mirrors/tinygrad
2
0
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-06-24 02:14:17 +00:00
Code Issues Projects Releases Packages Wiki Activity
11,106 commits 280 branches 15 tags 325 MiB
Commit graph

915 commits

Author SHA1 Message Date
chenyu
03ef5197fc
 		move get_contraction to helpers [pr] (#12594) 2025-10-10 04:28:57 -04:00
chenyu
af90dc00de
 		remove some View add logic [pr] (#12584) 
no longer simplify the case of v0+v1 where v0 has a mask
 2025-10-10 03:47:56 -04:00
chenyu
c8dfd10257
 		ShapeTracker.real_strides -> is_expanded [pr] (#12579) 
only keep the used part
 2025-10-09 22:52:45 -04:00
chenyu
678f83e41b
 		delete ShapeTracker to_valid_uop and substitute [pr] (#12563) 2025-10-09 05:06:10 -04:00
chenyu
cf8232ec6a
 		clean up more RANGEIFY flag (#12556) 2025-10-09 03:06:48 -04:00
chenyu
250f05a776
 		run some hashing test only on METAL (#12554) 
quite slow on CPU
 2025-10-09 02:39:49 -04:00
chenyu
ae51bdd06a
 		remove trivial use of RANGEIFY flag (#12550) 
some tests need update still
 2025-10-09 02:29:38 -04:00
chenyu
43bce1f39f
 		delete View minify [pr] (#12538) 2025-10-08 23:25:53 -04:00
chenyu
20d98b19c3
 		delete more unused ShapeTracker stuff (#12536) 2025-10-08 23:09:44 -04:00
George Hotz
0774575442
 		delete the old rangeify path and all the children stuff (#12524) 
* delete the old rangeify path and all the children stuff

* remove the on_stack stuff and any retries

* don't use the p word

* Revert "remove the on_stack stuff and any retries"

This reverts commit 49a2b328b9.
 2025-10-08 21:24:04 +08:00
qazal
b6835f4134
 		remove Ops.VIEW and related UOp methods (#12522) 
* remove Ops.VIEW and related UOp methods

* update abstractions2.py

* no ShapeTrackers in abstractions2.py

* it's a size 1
 2025-10-08 14:47:02 +03:00
George Hotz
3b0b3a2e64
 		fast RANGEIFY (#12504) 
* rtoposort is fast, can replace rangeify with this

* fast rangeify

* work

* fast rangeify works for mnist

* should work

* progress

* pad fix

* FAST

* tests passing

* don't delete those shape ops

* put in rangeify map

* ending ranges fix

* tests

* mstack/mselect no hacks

* move to indexing.py

* touch up tests + add comments

* disable failing test

* actually make the file readable

* failing

* error
 2025-10-08 19:38:06 +08:00
chenyu
ee0382ad99
 		remove ShapeTracker.invert (#12520) 2025-10-08 18:37:34 +08:00
chenyu
d5058427ea
 		remove ShapeTracker.real_size (#12519) 2025-10-08 06:15:29 -04:00
qazal
2e19354c1c
 		viz: reorder timeline graphs (#12498) 
* viz: reorder timeline graphs

* update test_viz with the new order
 2025-10-08 07:10:23 +03:00
Sieds Lykles
b465c17b56
 		Revert "UOp.factor and add chain sorting (#12413)" (#12492) 
This reverts commit e74be4a140.
 2025-10-08 03:20:23 +02:00
George Hotz
945cc46475
 		delete children tracking from uop (#12491) 
* delete children tracking from uop

* uop children no longer exists

* no tracked children

* that test is flaky too
 2025-10-08 09:04:14 +08:00
chenyu
a5484b767e
 		remove skipping cast in simplify_valid [pr] (#12472) 
* remove skipping cast in simplify_valid [pr]

unsupported statements are handled in uop_given_valid already. the test failed because (100%x) somehow got simplified

* better test
 2025-10-07 00:10:04 -04:00
qazal
f664bcc8bd
 		use recursive_property in UOp tracing (#12469) 
* test

