geohot
1a90d06c60
6 errors left
2026-08-04 16:40:59 -07:00
geohot
27f2465098
this crap is wrong
2026-08-04 16:34:09 -07:00
geohot
bf0e6d2e9b
prevent early removal
2026-08-04 16:09:24 -07:00
geohot
178f0f661e
simple index/stage
2026-08-04 15:59:07 -07:00
geohot
fb156d6f45
merge ranges
2026-08-04 15:54:19 -07:00
geohot
3a9b73a652
fix symbolic
2026-08-04 13:48:36 -07:00
geohot
8c2b43a8e4
fix dumb end removal
2026-08-04 11:26:28 -07:00
geohot
42dfd836ce
contig nonsense
2026-08-04 11:20:19 -07:00
George Hotz and GitHub
dfba290575
Merge branch 'master' into simple_rangeify
2026-08-04 11:08:22 -07:00
George Hotz and GitHub
f489f4bc36
add test_eye + color INDEX ( #17394 )
2026-08-04 11:08:08 -07:00
George Hotz and GitHub
4909868a2b
Merge branch 'master' into simple_rangeify
2026-08-04 11:01:27 -07:00
wozeparrot and GitHub
80d2073a11
fa: fix dq hazard with D=64 ( #17393 )
2026-08-04 09:38:01 -07:00
chenyu and GitHub
85e9440446
keep weak consts weak in symbolic [PR] ( #17392 )
2026-08-04 12:16:53 -04:00
chenyu and GitHub
0db63e13cf
update dtype_from_uop for IMAGE INDEX [PR] ( #17391 )
2026-08-04 11:50:24 -04:00
chenyu and GitHub
6eedca57ce
dtype_from_uop for CUSTOM, CUSTOMI, PYLITERAL [PR] ( #17390 )
...
these do not change dtype
2026-08-04 11:02:13 -04:00
b1tg and GitHub
0796853845
support symbolic shapes in allreduce ( #17364 )
2026-08-04 07:53:12 -07:00
geohot
568bfb6a37
Revert "usb: wait for PCIe link after power on ( #17380 )"
...
This reverts commit 104ee90ccf
2026-08-04 06:42:26 -07:00
qazal and GitHub
f993228b73
llama: accurate mxfp4 mfu ( #17388 )
...
* llama: accurate mxfp4 mfu
* train_llama3 import
2026-08-04 18:41:57 +09:00
qazal and GitHub
c9cd44b751
more custom kernel contig input edge case tests ( #17387 )
2026-08-04 18:20:50 +09:00
chenyu and GitHub
13fff4f258
const_like cleanups [PR] ( #17386 )
2026-08-03 23:57:20 -04:00
chenyu and GitHub
0170a3075a
use python bitcast in fold_bitcast [PR] ( #17385 )
2026-08-03 23:09:53 -04:00
George Hotz and GitHub
c21a552f3d
llm: bugfixes + warmup ( #17384 )
2026-08-03 18:23:14 -07:00
chenyu and GitHub
67dc02d7e7
bitcast in python for _bits_to_rand [PR] ( #17383 )
...
* bitcast in python for _bits_to_rand [PR]
const in mixin would be weak only without width, so not bitcast
2026-08-03 21:14:58 -04:00
George Hotz and GitHub
3cb786f447
llm: update test_llm_server tests ( #17382 )
2026-08-03 16:06:17 -07:00
geohot
87289a7410
hotfix: revert test_scalar_alu_index, violates spec
2026-08-03 15:25:51 -07:00
George Hotz and GitHub
c2625c78cb
scalar ALU index fix + llm: preserve_thinking ( #17381 )
...
