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

• 10ba96029c device: ImportError -> Exception George Hotz 2023-02-22 19:48:59 -08:00
• 76b4d0577d yolov8 works up to the MaxPool George Hotz 2023-02-22 19:32:01 -08:00
• 5190784cbb Fix Tensor random functions determinism with same seed (#580) Mischa UntagaandGitHub 2023-02-23 04:08:43 +01:00
• c8d89eb20e avg/max pool strides George Hotz 2023-02-22 18:00:48 -08:00
• ac8daaeea5 refactor device, credit martinloretzzz George Hotz 2023-02-22 17:28:04 -08:00
• a3ddc1d484 CLANG doesn't support float4 geohot 2023-02-22 10:41:15 -08:00
• c4c2c28738 a sustainable approach to float4 (#582) George HotzandGitHub 2023-02-22 09:45:08 -08:00
• 50e541174c fix image copyout geohot 2023-02-22 07:27:13 -08:00
• 628ce067a1 add tests to mypy geohot 2023-02-22 07:07:38 -08:00
• 104c3c5e73 oops, forgot that debug geohot 2023-02-22 06:58:27 -08:00
• c5e2126d49 move DEBUG to helpers geohot 2023-02-22 06:51:34 -08:00
• 14bb2c40a2 Fix yolov3 example (#577) Mischa UntagaandGitHub 2023-02-21 18:24:00 +01:00
• 8550b3e168 Fix kernel cache key (#570) Martin LoretzandGitHub 2023-02-21 12:53:07 +01:00
• 66b4b3bdd3 clang: linux needs fPIC George Hotz 2023-02-20 21:09:33 -08:00
• 4f4b7d05a7 clang needs -lm and is very slow geohot 2023-02-20 21:02:04 -08:00
• 4126bf2982 remove six (hopefully not needed) geohot 2023-02-20 20:44:23 -08:00
• efcb3f0cdd fix metal dep geohot 2023-02-20 20:43:15 -08:00
• 9670bf1fd1 Add unsqueeze (#574) Connor HendersonandGitHub 2023-02-20 23:14:59 -05:00
• cfad2902d5 clang: that might have been getting GCed geohot 2023-02-20 20:12:49 -08:00
• 506970414a added metal packages to setup and release metal buffers after del (#571) DiogoandGitHub 2023-02-20 23:01:07 -05:00
• d9fa47ecc9 use the TinyJit in the efficientnet runner, 200ms -> 20ms geohot 2023-02-20 19:58:16 -08:00
• 714bf4b108 clang backend (#572) George HotzandGitHub 2023-02-20 18:18:18 -08:00
• 353e81fa8f need hand opt with IMAGE==2 geohot 2023-02-20 10:18:19 -08:00
• b925d05e22 no networkx is okay geohot 2023-02-20 09:57:12 -08:00
• 4d232c7c95 optional networkx + DEBUGCL=2 geohot 2023-02-20 09:50:46 -08:00
• 8b0082540b openpilot compile cleanups geohot 2023-02-20 09:16:03 -08:00
• ea13504f35 fix METAL_XCODE geohot 2023-02-19 20:02:12 -08:00
• bbfec2fde7 8.46 TFLOPS geohot 2023-02-19 13:21:25 -08:00
• 1ba847963d reshape and retain metal_matmul geohot 2023-02-19 13:07:23 -08:00
• 60008e55cd sick of that failing geohot 2023-02-19 13:05:37 -08:00
• 0b3f686530 Good fast triton (#567) George HotzandGitHub 2023-02-19 12:21:55 -08:00
• cb679cd051 Fix weight initialization (#566) Jacky LeeandGitHub 2023-02-19 11:25:29 -08:00
• a508c2b429 small tweaks to the metal runtime (#562) DiogoandGitHub 2023-02-19 14:25:13 -05:00
• 7e9a5e3f31 Refactor graph (#560) Martin LoretzandGitHub 2023-02-19 19:41:30 +01:00
• 7944cfdadc Remove Tensor.data (#565) KirillandGitHub 2023-02-19 03:36:12 +03:00
• 4efe0169bb remove old metal implementation geohot 2023-02-18 13:51:48 -08:00
• 7e8b0305f3 Fix mnist gan example (#563) Jacky LeeandGitHub 2023-02-18 13:45:37 -08:00
• 9fd41632c6 Import get_parameters from tinygrad.nn (#559) Jacky LeeandGitHub 2023-02-17 15:22:26 -08:00
• fae7654924 fix sync issue geohot 2023-02-17 12:42:45 -08:00
• 5e6265be6e metal timing, fix speed test geohot 2023-02-17 12:31:05 -08:00
• 121bd03cbd metal globalcounters geohot 2023-02-17 12:02:54 -08:00
• 67d1df80ba gid is array, metal works geohot 2023-02-17 11:54:50 -08:00
• f9af0322e7 metal can add geohot 2023-02-17 11:45:33 -08:00
• e172f0087a BatchNorm2D -> BatchNorm2d (#558) Jacky LeeandGitHub 2023-02-16 12:31:49 -08:00
• 09315ef34f Add tinygrad.org reference in Readme. (#556) LiamandGitHub 2023-02-14 18:39:00 +01:00
• 82c257e8f5 more kernel search geohot 2023-02-12 10:34:38 -08:00
• c35fcc6964 Replace phrase for prompt (#555) Jacky LeeandGitHub 2023-02-12 09:04:44 -08:00
• 20a03d5017 woah, don't sync torch if it's not torch geohot 2023-02-12 07:48:56 -08:00
• de71c13934 test speed v torch uses jit geohot 2023-02-12 07:43:17 -08:00
