github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-01-25 23:08:06 -05:00
Code Issues Packages Projects Releases Wiki Activity
2,585 Commits 205 Branches 14 Tags
dece9958f87ce415c31304674078e3e6dbdb4582
Commit Graph

2585 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Francis Lam
dece9958f8 wmma: clean up to make WMMA arg order consistent (#2014) 
also add cache defeat to extra/gemm/simple_matmul.py
 2023-10-07 17:45:40 -07:00
George Hotz
cea4cbfc7a move image+kopt to features (#2015) 
* move image+kopt to features

* fix tests

* debug prints (unrelated)
 2023-10-07 15:41:08 -07:00
George Hotz
44ed94ef5c use the device abstraction in handcode_resnet50_opt 2023-10-07 13:22:20 -07:00
George Hotz
6ee9cae44f don't extract CIFAR every time / use the cache 2023-10-07 12:33:50 -07:00
nimlgen
d07ac379f9 add var_vals to kopt with symbolic (#2008) 
* add var_vals to kopt with symbolic again

* no copies
 2023-10-07 09:34:21 -07:00
George Hotz
121f7aa8c5 Schedule item (#2012) 
* ScheduleItem

* put var_vals in the schedule

* fix tests, wow that proliferated quickly

* not ready to be in the schedule
 2023-10-07 08:59:25 -07:00
George Hotz
f1f64bc88d remove val_vars from the linearizer (#2009) 
* remove val_vars from the linearizer

* no need to store var vals
 2023-10-07 07:47:28 -07:00
George Hotz
dea8bb0938 triton isn't tested, and allows this refactor (#2007) 
* triton isn't tested

* cuda buffer
 2023-10-07 07:29:59 -07:00
George Hotz
23de1db727 strip whitespace 2023-10-07 06:06:27 -07:00
Roelof van Dijk
26fcc8dff6 fix: remove runtime imports (#1982) 
fix: import what is used

probably monkeypatched

fix: import

revert selective import
 2023-10-07 05:23:08 -07:00
George Hotz
f54959e5cd move print tree into graph (#2003) 
* move print tree into graph

* add winograd profiling test

* change pre-commit to run ruff first
 2023-10-07 04:39:21 -07:00
Ahmed Harmouche
2114dc13d1 Allow multi-input model export (#1995) 
* Allow multi-input model export

* Add model export unit test

* Fix efficientnet compilation

* Only run model export test on JIT supported devices

* Skip export model test if not EXPORT_SUPPORTED_DEVICE
 2023-10-07 04:13:34 -07:00
George Hotz
ffa33d743a good changes from openpilot_compile2 (#2000) 
* good changed from openpilot_compile2

* float32 image type was wrong

* cleaner way to write that + a test
 2023-10-06 13:33:24 -07:00
chenyu
05be57f57f Fix llama with empty prompt (#1997) 
* fix llama with one token prompt

* llama is all_jitted
 2023-10-06 06:48:07 -07:00
George Hotz
7a68060422 Revert "allow local + grouped reduce in hand_coded (#1996)" (#1998) 
This reverts commit 219a1f7063.
 2023-10-06 06:43:28 -07:00
nimlgen
219a1f7063 allow local + grouped reduce in hand_coded (#1996) 
* allow local + grouped reduce in hand_coded

* allowed loop size based on global_dims

* fix const

* fix const one more time

* better divisor

* a bit fix

* can take 2, why not

* fix linter

* better comments

* start with 2

* not always pick group reduce

* fix images

* better images

* better
 2023-10-06 06:11:28 -07:00
George Hotz
fa9945dac0 remove stale tests 2023-10-06 02:14:56 -07:00
Vidhan Bhatt
94b21c41a7 ci: use mypy.ini (#1993) 2023-10-06 01:45:28 -07:00
George Hotz
e43d8977f8 Revert "chore: add py.typed marker. (#1991)" (#1994) 
This reverts commit 6d581e8911.
 2023-10-06 01:44:34 -07:00
Vidhan Bhatt
6d581e8911 chore: add py.typed marker. (#1991) 
* chore: add `py.typed` marker.

* fix: add comma
 2023-10-05 16:27:33 -07:00
chenyu
da2b3e55f4 simpler llama - don't shrink twice (#1981) 2023-10-05 14:31:46 -07:00
Roelof van Dijk
972d9ea215 fix: PRUNEGRAPH is unused (#1985) 2023-10-05 14:28:43 -07:00
George Hotz
21a2c5df73 fix up contiguous (#1978) 2023-10-05 07:22:05 -07:00
chenyu
c99fa58dd2 simplify gpt2 example (#1973) 
* simplify gpt2 example

* kernel_jitted_count and jit tests

* Revert "kernel_jitted_count and jit tests"

This reverts commit 31a3c26dd0.