* cstyle: scalar ALU index fix, serve: preserve_thinking, test: fix Handler import
- cstyle.py: return scalar directly when ALU buffer has 1 element
- cli.py: add preserve_thinking param to FallbackTemplate.render
- serve.py: pass preserve_thinking=True when rendering chat completions
- test_llm_server.py: fix import to use Handler from llm.serve
* real fix
2026-08-03 15:10:21 -07:00
chenyu and GitHub
33755a3465
improve threefry codegen [pr] ( #17379 )
...
decomp uint64 can handle part of it
2026-08-03 17:24:25 -04:00
YassineYousfi and GitHub
104ee90ccf
usb: wait for PCIe link after power on ( #17380 )
2026-08-03 14:19:05 -07:00
chenyu and GitHub
a2385ae21d
MAX_LINE_COUNT=26000 ( #17378 )
...
oh well
2026-08-03 15:37:58 -04:00
wozeparrot and GitHub
3331944547
gptoss: fix moe routing ( #17377 )
2026-08-03 11:16:44 -07:00
nimlgen and GitHub
7c1ce50f63
hcq2: epoch ( #17376 )
...
* hcq2: epoch
* x
* minor
2026-08-03 16:37:25 +03:00
qazal and GitHub
be5f62d269
llama: refactor amax stuff and skip in fp4 ( #17375 )
2026-08-03 20:05:52 +09:00
nimlgen and GitHub
e22935c758
hcq2: inputs table ( #17374 )
...
* revert this
* x
* simpler
* fix
2026-08-03 13:40:43 +03:00
chenyu and GitHub
314df72b5f
deflake test_hcq with MOCKGPU ( #17370 )
...
for MOCKGPU we compare with e2e wall time which would be device agnostic
2026-08-03 13:22:59 +03:00
chenyu and GitHub
23c7813f44
no hard coded dtype int for rangeify debuf [PR] ( #17373 )
2026-08-02 20:43:04 -04:00
George Hotz and GitHub
05bc7c6994
fix llm reasoning and Linear import ( #17372 )
2026-08-02 12:31:52 -07:00
chenyu and GitHub
09dabfe05e
minor pm_float_decomp cleanup [PR] ( #17371 )
...
make the rule order independently correct
2026-08-02 15:17:33 -04:00
chenyu and GitHub
59df317b12
use UOp.const to create new consts [PR] ( #17368 )
...
replace arg won't work with ConstArg
2026-08-02 13:31:55 -04:00
wozeparrot and GitHub
e14cadb1fb
gptoss: set ASM_GEMM ( #17363 )
2026-08-02 06:52:36 -07:00
chenyu and GitHub
0258c7fefc
minor argstr and alloc stack cleanup [PR] ( #17361 )
2026-08-01 22:37:06 -04:00
George Hotz and GitHub
fb607fb990
faster devectorizer with one line ( #17360 )
2026-08-01 13:06:35 -07:00
George Hotz and GitHub
15c936db01
merge devectorize + indexing ( #17354 )
...
* external benchmark schedule in 5 sec (codex slop)
* prune
* real?
* delete
2026-08-01 11:40:15 -07:00
wozeparrot and GitHub
98b700bad1
gptoss optim fixes ( #17356 )
2026-08-01 09:10:06 -07:00
chenyu and GitHub
8e524ca467
CONST related cleanups [pr] ( #17352 )
...
ConstFloat(nan) != nan should be False, and some Invalid bool cleanups
2026-08-01 09:54:28 -04:00
nimlgen and GitHub
665822ab34
hcq2: faster replace ( #17353 )
...
* hcq2: use runtime device for submit params
* hcq2: parameterize buffers in linear time
2026-08-01 16:32:04 +03:00
geohot
7727698486
test tiny passes
2026-07-31 23:59:17 -07:00
chenyu and GitHub
6c0ec39279
UOp.is_invalid [PR] ( #17351 )
...
helper to prep ConstArg
2026-08-01 02:36:32 -04:00
chenyu and GitHub
b502fc1367
more const arg -> val ( #17350 )
2026-08-01 02:18:10 -04:00
qazal and GitHub
161783d8f7
add _device_num back to ast.variables (kimi) ( #17327 )
2026-08-01 14:56:50 +09:00
George Hotz and GitHub
5a1c641f79
more arg -> val ( #17349 )
...
* more arg -> val
* kimi
* more
2026-07-31 22:44:24 -07:00
geohot
