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

• a72b1b6d65 sharding for llama (#3151) 3 people 2024-01-16 19:28:00 -08:00
• 14c010958b support for non-uniform sharding (#3154) 3 people 2024-01-16 20:33:32 -05:00
• 81ae4ea179 compile cache for several devices (#3148) nimlgenandGitHub 2024-01-16 22:45:26 +03:00
• 589c16756f hlb_cifar multi gpu training (#3150) chenyuandGitHub 2024-01-16 14:38:45 -05:00
• cc0de99751 hotfix: multilazybuffer can have only one lazybuffer geohot 2024-01-16 10:06:45 -08:00
• 228f30b96a multitensor jit (#3149) 3 people 2024-01-16 09:09:15 -08:00
• b9d470577c gelu -> quick_gelu in hlb_cifar (#3147) chenyuandGitHub 2024-01-16 02:03:37 -05:00
• ec5a212b0a modernize hlb_cifar (#3146) chenyuandGitHub 2024-01-16 01:35:11 -05:00
• 2088937206 run full hlb_cifar training in tinybox ci (#3145) chenyuandGitHub 2024-01-15 23:59:20 -05:00
• 22920a7e55 add LATEBEAM to hlb_cifar (#3142) chenyuandGitHub 2024-01-15 23:26:03 -05:00
• 766bd0bbe8 make _deepwalk a generator and not passing nodes around (#3141) chenyuandGitHub 2024-01-15 20:26:00 -05:00
• 120c8b1841 update llvm api + add cache key (#3140) George HotzandGitHub 2024-01-15 17:25:32 -08:00
• cec0a7bc37 use shard api to eval resnet fast (#3136) George HotzandGitHub 2024-01-15 16:49:38 -08:00
• ca0beeef38 Christopherm99 ptx (#3139) +212 people 2024-01-15 16:44:20 -08:00
• 1ee11411f1 s/lazydata/lazyop/ in print_tree (#3138) chenyuandGitHub 2024-01-15 19:38:27 -05:00
• a5d634a541 simplify dtype (#3137) George HotzandGitHub 2024-01-15 16:27:43 -08:00
• e4528543fa remove LLVMOPT geohot 2024-01-15 16:01:09 -08:00
• a464909d79 fast resnet eval (#3135) George HotzandGitHub 2024-01-15 14:15:18 -08:00
• b7b494e9b8 no numpy (#3134) Jyotirmaya MahantaandGitHub 2024-01-16 02:39:05 +05:30
• 6bb65cd02e fix off-by-one error in st_equal (#3131) Paul GustafsonandGitHub 2024-01-15 11:32:13 -08:00
• 44c05919c1 dtype fmt (#3132) George HotzandGitHub 2024-01-15 11:31:54 -08:00
• 5ec66938de remove np from metal graph (#3129) nimlgenandGitHub 2024-01-15 19:44:35 +03:00
• 2ef09ca641 update test_ptr_ne (#3130) Jyotirmaya MahantaandGitHub 2024-01-15 22:06:29 +05:30
• e39cd3e7f2 update env_vars.md (#3127) chenyuandGitHub 2024-01-15 01:06:56 -05:00
• 537fb8b0b8 separate try except blocks in onnx2torch in model benchmark (#3126) chenyuandGitHub 2024-01-15 00:39:33 -05:00
• 0dba34b81c Fix backward fn for < and == (#3037) 3 people 2024-01-15 04:39:52 +00:00
• db965a0c74 remove numpy from ops_torch (#3124) chenyuandGitHub 2024-01-14 22:46:57 -05:00
• 1f9aee8b6f remove numpy from device (#3123) George HotzandGitHub 2024-01-14 19:36:05 -08:00
• ea5824657d move fromcpu out of lazy.py (#3122) George HotzandGitHub 2024-01-14 18:21:08 -08:00
• 96345061d3 hotfix: ptrdtype compare was broken geohot 2024-01-14 18:08:22 -08:00
• 26e0faf656 make DType a dataclass (#3111) Jyotirmaya MahantaandGitHub 2024-01-15 06:45:59 +05:30
• c13d51da1d add device options for tests in multigpu (#3121) Yixiang GaoandGitHub 2024-01-14 15:17:47 -08:00
• 79f4627fbc fix conversation: llama generates token not prob now (#3120) chenyuandGitHub 2024-01-14 13:10:01 -05:00
• 152ef7fc79 minor cleanups of onnx_ops (#3116) chenyuandGitHub 2024-01-14 02:15:24 -05:00
• fb3f8f7597 move sample inside jit for beautiful_mnist (#3115) chenyuandGitHub 2024-01-14 01:36:30 -05:00
