George Hotz
18addc0a1d
process replay only get_program ( #13475 )
2025-11-27 08:18:18 -08:00
George Hotz
a8e005b095
enable process replay (non-checking) by default ( #13474 )
2025-11-27 07:28:44 -08:00
George Hotz
05cd2279d0
add cache on reshape ( #13466 )
...
* remove cache on divmod, way less objects
* _apply_reshape
* reshape
* no gc on realize
* wow that cache is fast
2025-11-26 18:57:40 -08:00
George Hotz
19228e8d37
test_graph is flaky
2025-11-26 16:37:42 -08:00
George Hotz
e4cd649ff0
remove kernelize to prepare for refactors ( #13463 )
...
* remove kernelize to prepare for refactors
* less kernelize
* last test
2025-11-26 14:18:50 -08:00
wozeparrot
ffc31a23f4
tk mi350 ( #13288 )
2025-11-25 15:49:44 -08:00
qazal
7238df7a94
viz: cleanup sort_fn ( #13454 )
2025-11-26 04:10:10 +08:00
wozeparrot
249553a119
tinyfs tweaks ( #13444 )
2025-11-24 18:07:32 -08:00
C T
2d53029be3
Whisper less flaky tests ( #13435 )
...
* use less flaky metric for whisper long transcription
* multiline long transcription 3 reference
* fix reference transcript
see https://homepage.ntu.edu.tw/~karchung/miniconversations/MC.htm
sanitized for whisper
* try lower wer threshold
* add test for wer metric
* extract TRANSCRIPTION_3_ALT
* rename test
* rename
* add tests for high WER difference
* move tests
* sync metric
2025-11-24 09:50:49 -08:00
Sieds Lykles
63a931ff76
Symbolic divisor fuzzer ( #13433 )
...
* render z3 range better
* working version
* rename
* add to workflow
* factor out variable_names
* smaller expressions
* smaller
* + back
2025-11-23 20:29:32 +01:00
George Hotz
9d7a17ee39
beautiful SQTT_PARSE=1 with color ( #13428 )
...
* beautiful SQTT_PARSE=1 with color
* linter
* linter 2
* a few more labels
* filter and or
* wave alloc
* a few more
2025-11-23 01:05:14 -08:00
chenyu
cb29265f23
add test that shows the validhack regression with bad rewrite order ( #13411 )
2025-11-21 13:48:30 -05:00
Sieds Lykles
114bb94c55
Fix load collapse MAX to ADD ( #13406 )
...
* add Ops.ADD to pattern
* add test
2025-11-21 12:26:14 +01:00
George Hotz
e1051d00d7
multi like on full_like as well as rand_like ( #13402 )
...
* multi like on full_like as well as rand_like
* add test and fix bug
* mismatch, optim match
* one line
2025-11-20 20:46:48 -08:00
chenyu
647fde64e6
no sym in pm_reduce [pr] ( #13398 )
...
* no sym in pm_reduce [pr]
* fix that
2025-11-20 16:49:09 -05:00
chenyu
0251a8e628
parse_valid minor cleanup [pr] ( #13385 )
...
* stricter parse_valid [pr]
* not stricter
* no VCONST
* Revert "no VCONST"
This reverts commit 330dbdf4060562596febcbf970bda6051a35012f.
2025-11-20 13:15:06 -05:00
qazal
9dcd52287a
add external_benchmark_pyrender ( #13378 )
...
* add external_benchmark_pyrender
* can ctrlc it
* cpu_profile exists
2025-11-20 17:38:28 +08:00
George Hotz
8919c994b7
Revert "AxisType.PLACEHOLDER in reshape to do less graph_rewrite ( #13373 )" ( #13375 )
...
This reverts commit ac7559e33d
2025-11-19 19:34:30 -08:00
George Hotz
ac7559e33d
AxisType.PLACEHOLDER in reshape to do less graph_rewrite ( #13373 )
...
* AxisType.PLACEHOLDER in reshape to do less graph_rewrite
* _apply_movement_op cache
2025-11-19 19:19:58 -08:00
George Hotz
ab7df42c78
bring back fold_divmod_general with bugfix and test [pr] ( #13369 )
...
* Revert "Revert "merge to fold_divmod_general [p] (#13359 )""
This reverts commit 05ccc69248#13363 )""
This reverts commit 90e5752199#13365 )""
This reverts commit 8e17bd6791
2025-11-19 14:51:51 -08:00
George Hotz
986d113024
symbolic fuzz failure ( #13367 )
...
* symbolic fuzz failure
* skip flaky test
2025-11-19 14:21:08 -08:00
George Hotz
05ccc69248
Revert "merge to fold_divmod_general [p] ( #13359 )"
...
This reverts commit 7711bbac7f
2025-11-19 14:18:09 -08:00
George Hotz
8e17bd6791
Revert "add cache to fold_divmod_general ( #13365 )"
...
This reverts commit b5309a5043
2025-11-19 14:18:08 -08:00
George Hotz
b5309a5043
add cache to fold_divmod_general ( #13365 )
2025-11-19 13:49:18 -08:00
George Hotz
7711bbac7f
merge to fold_divmod_general [p] ( #13359 )
...
