George Hotz
|
5d13dd1123
|
refactor to u.
|
2026-05-29 19:05:20 -07:00 |
|
George Hotz
|
058259e0ea
|
refactor to use uop
|
2026-05-29 19:01:10 -07:00 |
|
George Hotz
|
da9318e745
|
python prg is pkl uops
|
2026-05-29 18:54:55 -07:00 |
|
Christopher Milan
|
c377d01491
|
ci: run dsp on tinygrad[testing] (#16442)
|
2026-05-29 21:16:56 -04:00 |
|
wozeparrot
|
c23652e486
|
llama: minimize peak init mem (#16440)
|
2026-05-29 18:00:37 -07:00 |
|
Christopher Milan
|
d943493b79
|
ci: remove duplicate op compile test (#16441)
|
2026-05-29 19:20:31 -04:00 |
|
chenyu
|
8ac62b28e5
|
fix AffineGrid fusion (#16439)
|
2026-05-29 17:59:47 -04:00 |
|
Christopher Milan
|
ef50a49693
|
ci: macos dev matrix (#16436)
|
2026-05-29 17:40:32 -04:00 |
|
Christopher Milan
|
434cfa96a3
|
ci: no fetch in backend tests (#16438)
should make for less actions cache thrashing
|
2026-05-29 17:11:16 -04:00 |
|
chenyu
|
b7280705a7
|
limit CONST(UNIQUE) to invalids only (#16432)
|
2026-05-29 16:02:06 -04:00 |
|
George Hotz
|
9506b78d73
|
fix viz addrspace (#16437)
* fix viz addrspace
* revert that
|
2026-05-29 12:58:05 -07:00 |
|
nimlgen
|
d69aca41a9
|
hcq2: rework pm_bufferize (#16431)
|
2026-05-29 22:09:52 +03:00 |
|
George Hotz
|
e2a0434403
|
full derivation of addrspace (#16433)
* full derivation of addrspace
* w/e, it fixes it
|
2026-05-29 11:39:31 -07:00 |
|
wozeparrot
|
6787de9f52
|
llama: fix mp (#16434)
|
2026-05-29 11:21:43 -07:00 |
|
chenyu
|
2d7e5baab4
|
remove vec= from UPat.cvar [PR] (#16430)
|
2026-05-29 10:52:30 -04:00 |
|
chenyu
|
fa666cefe8
|
remove dead branch in UOp [PR] (#16429)
|
2026-05-29 10:38:49 -04:00 |
|
qazal
|
81bc00c006
|
do not require clearing method_cache in viz tests (#16428)
* update
* update test_dedup
|
2026-05-29 18:12:34 +09:00 |
|
qazal
|
54cfb794b8
|
viz: addrspace little colored box (#16427)
* return addrspace
* layout
* render
* addrspace encodes color
* update colors
* in input_ast all are params are green
* update stroke
|
2026-05-29 17:25:07 +09:00 |
|
qazal
|
814d414f41
|
viz: set label offset for asm (#16426)
|
2026-05-29 13:16:34 +09:00 |
|
wozeparrot
|
f86966af56
|
llama: optim amax margin (#16425)
|
2026-05-28 20:18:11 -07:00 |
|
Christopher Milan
|
6e0d5262dc
|
ci: autocancel outdated pr jobs (#16424)
|
2026-05-28 23:14:35 -04:00 |
|
Christopher Milan
|
69aa2054f6
|
rename clangjit to clang (#16423)
|
2026-05-28 22:41:58 -04:00 |
|
Christopher Milan
|
a909acb882
|
move llvmspeed to benchmarks (#16422)
|
2026-05-28 22:26:22 -04:00 |
|
George Hotz
|
1e7f1dcf49
|
add ParamArgs [pr] (#16421)
* add ParamArgs
* fix export
* cleanups
* fixes
* simpler
|
2026-05-28 19:17:17 -07:00 |
|
Christopher Milan
|
7d38edffdb
|
ci: dev matrix (#16420)
windows just runs test_tiny
|
2026-05-28 22:04:04 -04:00 |
|
wozeparrot
|
36c8ff70c1
|
llama: use old scale for dequant in optim (#16417)
|
2026-05-28 15:21:19 -07:00 |
|
George Hotz
|
c87f3433d1
|
use namespace runners (#16387)
Co-authored-by: Christopher Milan <chrismilan@ucla.edu>