• a313e63a9b add Tensor.var (#3114) chenyuandGitHub 2024-01-14 01:11:08 -05:00
• c658aa4fbf minor cleanup of test_disk_tensor (#3112) chenyuandGitHub 2024-01-13 20:54:58 -05:00
• 9c73d2724f cleanup ops_disk type annotation and redundant str cast (#3110) chenyuandGitHub 2024-01-13 16:56:48 -05:00
• a300fea2a4 failed test case due to cast resets shapetracker (#3109) chenyuandGitHub 2024-01-13 12:46:51 -05:00
• cf1d0a6704 no exceptions in __del__ when module creation is failed in hip/cuda (#3107) nimlgenandGitHub 2024-01-13 20:03:55 +03:00
• 12f28ac9d4 catch runtime error in search._time_program (#3106) chenyuandGitHub 2024-01-12 21:53:13 -05:00
• f018a55ea1 update NumNode.__hash__ to be hash(self.b) (#3105) chenyuandGitHub 2024-01-12 19:46:21 -05:00
• c3c35f9142 flag to profile mixtral - 1.7 tok/s now (#3104) chenyuandGitHub 2024-01-12 18:54:27 -05:00
• e078e2d060 add half @ half to mac benchmark (#3103) chenyuandGitHub 2024-01-12 16:38:41 -05:00
• ddbdb52f77 wmma: enable METAL half tensor cores and clean up cstyle (#3095) Francis LamandGitHub 2024-01-12 13:25:28 -08:00
• f96fc6e9d4 fix gpt2 with empty prompt take 2 (#3102) chenyuandGitHub 2024-01-12 14:46:36 -05:00
• ca46d3541b Revert "fix gpt2 with empty prompt" (#3101) chenyuandGitHub 2024-01-12 14:27:41 -05:00
• 1d7f01bc6d fix gpt2 with empty prompt (#3100) chenyuandGitHub 2024-01-12 14:18:17 -05:00
• 0c49d38ba7 replace with tensor op (#3099) SnakeOnexandGitHub 2024-01-12 20:13:40 +01:00
• f3a50b4e40 fix broadcasted logic if there's 0 in shapes (#3097) chenyuandGitHub 2024-01-12 13:32:43 -05:00
• 025fbf4e80 One hot in tensor.py (#3093) SnakeOnexandGitHub 2024-01-12 19:31:18 +01:00
• 7086d77db1 bugfix do not reset shapetracker of 0 size lazybuffer (#3096) chenyuandGitHub 2024-01-11 23:22:52 -05:00
• 13e872b53f add mutigpu support for llama attention (#3064) Yixiang GaoandGitHub 2024-01-11 16:31:02 -08:00
• dcf7ecaaff update jit type annotation post lazy rewrite (#3091) chenyuandGitHub 2024-01-11 15:49:30 -05:00
• 0fe6904351 use device from LinearizerOptions in kernel search (#3090) chenyuandGitHub 2024-01-11 14:46:03 -05:00
• 93e3f952aa use BEAM=2 instead of BEAM=4 in cuda ci gpt2 (#3089) chenyuandGitHub 2024-01-11 13:21:06 -05:00
• f502c9b08f minor cleanup of View.reshape (#3088) chenyuandGitHub 2024-01-11 13:05:54 -05:00
• f40299c3fe remove the third merging state in view._merge_dims (#3085) chenyuandGitHub 2024-01-11 12:07:43 -05:00
• 7f9590d357 hotfix disable flaky mac runner wino cifar (#3087) chenyuandGitHub 2024-01-11 11:57:05 -05:00
• adcc844755 cat works (#3086) Yixiang GaoandGitHub 2024-01-11 08:25:20 -08:00
• cdeab9ad97 mem_estimate is always int, not symbolic (#3083) chenyuandGitHub 2024-01-10 23:39:51 -05:00
• 162fa61a32 wmma: clean up device specific tensor core code (#3081) Francis LamandGitHub 2024-01-10 18:03:09 -08:00
• d218d13885 minor cleanups of lazy.py (#3080) chenyuandGitHub 2024-01-10 20:17:56 -05:00
• 56dda33fc6 Tensor.expand resolves the new_shape before shortcut return (#3078) chenyuandGitHub 2024-01-10 14:29:15 -05:00
• 6842476ca6 better test demonstration (#3077) Yixiang GaoandGitHub 2024-01-10 10:50:52 -08:00
• 507e0afba0 fix onehot and jit in examples/transformer (#3073) chenyuandGitHub 2024-01-10 02:22:41 -05:00