* all_jitted test in test_real_world
 2023-10-05 07:09:29 -07:00
George Hotz
2d0c1037b1 Fix up latest openpilot model (#1976) 
* fix gemv triggering for gemm

* fixup_openpilot

* external test issues
 2023-10-05 05:24:28 -07:00
George Hotz
1862e14a4f fix gemv triggering for gemm (#1975) 2023-10-05 05:23:00 -07:00
Francis Lam
0ba75c4370 optimizer: add matvec optimizations (#1972) 
* optimizer: add matvec optimizations

* renderer: fix alignment of shared memory in opencl
 2023-10-04 14:16:27 -07:00
George Hotz
3d5127038c don't create linearizer if we are in the method cache (#1969) 
* don't create linearizer if we are in the method cache

* remove unchecked properties

* that key isn't used

* fix default type is sticky
 2023-10-04 12:42:58 -07:00
George Hotz
de5d603ec1 corealize + remove realize from lazybuffer (#1968) 
* corealize + remove realize from lazybuffer

* fix multigpu

* fix graph
 2023-10-04 10:59:31 -07:00
George Hotz
88b6ed6945 disable broken optim_conv2d 2023-10-04 07:33:50 -07:00
George Hotz
d449b3bef1 think about removing realize from lazybuffer (#1965) 
* remove realize from lazybuffer

* okay fine, back that off

* fix tests maybe

* fix test
 2023-10-04 07:18:58 -07:00
nimlgen
2ea1dd3e87 no process() in Linearizer (#1966) 
* no process() in Linearizer

* more process() clean up
 2023-10-04 07:18:42 -07:00
George Hotz
0945848b5f schedule the loadops like everything else (#1964) 
* schedule the loadops like everything else

* unify loadops with other things we schedule

* delete all the ops

* fix symbolic jit
 2023-10-04 02:36:04 -07:00
Ahmed Harmouche
fb4d830a2a Fix cast error in render_load in wgsl (#1956) 
* Fix cast error in wgsl

* User render_cast intead of introducing new method

* Make it shorter

* Add back webgpu tests: efficientnet and dtypes
 2023-10-04 02:29:14 -07:00
George Hotz
6a79d4044a unrealized consts everywhere (#1963) 
* unrealized consts everywhere

* don't import device from lazy

* Device isn't in Lazy

* same issue

* disable jit random
 2023-10-04 01:48:10 -07:00
nimlgen
f04c1a63ae Rand works in jit (#1960) 
* rand works in jit

* better jitted rand creation

* Update realize.py

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2023-10-03 12:55:25 -07:00
George Hotz
f64d5b3ba8 move to realize.py (#1961) 
* move to realize.py

* run_schedule moved
 2023-10-03 07:25:40 -07:00
George Hotz
717451a244 Revert "optimizer: add matvec optimizations (#1753)" (#1959) 
This reverts commit f520323054.
 2023-10-03 00:28:42 -07:00
Francis Lam
f520323054 optimizer: add matvec optimizations (#1753) 
* optimizer: add matvec optimizations

* Update optimizer.py

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2023-10-03 00:01:59 -07:00
nimlgen
e1f2c2cc19 fix jitted dist (#1955) 2023-10-02 11:45:13 -04:00
Roelof van Dijk
35ac60775b simplify line (#1950) 
* no need to index here, zip automatically truncates

* enumerate is faster

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2023-10-02 03:19:15 -07:00
nimlgen
08e884217c metal batch executor (#1920) 
* metal batch executor

* no sym_infer in backends

* calc_stat in BasicBatchExecutor`

* run in batches of size 8

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2023-10-02 03:18:31 -07:00
George Hotz
d48a90859c use the opts from the default device (#1954) 2023-10-02 03:13:46 -07:00
nimlgen
c27971d51f fix llvm nan/inf const (#1951) 
* allow llvm

* llvm works with inf/nan

* enable some fast math back

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2023-10-02 03:08:57 -07:00
George Hotz
6a4ec4776e fix CI (#1953) 
* this work

* unauth

* update in all places
 2023-10-02 02:58:58 -07:00
Daniel Riege
579cabf668 Fix examples/train_efficientnet (#1947) 
* added missing colon

* bug fixes for cifar10 dataset loading
needed a reshape to work with conv layers and resolve fetched tensor to numpy since further code expects numpy array
 2023-10-02 02:23:38 -07:00
David Hou
d4671cd8e3 use schedule in more places in linearizer tests (#1946) 
* pass current linearizer opts to Linearizer in TestFloat4

* use schedule instead of exec_ast hook
 2023-10-02 02:22:56 -07:00
Roelof van Dijk
e7a49e84c8 perf: assert behind if is not optimized (#1847) 
* perf: assert behind if is not optimized

* Update helpers.py

---------

Co-authored-by: Roelof van Dijk <roelof.van.dijk@vitestro.com>
Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2023-09-29 11:07:24 -07:00
David Hou
8e9db88474 expand after expr_idxs in Linearizer.global_load (#1818) 
* small changes

* expand in terms of substitute, directly expand g_idxs g_valid

* delete expand_ops

* don't compare using hash

* any instead of in

thanks gijskoning

Co-authored-by: Gijs Koning <gijs-koning@live.nl>

* support tc

* testing code

* no more create_rednode

* maxsize none in view/node

* oops

* undo

* typing

* oops

* oops

* lmao

* lmao

* add expand multi test

* Node.iter_idxs

* type

* type

* delete checks!

* clean up a little?

* expand_idx in symbolic

* un-golf

* play around with types >.>

* test_substitute and also remove an incorrect test?

* get rid of range

* Update symbolic.py

* split out view cache change

* split out flat components change

* reduce diff

* reduce diff

* add some float4 tests

* fix

---------

Co-authored-by: Gijs Koning <gijs-koning@live.nl>
 2023-09-29 10:33:34 -07:00
nimlgen
692bec7b6f simplify CacheCollector (#1944) 
* rewrite cc

* fix

* fix tests

* fix all tests

* is it better

* better with shape

* cleaner

* linter fix

* no ;

* better comment

* better comments

* no thneed changes
 2023-09-29 10:13:04 -07:00