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

• 60590cf8b5 perf: create buffer only when needed (#1684) 3 people 2023-09-03 02:43:29 +02:00
• 66a6bbd029 codellama (#1702) 3 people 2023-09-02 10:45:12 -05:00
• a2745819f6 faster gpt2 jit path and gpt2 in test_real_world (#1738) chenyuandGitHub 2023-09-02 08:39:12 -07:00
• 89cd380bfc add nvidia CI (#1737) George HotzandGitHub 2023-09-01 22:02:30 -07:00
• 91258aa67f render const (#1736) George HotzandGitHub 2023-09-01 19:01:43 -07:00
• a96e54d8bb search for grouped reduces (#1732) nimlgenandGitHub 2023-09-02 00:21:10 +03:00
• cd844ec4b2 remove Token class (#1723) George HotzandGitHub 2023-09-01 12:53:07 -07:00
• 458eb89463 minor changes from prerender (#1734) George HotzandGitHub 2023-09-01 10:04:47 -07:00
• f964b9e5ee visitor pattern for sym_infer and unit tests (#1733) chenyuandGitHub 2023-09-01 09:47:45 -07:00
• bf05534c6e hip multidevice (#1728) wozeparrotandGitHub 2023-09-01 09:46:13 -04:00
• 024dd690fa Reactivate commavq/gpt2m benchmark (#1731) JaSpa99andGitHub 2023-09-01 15:45:08 +02:00
• 7780eb3c5a minor dimensions (#1730) George HotzandGitHub 2023-09-01 06:42:00 -07:00
• 5c403d43b9 New >3 indexing (#1729) George HotzandGitHub 2023-08-31 21:24:15 -07:00
• e3a062ad17 real matvec test geohot 2023-08-31 17:27:25 -07:00
• 453e437598 move stuff in the linearizer (#1726) George HotzandGitHub 2023-08-31 14:42:09 -07:00
• c18a497dde minor global dim cleanup (#1724) George HotzandGitHub 2023-08-31 12:23:39 -07:00
• 94b1257f5e Changed DEVICE to Device.DEFAULT in deep_determinist_policy_gradient (#1715) geohotstanandGitHub 2023-08-31 22:08:51 +08:00
• b5cf274da3 remove memory peak for quantized llama (#1720) nimlgenandGitHub 2023-08-30 23:32:30 +03:00
• e4eb5d55c7 critical realize for unjitted llama (#1718) chenyuandGitHub 2023-08-30 11:52:32 -07:00
• cd7ceed914 gpt2: print total instead of sync time George Hotz 2023-08-30 10:59:42 -07:00
• 62536d6000 perf: use enumerate where possible (#1692) 3 people 2023-08-30 19:41:51 +02:00
• a8aa13dc91 [ready] Replacing os with pathlib (#1708) Karan HandaandGitHub 2023-08-30 23:11:08 +05:30
• 355b02dc3f allow zerosized tensors (#1659) nimlgenandGitHub 2023-08-30 20:39:24 +03:00
• f9cb31fdc2 added visitor pattern (#1669) Max HahnandGitHub 2023-08-30 18:03:44 +02:00
• fdd7f282cb Reenable tensor cores for self-hosted Mac CI (#1717) George HotzandGitHub 2023-08-30 07:53:04 -07:00
• ac183568be llama JIT python runtime speedup (#1633) chenyuandGitHub 2023-08-30 07:51:05 -07:00
• 1682e9a38a Fix: Stable Diffusion index (#1713) Umut ZenginandGitHub 2023-08-30 07:21:10 +03:00
• 2f768e386d stable diffusion benchmark artifact (#1714) wozeparrotandGitHub 2023-08-29 21:08:40 -04:00
• 0ea22bf249 remove DEBUG=1 from stable diffusion AMD since jit cache is fixed geohot 2023-08-29 12:46:12 -07:00
• ab9b9ff3e2 pipefail benchmark (#1709) (#1710) 3 people 2023-08-29 08:15:02 -07:00
• aa7c98722b sd timing (#1706) George HotzandGitHub 2023-08-28 20:22:57 -07:00
• 8844a0a822 llvm jitted (#1652) nimlgenandGitHub 2023-08-29 06:22:44 +03:00
• 1c0449e190 add cache collector (#1595) nimlgenandGitHub 2023-08-29 05:59:55 +03:00