* simple passing
 2025-10-06 21:10:52 +03:00
Sieds Lykles
e74be4a140
 		UOp.factor and add chain sorting (#12413) 
* add ordering

* fix some tests

* fix more tests

* shorten comment

* update test

* add rule and test

* add rule and test

* remove check

* use fold_divmod_congruence instead of simplify

* adjust tests

* shorten line

* new algo

* add test

* add function to un-nest the div

* add UOp.factor

* test UOp.factor

* uop_given_valid tries to factor simplex expression

* shorten line

* symbolic_flat is back

* change that back

* fix those new tests

* new rule for ordering

* factor multiple factors

* no symbolic_flat

* symbolic_flat to there

* move that back

* fix imports

* merge correctly

* linter happy

* add rule

* add a test

* cleanup

* revert that for now

* UOp.factor returns self instead of None

* try all_candidates

* remove or_else

* post index symbolic

* add test

* maket this closer to the original

* increase mac hlb_cifar min step time

* add some ordering tests

* cleanup

* increase pytest timeout time

* check dtype
 2025-10-04 06:05:38 +02:00
George Hotz
c7849ac593
 		fix test lil model (#12437) 
* fix test lil model

* 4 not 3
 2025-10-03 02:28:37 -04:00
chenyu
bf99de7b1e
 		update a few more tests for RANGEIFY (#12434) 2025-10-03 00:16:58 -04:00
Sieds Lykles
16a65b4fd0
 		fix test_symbolic_gcd_div hang (#12427) 2025-10-03 04:21:16 +02:00
chenyu
7b3912d8e4
 		relax atol for some tests (#12422) 2025-10-02 05:04:44 -04:00
George Hotz
583553f467
 		split ranges (#12411) 
* split ranges

* simpler

* split ranges

* range str

* fix test

* oops

* faster

* no group 2

* tests

* dont_sub_ranges_for_image

* revert that
 2025-10-02 12:57:22 +08:00
nimlgen
3e0e0290ce
 		increase timeout in test_module_runs (#12408) 2025-10-01 22:01:44 +03:00
b1tg
57ad46c6e4
 		rangeify: increase atol for test_two_binops_no_rerun passing on real windows machine (#12389) 
CPU_LLVM=1
 2025-10-01 00:56:45 -04:00
chenyu
0662946fac
 		atol in test_two_binops_no_rerun (#12387) 
for RANGEIFY LLVM
 2025-10-01 00:05:47 -04:00
wozeparrot
4204edc60b
 		feat: skip test_long (#12383) 2025-09-30 20:07:39 -07:00
George Hotz
4c9a930de2
 		rangeify attn tests (#12377) 2025-10-01 09:59:19 +08:00
qazal
109c63b904
 		update Tensor unit tests for RANGEIFY (#12359) 
* update test_kernelize for RANGEIFY

* also kernelizes user contiguous

* skip that test

* tensor uop repr

* 4 kernels, still realizes a float
 2025-09-30 11:17:21 +03:00
George Hotz
7129419500
 		fix cifar training in RANGEIFY (#12355) 
* fix cifar training in RANGEIFY

* even more wino fuse

* bugfix

* test to show issue
 2025-09-30 15:59:19 +08:00
chenyu
86c5c969ea
 		linalg cosmetic change (#12356) 2025-09-30 03:00:59 -04:00
George Hotz
f522e83a02
 		fix rangeify elu fusion for openpilot (#12341) 
* fix rangeify elu fusion for openpilot

* flip the metadata

* copy over permuted contiguous support

* this is correct

* update that
 2025-09-30 11:41:52 +08:00
George Hotz
cdfa0f29fd
 		add rendering to index (#12338) 2025-09-30 09:18:05 +08:00
chenyu
9d2f2b8e34
 		skip test_mean_half_precision_overflow (#12331) 
it only works with SPLIT_REDUCEOP=1
 2025-09-29 05:15:04 -04:00
chenyu
76c87d81b3
 		delete test_backward_sum_acc_dtype (#12330) 
this test tests the wrong thing, it was only working because expand realize rule
 2025-09-29 04:46:17 -04:00
chenyu
17cec8d645
 		RANGEIFY winograd test (#12297) 
speed seems fine
 2025-09-24 23:42:32 -04:00
Sieds Lykles
45c7252aed
 		Better div nesting 2 (#11812) 
* remove check

