Commit graph

13,243 commits

Author SHA1 Message Date
Christopher Milan
2dccd4a3eb
am: autogen pmc (#16143)
* am: autogen pmc

* cleanup

* fix

* type
2026-05-11 19:22:12 -04:00
Christopher Milan
7ba55ad3ba
nv: autogen regs (#16139)
* nv: autogen regs

* flcn cot

* ci

* gen
2026-05-11 18:52:24 -04:00
chenyu
0b02fb6797
Revert "[pr] match torch rmsnorm (#16122)" (#16144)
This reverts commit 692257dd70.
2026-05-11 17:53:42 -04:00
chenyu
fbe8be0b8b
style cleanup to Tensor.qr and svd (#16142)
* style cleanup to Tensor.qr and svd

same kernels

* more

* enable
2026-05-11 17:16:59 -04:00
qazal
fc2cc1d77a
viz: call graph renderer example (#16141)
* work

* emits

* this

* cleaner repr for custom binaries

* --call-graph

* _ref

* this

* start

* this

* everything execpt the pyrender

* bring pyrender back
2026-05-12 05:07:30 +09:00
chenyu
f65e343fb3
spec.py cleanups (#16140)
removed END from shared_spec and NOOP from full_spec
2026-05-11 15:59:49 -04:00
Joshua James Venter
692257dd70
[pr] match torch rmsnorm (#16122)
* [pr] match rmsnorm torch

Signed-off-by: Joshua James Venter <venter.joshua@gmail.com>

* 1e-5

* ops.md

---------

Signed-off-by: Joshua James Venter <venter.joshua@gmail.com>
Co-authored-by: chenyu <chenyu@fastmail.com>
2026-05-11 14:36:41 -04:00
Sachith Shetty
59a81559d4
fix: add self.device to qr, svd, masked_select intermediates (#16131) 2026-05-11 11:22:54 -04:00
nimlgen
70c2480e71
hcq2 to extra (#16126)
* hcq2 in extra

* correct

* some revert from non-extra

* cln

* cpu

* x

* attach

* min

* remove attach

* linter
2026-05-11 17:17:30 +03:00
nimlgen
ad9738892c
get_buf() for Buffer (#16134)
* p

* mypy

* x
2026-05-11 16:36:14 +03:00
qazal
2dd84416bf
viz/cli: schedule renderer (#16101)
* simpler steps

* work

* work

* iterate

* faster

* better

* simplify more

* sys stdin

* less

* work

* work and mv

* better

* seen bufs

* all call graphs

* print query

* ux

* param to buffer / buffer_view

* work

* respect NO_COLOR in uop_to_json

* less

* render uops

* rm custom renderer

* call can't pyrender.

* unrelated diff

* assert

* 5
2026-05-11 01:56:16 +09:00
George Hotz
53f9587099 add canary 2026-05-10 09:38:18 -07:00
George Hotz
28cb7f1bcc update readme with contributing guidelines 2026-05-10 09:35:48 -07:00
George Hotz
daed602569
rename BUFFERIZE to STAGE (#16125) 2026-05-10 09:26:46 -07:00
qazal
39ce780907
viz/cli: emit all runs of selected kernel, json fixes (#16124)
* keep print

* --json in tests, sqtt --json err

* work

* import

* less

* line
2026-05-10 21:45:51 +09:00
qazal
51c7dafb0d
split viz cli test helpers (#16123) 2026-05-10 19:42:24 +09:00
chenyu
b2a682ec60
remove _shape check in pm_mops [pr] (#16120)
seems fine now
2026-05-09 17:54:22 -04:00
wozeparrot
026688f03f
llama: move to correct dir (#16118) 2026-05-08 19:42:16 -07:00
Christopher Milan
a7512e0d12
PYTHON: images have no alignment constraints (by default) (#16115) 2026-05-08 20:35:03 -04:00
Christopher Milan
105b037c3c
cl: image alignment in arch (#16106) 2026-05-08 19:33:33 -04:00
Charlie Kerfoot
71a8c0da09
fix: trailing space format string (#16005) 2026-05-08 16:31:10 -07:00
Pawan
4dd6ad3514
gradient: add TRUNC backward (#15925)
* gradient: add TRUNC backward