• f5f8b09c13 allow manual release (#1704) George HotzandGitHub 2023-08-28 17:54:25 -07:00
• 715047a1e4 fix release publish (#1703) George HotzandGitHub 2023-08-28 17:48:00 -07:00
• ee6d8de2dc Llama: load models in HuggingFace format (incl. indexed, safetensors) (#1583) Olivier ChafikandGitHub 2023-08-28 20:11:40 +01:00
• 3515ba4f23 add dtypes test (#1682) qazalandGitHub 2023-08-28 18:12:15 +03:00
• 50f669e43b [ready] perf: simpler Tensor init (#1679) 3 people 2023-08-28 04:18:03 +02:00
• b66f54e379 perf: avoid reshaping if not necessary (#1683) 3 people 2023-08-27 20:17:04 +02:00
• 328cf2e86a perf: remove cast and revert back to isinstance (#1694) 3 people 2023-08-27 20:15:52 +02:00
• 8b354b3f73 feat: version bump! (#1687) v0.7.0 wozeparrotandGitHub 2023-08-27 12:38:58 -04:00
• abaa605f71 [ready] perf: start enumerate at 1 instead of checking all i (#1691) 3 people 2023-08-27 18:00:32 +02:00
• 2730ed657f perf: faster lazyop eq (#1693) 3 people 2023-08-27 17:17:02 +02:00
• 6ca509a485 perf: constant in while in for in busy func (#1688) 3 people 2023-08-27 17:13:16 +02:00
• b89d81330f fix: restore old behaviour (#1689) 3 people 2023-08-27 16:45:53 +02:00
• 66fbf4800b fix symbolic_ops tests with Tensor.training=True (#1686) chenyuandGitHub 2023-08-26 20:19:56 -07:00
• 6c5dc9c153 [ready] perf: faster lazyop init (#1673) 3 people 2023-08-27 04:59:10 +02:00
• f61d0657d1 document new envvars (#1676) wozeparrotandGitHub 2023-08-26 20:17:02 -04:00
• 9d93a82354 remove FAKEDATA (#1685) Yixiang GaoandGitHub 2023-08-26 19:15:54 -05:00
• b5d700adae update openpilot supercombo.onnx to 0.9.4 (#1681) chenyuandGitHub 2023-08-26 16:16:08 -07:00
• 89b529c07f [ready] ci: add py38 to linters (#1674) 3 people 2023-08-26 15:34:15 +02:00
• 25be7f745d Tensor.uniform with dtype=int bug fix (#1593) Jordan WrightandGitHub 2023-08-26 01:59:53 -04:00
• f702a8f497 [ready] avoid in-function graph imports in lazy.py (#1666) 3 people 2023-08-25 19:56:28 +02:00
• 02e64da678 refactor: tuples can be concatenated with + (#1671) 3 people 2023-08-25 18:37:13 +02:00
• 173850f599 fix CIFAR jit (#1657) Yixiang GaoandGitHub 2023-08-24 19:14:40 -04:00
• f00325e77d ops_metal newCommandQueueWithMaxCommandBufferCount_(1024) (#1664) chenyuandGitHub 2023-08-24 15:42:00 -07:00
• 1ba8f0dca3 Quickstart: Upgrade section "Training" to new code (#1663) 3 people 2023-08-24 23:12:16 +02:00
• 29adae84eb Quickstart: Use tensors to compute train accuracy (#1662) 3 people 2023-08-24 23:09:12 +02:00
• d37d092c14 split linearizer into 3 files (#1654) George HotzandGitHub 2023-08-23 14:58:47 -07:00
• 1b8c40234f Uast start (#1650) George HotzandGitHub 2023-08-23 12:00:06 -07:00
• 484708da87 #1615 fix (#1616) geohotstanandGitHub 2023-08-23 11:51:05 -07:00
• b57c374164 add accelerator links to readme (#1649) Pavol RusnakandGitHub 2023-08-23 20:47:55 +02:00
• 82623697a8 Move asm renderer (#1648) George HotzandGitHub 2023-08-23 10:06:43 -07:00
• a89363574d teeny changes (#1647) George HotzandGitHub 2023-08-23 09:53:39 -07:00
• a6d842af7a move device to ops (#1646) George HotzandGitHub 2023-08-23 08:30:17 -07:00
• a65ae1198b do replace div->mul for non-floats (#1644) nimlgenandGitHub 2023-08-23 17:34:31 +03:00