* merge to fold_divmod_general [p]
* merge more
* merge more
* merge more
2025-11-19 11:37:45 -08:00
George Hotz
6fdbd03104
more divmod cleanup [p] ( #13358 )
...
* more divmod cleanup [p]
* lil cleanups, faster
2025-11-19 10:35:15 -08:00
George Hotz
957cf717e7
Python speed ( #13355 )
...
* skip process replay by default
* work on python speed
* fix names of rewrite rules
* fix that test
2025-11-19 09:03:00 -08:00
George Hotz
385618d45b
skip process replay by default ( #13353 )
2025-11-19 08:25:34 -08:00
Christopher Milan
a438c277de
autogen tests for 3.14 ( #13343 )
2025-11-18 22:16:59 -05:00
George Hotz
1afa3c0877
vmap on full model ( #13340 )
...
* vmap on full model
* vmap gemm
* reduce sums on end
* outer reduce
* only if there's ranges
* put those rules in symbolic
* ranges
* do opt later
* add zero range
2025-11-18 16:06:06 -08:00
George Hotz
9c59b3d19e
vmap grad needs reduce_backward ( #13336 )
...
* vmap grad needs reduce_backward
* fuse and outer
2025-11-18 10:08:30 -08:00
George Hotz
06e39a88a9
outer vmap works ( #13334 )
...
* outer vmap works
* fuse works
* vmap outer works
* outer ranges work
* grad work
* should be good to merge
2025-11-18 09:27:48 -08:00
George Hotz
583560ab72
this is the right way to write vmap ( #13328 )
2025-11-17 20:20:52 -08:00
George Hotz
e4fead8a86
write scan in uops ( #13321 )
...
* write scan in uops
* ops range
* no need for variable
* meh, later
* shorter
2025-11-17 16:58:08 -08:00
George Hotz
6d3385c284
print special ops in postrange ( #13318 )
...
* print special ops in postrange
* fix on OSX
2025-11-17 14:43:23 -08:00
wozeparrot
33773fda87
tk initial mi350 ( #13289 )
2025-11-17 11:46:32 -08:00
nimlgen
9bb17c53ea
amd: timer fix ( #13267 )
2025-11-17 13:59:03 +03:00
George Hotz
cabd4add48
more work parsing SQTT, separate VIZ/PROFILE ( #13308 )
...
* more work parsing SQTT
* more minimal runner
* sep VIZ/PROFILE
* parse print new
* improve parser
* more filter
* that
* split them
* lil cleanup
* skip flaky test
* AQL in mmapeak
2025-11-16 10:40:39 -08:00
George Hotz
295600dc5a
saturday coffee shop work parsing the att format ( #13295 )
...
* saturday coffee shop work parsing the att format
* add examples
* parser
* classes of packets
* fully vibe coded parser
* vibing
* empty
* some vibe names
* vibes
* most of these are wrong
* more vibes
* better names
* parsing
* parse
* cleanup parser
* touchups
2025-11-16 08:25:51 -08:00
chenyu
8f0e747b3a
Tensor._tri with arange ( #13297 )
2025-11-16 10:21:16 -05:00
chenyu
e8844853ed
Tensor.eye with arange ( #13287 )
...
with rangify we can write these with arange
2025-11-15 12:32:27 -05:00
George Hotz
22c08b470c
fold using outerworld range ( #13286 )
...
* scan using outerworld range
* almost
* sched
* simple range
* mypy
* woooo outer range
* spec passes
* print the numbers
* lol it runs
* real test
2025-11-14 20:43:41 -08:00
George Hotz
567066f51f
tests for cast there and back ( #13195 )
...
* fix cast folding in llama
* dtypes that work everywhere
* Skip test_cast_there_and_back for backend casts
Skip test due to backend casting issues.
2025-11-14 16:56:09 -08:00
George Hotz
6c5fa349e1
add (unused) outer range ( #13285 )
2025-11-14 16:47:52 -08:00
George Hotz
e5351699bd
openpilot warp ( #13283 )
...
* openpilot image warp test
* 0.4 ms on metal, 1 ms on CPU
* new inputs each time
* reshape
2025-11-14 13:55:32 -08:00
chenyu
888aaab151
test_tiny cleanup ( #13276 )
2025-11-14 11:11:32 -05:00
nimlgen
3e63831b98
nv: support 580+ drivers ( #13269 )
...
* nv: 580+ support
* start
* f
* fake
* fix
2025-11-14 21:44:16 +08:00
nimlgen
c80d459d99
autogen: fix packed args structs ( #13274 )
...
* autogen: fix packed args structs
* and test this
2025-11-14 20:24:06 +08:00
nimlgen
14eb48b13a
autogen: rename nv_gpu to nv_570 ( #13273 )
...
* autogen: rename nv_gpu to nv_570
* rename
2025-11-14 20:07:19 +08:00
nimlgen
f72b1fbca4
nv: read numClasses ( #13271 )
...
* nv: read numClasses
* fix
* d
2025-11-14 19:43:25 +08:00
