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

• 4bf54335a6 add test_hgemm to test_tiny geohot 2026-07-16 12:02:32 -07:00
• 2b1146b3f4 further clean up wmma (#17048) George HotzandGitHub 2026-07-16 11:43:23 -07:00
• f6a92d0a16 sum_acc_dtype(weak) is weak (#17047) chenyuandGitHub 2026-07-16 14:32:37 -04:00
• 61e104bdfb use UOp.wmma everywhere (#17045) George HotzandGitHub 2026-07-16 10:40:48 -07:00
• 5a4156c5d1 bitcast and element_size raise for weak dtypes (#17046) chenyuandGitHub 2026-07-16 13:07:45 -04:00
• 7eb197b1bb nv: always wait for reset (#17043) nimlgenandGitHub 2026-07-16 16:35:12 +03:00
• dba8b6b505 allow weak alu operands (#17044) chenyuandGitHub 2026-07-16 09:33:20 -04:00
• e33e96415f hcq2: tiny cleanupg (#17042) nimlgenandGitHub 2026-07-16 16:14:54 +03:00
• 810d8732f9 fix n^2 in limit_bufs by memoizing reachable loads [PR] (#17017) 3 people 2026-07-16 02:54:04 -04:00
• 1c74e044a4 search /usr/lib/wsl/lib first for linux (#17027) 3 people 2026-07-16 06:28:17 +00:00
• 8b0dd870ce use wmma helper (#17038) George HotzandGitHub 2026-07-15 23:25:17 -07:00
• 783042d216 viz: graph stays in place when sidebars resize (#17037) qazalandGitHub 2026-07-16 10:47:22 +08:00
• e8d3047a50 dtype_from_uop cleanup [PR] (#17036) chenyuandGitHub 2026-07-15 21:52:21 -04:00
• 6b7fee7d9f minor lower_alu_dtype cleanup [PR] (#17034) chenyuandGitHub 2026-07-15 17:42:23 -04:00
• be075b200a weak dtypes in dtype_from_uop [PR] (#17032) chenyuandGitHub 2026-07-15 16:54:31 -04:00
• 3ffb4dc4bc unify lower index in lower_alu_dtype [PR] (#17033) chenyuandGitHub 2026-07-15 16:38:42 -04:00
• d6fddb066f usb: keep only custom (#17029) nimlgenandGitHub 2026-07-15 22:40:52 +03:00
• 0d30f97584 mlperf: make v6.1 dir (#17031) wozeparrotandGitHub 2026-07-15 10:40:55 -07:00
• c23d8188e1 remove _ensure_float [pr] (#17030) chenyuandGitHub 2026-07-15 11:19:20 -04:00
• 0d19970edc least_upper_dtype in dtype_from_uop [PR] (#17028) chenyuandGitHub 2026-07-15 09:27:53 -04:00
• ebe26420a7 update where Invalid rules [pr] (#17026) chenyuandGitHub 2026-07-15 00:00:30 -04:00
• 06169f5013 gptoss: small fixes (#17025) wozeparrotandGitHub 2026-07-14 23:40:23 -04:00
• 47ddf94f17 remove InvalidType lt and gt (#17023) chenyuandGitHub 2026-07-14 21:59:15 -04:00
• c9baa2ef79 use pattern matcher in contiguous_view_offset [PR] (#17022) sirhcmandGitHub 2026-07-14 19:37:36 -04:00
• 4257939e50 remove copyin/copyout from Buffer (#17020) nimlgenandGitHub 2026-07-14 19:47:22 +03:00
• 939f28d571 fused qkv rope custom kernel (#17021) qazalandGitHub 2026-07-15 00:08:42 +08:00
• 82fbca43c5 fix Tensor(np) dtype and support fp8 safetensor (#17019) chenyuandGitHub 2026-07-14 09:31:21 -04:00
• 872225e47d update dtype tests for small dtypes (#17016) chenyuandGitHub 2026-07-14 08:07:00 -04:00
• edfef062ed skip viz.cli -t in null device (#17018) qazalandGitHub 2026-07-14 18:24:58 +08:00
• 55bb251130 add pm_manual_bf16_cast to Metal [pr] (#17015) chenyuandGitHub 2026-07-13 21:53:47 -04:00
• a9fbc7db7b expect _offset support, CL and WEBGPU are outliers (#17014) sirhcmandGitHub 2026-07-13 18:53:32 -04:00