* use fold_divmod_congruence instead of simplify

* adjust tests

* shorten line

* new algo

* add test

* cleanup

* update tests

* ALLOWED_GATED_READ_IMAGE from 16 -> 12

* only remove the call to simplify

* add option to simplify with factor_remainder

* Allowed readimage gates back to 16
 2025-09-24 04:50:26 +02:00
Sieds Lykles
6146c64d81
 		lower the invalid gate last (#12164) 
* lowering invalid gate is part of lower_index_dtype

* update test

* remove import

* put that back

* reduce_collapse uses invalid

* fix that pattern to use invalid_pat

* valid creates the right dtype count

* seperate rule for lowering invalid gate

* dont unvectorize Invalid gate

* image_fixup uses Invalid

* update tests

* cleanup

* update split_load_store

* add .scalar() there
 2025-09-24 04:27:35 +02:00
chenyu
b54cb272d0
 		move test_qcom to test/device (#12272) 2025-09-22 21:07:10 -04:00
qazal
4756971c88
 		skip test_bf16_disk_write_read on CL=1 (#12256) 2025-09-20 17:11:06 +03:00
Sieds Lykles
cc038b31b6
 		Shrink instead of reshape to unregister symbolic (#12241) 
* Slice to unbind symbolic

* use vmax for now

* assert shape in reshape is valid

* update test_symbolic_ops to use shrink instead of reshape

* remove infer_with_bound_values for npw

* symbolic output doesnt have symbolic strides

* symbolic jit tests use shrink to unregister symbolic

* update test

* update more tests

* wrap vmax in int()

* only create a new st if the store is not an assigne

* unwrap st

* comments
 2025-09-19 06:04:35 +02:00
Sieds Lykles
8d703a6369
 		z3 xor doesnt use bitcast (#12243) 2025-09-19 00:31:44 +02:00
chenyu
7487c13b61
 		truncate_fp16 -> float_to_fp16 (#12234) 
match float_to_bf16 and float_to_fp8
 2025-09-18 09:48:27 -04:00
b1tg
54c15d74a4
 		python float8 support (#11960) 
* basic support

* alu

* nan in exec_alu

* rand_for_dtype

* inf + 0.0

* finfo

* revert rand_for_dtype

* clean

* truncate fp8s inf

* spec ok

* float_to_fp8 nan/inf

* least_upper_dtype

* clean up

---------

Co-authored-by: b1tg <b1tg@users.noreply.github.com>
 2025-09-18 09:17:09 -04:00
Sieds Lykles
158506b91e
 		Upgrade some divmod folding for symbolic divs (#12216) 
* use const_factor() instead of arg

* add test

* change div min_max

* add tests

* add divide_by_symbolic_gcd

* add tests

* one more test

* Slice to unbind symbolic

* deal with const factor properly

* minor cleanup

* divide_by_symbolic_gcd becomes UOp.gcd and UOp.divide_exact

* add tests

* add gcd_without_const

* fix divide_exact bug

* add factor_remainder

* add tests

* fix imports

* elif -> if

* remove expectedFailure

* add more tests

* add more unwrap

* fix signature of pop_const

* remove that

* remove that
 2025-09-17 03:00:50 +02:00
qazal
a388d2cb1a
 		remove PROFILE=1 option, it's just VIZ=1 [pr] (#12176) 
* remove PROFILE=1 option, it's just VIZ=1 [pr]

* sqtt

* sqtt 2

* return last

* rename
 2025-09-15 12:51:50 +03:00
chenyu
15b166ce6d
 		bump test_module_runs to 30 seconds (#12174) 
25 seconds sometimes
 2025-09-14 16:48:40 -04:00
chenyu
d09c0f28c5
 		increase test_module_runs (#12173) 
timed out on ci windows llvm
 2025-09-14 15:19:21 -04:00