mirrors/tinygrad
2
0
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-06-24 02:14:17 +00:00
Code Issues Projects Releases Packages Wiki Activity
13,427 commits 280 branches 15 tags 325 MiB
Commit graph

5,682 commits

Author SHA1 Message Date
George Hotz
2c8867c3e9 all cpu tests pass local 2026-05-26 19:28:46 -07:00
George Hotz
760acdf554 really fix llvm 2026-05-26 19:16:48 -07:00
George Hotz
f54a1d70d6 unneeded 2026-05-26 19:05:04 -07:00
George Hotz
156a4438d9
 		rename BUFFER_VIEW to SLICE (#16391) 
* rename BUFFER_VIEW to SLICE

* fix comments
 2026-05-26 18:15:00 -07:00
Christopher Milan
3adf7f5d95
 		disable flaky cl test (#16388) 2026-05-26 19:56:57 -04:00
Christopher Milan
d23659d38b
 		cleanup some old test skips (#16384) 2026-05-26 19:07:22 -04:00
George Hotz
fd963038a0
 		remove allow_any_len from store (#16385) 
* remove allow_any_len from store

* a few more

* no bv there

* more fixes

* fixes

* oh that
 2026-05-26 15:26:53 -07:00
chenyu
0b88827482
 		remove CONST(UNIQUE) (#16383) 2026-05-26 14:45:22 -04:00
chenyu
d861c50dce
 		remove unique_const (#16382) 2026-05-26 13:53:31 -04:00
George Hotz
bac82d4949
 		fix emu bug in gfx950 (#16381) 
* fix emu bug in gfx950

* fix renderer
 2026-05-26 10:32:03 -07:00
chenyu
9b00defc8c
 		Revert "remove unique_const (#16372)" (#16380) 
This reverts commit 09019d6761.
 2026-05-26 12:30:07 -04:00
chenyu
09019d6761
 		remove unique_const (#16372) 
* remove unique_const

* fix SDWA thing

* that?
 2026-05-26 12:18:03 -04:00
George Hotz
7f1b02854e
 		bufferview offset is units of input dtype (#16378) 2026-05-26 08:49:31 -07:00
qazal
846a809af7
 		viz: add +- toggle for hidden UOps (#16368) 
* first

* remove

* move src toggles to client side

* line

* update viz server tests

* remove those

* logic

* cleanup

* call matches

* fix const arg

* add labels

* keep changes

* the stack on movement ops hiding change

* structure

* rename to expandedNodes

* work

* test intention
 2026-05-26 22:31:54 +09:00
wozeparrot
76fc39ccc0
 		gather to single device (#16354) 2026-05-25 17:27:08 -07:00
Christopher Milan
8ddd1328df
 		remove getenv(CI) (#16365) 
gone everywhere except test_interop, because torch MPS does not work in actions
 2026-05-25 20:23:33 -04:00
George Hotz
689ab6a49f
 		move buffer view offset to src (#16364) 
* this work?

* failed
 2026-05-25 17:07:55 -07:00
Christopher Milan
d8f86be613
 		webgpu: shader-f16 support in arch (#16370) 2026-05-25 19:20:59 -04:00
qazal
b73d2d17b9
 		viz/cli: add --interval (#16363) 
* interval support

* add test_interval

* llama uses interval
 2026-05-26 03:35:06 +09:00
chenyu
5d5e02871f
 		remove Tensor.from_uop (#16344) 
and no device for const in Tensor init
 2026-05-24 18:53:09 -04:00
chenyu
926d125a63
 		update test_stack (#16345) 
also skip COMPILE_ONLY, it was comparing 0==0
 2026-05-23 10:42:35 -04:00
chenyu
149a87dac2
 		deviceless const cleanups (#16341) 2026-05-22 20:11:01 -04:00
Christopher Milan
451f38155c
 		start cleanup of the slowest tests (#16339) 2026-05-22 18:39:36 -04:00
qazal
bbfe4f80ec
 		quantize_fp8 kernels in uops (#16288) 
* add tests

* simple UOp kernel is n^2

* fast kernel matching c++, opts_to_apply=()

* remove cpp

* simple o(n) kernel, two passes

* fuse the loops

* works on DEV=CPU

* multi regression test

* fix multi, this can possibly be its own bugfix

* test cleanups

* minimal diff

* match C in UOps

* Revert "match C in UOps"

This reverts commit 0bef740c30.