• 9ce96c2628 fix subnormal in test_dtype (#17013) chenyuandGitHub 2026-07-13 18:53:13 -04:00
• c898dfe150 remove UOp.contiguous override [PR] (#17012) chenyuandGitHub 2026-07-13 16:10:58 -04:00
• 681a5e0cfd remove UOp cast and bitcast override [PR] (#17011) chenyuandGitHub 2026-07-13 14:23:11 -04:00
• 0410c9325d make test/null follow the SPEC (#17010) chenyuandGitHub 2026-07-13 14:01:41 -04:00
• e4bdc529c4 hcq2 ci (#17008) nimlgenandGitHub 2026-07-13 19:29:08 +03:00
• 4536a57f79 hcq rename map (#17009) nimlgenandGitHub 2026-07-13 19:23:12 +03:00
• 62ad646d1c llama: gemm/fa backward speedups (gpt 5.6) (#17007) qazalandGitHub 2026-07-13 23:27:00 +08:00
• 4d2becddf8 hcq2: spec=2 (#17006) nimlgenandGitHub 2026-07-13 18:13:33 +03:00
• ab9dde04a9 amd: do not spam with traps (#17004) nimlgenandGitHub 2026-07-13 16:36:01 +03:00
• 223c6d74c3 remove unused get_empty_input_data (#17002) chenyuandGitHub 2026-07-12 22:13:30 -04:00
• 006637cf73 remove unmeasured CIFAR micro-optimizations cifar_30_clean geohot 2026-07-12 22:51:19 +00:00
• f072c2bf97 qcom fast matmul crap from gpt 5.5 Comma Device 2026-07-12 22:44:05 +00:00
• dde2e736e5 fix disable_gc decorator reentrancy (#16999) George HotzandGitHub 2026-07-12 15:32:54 -07:00
• 1f335c6eef Merge remote-tracking branch 'origin/master' into cifar_30_clean geohot 2026-07-12 22:17:27 +00:00
• 434f076c4a delete unused UOp.wait and UOp.wmma [PR] (#16998) chenyuandGitHub 2026-07-12 17:30:03 -04:00
• 406e314d29 simplify CIFAR speedups geohot 2026-07-12 16:50:28 +00:00
• 03ecad9486 full removal of dtype.vec (#16996) George HotzandGitHub 2026-07-12 09:25:50 -07:00
• bd1dc49112 Merge branch 'master' into cifar_30_clean George HotzandGitHub 2026-07-12 09:20:47 -07:00
• 246eb51523 more dtype.count removal from x86 and others (#16989) George HotzandGitHub 2026-07-12 09:10:02 -07:00
• d799b4108a 20 sec geohot 2026-07-12 15:11:11 +00:00
• 5a0751076d simpler vconst_like (#16990) chenyuandGitHub 2026-07-12 08:21:57 -04:00
• bf919c9430 30s speedrun! geohot 2026-07-11 21:29:25 +00:00
• 9c0460e4ee Revert "tile large transposed AMD GEMMs across six waves" geohot 2026-07-11 12:06:22 +00:00
• 1da9d8386d tile large transposed AMD GEMMs across six waves geohot 2026-07-11 11:58:35 +00:00
• 6eb762f5b3 speed up AMD channel reductions and irregular GEMM geohot 2026-07-11 11:32:45 +00:00
• 4a4fccb317 fuse shared reductions and large AMD graphs geohot 2026-07-11 09:59:34 +00:00
• 50a98c0f20 deduplicate optimized kernel lowering geohot 2026-07-11 08:52:27 +00:00
• 40b4377a14 tune long AMD GEMMs and reject irrelevant store hazards geohot 2026-07-11 08:11:46 +00:00
• 2e5354af22 deduplicate codegen and tile irregular AMD GEMMs geohot 2026-07-11 07:14:06 +00:00
• 24267914f4 compact AMD GEMM sources and tune long reductions geohot 2026-07-11 06:48:11 +00:00
• d9610230df reduce registers in long AMD GEMM geohot 2026-07-11 06:00:53 +00:00
• 2908c33a13 compile cooperative AMD kernels at O2 geohot 2026-07-11 05:56:39 +00:00
• a030286672 speed up AMD GEMM lowering and codegen geohot 2026-07-11 05:53:53 +00:00
