr] (#14429) 
explicitly check the structure
 2026-01-29 18:40:07 -05:00
sirhcmandGitHub e47f12f671 ci: replace testing_minimal with testing_unit (#14427) 2026-01-29 18:02:43 -05:00
wozeparrotandGitHub c2fb8b208f fa: 32 block size (#14416) 2026-01-29 13:59:13 -08:00
chenyuandGitHub a979fafae5 cleanup around disk buffer [pr] (#14428) 
style change, prep for refactor
 2026-01-29 16:18:44 -05:00
nimlgenandGitHub dc977a03b0 nv_pma: bw decoder (#14424) 
* nv_pma: bw decoder

* decoder fix

* better
 2026-01-30 00:12:39 +03:00
chenyuandGitHub ddc041854b failed test case for disk setitem (#14426) 
strided setitem is wrong
 2026-01-29 14:54:19 -05:00
chenyuandGitHub 31706bf6bc add few more types [pr] (#14425) 2026-01-29 14:04:09 -05:00
nimlgenandGitHub 2d5c24879f nv: pma for 5090 (#14420) 
* nv: pma for 5090

* hm

* 4090
 2026-01-29 20:06:01 +03:00
nimlgenandGitHub c8dc6332d2 memory: read_fields is not universal (#14348) 2026-01-29 20:00:00 +03:00
chenyuandGitHub dbe8f034a7 pass z3.Context in validate ctx [pr] (#14423) 
does not need to pass the whole solver
 2026-01-29 11:11:47 -05:00
chenyuandGitHub 033ce1b885 types for validate.py (#14422) 2026-01-29 10:56:50 -05:00
nimlgenandGitHub 230d08ec70 test for am recovery and faults handling (#14421) 
* test for am recovery and faults handling

* linter
 2026-01-29 17:11:24 +03:00
George HotzandGitHub 793afbd473 simplify nn.Embedding, support AFTER in CUSTOM_KERNEL (#14419) 2026-01-29 17:22:13 +08:00
sirhcmandGitHub 0c855d6149 ci: remove unused pydeps (#14418) 2026-01-29 01:51:26 -05:00
wozeparrotandGitHub 4845e42135 llama3 gradacc fixes (#14414) 2026-01-28 19:12:39 -08:00
chenyuandGitHub 37cde4a01a add one line mypy report (#14415) 2026-01-28 20:39:32 -05:00
chenyuandGitHub 15aed51544 return types for all math.py function (#14413) 
calling int() on sint -> int, i think it's better support since some UOp can be safely cast to int
 2026-01-28 20:10:11 -05:00
nimlgenandGitHub aec1ae0de1 llama: set manual_seed (#14409) 2026-01-28 14:40:00 -08:00
chenyuandGitHub 0870ed28b1 add Self type to MathMixin (#14411) 
these don't cause error
 2026-01-28 16:59:38 -05:00
chenyuandGitHub 079f33c208 fix type in Tensor.mean and Tensor.var (#14410) 
use Tensor.from_uop to wrap UOp from symbolic shape, kernels are the same
 2026-01-28 15:24:02 -05:00
chenyuandGitHub 2b5e99ccc1 minor type cleanups [pr] (#14408) 
mypy --warn-redundant-casts has false negative
 2026-01-28 14:11:50 -05:00
chenyuandGitHub 726415dbc8 import sint directly in movement.py TYPE_CHECKING (#14406) 
avoid creating string TypeAlias, fixed warning in `TYPED=1 python test/test_tiny.py`
 2026-01-28 12:47:26 -05:00
nimlgenandGitHub acb2fc36ba nv_pma: add decoder (#14404) 
* nv_pma: add decoder

* cl
 2026-01-28 20:44:02 +03:00
chenyuandGitHub 7b9bc1d8cf _MockMemoryviewMeta for mockgpu (#14405) 
fixed `PYTHONPATH=. TYPED=1 DEV=AMD MOCKGPU=1 python test/test_tiny.py`. basically make `isinstance(TrackedMemoryView_instance, memoryview)` true
 2026-01-28 11:59:00 -05:00
chenyuandGitHub 93793a645b use cl.cl_mem instead of internal ctypes._CData (#14403) 
fixed `CHECK_OOB=0 DEV=CL TYPED=1 python test/test_tiny.py`
 2026-01-28 10:56:41 -05:00
chenyuandGitHub a9b44070a8 fix webgpu runtime types (#14402) 
`CHECK_OOB=0 DEV=WEBGPU TYPED=1 python test/test_tiny.py` passed, also skip tests that failed locally
 2026-01-28 10:37:25 -05:00
George HotzandGitHub 0c6b3f50aa add marker to llama training (#14401) 2026-01-28 22:44:28 +08:00
Jakob SachsandGitHub 2b7c00d3d2 fix sd-example dtype for CLIP embeddings (#14397) 2026-01-28 09:07:19 -05:00