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

• d26bd73c1e have to ignore that type George Hotz 2022-09-07 07:24:27 -07:00
• b7783565af cpu line savings and cleaner geohot 2022-09-06 21:24:22 -07:00
• 1c92a6da22 make gpu code readable geohot 2022-09-06 21:17:36 -07:00
• 790af99a48 fix slice one multi, and linear can be simpler with new broadcasting George Hotz 2022-09-06 19:51:33 -07:00
• 4f4ecbec97 add div to operators George Hotz 2022-09-06 17:39:26 -07:00
• 5a76e652b8 simpler movement op George Hotz 2022-09-06 17:27:33 -07:00
• 896f9f74a9 hmm, need this with broadcast change George Hotz 2022-09-06 16:54:01 -07:00
• a18a6a0773 fix sd with TORCH=1 George Hotz 2022-09-06 16:51:08 -07:00
• 5aad460c7a broadcast from right to left (#375) YassineYousfiandGitHub 2022-09-06 16:36:13 -07:00
• 0516359af8 fix stupid OPENCL=1 OOM George Hotz 2022-09-06 14:29:23 -07:00
• f215534a64 1100 lines, but sane linter rules George Hotz 2022-09-06 13:47:45 -07:00
• 682dc64430 works at work George Hotz 2022-09-06 08:06:11 -07:00
• f683b26eef bring back native exp log George Hotz 2022-09-06 07:59:04 -07:00
• d6f499fd69 improve opencl, why is it OOMing geohot 2022-09-05 20:14:31 -07:00
• 0ba6179de7 stable diffusion in readme geohot 2022-09-05 18:51:56 -07:00
• c1d5af8b0c stable diffusion cleanups geohot 2022-09-05 18:34:13 -07:00
• 3728ef6d02 better alphas geohot 2022-09-05 16:48:26 -07:00
• 0fda854b3e other prompt example geohot 2022-09-05 16:14:16 -07:00
• 16cb4290c4 cat horse winning ❗ geohot 2022-09-05 16:05:14 -07:00
• 1043fa067a it renders something geohot 2022-09-05 15:52:14 -07:00
• 5a685b93ac brown img geohot 2022-09-05 15:20:18 -07:00
• 98d6264987 all models match geohot 2022-09-05 12:27:54 -07:00
• b8bd34b5d2 fix last bug in unet probz geohot 2022-09-05 11:32:44 -07:00
• 3df67aa0af fix transformer bugs geohot 2022-09-05 11:26:28 -07:00
• 2ed3bb6223 clip model is running geohot 2022-09-05 09:50:13 -07:00
• 2c6f4e4c66 Make creation helpers use fp32 by default (#374) Ollin Boer BohanandGitHub 2022-09-04 13:47:27 -07:00
• 1a54ea2417 runs on torch cpu geohot 2022-09-04 12:06:42 -07:00
• 9590d92750 stable diffusion compiles (add no_init) geohot 2022-09-04 11:40:50 -07:00
• 172683c314 work geohot 2022-09-04 11:21:09 -07:00
• bcb867cdd6 better idea for numbers, do the division in python geohot 2022-09-03 16:23:39 -07:00
• 39e1d23c88 from_number_like to fix div issue geohot 2022-09-03 16:19:16 -07:00
• c2a030fe55 one liner that's more clear geohot 2022-09-03 16:08:48 -07:00
• 4a3ed58edb more readable actually geohot 2022-09-03 16:00:35 -07:00
• 633f31dc73 easier to read geohot 2022-09-03 15:53:13 -07:00
• 6578e08919 cleanups for Mid geohot 2022-09-03 15:50:33 -07:00
• 852de7c66c remove ugly parens geohot 2022-09-03 15:41:37 -07:00
• 6b190c2fa5 stable diffusion works geohot 2022-09-03 13:55:36 -07:00
• 871b2a7b52 fix check geohot 2022-09-03 13:37:40 -07:00
• 947e10dab0 yolo geohot 2022-09-03 12:39:48 -07:00
• 033a3ecccf found tinygrad bug geohot 2022-09-03 12:32:43 -07:00
• 114728d363 torch bs geohot 2022-09-03 11:57:23 -07:00
• 356732515b stable_diffusion: add attn and layernorm geohot 2022-09-03 11:02:27 -07:00
• 4dadd95e3c fix tests hopefully, more stable diffusion geohot 2022-09-03 10:38:31 -07:00
• c01a8c5c2d stable diffusion start geohot 2022-09-03 10:08:42 -07:00
• c07bf72d6e save free 200ms Comma Device 2022-08-31 20:31:42 -04:00
• 2e9b7637b3 don't save input buffers George Hotz 2022-08-31 15:37:38 -07:00
• a3fc64a585 fix batchnorm folding in openpilot compile George Hotz 2022-08-31 13:04:49 -07:00
• a734df98fa TEST_ENET for openpilot compiler Comma Device 2022-08-31 13:23:36 -04:00
• d919ac32af fix wrong size input George Hotz 2022-08-31 09:07:34 -07:00
