sirhcm/tinygrad
1
0
Fork 0
forked from tinygrad/tinygrad
Code Pull Requests 2 Actions Activity
12,065 Commits 336 Branches 15 Tags
kernel_is_call
Commit Graph
12065 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
George HotzandGitHub 771a395240 Merge branch 'master' into kernel_is_call 2026-02-06 09:15:03 +08:00
wozeparrotandGitHub f73468d516 fa: block skipping for fa kv bwd (#14569) 2026-02-05 16:13:53 -08:00
chenyuandGitHub b7ef775677 more cleanup in create_schedule [pr] (#14566) 
fixed wrong comments and simplified queue building
 2026-02-05 16:12:17 -05:00
Garret CastroandGitHub cee7ef7ab2 disable threads (#14555) 2026-02-05 16:11:32 -05:00
chenyuandGitHub 79b7799dba clean up linearize schedule [pr] (#14565) 
* clean up linearize schedule [pr]

don't mix ScheduleItem and UOp in schedule queue

* ok
 2026-02-05 15:24:09 -05:00
chenyuandGitHub 41a179f542 fix test_xlm_roberta_large (#14564) 
onnxruntime does not allow symlink that's outside model dir. update snapshot_download to use local_dir instead of cache_dir. some ad hoc migration step to copy the existing model too
 2026-02-05 14:56:06 -05:00
sirhcmandGitHub aa9dc50577 dtype decomps don't require bitshifts (#14542) 
* dtype decomps don't require bitshifts

* simplify shr/shl

* ruff
 2026-02-05 14:42:30 -05:00
sirhcmandGitHub b47397ab17 list ml_dtypes as dependency for DSP (#14562) 
* pin onnxruntime to 1.23.2 for DSP

* list ml_dtypes instead

This reverts commit 84bb2cc0fc.
 2026-02-05 14:27:50 -05:00
chenyuandGitHub 2b47a9a1b5 skip test_xlm_roberta_large (#14563) 
symlink model not allowed in latest onnxruntime
 2026-02-05 14:00:24 -05:00
chenyuandGitHub 42c18da88a add Ops asserts in toposort sched_sink [pr] (#14561) 
more explicit
 2026-02-05 12:40:02 -05:00
nimlgenandGitHub 483bba4f05 nv: use prof_exec_counter (#14559) 2026-02-05 19:00:14 +03:00
qazalandGitHub 190042358f llama: faster bf16 matmul / rope backward (#14558) 2026-02-05 23:57:25 +09:00
George HotzGitHubsirhcm
 		b398335f62 assembly/amd: fix saturation in python remu (#14557) 
* PYTHONREMU: failing test for V_SUB_NC_U32_E64 clamp

* fix saturation in PYTHON_REMU

* simpler

* more tests, less lines

---------

Co-authored-by: Christopher Milan <[email protected]>
 2026-02-05 18:35:57 +08:00
wozeparrotandGitHub c1ea6687e5 fa: simpler is faster (#14548) 2026-02-05 01:13:17 -08:00
George HotzGitHubqazalqazal
 		43e7eda4e7 grad_b uses custom gemm (#14550) 
* grad_b uses custom gemm

* fix multi backward, acc is in float32

* test_gemm_batched

* square gemm

---------

Co-authored-by: qazal <[email protected]>
Co-authored-by: qazal <[email protected]>
 2026-02-05 15:22:27 +09:00
geohot d83ddc05c8 resolve_call 2026-02-05 12:57:31 +08:00
qazalandGitHub f9cfb64cd9 test asm_gemm in CI (#14551) 
* test asm_gemm in CI

* default float16

* use a smaller shape for multi

* smaller size

* smaller for CI

* smaller for ci

* need half
 2026-02-05 13:32:22 +09:00
geohot 8e8cac4b0f don't use tag, use KernelInfo 2026-02-05 12:31:14 +08:00
geohot 57199fd9de keep the all buffers on same device check 2026-02-05 12:17:32 +08:00
geohot 2193d0edfa fix arg order 2026-02-05 12:03:45 +08:00
geohot 77adccb925 use call for kernel 2026-02-05 11:48:50 +08:00
chenyuandGitHub c0ca7f9c51 use more UOp.sum and UOp.prod [pr] (#14549) 2026-02-04 22:05:20 -05:00
chenyuandGitHub e8dace41b6 clean up UOp.vars [pr] (#14547) 2026-02-04 20:52:25 -05:00
sirhcmandGitHub 232848d086 PYTHONREMU: VOP3P integer operations with constants don't cast to fp16 (#14546) 
* PYTHONREMU: VOP3P integer operations with constants don't cast to fp16

