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

• ec03eb44bd tinygrad does forward pass convs on GPU geohot 2020-11-07 10:15:52 -08:00
• 23c39d9f52 fetch() 'ran out of input' fix (#65) Manjeet SinghandGitHub 2020-11-07 23:15:10 +05:30
• bc7758cc5b getting convs to work on gpu geohot 2020-11-07 09:17:57 -08:00
• 3302286e68 yayay test_sgd_gpu passes geohot 2020-11-07 08:48:17 -08:00
• 98d1a4f740 GPU Selection works when len(answers)>ngpus (#64) Ryan NephandGitHub 2020-11-07 08:06:59 -08:00
• ce0c80e1bd ugh, corrupt downloads. when did this break? geohot 2020-11-07 08:04:43 -08:00
• 9332fe1989 weee, logsoftmax on GPU. very slow geohot 2020-11-07 07:59:37 -08:00
• 38e112cccd logsoftmax test geohot 2020-11-07 07:26:53 -08:00
• 237bbe3625 sigmoid on GPU geohot 2020-11-07 07:23:43 -08:00
• f40dbd791c Use --upgrade since its in active dev (#63) Rakib FihaandGitHub 2020-11-08 00:15:05 +09:00
• cd54697fd8 fix gpu sum forward (#61) Rene DelgadoandGitHub 2020-11-06 02:59:16 -03:00
• aea1069f63 Div on CPU (#58) gallanoeandGitHub 2020-11-05 21:58:37 -08:00
• cc605da36d Stable Sigmoid op (#59) NeuralLinkandGitHub 2020-11-06 11:27:50 +05:30
• cc852efa64 make OpenCL work elsewhere George Hotz 2020-11-05 07:59:55 -08:00
• 22a5f9975d pow+div on GPU (#57) Ryan NephandGitHub 2020-11-05 07:49:45 -08:00
• b16fadc5c6 Speed up sum backward pass (#54) 3 people 2020-11-05 03:55:45 +00:00
• 0be3d2cdba install pyopencl George Hotz 2020-11-04 11:22:11 -08:00
• 3efb4f4df4 chicken.jpg George Hotz 2020-11-04 11:20:22 -08:00
• c69582744d Update efficientnet.py (#53) Marcel BischoffandGitHub 2020-11-04 12:36:27 -05:00
• c40862241d Revert "choose aspect ration dependent on landscape or portrait (#48)" (#51) Marcel BischoffandGitHub 2020-11-02 22:31:07 -05:00
• 2808411421 choose aspect ration dependent on landscape or portrait (#48) Anders Aaen SpringborgandGitHub 2020-11-03 03:10:22 +01:00
• d24363f421 Update setup.py (#49) Marcel BischoffandGitHub 2020-11-02 21:09:31 -05:00
• 940e14c6ca more readme geohot 2020-11-02 08:33:48 -08:00
• 1e6bbdf4f8 readme updates geohot 2020-11-02 08:30:43 -08:00
• f178d23ff3 gpu relu is good geohot 2020-11-02 08:25:32 -08:00
• 8766346187 fix gpu sub geohot 2020-11-02 08:18:58 -08:00
• 231c1134bd cute trick for GPU test geohot 2020-11-02 08:17:17 -08:00
• 6a7896f3e9 opsgpu cl program caching geohot 2020-11-02 07:46:26 -08:00
• 2e7f16bf3f the power of cheating geohot 2020-11-02 07:42:11 -08:00
• fc358a07ad print sloccount geohot 2020-11-02 07:38:13 -08:00
• 3da64acd6b the power of small geohot 2020-11-02 07:36:34 -08:00
• 82fc842b40 in_place_op geohot 2020-11-02 07:26:13 -08:00
• 1d793b8571 gpu SGD runs now geohot 2020-11-02 07:19:12 -08:00
• 621c61b30d sgd now operates on Tensors geohot 2020-11-02 07:14:54 -08:00
• 355402504e ops work geohot 2020-11-02 07:03:23 -08:00
• af5a4e0f5a alternative to einsum which is slow in extreme cases (#44) 3 people 2020-11-02 09:36:07 -05:00
• 5201a8e89f matmul on GPU geohot 2020-11-01 08:54:20 -08:00
• 41e7d59aed test dot geohot 2020-11-01 07:51:35 -08:00
• 1f544d6ece test mnist on GPU geohot 2020-11-01 07:46:17 -08:00
• 499604d69b convert tensor to gpu geohot 2020-11-01 07:34:50 -08:00
• bac6bfda11 with that at least it's only 10x slower geohot 2020-11-01 07:30:19 -08:00
• 9ac1ad40d6 Add GPU Support! (do not merge yet) (#41) George HotzandGitHub 2020-11-01 07:00:49 -08:00
• c06a4fcc80 Fix temp location for windows (#39) liej6799andGitHub 2020-11-01 08:21:14 +08:00
• 7c0dc8f48b more whitespace geohot 2020-10-31 11:05:11 -07:00
• e01e35e545 14 ops to write for GPU geohot 2020-10-31 10:59:30 -07:00
• 06928cf3cc ENET WORKS geohot 2020-10-31 10:42:58 -07:00
• 68cba88e8f fix enet bugs, now is mousetrap geohot 2020-10-31 10:28:07 -07:00
• a852143572 simpler conv2d geohot 2020-10-31 09:20:20 -07:00
• e5153d401f Conv2D forward without loops much faster, backward slightly faster (#38) 3 people 2020-10-31 12:08:32 -04:00
• 053c319708 optional look at cat geohot 2020-10-30 09:30:15 -07:00
