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

• 537a9eb414 fix termcolor import geohot 2022-11-07 21:19:08 -08:00
• 2cc1d970c6 updates from the chonker branch geohot 2022-11-07 21:12:08 -08:00
• d878065ece Gemm (#416) 3 people 2022-11-06 10:07:28 -08:00
• caea34c529 1s are always mergable geohot 2022-11-03 10:50:48 -07:00
• c48fc47d01 fix type error geohot 2022-10-31 09:56:56 -07:00
• 9585b6c0cf comments and readability in lazy.py geohot 2022-10-30 19:50:48 -07:00
• db2da22a04 stop blowing up floats geohot 2022-10-30 16:47:16 -07:00
• 8afc643bb1 fix bug in ops test, it was cheating somehow geohot 2022-10-30 16:43:24 -07:00
• b7a115e5e5 rewrite some strideds into reshapes geohot 2022-10-30 16:31:27 -07:00
• 8c849e637c that was in there twice, DEBUG>=4 to see loop opt geohot 2022-10-30 15:31:39 -07:00
• cfdf803b52 fix llvm vectorization by add analysis passes from the target machine geohot 2022-10-30 15:28:36 -07:00
• 2f602a92ff seperate STRIDED and EXPAND geohot 2022-10-30 13:23:58 -07:00
• 544cb0a069 oops, remove while(1) geohot 2022-10-29 14:05:13 -07:00
• 4b6097f81d more amx notes geohot 2022-10-29 14:04:10 -07:00
• fdb43fe553 gemm is 1.7 TFLOPS on a single M1 core geohot 2022-10-29 13:42:33 -07:00
• 52bfbc31be vectorization geohot 2022-10-29 12:47:52 -07:00
• e473d35f90 llvm doesn't vectorize geohot 2022-10-29 11:59:48 -07:00
• 86eb06eb76 accurate flop estimation geohot 2022-10-28 19:13:20 -07:00
• 7909786dbf one more opt test geohot 2022-10-28 18:37:53 -07:00
• dd543fbc7a MovementOps is unused geohot 2022-10-28 18:26:08 -07:00
• 71b336503f no RESHAPEs in the AST geohot 2022-10-28 18:25:30 -07:00
• 294ab9e2f8 more test opt geohot 2022-10-28 18:04:12 -07:00
• f885ceb695 test speed w/o bias geohot 2022-10-28 11:22:15 -07:00
• 3735e26492 very minor geohot 2022-10-28 09:39:30 -07:00
• c0050fab8f clean up movement_op in cpu and torch geohot 2022-10-28 09:29:12 -07:00
• df31dde174 hasattr and DeviceBuffer type fixups geohot 2022-10-28 09:05:45 -07:00
• e6b65f8e01 fix graph in openpilot/compile.py geohot 2022-10-28 08:55:34 -07:00
• 1013540370 fix flake8 geohot 2022-10-28 08:52:53 -07:00
• 804b2dd001 move into graph.py geohot 2022-10-28 08:50:11 -07:00
• 8517b69bfb lazy cleanups geohot 2022-10-28 08:43:43 -07:00
• d02f8f9bc0 can we lose the lines with E701 still there? geohot 2022-10-28 08:36:03 -07:00
• ef62db3186 cleanups, remove E701 geohot 2022-10-28 08:28:56 -07:00
• b65b70812a Exec AST (#404) George HotzandGitHub 2022-10-28 08:27:03 -07:00
• 6a15fd3844 LLVM Backend take 2 (#403) George HotzandGitHub 2022-10-26 20:32:31 -07:00
• 10921a60c4 more imports from llvm branch George Hotz 2022-10-26 18:02:36 -07:00
• 463995e64f relu simpler backward pass George Hotz 2022-10-26 17:57:32 -07:00
• 6a8fb53304 move ops.py into lazy.py (#402) George HotzandGitHub 2022-10-25 13:58:03 -07:00
• 8e22d5ee67 replace networkx with defaultdict George Hotz 2022-10-20 19:36:43 -07:00
• 3b9b7eda48 remove run_thneed dead code George Hotz 2022-10-20 17:24:18 -07:00
• 63f9c55156 really dumb bug George Hotz 2022-10-20 17:07:47 -07:00
• 1bec4651b3 fix nonstatic weights George Hotz 2022-10-20 17:04:14 -07:00
• 59143bbb3b raise, don't assert George Hotz 2022-10-20 16:32:34 -07:00
• 9f8c414589 might fix tests George Hotz 2022-10-20 16:27:11 -07:00
• fd6ba8e7ac don't recopy backing George Hotz 2022-10-20 16:06:11 -07:00
• 62affbd9ce add CONTIGUOUS loadop George Hotz 2022-10-20 15:55:19 -07:00
• bb288e6938 safe_numpy and warning for broken matmul George Hotz 2022-10-20 15:40:22 -07:00
• 50c95c7d9a add assert to catch issue in attention George Hotz 2022-10-20 15:13:00 -07:00
• 26c78ccf7d remove useless buffer George Hotz 2022-10-20 14:07:28 -07:00
• a18c1f3178 zero out the inputs George Hotz 2022-10-20 13:46:52 -07:00
• 61ee428e4c rerun George Hotz 2022-10-20 13:29:14 -07:00
• 5dae64b7b0 read input shapes and break down the layers George Hotz 2022-10-20 13:11:24 -07:00