• 8fedf152b1 10.6 seconds geohot 2026-07-10 18:04:29 +00:00
• f38cba1c08 16 seconds geohot 2026-07-10 14:32:30 +00:00
• fab262fb98 fix codegen correctness and lifetime regressions geohot 2026-07-10 02:00:42 +00:00
• 09b9e5e57d speed up CIFAR-10 example geohot 2026-07-10 01:44:30 +00:00
• a69186bdf4 speed up code generation and sorting geohot 2026-07-10 01:44:30 +00:00
• cae6696d75 llama: split current and next amax state (#16993) qazalandGitHub 2026-07-12 17:52:33 +08:00
• 9ea7de020b llama: fuse grad scale in gemm epilogue (#16992) qazalandGitHub 2026-07-12 17:15:01 +08:00
• 73b38854b4 llama: aT_b fp8 gemm (#16991) qazalandGitHub 2026-07-12 15:57:10 +08:00
• 7252a38b05 Merge branch 'master' into x86_numel x86_numel George HotzandGitHub 2026-07-11 16:54:49 -07:00
• 0af1afc43c remove dtype.count from x86 and others geohot 2026-07-11 16:54:17 -07:00
• e69ce4be7f switch x86 to use numel instead of dtype.count (#16985) George HotzandGitHub 2026-07-11 15:26:57 -07:00
• 047a467bf9 delete UOp._stack and UOp.vectorize [PR] (#16988) chenyuandGitHub 2026-07-11 15:26:55 -04:00
• 9d47014fd8 first class STACK [PR] (#16986) chenyuandGitHub 2026-07-11 13:55:32 -04:00
• bc68e54447 works? geohot 2026-07-11 09:45:30 -07:00
• ff9dd57a05 switch x86 to use numel instead of dtype.count geohot 2026-07-11 08:11:58 -07:00
• afeb5c708f x86 simplification (#16983) George HotzandGitHub 2026-07-11 08:11:05 -07:00
• 928af24b74 hcq2: mini speed ups (#16984) nimlgenandGitHub 2026-07-11 15:48:07 +03:00
• f1ccb85a27 ci: use llvm-20 in amd tests (#16982) sirhcmandGitHub 2026-07-11 02:26:10 -04:00
• a7b74ee593 remove slice from rangeify [PR] (#16981) sirhcmandGitHub 2026-07-11 02:01:38 -04:00
• 22c3f069ef Revert "clean up is_address" grok_x86_novec geohot 2026-07-10 20:45:26 -07:00
• e31d3dd1b1 Revert "works" geohot 2026-07-10 20:45:25 -07:00
• 57518e57d4 Merge branch 'master' into grok_x86_novec George HotzandGitHub 2026-07-10 20:42:55 -07:00
• a4079c2492 works geohot 2026-07-10 20:40:30 -07:00
• 43ad225d36 nv_610 support (glm) (#16979) George HotzandGitHub 2026-07-10 20:37:30 -07:00
• 75a4bfddc9 fp8 gemm tests including fused scales (#16980) qazalandGitHub 2026-07-11 11:33:21 +08:00
• 6e3f7e6a84 clean up is_address geohot 2026-07-10 20:11:23 -07:00
• 5e869b4a08 fix geohot 2026-07-10 16:44:23 -07:00
• 55bb64c427 remove vec from x86 backend (grok) [pr] geohot 2026-07-10 16:25:32 -07:00
• ff4e97c432 is grok good? no_vec_x86_4 geohot 2026-07-10 16:02:19 -07:00
• 96d5dcc80e gpt...ish geohot 2026-07-10 15:48:22 -07:00
• 8b2916344c Merge branch 'master' into no_vec_x86_4 George HotzandGitHub 2026-07-10 15:16:55 -07:00
• e3c8c98a44 cleanups geohot 2026-07-10 15:15:05 -07:00
• c7f011f881 fixes vecless_x86_glm geohot 2026-07-10 22:12:32 +00:00
• 4234a9d727 empty _get_clause is True [PR] (#16975) chenyuandGitHub 2026-07-10 18:03:39 -04:00
• 40de90ab19 lil changes to cifar (#16972) George HotzandGitHub 2026-07-10 14:19:46 -07:00
• 3aa9b07ca8 Merge branch 'master' into no_vec_x86_4 George HotzandGitHub 2026-07-10 14:10:04 -07:00