• 693d4b89a4 fixup TRITON backend to use new APIs geohot 2023-02-12 06:57:49 -08:00
• 191c76cfd7 hlb_cifar10 torch version geohot 2023-02-11 18:04:40 -08:00
• 9057d98d36 no lr decay in cifar. test this in torch tomorrow geohot 2023-02-11 17:42:54 -08:00
• dd7accb9cc decay LR, little bugfix geohot 2023-02-11 17:34:15 -08:00
• ba3bf5bdf7 cifar stops learning geohot 2023-02-11 17:21:42 -08:00
• 40f3949742 fancier KOPT geohot 2023-02-11 16:40:25 -08:00
• 446442dbb3 fix tests symbolic geohot 2023-02-11 15:16:47 -08:00
• 7a7046f264 sum_combine_num geohot 2023-02-11 14:48:31 -08:00
• a4f5f2ff8b Add missing packages to setup.py (#554) KirillandGitHub 2023-02-12 01:41:56 +03:00
• 20a351a3c6 hand optim CONVW geohot 2023-02-11 14:41:08 -08:00
• 89499b303d oops, bad else. why didn't linter catch geohot 2023-02-11 12:02:09 -08:00
• 7d33f2d659 CL.CACHE is over, GlobalCounters.cache is it geohot 2023-02-11 12:00:14 -08:00
• b9eae94ae9 move Device back into lazy geohot 2023-02-11 11:26:53 -08:00
• 9152bb5b4a momentum support in SGD geohot 2023-02-11 10:22:37 -08:00
• 0a2035e015 oops, GPU isn't defined geohot 2023-02-11 10:10:02 -08:00
• 3421d4af10 the jit has a test geohot 2023-02-11 10:04:03 -08:00
• 031edd01e6 switch openpilot compile to TinyJit geohot 2023-02-11 09:51:44 -08:00
• 8f912c3966 added deep deterministic policy gradient example (#531) jspielerandGitHub 2023-02-11 17:10:46 +01:00
• 608fd730d3 put the JIT in extra geohot 2023-02-11 00:35:18 -06:00
• ed8ae7522a tinyjit geohot 2023-02-11 00:22:36 -06:00
• 4c90a15689 make the fake data actually learnable geohot 2023-02-10 23:35:21 -06:00
• 07629d7476 fakedata and move to new cache geohot 2023-02-10 23:32:31 -06:00
• 63fa7daf30 wrong place for CL geohot 2023-02-10 23:22:24 -06:00
• 6f9b103878 fix opencl types geohot 2023-02-10 23:18:39 -06:00
• fed95119dc CL.mem_used -> GlobalCounters.mem_used geohot 2023-02-10 23:13:29 -06:00
• 51037815b9 add comment so we don't remove self.t tensor again geohot 2023-02-10 23:07:07 -06:00
• c0ea538ba0 Revert "revert t as tensor, constant folding should be done better" geohot 2023-02-10 23:06:00 -06:00
• 1d800a94ad revert t as tensor, constant folding should be done better geohot 2023-02-10 22:58:39 -06:00
• 77988e3236 fix str() line count bug in scc geohot 2023-02-10 22:53:30 -06:00
• 1fb5b8069b simpler processed check geohot 2023-02-10 22:49:20 -06:00
• 609477656e clean up lazy processing_op geohot 2023-02-10 22:41:52 -06:00
• a4cb161bd4 log_kernel geohot 2023-02-10 21:51:53 -06:00
• b9f02671d3 oops, broke torch speed test geohot 2023-02-10 16:13:53 -06:00
• 0efe1e435f no need to render to check valid geohot 2023-02-10 15:35:12 -06:00
• 27154db99a Downloads weights in examples/stable_diffusion.py (#537) KirillandGitHub 2023-02-10 23:37:04 +03:00
• 4459cde68b minor matmul location cleanup geohot 2023-02-10 14:12:42 -06:00
• a007145ac4 oops, should be __itruediv__. we should add test for this geohot 2023-02-10 14:04:40 -06:00
• 5c51ae8dbf Show where tinygrad is faster in speed test vs torch (#549) Jacky LeeandGitHub 2023-02-10 12:01:07 -08:00
• 87a7717222 LLVM backend uses shapetracker geohot 2023-02-10 13:53:33 -06:00
• c3cf17c6d0 Symbolic render (#550) George HotzandGitHub 2023-02-10 13:22:26 -06:00
• 5ed3622965 add dump to kernel_search geohot 2023-02-10 12:13:30 -06:00
• f08187526f Fix examples (#540) 5 people 2023-02-10 10:09:37 -08:00
• 56a06280c5 Testing/utils (#548) Lucas KellerandGitHub 2023-02-10 12:08:20 -06:00
• 1257b0433a should fix tests geohot 2023-02-09 13:12:14 -06:00
• e6f19d4ce2 assume all generic exec ast have ProcessingOp geohot 2023-02-09 13:03:48 -06:00
• 78795e3507 reduce line count by simplifying DeviceBuffer geohot 2023-02-09 12:51:50 -06:00
• 5de850f6d5 assign buffer reuse (#547) George HotzandGitHub 2023-02-09 11:53:02 -06:00
• 473bbd3e35 fix graphs geohot 2023-02-09 09:40:46 -06:00
• 16a7edc775 move base_fxn_for_op to ops_cpu geohot 2023-02-08 18:23:48 -06:00
• c642f5e72b less lines for torch geohot 2023-02-08 18:15:59 -06:00
• 58a03eb693 generic processing op geohot 2023-02-08 18:09:17 -06:00