• e00601faea fix thneed self test George Hotz 2022-10-20 12:55:02 -07:00
• ace8db29f8 ReduceSum George Hotz 2022-10-20 12:48:14 -07:00
• c400ee0beb refactoring thneed (#400) George HotzandGitHub 2022-10-20 12:35:59 -07:00
• 0514594083 fix openpilot test George Hotz 2022-10-20 11:56:26 -07:00
• ae0f9b17df openpilot: new models and onnx ops (#401) 3 people 2022-10-20 11:49:19 -07:00
• a4ad1d774a enable tests in test_ops.py that are disabled but now work. (#396) Drew HintzandGitHub 2022-10-13 11:58:53 -05:00
• 165fb4d631 remove redundant list comprehension from inside all. (#397) Drew HintzandGitHub 2022-10-13 11:58:35 -05:00
• 793edf8900 touchup George Hotz 2022-10-10 16:13:34 -07:00
• d54a45b50d measure speed vs torch George Hotz 2022-10-10 16:06:00 -07:00
• b7f748c15a Fix GPU 2**31 virtual size limit (#392) George HotzandGitHub 2022-10-04 21:55:20 -07:00
• 392e57aea7 ugh, why did that fail geohot 2022-10-01 13:38:43 -04:00
• 8382c51c12 always MATMUL, test the ops in OPENCL geohot 2022-10-01 13:31:29 -04:00
• 7a61dc7ee9 test_sd_big_conv geohot 2022-10-01 13:26:05 -04:00
• 178ba50c03 some args for stable diffusion geohot 2022-09-29 01:52:04 -04:00
• 3b1767e013 Fix OpenCL Metal texture issues (#378) Ollin Boer BohanandGitHub 2022-09-28 22:21:54 -07:00
• e737513c52 external_test_opt geohot 2022-09-28 23:29:41 -04:00
• 650c011646 notrain test geohot 2022-09-28 23:27:20 -04:00
• af87d692e4 should this be 10? geohot 2022-09-28 23:25:52 -04:00
• 0fd459b24e ugh, global state geohot 2022-09-28 23:10:49 -04:00
• fa4eff9cc1 Device.GPU isn't definied geohot 2022-09-28 23:00:15 -04:00
• 0b6537a572 fix tests geohot 2022-09-28 22:57:58 -04:00
• 726cca78cd fix bn folding issue, add new test geohot 2022-09-28 22:52:18 -04:00
• a0d169eb59 fix efficientnet George Hotz 2022-09-28 14:23:01 -07:00
• dec5334da9 revert layernorm to have axis param geohot 2022-09-26 10:11:38 -04:00
• dc80bf6f85 layernorm is all axis but the first geohot 2022-09-25 17:55:48 -04:00
• 60df954377 Fix weight init: this work? (#391) George HotzandGitHub 2022-09-25 16:46:33 -04:00
• ff11c4316b move get_parameters to optim.py geohot 2022-09-25 13:16:58 -04:00
• a0c0239ff1 fix mnist load from other dirs geohot 2022-09-25 12:50:22 -04:00
• 2c01a66265 Reshape dataset from fetch_mnist (#390) Jacky LeeandGitHub 2022-09-24 18:16:29 -07:00
• acae9a20c1 clipnorm support geohot 2022-09-24 13:26:38 -04:00
• 271446e3eb set requires_grad to None (#387) George HotzandGitHub 2022-09-21 11:16:02 -04:00
• 29ae21bb0d import tests from CL metal texture fix geohot 2022-09-19 20:01:47 -04:00
• a8aa1f9589 that's simpler geohot 2022-09-18 20:40:46 -04:00
• 57e804a9bf add min support geohot 2022-09-18 20:39:41 -04:00
• 2f0f91ba3d support float16 onnx weights (#384) YassineYousfiandGitHub 2022-09-15 06:12:18 -07:00
• 75f937227a add barrier Comma Device 2022-09-13 11:39:48 -04:00
• 3c3534736e fix matmul kernel and tests geohot 2022-09-13 08:31:04 -07:00
• 62e9419206 fix test failure on MATMUL=1 backward pass Comma Device 2022-09-13 11:18:52 -04:00
• 3b82afc6a0 simple on device failing test Comma Device 2022-09-13 10:59:15 -04:00
• 4efde1ba0a test_matmul geohot 2022-09-13 07:51:33 -07:00
• 894a7cee79 forgot a few geohot 2022-09-12 09:21:46 -07:00
• 801ecd4a07 cleanup clip tokenizer geohot 2022-09-12 09:20:12 -07:00
• ff0da4c802 Added standalone CLIP tokenizer (#382) Fernand PajotandGitHub 2022-09-12 10:12:55 -06:00
• a1810c8617 update serious_mnist.py (#380) David RedmonandGitHub 2022-09-11 16:37:40 -04:00
• ce348f0c92 Revert "change default opt to 2" geohot 2022-09-11 13:35:42 -07:00
• 726f4e98e9 change default opt to 2 geohot 2022-09-09 07:50:25 -07:00
• 1a7bdc51f8 support more onnx ops (#376) YassineYousfiandGitHub 2022-09-07 15:15:24 -07:00
• 0b8c2221b5 relax mnist test a tiny bit George Hotz 2022-09-07 07:52:05 -07:00
• ecc1a0470d add Linear to tinygrad.nn George Hotz 2022-09-07 07:40:48 -07:00