Commit graph

11,106 commits

Author SHA1 Message Date
chenyu
e18922f111
limit AND const min max to ints [pr] (#12918) 2025-10-25 16:07:52 -04:00
nimlgen
92324172be
amd: refactor usb into usbdevice (#12916)
* amd: refactor usb into usbdevice

* nu

* my bad

* ops

* my bad
2025-10-26 01:00:19 +08:00
qazal
3b192f5eac
split viz graph rendering from dag layout (#12914) 2025-10-25 15:36:44 +08:00
George Hotz
6415e3e8a7
use Ops.GROUP instead of Ops.NOOP for merging stores (#12912)
* use Ops.GROUP instead of Ops.NOOP for merging stores

* fs noop
2025-10-25 12:26:12 +08:00
George Hotz
b4f6a2c7a3
add kernel spec (#12911)
* add kernel spec

* fix kernel spec
2025-10-25 11:49:20 +08:00
George Hotz
8a941d95a4
SPEC=2 is full spec, SPEC=1 is default (#12910)
* SPEC=1 passes all tests

* just use SPEC, not __debug__
2025-10-25 11:10:43 +08:00
wozeparrot
456560c1ff
stateless tinyfs copyin (#12908) 2025-10-24 19:18:38 -07:00
wozeparrot
a5b0f57067
clean: cleanup tinyfs copyout (#12907) 2025-10-24 18:32:55 -07:00
chenyu
4b7329001d
clean up test_avg_pool3d (#12905) 2025-10-24 14:31:36 -04:00
George Hotz
6b35467f53
stores don't end ranges (#12902)
* early endrange

* bugfixes
2025-10-24 23:05:03 +08:00
nimlgen
5b5ba31a86
amd: make sqtt bufs uc (#12898) 2025-10-24 18:55:14 +08:00
Sieds Lykles
e1f8c82938
Onnx Layer/Group/RMS/Batch-Norm ReduceL2 fp32 intermediates for fp16 (#12109)
* match onnx spec

* use least_upper_dtype

* promote the square

* just cast before the square
2025-10-24 12:26:11 +02:00
George Hotz
0bde87d8d7
cleanups from flash attention branch (#12897) 2025-10-24 14:14:56 +08:00
wozeparrot
9dac505565
variable bs keccak (#10731) 2025-10-23 14:10:21 -07:00
chenyu
154b4f9f40
test FUSE_OPTIM=1 test/test_optim.py (#12895) 2025-10-23 15:54:27 -04:00
chenyu
6e4ee8deea
small heuristic cleanup [pr] (#12892) 2025-10-23 10:50:15 -04:00
nimlgen
f835566e27
sqtt: correct header (#12891)
* sqtt: correct header

* f
2025-10-23 22:37:17 +08:00
Sieds Lykles
c1db62ff7c
move reduce collapse to rangeify (#12845) 2025-10-23 15:44:17 +02:00
Sieds Lykles
04b3e51f1b
remove old reduce collapse rule (#12889)
* comment this out

* remove
2025-10-23 13:51:49 +02:00
qazal
cdfb8e31ae
hotfix: correct viz rewrite step counter reset (#12890) 2025-10-23 19:47:16 +08:00
George Hotz
6df19a4ac6
lil qol improvements to viz (#12887) 2025-10-23 18:41:07 +08:00
George Hotz
ff68a6263b
move locals into codegen (dedup works) (#12885)
* move locals into codegen (dedup works)

* move in optimize
2025-10-23 17:07:39 +08:00
George Hotz
ddb53d1d48
PCONTIG=3 both saves ram and flops (#12884)
* PCONTIG=3 both saves ram and flops

* group

* gate locals

* should be correct
2025-10-23 16:37:26 +08:00
qazal
2a5c22436e
remove outdated docs (#12881) 2025-10-23 12:52:36 +08:00
qazal
bcc30e5e10
viz: add linearized UOp list view (#12883)
* viz: add linearized UOp list view

* lang
2025-10-23 12:52:14 +08:00
George Hotz
e85cee0aad
flip Ops.END srcs (#12882)
* flip Ops.END srcs

* backward

* late end split
2025-10-23 12:47:50 +08:00
George Hotz
74b4cfe44b
Ops.GROUP + range check (#12880)
* simpler

