Commit Graph

Select branches

Hide Pull Requests

abstractions_4

add_local_buffer

after_qol

algebraic_upat

all_bv_no_idx

all_sink

amd_isel

amd_refac

amd_smu_reset_patch

amd_sqtt

amd_uop

amd_uop_cleanups

assign_copy

assign_try_3

before2

better_spec

better_syn

bitcast_spec

broadcast_shape_2

bt_ei

buf_heur

bump_nv_2404

cache_reshape

call_inline

call_is_kernels

callify

cast_vec_as_cast

cat_mop

changes_from_rangify

check_clsetarg

check_dims

cifar_30

cifar_30_clean

clean_load

clean_spec

clone_tg

codegen_try_2

contig_on_after

continue_mat

deepsek

default_late_allreduce

delete_kernel_py

delete_ones

delete_slow_rangeify

delta_attention

delta_attention_scan

demote_op

disable_rules

dsp_search

dsp_search_merged

dtype_shape

earlier_gater

emu_cdna

emulate_ctx

end_all_ranges

endrange

er_prioity

execution_unit

failing_assign

fancy_usb

fast_llm

fix_fa_fwd_clang_22

fix_fuse

fix_llvm_cast

fix_range_merging_sd

fix_rangeify_tests

fix_rc

fix_sched_copy

fix_slice_store

fix_src_spec

fixup_fix_kernel

full_spec

fun_w_egraphs

gate-on-load-store

gate_rewrite

gemma4_gpt

gen_pdf_fast

gh-pages

gl_dims

gldims

glm_flash

grad_outerworld

grok_x86_novec

group_div_rules

hcq2_vram_copies

heur_4096

heur_group

hk_gemm

http_threading

hwtest_fixes

id_store

if_range_cleanups

ignore_oob

image_idx_xy

image_in_mc

image_no_vec

image_refactors_stack

index_to_slice

invalid_try_3

jinja_template

kernel4

kernel_is_call

kernelless

kimi_slop

kitten_matmul

late_Locals

late_add_load

late_allreduce

late_codegen

late_decomp_is_late

late_lower_idx

late_view

late_weakint

lds_double_buffering

less_wmma_arg

lil_dg_change

lil_rangeify_td

llama_trainer

llm_262k_minimal

llm_flash_attn

llm_qwen

llm_serve_fixes

load_view_late

local_cache

locals_have_device

loop_end_op

loop_is_rng

lowerer_hacks

mac_pytest

manual_kernel_3

master

mega_lowerer

min_new_lin

min_outer_range

minor_views

mixins7

mlperf_training_v4.1

mlperf_training_v5.0

mlperf_training_v6.0

modern_scan

more_apps

more_early_comps

more_folding_rules

more_kernelcount

more_pcode

more_rangeify_pm_tests

more_ren_clean

more_uop_gc_test

more_uop_mm

more_vmap

move_devec

move_dtype_decomps

move_expander

move_gates_to_load_store

move_mop_cleanup

move_schedule

move_simp_views

moveleftright

mt_typing

multi_range

multioutput

new_ck

new_devec

new_invalid_and_image

new_linearizer

new_shape

new_test

no_assign_2

no_buf_0

no_decimals

no_late_simplify_marg

no_merge_ends

no_merge_views

no_mstack_in_allreduce

no_one_ranges

no_sip_usbgpu

no_tsink_base

no_tuplize

no_user_ck_contig

no_vec_x86_4

no_weakint_variable

noop_index

novecimg

num_batches_tracked_shape

olmo3_support

one_compiler

one_hcopt

only_assign_buffers

only_loop_backends

only_reg_emu

only_reg_emu2

outer_range

outerworld_work

pad_invalid_work

pad_mixin

param_call_is_function

parse_sqtt_more

pci_iommu

port_nir

postrange_hcopts

precompile_backward

prefill_qwen_fast

print_special_ops

ps_clean

pyremu_try

python_prg

python_speed

q4k_support

qcom-openpilot-13ms

qcom_830

qcom_mmapeak

qcom_speed_codex

qol_string

qwen36_27b_amd_900

qwen36_27b_amd_minimal

qwen36_amd_bugfixes

qwen_mergable

r_cleanups

r_new_opt

ranged_call

rangeify_2

rangeify_const_folding

rdna3_vibes

rdna4_asm

rdna4_gemm

real_pyrender