• 040640a580 fix cl import error George Hotz 2022-08-31 08:43:33 -07:00
• e194ae0c1d typos George Hotz 2022-08-30 19:52:21 -07:00
• 3af650b028 Rewrote Tensor.cat to be shorter and (hopefully) clearer (#372) Mitchell GoffandGitHub 2022-08-30 16:15:07 -07:00
• db56297011 line count George Hotz 2022-08-30 15:23:35 -07:00
• 33ac355bcd still broken George Hotz 2022-08-29 19:08:07 -07:00
• 5efab7cf1d add reciprocal George Hotz 2022-08-29 18:00:24 -07:00
• 880707f2d2 no torch test if no torch George Hotz 2022-08-29 15:29:19 -07:00
• 5eba228844 print inputs George Hotz 2022-08-29 08:56:04 -07:00
• fa33ba716e flip that George Hotz 2022-08-28 11:27:33 -07:00
• dd587d26e3 oops, compare with abs George Hotz 2022-08-28 11:23:21 -07:00
• dc7af8c3ac thneed run float32 George Hotz 2022-08-28 11:03:23 -07:00
• f0d11f29c7 float32 in image desc Comma Device 2022-08-28 08:47:43 -07:00
• 9678cb8a1a hmm, the native exp/log breaks it too much Comma Device 2022-08-22 17:13:08 -07:00
• 0c1378e7db native_exp is way faster on qcom Comma Device 2022-08-22 16:58:50 -07:00
• 2162cd3383 fix typing George Hotz 2022-08-22 16:25:15 -07:00
• e0a8d0f836 image input works Comma Device 2022-08-22 16:04:17 -07:00
• 18340e7d30 remove from_image George Hotz 2022-08-22 15:52:26 -07:00
• 1b5f4e52d9 refactor getters Comma Device 2022-08-22 13:29:08 -07:00
• 13646ae07a opencl can't optimize that Comma Device 2022-08-22 10:26:42 -07:00
• 227937d6d2 fix test maybe George Hotz 2022-08-22 09:50:14 -07:00
• 11626053b0 run_thneed with test George Hotz 2022-08-22 09:44:32 -07:00
• ec5d9b355c use functools.partialmethod (#369) 3 people 2022-08-21 12:13:31 -07:00
• e7a4cd91ba fix cpu thneed running George Hotz 2022-08-21 12:11:07 -07:00
• 0aca848d10 enable the openpilot test George Hotz 2022-08-21 12:05:48 -07:00
• a8734df030 add openpilot tests to tinygrad George Hotz 2022-08-21 12:03:37 -07:00
• 7f15779942 t.assign in optim geohot 2022-08-20 14:04:33 -07:00
• 5c8466f475 tune inter_red geohot 2022-08-20 12:50:26 -07:00
• c5433db050 2 stage reduce geohot 2022-08-20 12:20:32 -07:00
• b020fc2225 less needless reshaping geohot 2022-08-20 08:43:38 -07:00
• ccd539e93e maybe that's a better way to do this geohot 2022-08-20 08:25:12 -07:00
• b7d782b921 hmm, with the new reduce, we have to opt 3 for memory usage geohot 2022-08-20 08:16:15 -07:00
• e34eb855fe much simpler reduce geohot 2022-08-20 08:12:43 -07:00
• 1eb12dafbc reduce axis at the end geohot 2022-08-20 07:40:56 -07:00
• 85453288d7 run_onnx_torch Comma Device 2022-08-18 08:30:12 -07:00
• 1f23517d92 fixup run thneed Comma Device 2022-08-18 08:22:53 -07:00
• b132de677d tinygrad.nn (#367) George HotzandGitHub 2022-08-18 07:41:00 -07:00
• 783c120a8c rawcpu (#365) George HotzandGitHub 2022-08-17 11:33:20 +02:00
• 57e5df9f28 ane: procPath issue. don't waste more time with this, focus on core tinygrad geohot 2022-08-16 10:36:13 +02:00
• bdfdbc8f8d broken amfi patch geohot 2022-08-13 10:41:25 +02:00
• 262efe5784 update readme geohot 2022-08-09 11:08:52 +02:00
• 6267a3c8c2 notes geohot 2022-08-09 00:42:14 +02:00
• f4ff130947 docs geohot 2022-08-09 00:06:24 +02:00
• 01de17eeb8 amfi note geohot 2022-08-08 13:17:36 +02:00
• 136706169d fix ane on new mac os x geohot 2022-08-06 19:10:22 +00:00
• f300caa486 notes geohot 2022-08-06 15:21:26 +00:00
• 27f209b80b add time sum geohot 2022-08-06 14:35:21 +00:00
• 64c0d2dedb add gflop estimate geohot 2022-08-06 14:21:48 +00:00
• 94d526f8fc fix op estimate geohot 2022-08-06 14:15:50 +00:00
• f2847cb710 remove useless init, add ops counter geohot 2022-08-06 14:05:25 +00:00
• 18fde22dac fix that soon geohot 2022-07-20 09:07:09 -07:00
• 187d581dc6 fix options on old pyopencl geohot 2022-07-20 08:24:05 -07:00