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

autogen-comgr-2

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

ci-deps

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

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

explicit-dev

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

gitea-test-ci

gl_dims

gldims

glm_flash

grad_outerworld

grok_x86_novec

group_div_rules

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-chat-template

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

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

onnx-cpu-count

outer_range

outerworld_work

pad_invalid

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

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-lp

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

view_after_const

view_in_codegen

viz_cbroadcast

warp_fun

wgpu-f16-packed

working_postopt

x86_moves

x86_numel

#1

#2

#3

#4

#5

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

• dffde3de5a support both asymmetric and negative padding geohot 2022-06-26 17:59:25 -07:00
• 49c954b389 comments geohot 2022-06-26 17:20:25 -07:00
• 8c483fbdc9 maxpool lazy fix geohot 2022-06-26 17:07:03 -07:00
• f607f18006 fix backward geohot 2022-06-25 00:00:53 -07:00
• ec30f0402f improve benchmark_train_efficientnet geohot 2022-06-24 23:46:38 -07:00
• 3a147137ee CL_DEVICE option geohot 2022-06-24 23:22:10 -07:00
• d748353ce5 err, okay, a bit more off geohot 2022-06-24 22:44:57 -07:00
• bdde95f16e CACHE_LAZYBUFFERS options + benchmark. only a couple x from torch geohot 2022-06-24 22:33:53 -07:00
• 6847eaf5b6 comments geohot 2022-06-22 09:37:50 -07:00
• 1d4fb3527e cleanups to Tensor class geohot 2022-06-22 09:33:30 -07:00
• 3e13e3330a UNSAFE_FLOAT4 env geohot 2022-06-22 08:20:29 -07:00
• 73415e20ab this fixes 2 of the conv recomputes...but it's ugh geohot 2022-06-22 08:18:12 -07:00
• b2d5df6049 3 convs are being recomputed geohot 2022-06-22 07:54:52 -07:00
• ba2defcdef elif False geohot 2022-06-21 23:54:09 -07:00
• 9cb0522574 noargs geohot 2022-06-21 23:48:58 -07:00
• 1074dfbb71 unstrided geohot 2022-06-21 23:42:21 -07:00
• 9ae01290ba pass in shorts geohot 2022-06-21 23:33:23 -07:00
• 18d74c01b1 float4 opt geohot 2022-06-21 21:27:51 -07:00
• ff3d5fe962 debugging while we compile geohot 2022-06-21 21:12:04 -07:00
• b12985b013 openpilot compiler geohot 2022-06-21 20:31:18 -07:00
• 98a730dd00 benchmark on different inputs geohot 2022-06-21 20:20:58 -07:00
• 9d06a86f7f CL class, debugging geohot 2022-06-21 20:16:29 -07:00
• 0b820f7966 FOLD_CONSTANTS_INTO_KERNELS and shapetracker OOB tweak geohot 2022-06-21 19:47:15 -07:00
• 83d50e2687 move to extra.onnx geohot 2022-06-21 19:43:44 -07:00
• 1ebc2b5545 lazy opencl works geohot 2022-06-21 19:41:08 -07:00
• c833886bf5 improved shapetracker geohot 2022-06-21 19:16:55 -07:00
• c53c91f949 opencl tests passed (#347) George HotzandGitHub 2022-06-21 18:57:09 -07:00
• 8fbe2e4aed No ctx in llops (#345) George HotzandGitHub 2022-06-21 10:07:49 -07:00
• 159a2d1a80 Simple Lazy (#340) George HotzandGitHub 2022-06-20 22:45:11 -07:00
• a3538e225a Simple Lazy Pieces (#343) George HotzandGitHub 2022-06-20 20:28:10 -07:00
• 2ee85812f7 intel opencl (#342) George HotzandGitHub 2022-06-20 19:25:55 -07:00
• 3e7416163d batch from lazy branch (#341) George HotzandGitHub 2022-06-20 17:42:35 -07:00
• a7131b6a46 Non contig (#339) George HotzandGitHub 2022-06-19 22:40:48 -07:00
• d05e7c291a contiguous_view (#336) George HotzandGitHub 2022-06-19 20:37:28 -07:00
• fb72ea3fbd gpu uses shapetracker (fix tests) (#335) George HotzandGitHub 2022-06-19 17:32:07 -07:00
• ce2e20b768 fix test geohot 2022-06-19 17:07:09 -07:00
• f5f21ecb86 gpu buffer is shapetracker geohot 2022-06-19 17:02:24 -07:00
• 6b652dafb2 touchups geohot 2022-06-19 16:57:14 -07:00
• e364849b3b stuff from lazy geohot 2022-06-19 09:57:16 -07:00
• 2069fef292 unnecessary assign add in cpu processing_op (#334) Tim LüggerandGitHub 2022-06-19 16:41:40 +02:00
• 8d08e41c21 print time in test geohot 2022-06-19 00:59:09 -07:00
• 395eb60f46 less lines, and oddly faster geohot 2022-06-18 21:48:42 -07:00
• aa164d901e remove ctx from buffers (#333) George HotzandGitHub 2022-06-18 17:27:10 -07:00
• 77f5cef8a6 First batch from lazy branch (#332) George HotzandGitHub 2022-06-18 17:26:59 -07:00