recurse_sub

reg_coalese

relu_pattern

remove_assign

remove_cast_hack

remove_ck_contig

remove_define_var

remove_forced_re

remove_on_stack

remove_programspec

remove_pylint

remove_slow_tests

remove_vec_2

remove_vec_3

renderers_use_shapes_2

replace_if_with_range

reprocess_node_experiments

reshape_trait

resnet-ci

revert-11547-fix_mismatch_red

revert-11715-revert-11711-amd_llvm_tc_cast

revert-11809-no_ki_in_gpudims

revert-11811-better_div_nesting

revert-11876-memplan_vbig

revert-12104-simplify_valid_in_symbolic

revert-12308-bufferize_early

revert-12731-jitlink_init_check

revert-12813-after_cleanups

revert-13046-better_cleanup_arange_buffers

revert-13311-prof_link_sqtt

revert-13583-rocr_desc_2

revert-14734-tload

revert-17141-full_rewrite_to_sink

revert-17408-_broadcasted-weak

revert-9438-simple_matmul_np_rand

revert_wrong_ck_fix

rewrite_rangeify2

rtoposort

scan_assign

scan_op_8

sched_cache

sched_lin

sd_load_simple

self_type

set_testpath

shrink_in_render

simp_spec

simple_gidxs

simple_pool

simple_priority

simple_rangeify

simpler-sd-compile

simpler_end

simpler_fusion

simpler_image

simpler_postrange

simpler_reduce

slice_like_shrink

spec_3

spec_cleanup

spec_ren

speed_one_linear

split_buf_idx

split_decomp_pass

split_sub

stack_0

store_is_void

string_ranges

support_opts_in_contig

swizzle_in_kernel

sym_llm

sym_work

symbolic-chunk-split

szdiff_bot_test

tc_is_heur

tensor_graph_single_sink

test_amd_crash

test_fa

test_from_postopt

test_hgemm

test_unit

tiny_scan

tmr

tool_call_support

topoprop

ttomsa-new_linearizer

tuplegtuple

two_stage_remove

typed_checks

typing_work

unused_permute_arg

uop_prg

upd_test4

update_benchmark

update_benchmark_search

update_mlperf

user_failure

vecless_x86_glm

verify_sink_dims_clean

vf_ish

view_after_const

view_in_codegen

viz_cbroadcast

warp_fun

wgpu-f16-packed

working_postopt

x86_moves

x86_numel

v0.10.0

v0.10.1

v0.10.2

v0.10.3

v0.11.0

v0.12.0

v0.13.0

v0.4.0

v0.5.0

v0.6.0

v0.7.0

v0.8.0

v0.9.0

v0.9.1

v0.9.2

Mono Color

• 0055f0c2b3 touchups George Hotz 2023-02-28 20:36:11 -08:00
• f4aa3868e3 remove save_for_backward, there's still lines to save in the simplest places George Hotz 2023-02-28 20:11:05 -08:00
• 7ff92550bb slice -> pad, shrink George Hotz 2023-02-28 19:57:45 -08:00
• ea3fa07c2a bump tinygrad to 0.5, move reshape logic from mlops George Hotz 2023-02-28 18:07:03 -08:00
• e9e71fbfc4 remove mlop (#619) George HotzandGitHub 2023-02-28 17:58:24 -08:00
• 6b423b675d fix mulacc when both strides are 0 George Hotz 2023-02-28 17:27:52 -08:00
• 4c4d88aad4 fix the last bug, and make HLOP the default George Hotz 2023-02-28 17:04:28 -08:00
• fde6c2d62b fix image grouping George Hotz 2023-02-28 16:50:46 -08:00
• 17c55f051d fix test symbolic George Hotz 2023-02-28 16:37:08 -08:00
• 28f52f7c24 improve symbolic George Hotz 2023-02-28 16:21:58 -08:00
• 1702a5779f remove hacks from can_merge George Hotz 2023-02-28 15:30:20 -08:00
• e21df1701b distribute + refactor merge_views George Hotz 2023-02-28 14:57:56 -08:00
• 7e6edfbc64 unbreak onnx conv padding George Hotz 2023-02-28 13:55:03 -08:00
• 7d556ca7e0 avg/max pool work in N-D George Hotz 2023-02-28 13:38:27 -08:00
• dcb50a3a9f better hlop image conv George Hotz 2023-02-28 13:06:58 -08:00
• 9d539b8ebb more intuitive output shape from _pool George Hotz 2023-02-28 11:41:48 -08:00