• 4342fccc83 filter_strides -> canonicalize_strides (#3072) chenyuandGitHub 2024-01-10 01:06:48 -05:00
• 023f5df0e9 simpler idxs_to_idx (#3071) chenyuandGitHub 2024-01-10 00:30:10 -05:00
• 2495ca95c7 early gate the graph (#3070) George HotzandGitHub 2024-01-09 20:17:13 -08:00
• ff0d6e4551 jit autorealizes output (#3069) George HotzandGitHub 2024-01-09 20:10:22 -08:00
• ae83733431 hotfix: examples/transformer.py geohot 2024-01-09 19:25:37 -08:00
• 145718a90f unbind view or shapetracker also returns var_val (#3067) chenyuandGitHub 2024-01-09 21:45:05 -05:00
• ef3aa6d7fb update gh actions (#3033) 3 people 2024-01-10 01:52:22 +00:00
• 3f80c1a098 speedtweaks3: apply shouldn't use the tensor constructor (#3065) George HotzandGitHub 2024-01-09 17:42:33 -08:00
• 0abe72b677 hotfix: use is for enum compare, a few more geohot 2024-01-09 16:53:13 -08:00
• b2b5849f74 hotfix: use is for enum compare geohot 2024-01-09 16:47:27 -08:00
• ac3f246c11 cached size (#3060) George HotzandGitHub 2024-01-09 16:37:37 -08:00
• 73b72b8de2 test scaled dot product attention (#3063) Yixiang GaoandGitHub 2024-01-09 14:30:57 -08:00
• 55ac2a2cf7 Tensor.cat with 0 shape tensors (#3062) chenyuandGitHub 2024-01-09 16:54:06 -05:00
• f0d7ad8aaa fix gpt2 attention with start_pos = 0 (#3061) chenyuandGitHub 2024-01-09 16:14:55 -05:00
• 39b91131bc Speed tweaks (#3059) George HotzandGitHub 2024-01-09 11:34:17 -08:00
• bf6281f316 hotfix: remove useless slow assert from ShapeTracker geohot 2024-01-09 10:56:36 -08:00
• 4b687af98f explicit lazybuffer caching (#3058) George HotzandGitHub 2024-01-09 10:52:37 -08:00
• 2c6f2e899d No extra vars call (#3054) v0.8.0 George HotzandGitHub 2024-01-09 09:52:58 -08:00
• 259bf9bffc add multigpu test for RMSNorm (#3056) Yixiang GaoandGitHub 2024-01-09 09:52:51 -08:00
• dab8214103 unit tests for Device.canonicalize (#3055) chenyuandGitHub 2024-01-09 12:47:20 -05:00
• 374f7659a7 remove unused reciprocal (#3053) George HotzandGitHub 2024-01-09 08:59:04 -08:00
• a686663657 make Embedding device aware for multigpu (#3051) Yixiang GaoandGitHub 2024-01-08 20:09:26 -08:00
• 19298e7a3f Device._buffers -> Device._devices (#3052) chenyuandGitHub 2024-01-08 21:30:38 -05:00
• 4f4e8634b8 use in_features directly in nn.Linear.__init__ bound check (#3050) chenyuandGitHub 2024-01-08 19:32:35 -05:00
• ee6a73826b clean up test_nn.py (#3049) chenyuandGitHub 2024-01-08 18:45:03 -05:00
• 3eb3664074 fix nn.Embedding with empty length input (#3048) chenyuandGitHub 2024-01-08 18:08:36 -05:00
• 7ea2e0035b move children for speed (#3047) George HotzandGitHub 2024-01-08 15:02:32 -08:00
• 655c6f61d3 St real size (#3046) George HotzandGitHub 2024-01-08 14:44:53 -08:00
• 1d730b8853 remove ACCUM_FP32 in simple_matmul.py (#3045) chenyuandGitHub 2024-01-08 17:37:57 -05:00
• 47d67da830 track the size in the lazybuffer (#3044) George HotzandGitHub 2024-01-08 13:44:55 -08:00
• c003be7309 Revert "track size in shapetracker" (#3043) George HotzandGitHub 2024-01-08 13:13:39 -08:00
• 50754f1494 add caches there (#3042) George HotzandGitHub 2024-01-08 13:02:16 -08:00
• c5a941d466 webgl backend in extra (#3041) 3 people 2024-01-08 09:29:13 -08:00
• 8cbcd1b342 Remove webgpu, back to 5k lines (#3040) George HotzandGitHub 2024-01-08 09:10:07 -08:00