• 3faf8353ca remove out_shape from processing_op George Hotz 2022-06-16 17:07:57 -07:00
• a11deb5150 shapetracker check for noop George Hotz 2022-06-16 16:29:18 -07:00
• 52505faaf4 minor George Hotz 2022-06-16 15:53:45 -07:00
• d5b3e18540 Accelerate with CL (#325) George HotzandGitHub 2022-06-16 15:40:52 -07:00
• bd7068f635 fix tests hopefully George Hotz 2022-06-16 14:07:37 -07:00
• 9306759cbc put the allocations back in the ops George Hotz 2022-06-16 12:12:55 -07:00
• ce15bf2bdb the big memory gradient didn't even need to be computed George Hotz 2022-06-16 11:41:29 -07:00
• 2e58948f6a Revert "can put that test back" George Hotz 2022-06-16 11:25:49 -07:00
• 51b082b41a can put that test back George Hotz 2022-06-16 11:18:14 -07:00
• 73bc181fbe cleaner output shape geohot 2022-06-16 10:24:03 -07:00
• b5796ae4f9 remove useless reshape geohot 2022-06-16 10:15:43 -07:00
• 89db797e57 get rid of reduce using channels geohot 2022-06-16 10:01:54 -07:00
• 38d6cfec2a remove the expand geohot 2022-06-16 09:54:56 -07:00
• bcfbb4c81b minor cleanups geohot 2022-06-15 22:27:46 -07:00
• 3667200df5 remove unused unstride George Hotz 2022-06-15 20:03:43 -07:00
• ff648e9510 remove convt and compute dx with conv George Hotz 2022-06-15 19:54:15 -07:00
• 142c88f2e3 move to mlops George Hotz 2022-06-15 18:06:07 -07:00
• 85fe25e27b add stride support to shapetracker George Hotz 2022-06-15 17:48:41 -07:00
• 827e8f67eb comment George Hotz 2022-06-15 17:31:27 -07:00
• 3d4657167b fix tests hopefully George Hotz 2022-06-15 17:26:37 -07:00
• e4ab57e39d oops, only stride George Hotz 2022-06-15 15:25:58 -07:00
• 86f55b078d transpose dilation was simple George Hotz 2022-06-15 15:20:51 -07:00
• 2a14befb74 support padding George Hotz 2022-06-15 14:46:44 -07:00
• 6d98366214 move CONVDW out of llops George Hotz 2022-06-15 12:05:11 -07:00
• fef6c82491 wow dilation support was simple George Hotz 2022-06-15 11:38:23 -07:00
• 0b182029dd support dilated convolution in torch George Hotz 2022-06-14 18:03:35 -07:00
• a690ba4588 add test for padding George Hotz 2022-06-14 17:41:22 -07:00
• e057ca23bb add flip George Hotz 2022-06-14 17:28:43 -07:00
• a8aeebfb0c use shapetracker to combine adj reduce axis George Hotz 2022-06-14 17:08:12 -07:00
• 906cce9916 reduce with loops George Hotz 2022-06-14 16:38:33 -07:00
• 6261a0639b ShapeTracker (#328) George HotzandGitHub 2022-06-14 16:08:22 -07:00
• e58b5711ec simpler convdw George Hotz 2022-06-13 17:56:54 -07:00
• dcbca4fdf1 Expand Operator (#327) George HotzandGitHub 2022-06-12 12:31:48 -07:00
• 5cf7649eda register the operators outside geohot 2022-06-12 10:26:34 -07:00
• 33f18c61a1 test_broadcasted_add geohot 2022-06-12 10:19:58 -07:00
• d47a421970 add cout to conv_args, don't change the first 12 geohot 2022-06-12 00:10:15 -07:00
• af300b121b refactor to pass conv args into llops geohot 2022-06-11 23:08:46 -07:00
• d747a4b9e2 add padding to conv2d function, other minor things geohot 2022-06-11 22:29:42 -07:00
• 9a3c048724 skip broken tests, no float64 allowed geohot 2022-06-11 17:12:04 -07:00
• 3461e1a467 A is actually a unary op geohot 2022-06-11 17:05:29 -07:00
• 296f391403 delete old graph engine to reach line count geohot 2022-06-11 17:02:53 -07:00
• a4d0d3f17a stupid numpy hack geohot 2022-06-11 17:01:56 -07:00
• c03c835d75 oops geohot 2022-06-11 16:53:34 -07:00
• 361123490a remove a lot of useless returns geohot 2022-06-11 16:50:06 -07:00
• 35e55afe17 and processing op geohot 2022-06-11 16:46:38 -07:00
• 6d5591f7a3 movement op to SSA geohot 2022-06-11 16:44:24 -07:00
• 6685807df7 reduce_op in SSA format too geohot 2022-06-11 16:40:14 -07:00
• bbf231da34 move unary and binary op mem alloc to Ops class geohot 2022-06-11 16:35:03 -07:00
• 1511cbf9c5 graphs of llops, sorry about the line count geohot 2022-06-11 16:28:39 -07:00
• 9ebd472375 move ops to ops.py geohot 2022-06-11 15:58:56 -07:00
• b5b68e75ff simpler onnx geohot 2022-06-11 15:35:45 -07:00
• 2305a5347b test_onnx works with enet also geohot 2022-06-11 14:30:26 -07:00
• 6fdb276886 flip batchnorm function order geohot 2022-06-11 13:20:41 -07:00
• 85d17a2acd running resnet onnx geohot 2022-06-11 13:17:15 -07:00
• 0225360191 fixed with one return x geohot 2022-06-11 12:08:53 -07:00
• 094b304348 very weird += is broken geohot 2022-06-11 11:58:50 -07:00