• d722ffbd04 _pool2d -> _pool George Hotz 2023-02-28 11:35:19 -08:00
• 3c8da6bd03 add typing George Hotz 2023-02-28 10:54:46 -08:00
• 922f96e527 DeviceBuffer : shape can be correct type now George Hotz 2023-02-28 10:00:48 -08:00
• a8bbcccc16 debug print shapetrackers geohot 2023-02-28 08:11:40 -08:00
• cfa5a12f13 simplify in shapetracker geohot 2023-02-28 00:34:49 -08:00
• 8478a61cdb simplify in shapetracker geohot 2023-02-28 00:31:12 -08:00
• 262f81d795 applegpu everywhere geohot 2023-02-27 22:54:59 -08:00
• d584bae5c0 fine, openpilot can have 197 kernels geohot 2023-02-27 11:48:36 -08:00
• 7b999add1d all onnx model tests pass geohot 2023-02-27 11:22:45 -08:00
• 652d48ccec onnx : openpilot expand issue was fixed yesterday. remove hack geohot 2023-02-27 11:04:42 -08:00
• 9d6b63f043 add ConstantOfShape geohot 2023-02-27 10:57:50 -08:00
• 082134952b CastLike works with one type hack geohot 2023-02-27 10:51:26 -08:00
• 1ffe8d68d5 Add more onnx ops (#615) Jacky LeeandGitHub 2023-02-27 10:43:41 -08:00
• 643e8b0388 fix tests, test bn evaluate too geohot 2023-02-27 10:39:47 -08:00
• 2f17d151b3 fix batchnorm not realizing geohot 2023-02-27 10:19:54 -08:00
• c9252d38b2 mypy cache breaks if you sometimes check untyped defs, no checking tests for now geohot 2023-02-27 09:57:33 -08:00
• e74779f19d typing fixup geohot 2023-02-27 09:52:04 -08:00
• 9aaa7edd74 no rnn support in onnx geohot 2023-02-27 08:25:48 -08:00
• 07e643431c added onnx group norm (#614) DiogoandGitHub 2023-02-27 11:11:01 -05:00
• edc8fbfff2 woah, why isn't OPT=2 geohot 2023-02-27 08:03:31 -08:00
• a52913b242 test conv shapetracker has one view geohot 2023-02-27 07:54:47 -08:00
• 57aec11736 minor randomness touchup geohot 2023-02-27 07:03:08 -08:00
• f10ccf7ec1 fix up eye and fix gc test geohot 2023-02-27 06:53:18 -08:00
• 686a74de92 fast zeros and ones geohot 2023-02-27 06:46:26 -08:00
• e68fa18c9b layer norm support in onnx (#607) DiogoandGitHub 2023-02-27 01:04:02 -05:00
• c4856aa193 fix yolo webcam geohot 2023-02-26 17:24:05 -08:00
• 3a2a500e90 prevent race condition, external yolo test for now geohot 2023-02-26 17:08:24 -08:00
• 0f58c4c648 Cleanup yolo and remove stateless classes (#604) Jacky LeeandGitHub 2023-02-26 16:55:21 -08:00
• 71ae6e5605 fix: avgpool without counting padding (#605) SohaibandGitHub 2023-02-26 16:13:00 +01:00
• 2b6d8e788d pushing permutes is done geohot 2023-02-25 21:37:20 -08:00
• f4ee7d2cad back to 196 kernels geohot 2023-02-25 18:25:34 -08:00
• 187738f64f test fix, wow fixing contig exposed so much geohot 2023-02-25 18:01:14 -08:00
• c3d19101cb more debug info geohot 2023-02-25 17:52:46 -08:00
• 6e98a172a0 fix broken contiguous geohot 2023-02-25 17:41:49 -08:00
• a44e8e4385 discard children on mop shuffle, 200 -> 196 kernels geohot 2023-02-25 10:51:07 -08:00
• 1d01842232 remove fake test geohot 2023-02-25 10:21:07 -08:00
• 8b96522e1d instant identity removal geohot 2023-02-25 09:46:04 -08:00
• a8de233e12 only div, no reciprocal (#601) George HotzandGitHub 2023-02-25 09:35:03 -08:00
• d581a99d90 onnx: lrn (#602) 3 people 2023-02-25 18:24:53 +01:00
• 94bec40110 moved extras/jit.py -> tinygrad/jit.py (#599) voidzandGitHub 2023-02-25 22:02:33 +05:30
• 7348e9a6c6 add restrict qualifier to inputs in c backend (#593) Benedikt MandelkowandGitHub 2023-02-25 17:32:21 +01:00
• 2c5e13a513 Reluless (#600) George HotzandGitHub 2023-02-25 01:21:16 -08:00