* put that back

* cleaner

* do that once
 2026-02-04 20:10:59 -05:00
wozeparrotandGitHub 2966619834 feat: llama uses enable_gqa during training (#14545) 2026-02-04 16:22:31 -08:00
chenyuandGitHub 664f1bf76d minor ops/jit cleanups [pr] (#14543) 2026-02-04 17:21:34 -05:00
chenyuandGitHub 03d0fa9c3f merge as_buf into buf_uop [pr] (#14541) 2026-02-04 16:32:23 -05:00
chenyuandGitHub 43ef24a8af remove buf_target [pr] (#14540) 
not really needed
 2026-02-04 15:03:47 -05:00
chenyuandGitHub 8b7343b950 clean up is_realized [pr] (#14538) 
base cannot be Ops.MULTI since MULTI is a view now
 2026-02-04 14:24:10 -05:00
sirhcmandGitHub 5338ce6b74 test S_PACK in extra/assembly/amd/test/hw (#14537) 
* S_PACK_LL_B32_B16 in test/hw

* add rest of S_PACK instructions
 2026-02-04 14:17:16 -05:00
chenyuandGitHub 9052db678f remove allow_shape_mismatch in Tensor.replace (#14536) 
move all logic to torch_backend and not hacking Tensor method
 2026-02-04 12:38:18 -05:00
nimlgenandGitHub ec2b6bbda8 hcq: update signal logic (#14531) 2026-02-04 19:32:56 +03:00
nimlgenandGitHub 62786d488a am: mi3xx perf (#14529) 2026-02-04 19:32:43 +03:00
chenyuandGitHub d57d24c7d4 Buffer.as_buffer -> Buffer.as_memoryview [pr] (#14535) 
it casts to memoryview. also inline the as_typed_buffer checks to Tensor._data
 2026-02-04 11:31:11 -05:00
chenyuandGitHub 024f57ecf5 jit input_buffers cleanup [pr] (#14532) 2026-02-04 10:14:38 -05:00
chenyuandGitHub 67f91e897b UOp.is_contiguous -> UOp.has_buffer_identity [pr] (#14530) 
one more confusing buffer related method, but it's definitely not is_contiguous
 2026-02-04 09:21:26 -05:00
George HotzandGitHub fb9df1e031 pretty print binary (#14520) 2026-02-04 18:04:35 +08:00
sirhcmandGitHub 8c3c026d86 decomp float16 to float32 (#14417) 
* decomp float16 to float32

* denormals arent zero

* add test

* denormals are zero

* fix

* oops

* bitcast works

* fix LOADs

* test_dtype passing

* cleanup

* mypy

* debug print

* only emulate if EMULATED

* very ugly, but passes spec

* add test_dtype_alu tests

* Revert "very ugly, but passes spec"

This reverts commit fdc3999b654d630678bf208927ab2f55e026b4ca.

* bottom up decompositions

* that should have symbolic

* simplify a bit

* SPEC really works

* run with DEBUG

* debug=4

* rm debug
 2026-02-04 01:37:47 -05:00
sirhcmandGitHub ecbce5269e PYTHONREMU properly supports S_PACK_LL_B32_B16 (#14527) 
* PYTHONREMU properly supports S_PACK_LL_B32_B16

* default
 2026-02-03 23:45:33 -05:00
wozeparrotandGitHub 720c9597a9 feat: llama uses is_causal on sdpa during training (#14528) 2026-02-03 20:24:30 -08:00
chenyuandGitHub 9c2fc118ef relax setitem target check (#14526) 
old check was too conservative
 2026-02-03 22:32:49 -05:00
qazalandGitHub d1bfbe9ce3 isolate slow llama gemm (#14525) 2026-02-04 12:20:10 +09:00
nimlgenandGitHub 2f55005ad9 qcom: sync cpu cache when from_blob (#14518) 
* um

* fx

* d

* x

* x

* x

* x

* f

* ren
 2026-02-03 21:51:03 +03:00
chenyuandGitHub ee9d6a1f36 remove DEFINE_VAR in to_define_global [pr] (#14522) 
not needed
 2026-02-03 10:12:33 -05:00
Nino RisteskiandGitHub af4c74bb41 delete extra cast (#14517) 2026-02-03 08:29:04 -05:00
chenyuandGitHub 9d1e9e643e removed a duplicated remove_bufferize rule [pr] (#14519) 2026-02-03 08:28:07 -05:00
George HotzandGitHub d59e6e7a37 move more tests to test/null, split some existing ones (#14512) 
* move more tests to test/null, split some existing ones

* null work

* null work

* move more

* fixes

* move PIL

* PIL in CLIP

* don't move that
 2026-02-03 20:20:20 +08:00
qazalandGitHub a98c53769a ASM_GEMM=1 runs the UOp gemm on non cdna (#14516) 
* ASM_GEMM=1 runs the UOp gemm on non cdna

tests run on mac in 3 seconds

* min diff
 2026-02-03 20:42:02 +09:00
qazalandGitHub 5c1d21349e viz: profiler command line tool (#14515) 2026-02-03 19:51:25 +09:00
George HotzandGitHub dd2de4f838 rename all DEFINE_GLOBAL to PARAM (#14511) 2026-02-03 15:09:38 +08:00