|
2026-05-28 18:05:46 -04:00 |
|
George Hotz
|
c9adde72c1
|
addrspace property (#16418)
* addrspace property
* movement addrspace
* regs
|
2026-05-28 14:39:25 -07:00 |
|
Christopher Milan
|
c8af163d2b
|
disable process replay by default (#16419)
enable process replay with [pr] and assert with [PR]
process replay no longer captures on master
|
2026-05-28 17:36:28 -04:00 |
|
nimlgen
|
b0e49afaf1
|
hcq2: new multi (#16413)
* hcq2: new multi
* op
|
2026-05-28 22:16:10 +03:00 |
|
George Hotz
|
edca5df25a
|
flip offset and shape in pad and shrink (#16414)
* flip offset and shape in pad and shrink
* dumb test
|
2026-05-28 11:58:19 -07:00 |
|
chenyu
|
d72d8ee065
|
.const() should not ignore dtype (#16412)
fixed a bug in postrange, also cleaner
|
2026-05-28 10:49:15 -04:00 |
|
Christopher Milan
|
0ae957bb0a
|
refactor webgpu (#16406)
|
2026-05-27 23:13:08 -04:00 |
|
qazal
|
202adc644e
|
viz: make call toggle easier to click on (#16411)
* call tag is a rect
* details
* colors
* simplify, better comment
|
2026-05-28 11:53:36 +09:00 |
|
George Hotz
|
5ee6b6b79e
|
fix slice store to remove the index (#16410)
* fix slice store to remove the index
* fix spec
|
2026-05-27 19:17:53 -07:00 |
|
qazal
|
88e88d63d6
|
viz: click on +- toggles sources (#16409)
|
2026-05-28 09:12:43 +09:00 |
|
George Hotz
|
b21afb4883
|
marg line cleanup (#16408)
* marg line cleanup
* bitcast is a mop
|
2026-05-27 16:41:04 -07:00 |
|
wozeparrot
|
dac3743d75
|
llama: delayed scaling in optim (#16407)
|
2026-05-27 15:40:03 -07:00 |
|
George Hotz
|
8ee3a37524
|
shrink/pad use (new_shape, offset) (#16405)
* shrink uses offset and shape
* pad does too
* fix
|
2026-05-27 15:13:08 -07:00 |
|
Christopher Milan
|
171401e8df
|
skip modulo by zero in test_dtype_alu (#16404)
|
2026-05-27 17:09:05 -04:00 |
|
qazal
|
452c7d4230
|
llama: don't allocate grad_xw13 in bf16 (#16359)
|
2026-05-28 04:33:07 +09:00 |
|
nimlgen
|
0c385e31c6
|
hcq2 rewrite (#16375)
* hcq2 rewrite
* fi
* x
* simpler
|
2026-05-27 22:25:35 +03:00 |
|
chenyu
|
c33b767407
|
bring back test and torch backend change for unique const (#16403)
|
2026-05-27 15:16:08 -04:00 |
|
Christopher Milan
|
bacabf0866
|
webgpu: fix enums (#16402)
|
2026-05-27 13:09:50 -04:00 |
|
chenyu
|
6da785562b
|
test_custom_kernel_precompile_multidevice (#16401)
add a test to show what invalids need
|
2026-05-27 11:19:16 -04:00 |
|
chenyu
|
3e80f375ee
|
skip test_setitem_fancy_on_unrealized_view (#16400)
crashes in linux llvm ci
|
2026-05-27 09:50:26 -04:00 |
|
chenyu
|
945ed4f689
|
revert const unique changes (#16395)
|
2026-05-27 00:06:41 -04:00 |
|
Christopher Milan
|
aacc8addf4
|
ci: use ubuntu 24.04 (#16393)
|
2026-05-26 23:22:01 -04:00 |
|
chenyu
|
fa14cde05c
|
test update for arange and eye (#16394)
these will need explicit clone to make a buffer
|
2026-05-26 22:48:34 -04:00 |
|
wozeparrot
|
3a7a6da7d5
|
llama: fakedata uses real vocab size (#16389)
|
2026-05-26 18:58:55 -07:00 |
|