• 2c7e75d733 group conv: forward pass works (#34) George HotzandGitHub 2020-10-30 09:19:20 -07:00
• 339a35b081 div needs help geohot 2020-10-30 08:32:16 -07:00
• 71aedc2309 it thinks it's washer. it's cat. bad net. you do bad. geohot 2020-10-30 08:28:05 -07:00
• c14473f87d unit test for batchnorm2d geohot 2020-10-30 08:19:58 -07:00
• 843b1cb7d6 only warn once geohot 2020-10-30 08:04:48 -07:00
• 16d324064e streamlined numerical_jacobian (#32) 3 people 2020-10-30 05:20:12 -04:00
• 127f878fa9 Clean init function (#29) LiamandGitHub 2020-10-30 01:45:55 +01:00
• c13b255847 Update gradcheck.py (#30) Afshin1354andGitHub 2020-10-29 20:45:34 -04:00
• d0c3912517 fix for invalid GPU error caused by (test/test_net_speed.py::TestConvSpeed::test_mnist) when testing in CI (#31) Omar Ben BrahimandGitHub 2020-10-30 01:45:16 +01:00
• 1dde4ce609 oops, stay in float32 geohot 2020-10-29 08:24:12 -07:00
• 5e7e359706 fix tests geohot 2020-10-29 08:19:07 -07:00
• 9ae3e9daf3 shape has to be a kwarg now, idk why this didn't break before geohot 2020-10-29 08:13:05 -07:00
• 2db670ef26 enet refactor + no sigmoid warning geohot 2020-10-29 08:08:21 -07:00
• 17fa74c15b factor out a couple nn ops geohot 2020-10-29 08:01:12 -07:00
• f84f6c1edd write sqrt and div using pow geohot 2020-10-29 07:57:16 -07:00
• ec7bdb87ca clean up CI (#28) Omar Ben BrahimandGitHub 2020-10-29 15:43:31 +01:00
• cc17e3271a try to recognize cat. do not succeed geohot 2020-10-27 21:41:52 -07:00
• 03d9c98f5b enet work geohot 2020-10-27 21:23:02 -07:00
• 0ec279951f enet weight loading geohot 2020-10-27 21:01:48 -07:00
• e84ad3e27d enet runs geohot 2020-10-27 20:16:29 -07:00
• 9166eb58bb more enet geohot 2020-10-27 19:37:21 -07:00
• 41828d768f sigmoid and more enet geohot 2020-10-27 19:13:47 -07:00
• 09d1ebcdaa examples is better geohot 2020-10-27 18:57:00 -07:00
• 4b163ee270 efficient version of adam (#20) Göktuğ KarakaşlıandGitHub 2020-10-28 01:54:40 +03:00
• 8d75c9e4c4 we have matmul too geohot 2020-10-27 08:57:17 -07:00
• f9788eba14 parameters, and start on efficientnet geohot 2020-10-27 08:53:35 -07:00
• 0b68c08de0 literally just bump version for picture on pypi geohot 2020-10-27 08:14:22 -07:00
• 6b5982b6b3 push pypi geohot 2020-10-27 08:13:15 -07:00
• 716f86a572 sections geohot 2020-10-27 08:10:51 -07:00
• 4a663d31da less than 1000 lines (#24) LiamandGitHub 2020-10-27 16:07:15 +01:00
• dcf630dbeb Merge pull request #22 from tsmcalister/master George HotzandGitHub 2020-10-26 16:41:10 -07:00
• 6b44a7f729 adds beautiful and meaningful logo = 2020-10-26 18:12:49 +01:00
• c5db2cf517 readme more geohot 2020-10-26 09:21:48 -07:00
• 43591a1e71 make the example simpler geohot 2020-10-26 09:19:20 -07:00
• 1f0514e5df pip geohot 2020-10-26 09:15:31 -07:00
• 64bd4f7936 lol, it's not 1.0 geohot 2020-10-26 09:11:32 -07:00
• dec2f1c31a Merge pull request #18 from goktug97/master George HotzandGitHub 2020-10-26 09:09:31 -07:00
• 81040d926d import only the module Göktuğ Karakaşlı 2020-10-26 19:02:06 +03:00
• 8d80726207 two spaces Göktuğ Karakaşlı 2020-10-26 18:54:55 +03:00
• 1654008c1f conv stride support geohot 2020-10-26 08:54:43 -07:00
• 2a55d7402b clean up ops, refactor pool backward. add stride test geohot 2020-10-26 08:47:11 -07:00
• 93dceb4bee fix kernel_size bug, name like torch, add test geohot 2020-10-26 08:38:53 -07:00
• cc9bd45b44 add setup.py and change imports to relative Göktuğ Karakaşlı 2020-10-26 18:19:50 +03:00
• d0c9be3e25 Merge pull request #17 from tsmcalister/master George HotzandGitHub 2020-10-26 06:35:34 -07:00
• 15e5988323 make default parameters work for functions Timothy Mc Alister 2020-10-26 12:43:36 +01:00
• 2d37fd686b test ops geohot 2020-10-25 19:03:49 -07:00
• 2eebbd32c6 ops test speed geohot 2020-10-25 19:01:02 -07:00
• b27bcbe4b4 avgpool and test refactor geohot 2020-10-25 18:40:01 -07:00
• 4c42676cb6 400 -> 200 geohot 2020-10-25 17:19:59 -07:00
• 567707a5f6 rename max_pool2d to match torch, remove more fast conv crap geohot 2020-10-25 17:16:47 -07:00