• da694d4241 move that image import geohot 2023-08-22 21:30:55 -07:00
• 41e83be3dd simple where broadcast (#1643) George HotzandGitHub 2023-08-22 21:24:49 -07:00
• c831218139 Optional: Reduce line count and simplify the LazyBuffer interface (#1642) George HotzandGitHub 2023-08-22 21:01:10 -07:00
• d25046e66a matvec tests (#1634) George HotzandGitHub 2023-08-22 17:33:58 -07:00
• 643cbdfd50 make embedding and GPT-2 fast (#1631) George HotzandGitHub 2023-08-22 15:14:38 -07:00
• a7752ad65d Fix link to state.py in quickstart (#1632) Niklas DandGitHub 2023-08-22 23:39:30 +02:00
• c9c40bb16f Import whole math module in tensor.py (#1628) c143andGitHub 2023-08-22 23:07:46 +02:00
• 6fcfa50b35 [ready] perf: no noop cast just to make mypy happy (#1626) 3 people 2023-08-22 23:07:22 +02:00
• f04a6d7882 perf: faster partition (#1625) 3 people 2023-08-22 20:56:41 +02:00
• d3c401ba3c llama quantize: scale uses mul, not div geohot 2023-08-22 11:48:56 -07:00
• 696e4d20a1 fix KOPT=2 with variable shape geohot 2023-08-22 11:34:34 -07:00
• de1fcc418f no more toCPU path (#1624) George HotzandGitHub 2023-08-22 11:07:26 -07:00
• 463dece63e auto arg dtypes (#1623) George HotzandGitHub 2023-08-22 10:22:40 -07:00
• db8344ab83 add noalias to llvm (#1622) George HotzandGitHub 2023-08-22 09:26:01 -07:00
• 89e13f2f04 support symbols in shrink (#1611) chenyuandGitHub 2023-08-22 09:08:21 -07:00
• 718ced296c move state to nn/state (#1619) George HotzandGitHub 2023-08-22 07:36:24 -07:00
• 1e93fd5449 Readability for unreadable functions (#1610) Umut ZenginandGitHub 2023-08-22 17:09:08 +03:00
• 86a32ffb1a lt sum (#1617) George HotzandGitHub 2023-08-21 21:19:16 -07:00
• c64c47a6ae test arange simple geohot 2023-08-21 20:16:17 -07:00
• 4f459841bc Symbolic JIT for GPT2 (#1613) George HotzandGitHub 2023-08-21 19:44:57 -07:00
• 4f02491cd4 add cpu if torch tensor (#1609) Yixiang GaoandGitHub 2023-08-21 18:57:59 -05:00
• f720682beb np.argmax to Tensor.argmax (#1608) Umut ZenginandGitHub 2023-08-22 01:22:29 +03:00
• 4ea00bad38 track down llama bug George Hotz 2023-08-21 15:14:21 -07:00
• b02f77b354 perf: faster broadcasted (#1601) 3 people 2023-08-21 23:21:46 +02:00
• 4d54afb6df sparse cat cross entropy (#1597) Yixiang GaoandGitHub 2023-08-21 16:14:54 -05:00
• 109100656f refactor: no len if it is not needed (#1598) 3 people 2023-08-21 23:06:32 +02:00
• 2c8f8ac611 perf: no ret needed (#1604) 3 people 2023-08-21 23:05:13 +02:00
• 750714c386 perf: namedtuples are hashable, don't need a key (#1607) 3 people 2023-08-21 23:01:10 +02:00
• aaa6fdf347 this was unused code (#1600) George HotzandGitHub 2023-08-21 12:02:58 -07:00
• 8e8724d3a8 perf: if argument order (mops) (#1599) 3 people 2023-08-21 20:20:39 +02:00
• 2e60920317 Revert "sparse cat cross entropy (#1591)" (#1596) George HotzandGitHub 2023-08-21 10:04:26 -07:00
• f0ee850e98 sparse cat cross entropy (#1591) Yixiang GaoandGitHub 2023-08-21 11:56:41 -05:00
• 8d6662a741 .cpu().numpy() -> .numpy() (#1594) Yixiang GaoandGitHub 2023-08-21 11:53:29 -05:00
• 35bf21276f Argmax/Argmin Feature (#1576) Umut ZenginandGitHub 2023-08-21 04:46:46 +03:00