* edit test

* match speed with C try 2

* needs_second_gpu

* cleanup
 2026-05-22 20:54:06 +09:00
chenyu
3115952266
 		more unique const removal prerequisite (#16328) 2026-05-21 23:51:40 -04:00
Christopher Milan
c2d06570a5
 		remove getenv(CI) from core tinygrad (#16326) 2026-05-21 22:20:33 -04:00
Christopher Milan
150a82de1f
 		start cleaning up dtype tests (#16324) 2026-05-21 21:11:49 -04:00
chenyu
31424cda71
 		Tensor.requires_grad -> is_param (#16325) 
for optimizer
 2026-05-21 19:39:57 -04:00
chenyu
720a27bed8
 		remove many requires_grad= args (#16321) 
* remove many requires_grad= args

* doc and example

* not cifar
 2026-05-21 18:37:11 -04:00
chenyu
73ea36f4ac
 		full(buffer=True) (#16311) 
make full a buffer with flag to turn off
 2026-05-21 16:34:44 -04:00
George Hotz
6815f28849
 		dtype.vec shapes (#16287) 
* dtype.vec shapes

* something

* Closer

* more passes

* shape is in spec

* fix reduce

* image dtype shape correct

* lil

* use reshape on image

* need BUFFER there

* remove that test

* fix ptx + x86

* fix nir

* x86 fix maybe

* x86 fixups

* x86 fix

* don't check that for NOOP
 2026-05-21 11:56:49 -07:00
George Hotz
ec547250ef
 		don't use dtype vec for image idx (#16298) 
* don't use dtype vec for image idx

* double gate

* y/x confused

* upd

* fix nir

* simplify_valid_image_load
 2026-05-20 18:45:13 -07:00
Christopher Milan
172f9493e1
 		move is_dtype_supported to renderer (#16226) 2026-05-20 21:19:37 -04:00
chenyu
beea4633fc
 		UOp.clone [pr] (#16295) 
generates the store after structure
 2026-05-20 17:47:49 -04:00
George Hotz
58d58c1659
 		remove DEVECTORIZE (#16290) 
* remove DEVECTORIZE

* fully remove DEVECTORIZE
 2026-05-20 13:25:49 -07:00
Philipp Braun
a01d5918af
 		fix: qlinearconv quant params (#16234) 
* fix: qlinearconv quant params

* fix: simplify reshape

---------

Co-authored-by: Philipp Braun <braunphilipp@users.noreply.github.com>
 2026-05-20 11:31:41 -07:00
chenyu
4dbe6a2ee7
 		remove _force_unique from Tensor init (#16277) 2026-05-20 14:13:05 -04:00
qazal
1e0fffe256
 		fused ce llama kernel in UOps (#16263) 
* work

* using uops

* delete things

* work

* work

* higher level uops

* cleanups
 2026-05-20 19:45:28 +09:00
chenyu
e1715b3b92
 		extent jit const error to deviceless inputs (#16276) 2026-05-20 02:02:45 -04:00
chenyu
170b857da9
 		clean up deviceless const _buffer (#16274) 
process on CPU similar to multi
 2026-05-19 22:47:45 -04:00
chenyu
188d7ec15e
 		clone can take device (#16271) 
useful to materialize const on a specific device
 2026-05-19 21:29:27 -04:00
George Hotz
55515747b7
 		Remove Ops.VCONST (#16267) 
* start removing vconst

* remove a lot of vconst

* const folding + strict ordering

* update tests

* spec from minigen

* move that
 2026-05-19 16:35:24 -07:00
Christopher Milan
7cdd9cbdeb
 		PYTHONREMU: V_CVT_PK_BF8_F32 saturation (#16268) 2026-05-19 19:29:59 -04:00
Christopher Milan
bb2a51f1ea
 		fix mypy mockgpu and add tinygrad.renderer.isa to packages (#16265) 2026-05-19 16:45:03 -04:00
chenyu
890b731b1e
 		more prerequisuite test changed for deviceless const (#16264) 2026-05-19 15:43:45 -04:00
ttomsa
aa1e59ab97
 		X86 with Ops.INS (#14873) 
* draft