* test: move round quantization gradient to test_ops
2026-05-08 16:27:55 -07:00
chenyu
5152ff95e7
_pad_constant and avg_pool2d cleanups (#16110) 2026-05-08 18:09:47 -04:00
chenyu
e6584532f4
minor elementwise cleanups (#16102) 2026-05-08 13:38:34 -04:00
nimlgen
49b55af619
jit: simpler free_intermediates (#16099) 2026-05-08 19:08:33 +03:00
chenyu
0f46c08582
div mixin cleanups (#16100) 2026-05-08 12:05:37 -04:00
chenyu
235044c9d8
Ops.IDIV -> Ops.CDIV, Ops.MOD -> Ops.CMOD (#16093)
* Ops.IDIV -> Ops.CDIV, Ops.MOD -> Ops.CMOD

* ruff
2026-05-07 23:18:15 -04:00
Christopher Milan
faabe6aa42
nv: remaining firmware from /lib/firmware (#16088) 2026-05-07 23:07:43 -04:00
b1tg
7ef901a81d
llm: moe speedup (#16059) 2026-05-07 19:06:35 -07:00
George Hotz
80da8a4b9c
add spec to main tinygrad repo (#16092) 2026-05-07 18:52:49 -07:00
June
83eaefcd0f
onnx: deduplicate simple proto parsers (#16085)
Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
2026-05-07 18:44:27 -07:00
George Hotz
c106c73e51
remove the gate from index (#16081)
* remove the gate from index

* gpt says this works

* remove hanging casts

* simplify

* move that down

* move gates

* ptr

* remove that simplify

* move that
2026-05-07 18:42:00 -07:00
wozeparrot
d11f4d0ec2
fix: don't copy on slice of DP weight (#16089) 2026-05-07 17:58:01 -07:00
George Hotz
1d1b726cf6 hotfix: disable flaky framework pytest 2026-05-07 17:05:06 -07:00
Christopher Milan
9a6f7f7576
nv: look for fmc firmware in /lib/firmware (#16080) 2026-05-07 18:08:27 -04:00
George Hotz
b796bbae87
fix valid in indexing tests (#16087) 2026-05-07 14:11:28 -07:00
wozeparrot
4d1a9dca41
fix: don't copy precompiled custom kernel outputs (#16084) 2026-05-07 14:02:38 -07:00
qazal
f9083cf901
use subactions for benchmark.yml process replay [pr] (#13396) 2026-05-08 03:46:25 +09:00
nimlgen
2f0aa884d5
tinygpu: minimal is macos13 for resets (#16075) 2026-05-07 21:25:56 +03:00
chenyu
072db9924c
div to mixin (#16078)
also deleted idiv method
2026-05-07 12:52:37 -04:00
chenyu
516b00e286
mod and fmod to mixin (#16077) 2026-05-07 12:13:39 -04:00
qazal
a9a87ad8fd
viz/cli: less flags (#16076)
* viz/cli: merge -s and -i flags

* only -t

* merge parser

* fix
2026-05-08 00:22:40 +09:00
qazal
f813a04b3f
viz: pickle path in str (#16073) 2026-05-07 18:49:21 +09:00
wozeparrot
730fa66bf3
llama speed 6 (#16071) 2026-05-06 20:51:03 -07:00
Christopher Milan
7b91f7c90c
nv: look for gsp firmware in /lib/firmware (#16068) 2026-05-06 21:35:47 -04:00
George Hotz
8e84317743
the renderer part of gate moving from index to load/store (#16064)
* the renderer part of gate moving from index to load/store

* fixed

* fix gated stores

* fix spec

* better?

* Where after gated load becomes alt value

* cleaner expression

* fix python backend

* remove dead code
2026-05-06 13:47:04 -07:00
chenyu
ef085304bc
stronger divmod_recombine (#16066) 2026-05-06 15:41:54 -04:00
qazal
d7d32d82ee
viz/cli: print first uop with DEBUG=6 (#16065)
* viz/cli: print first uop with DEBUG=6

* rename fmt to emit

* define inst
2026-05-07 03:39:34 +09:00
chenyu
af4140f3be
fix divmod recombine for floordiv (#16062) 2026-05-06 14:22:42 -04:00
chenyu
c6ad3d3ac2
better divmod late rewrite (#16061)
better order
2026-05-06 11:31:48 -04:00