* fix that

* Ops.GROUP + range check

* fix bugs

* fix linter

* fix test
2025-10-23 12:05:21 +08:00
Sieds Lykles
914defd55d
give endrange priority (#12870)
* uncomment line

* try giving endrange priority
2025-10-23 05:19:13 +02:00
qazal
2f95c10702
remu new instructions / use volatile in emulator tests (#12862)
* remu new instructions

* start moving to volatile

* test_simple works

* test_exec_mov works and lid is still here

* test_exec_cmp_vopc

* clang did s_mov_b32 exec_lo, 1

* don't hardcode v1

* support volatile in tests

* hw_test passes

* only the volatile version

* subrev saturating behavior
2025-10-23 11:13:43 +08:00
George Hotz
e718254004
simpler end (#12879)
* simpler

* fix that
2025-10-23 10:35:58 +08:00
wozeparrot
6e00dec95d
feat: pin openpilot 0.10.1 models (#12878) 2025-10-22 14:57:54 -07:00
wozeparrot
3a9aa05359
feat: extra nvcc options (#12876) 2025-10-22 13:21:11 -07:00
chenyu
f0831c8c30
add 0.10.0 to comma benchmark (#12875)
* add 0.10.0 to comma benchmark

disabled the 0.10.1 ones which are pinned to master. it does not work because benchmark uses the cached old version

* that's pinned
2025-10-22 15:18:21 -04:00
nimlgen
e7e535cd53
amd: sqtt for gfx9 (#12844)
* amd: start sqtt for gfx9

* writes something, but sometimes zeroes

* HEADER!

* w

* tiny

* mypy
2025-10-23 02:31:07 +08:00
b1tg
81108f91ee
amd tc: 16x16x32 (#12874)
* amd tc: 16x16x32

* test

* clean, test amd_cdna4
2025-10-22 13:48:01 -04:00
George Hotz
bf173c0a37
we don't support multi end yet (#12869) 2025-10-22 23:43:32 +08:00
nimlgen
a7bc0104c2
amd: clean up sqtt_stop (#12872) 2025-10-22 22:17:03 +08:00
nimlgen
b6eb9172ea
amd: fix ip offsets (#12867) 2025-10-22 20:50:18 +08:00
George Hotz
174811fc0f hotfix: slightly looser load spec for AMD bfloat16 2025-10-22 19:55:59 +08:00
George Hotz
7762b3558b
clean up the spec (#12868)
* tighten up the spec

* move validate into a different file

* that moved to validate

* after(barr)
2025-10-22 19:50:42 +08:00
George Hotz
726988fa4b
late ifs try 2 (#12865)
* late ifs try 2

* fix image

* fix that test

* panic

* ptx fixups

* preserve toposort

* those pass locally

* Revert "those pass locally"

This reverts commit 063409f828.

* no ls

* make that explicit
2025-10-22 18:49:27 +08:00
George Hotz
6abe90fb7c
fix linearizer non-determinism (#12866) 2025-10-22 17:51:35 +08:00
qazal
cebc2b5721
cleanup viz profiler metadata ui (#12860)
* cleanup viz profiler metadata ui

* text

* select over .args

* space
2025-10-22 17:31:12 +08:00
Sieds Lykles
8d0256c46b
Move gate to load for loaded index (#12861)
* change condition

* change test to better represent how the uop looks irl
2025-10-22 09:53:07 +02:00
chenyu
6d86e962c7
update ASSERT_MIN_STEP_TIME (#12857)
0.10.1 driving_policy is good now, still need driving_vision and dmonitoring to be fast
2025-10-21 22:46:07 -04:00
George Hotz
92778c7a8b
rename opts to ren, add store ranges back (#12856)
* rename opts to ren

* fix docs and bring store back
2025-10-22 09:15:38 +08:00
chenyu
c5cee74706
remove BLOCK_REORDER (#12854)
not used
2025-10-21 19:10:14 -04:00
chenyu
0b673eddec
simpler newton_schulz transpose (#12853) 2025-10-21 17:21:45 -04:00
b1tg
60d7e232f2
cuda fp8 (#12782)
* cuda fp8

* tensor core

* tc test

* clean

* clean pm
2025-10-21 15:05:25 -04:00
Harald Schäfer
587ccc0e5c
compile3: make selftests opt-in (#12851) 2025-10-21 11:32:27 -07:00