* cleanup test_encodings

* cleanup test_isel

* model flag state and support rematerialization

* woops

* add vbroadcastss instruction

* don't fuse load if used multiple times in src

* add movabs instruction and fix idiv

* fixes

* add x86 backend to tests

* float16 fix

* rm TwoAddress2nd

* add BARRIER

* test windows ci

* yup isel fixes the mask stuff too and its beautiful

* add cmoves to the spec

* support storing imms

* no TUPLE_ORDER, breaks tests

* fix remaining seg faults

* add float max

* always fuse index

* minor

* fix DEFINE_VAR/SPECIAL and enable multithreading

* linter

* more linter

* more

* more

* more

* let's try this

* perhaps

* start new scheduler

* more scheduling info

* cleaner shuffle functions

* fixup isel tests

* skip bounds check when NOOPs exist

* skip inf rewrite tests

* fix const tag hack and add x86ops to _shape

* fix

* skip a few tests

* func arg order independent from op value

* x86 goes in own linearize

* switch to PARAM

* more

* add min x86op and neg in decomps

* do mulacc in isel

* use def_reg in test_encodings

* enable emulated int64 tests

* how much does this fix

* Ops becomes OpType

* fix

* rm noqa

* rm machine scheduler stuff

* and this

* allow for extending enums and move X86Ops out of uop

* fix imports

* rm X86GroupOp from ops.py

* spacing

* tell mypy to shut up

* more linter

* add x86op test

* allow set[X86Ops] in upat

* move NOOPs to pre_isel_matcher and rm NOOP from spec

* more asserts

* also this

* cleanup encode

* simplify live range

* fix idiv

* add Ops.INS to x86

* more changes

* more changes

* more changes

* fix

* fix

* fix

* fix

* print formatted assembly

* fix 8bit idiv?

* oops

* enable float16  and unaligned vector load/store

* actually no

* move x86 tests

* no more bool cast

* fix

* linter

* linter

* move X86Ops to x86.py

* fix vpbroadcast

* cleanups

* linter

* print correct reg names

* canonical max

* move max/min and add test

* support float16 vector load/store

* rm bad rewrite

* vpsrldq can't access memory

* regalloc takes renderer

* enable vector load/store on all dtypes

* more isel tests

* rm this for now

* a lot better

* fix

* fix

* fix

* deal with flags correctly

* fix

* enable gep noop rule

* fix

* fix

* fix

* add callee saved registers

* use Ops.CONST instead of X86Ops.IMM

* fix

* enable TUPLE_ORDER

* fix

* rm x86 code in linearizer

* fix

* fix

* fix

* move isa rewrites to codegen

* fix

* fix

* skip test_linearizer.py

* skip more tests

* fix

* fix for idiv/mod changes

* fix

* don't use fmadd if it duplicates fused op

* hacky

* fix

* cleanups

* cleanups

* fix

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2026-05-19 12:42:54 -07:00
Sachith Shetty
74567c1958
 		fix: pass input device to ONNX helper internal tensors (#16242) 
* fix: pass input device to onnx methods internal tensors

* test: onnx helper internal tensors use input device
 2026-05-19 11:16:33 -07:00
Christopher Milan
a178301dbe
 		PYTHONREMU: fix CDNA VOP3 conditional writes (#16258) 2026-05-19 13:31:31 -04:00
George Hotz
a120709671
 		tighten shape spec for broadcasting (#16206) 
* tighten shape spec for broadcasting

* use IndexError, not ValueError

* needs size
 2026-05-18 22:12:04 -07:00
George Hotz
3f2d401464
 		all tests pass with NOOPT=1 (#16257) 
* all tests pass with NOOPT=1

* fix a few more

* noopt 100% pass

* noopt 100% pass
 2026-05-18 20:39:51 -07:00