• 176ad29974 retain support for old onnx George Hotz 2023-02-24 22:29:54 -08:00
• f3386c7f09 improve symbolic, hlop conv output is simple now George Hotz 2023-02-24 22:20:40 -08:00
• 9b5b17db54 shapetracker : always 0 out the strides if the shape is 1 George Hotz 2023-02-24 21:55:38 -08:00
• f8f026e8bb oversized expand for HLOP convs George Hotz 2023-02-24 21:48:47 -08:00
• 2edfe64512 improve shapetracker tests George Hotz 2023-02-24 21:07:53 -08:00
• 6d97d62ab3 Add PyCharm's .idea to .gitignore (#597) Marcello FuschiandGitHub 2023-02-25 05:14:38 +01:00
• da5643d024 rest of tests shouid be made to pass geohot 2023-02-24 12:52:23 -08:00
• 85452fbaf3 onnx 58/109/208 geohot 2023-02-24 12:19:05 -08:00
• e8a153e4e9 onnx : add a whole bunch of ops geohot 2023-02-24 12:00:03 -08:00
• f2486a7248 more onnx ops geohot 2023-02-24 10:55:58 -08:00
• 4d0a3dd653 openpilot expand is bugged geohot 2023-02-24 10:25:59 -08:00
• 2e56a4793e rename log_softmax, support dim, fix onnx Softmax geohot 2023-02-24 10:11:24 -08:00
• 5cdfeffe2c fix shape test geohot 2023-02-24 09:36:32 -08:00
• 3becefa218 fix onnx tests geohot 2023-02-24 09:27:18 -08:00
• e263c0c628 onnx : another model test is passing geohot 2023-02-24 09:22:58 -08:00
• d3feea302d much cleaner way to write onnx ops geohot 2023-02-24 08:46:28 -08:00
• d3029c91c5 no rng for op test geohot 2023-02-24 00:23:20 -08:00
• f6d946853c more bugfixes geohot 2023-02-24 00:21:29 -08:00
• b1b2d8f440 onnx : some op tests working geohot 2023-02-23 23:58:13 -08:00
• b287b1d529 fix yolov8 to get to ConvTranspose geohot 2023-02-23 22:46:48 -08:00
• 2d59b25ead onnx backend test : enable only the model tests geohot 2023-02-23 22:36:26 -08:00
• 5b10dfcab8 onnx tests : 22/175/208 geohot 2023-02-23 22:00:16 -08:00
• d8b6f241f1 external_test_onnx_backend geohot 2023-02-23 21:55:07 -08:00
• edaf878339 put rng with creation ops geohot 2023-02-23 20:38:36 -08:00
• 8137e0b400 typo in __imatmul__ -> cleanup geohot 2023-02-23 20:12:55 -08:00
• 4c54adeb18 adding tuples is fine geohot 2023-02-23 19:42:48 -08:00
• 661812ffef don't ignore type geohot 2023-02-23 19:38:22 -08:00
• 10c6ccf7e0 better order in Tensor class geohot 2023-02-23 19:33:37 -08:00
• f2ca81c66d reduce cleanup geohot 2023-02-23 19:19:49 -08:00
• 8446f77fe9 image_conv2d_decorator geohot 2023-02-23 18:49:23 -08:00
• 82caa2d5b7 MULACC -> FusedOp geohot 2023-02-23 18:17:57 -08:00
• 758515dcc0 conv2d is an hlop (#589) George HotzandGitHub 2023-02-23 17:52:31 -08:00
• 8835df7a5c upgrade onnx to 1.13.0 (#588) 3 people 2023-02-23 22:59:23 +01:00
• 81f7c6800a Added info on simdgroup availability (#586) calleditandGitHub 2023-02-23 22:59:02 +01:00
• d22e19536b onnx: support low quality Resize. stuck on ConvTranspose will have to wait for convless geohot 2023-02-23 09:05:23 -08:00
• ab3a2ae9a2 fix test_resnet in onnx now that maxpool works geohot 2023-02-23 08:41:47 -08:00
• fd6082dcef support all _pool2d. conv will eventually be an hlop geohot 2023-02-23 08:19:47 -08:00
• 94ccab941e compile_tensorflow: no cast required George Hotz 2023-02-22 21:14:21 -08:00
• 135d0ddb78 compile_tensorflow: read weights from disk George Hotz 2023-02-22 21:12:35 -08:00
• 0615dcffe7 compile_tensorflow: save the weights George Hotz 2023-02-22 21:05:45 -08:00
• c537fd0614 compile_tensorflow: add initialize and tests George Hotz 2023-02-22 20:50:53 -08:00
• dc914cde50 compile_tensorflow George Hotz 2023-02-22 20:08:58 -08:00