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

• de13db7e6a lil geohot 2026-08-04 01:19:59 +00:00
• 67dc02d7e7 bitcast in python for _bits_to_rand [PR] (#17383) chenyuandGitHub 2026-08-03 21:14:58 -04:00
• ca78469398 cleanups geohot 2026-08-04 01:08:49 +00:00
• 70ed0a70e0 llm: minimize recurrent prefill integration geohot 2026-08-03 23:47:39 +00:00
• 69834abcc4 Merge remote-tracking branch 'origin/master' into qwen36_27b_amd_900 geohot 2026-08-03 23:11:44 +00:00
• 94e4d3c92c llm: move backend policy behind kernel API geohot 2026-08-03 23:11:39 +00:00
• 3cb786f447 llm: update test_llm_server tests (#17382) George HotzandGitHub 2026-08-03 16:06:17 -07:00
• d5d500a1fc Merge remote-tracking branch 'origin/master' into qwen36_27b_amd_900 geohot 2026-08-03 22:44:23 +00:00
• e8328c7498 llm: keep recurrent prefill portable geohot 2026-08-03 22:43:08 +00:00
• 87289a7410 hotfix: revert test_scalar_alu_index, violates spec geohot 2026-08-03 15:25:51 -07:00
• c36ccead6e llm: use upstream scalar index fix geohot 2026-08-03 22:15:06 +00:00
• 22bc5b40b2 Merge remote-tracking branch 'origin/master' into qwen36_27b_amd_900 geohot 2026-08-03 22:13:36 +00:00
• c2625c78cb scalar ALU index fix + llm: preserve_thinking (#17381) George HotzandGitHub 2026-08-03 15:10:21 -07:00
• 33755a3465 improve threefry codegen [pr] (#17379) chenyuandGitHub 2026-08-03 17:24:25 -04:00
• 104ee90ccf usb: wait for PCIe link after power on (#17380) YassineYousfiandGitHub 2026-08-03 14:19:05 -07:00
• 74fd4c2aa1 simpler geohot 2026-08-03 21:01:07 +00:00
• 69008bcaa5 Merge remote-tracking branch 'origin/master' into qwen36_27b_amd_900 geohot 2026-08-03 20:30:39 +00:00
• 1b1bc4ce4c simpler geohot 2026-08-03 20:26:26 +00:00
• a2385ae21d MAX_LINE_COUNT=26000 (#17378) chenyuandGitHub 2026-08-03 15:37:58 -04:00
• 3331944547 gptoss: fix moe routing (#17377) wozeparrotandGitHub 2026-08-04 02:16:44 +08:00
• f7579a2040 Merge remote-tracking branch 'origin/master' into qwen36_27b_amd_900 geohot 2026-08-03 16:33:10 +00:00
• f26128e8d2 400 lines geohot 2026-08-03 16:28:20 +00:00
• 79ef0a16c0 +499 lines geohot 2026-08-03 15:02:58 +00:00
• 7c1ce50f63 hcq2: epoch (#17376) nimlgenandGitHub 2026-08-03 16:37:25 +03:00
• be5f62d269 llama: refactor amax stuff and skip in fp4 (#17375) qazalandGitHub 2026-08-03 19:05:52 +08:00
• e22935c758 hcq2: inputs table (#17374) nimlgenandGitHub 2026-08-03 13:40:43 +03:00
• 314df72b5f deflake test_hcq with MOCKGPU (#17370) chenyuandGitHub 2026-08-03 06:22:59 -04:00
• 0dda620216 llm: fix recurrent rangeify and client cache reuse geohot 2026-08-03 03:30:32 +00:00
• 6414b1b2e1 Merge remote-tracking branch 'origin/master' into qwen36_27b_amd_900 geohot 2026-08-03 03:03:46 +00:00
• 23c7813f44 no hard coded dtype int for rangeify debuf [PR] (#17373) chenyuandGitHub 2026-08-02 20:43:04 -04:00
• de37fe0d8f kernel cleanups geohot 2026-08-02 22:00:18 +00:00
• 269f821421 llm: retain master model formatting geohot 2026-08-02 20:25:41 +00:00
• 41b7a02103 Merge remote-tracking branch 'origin/master' into qwen36_27b_amd_900 geohot 2026-08-02 20:12:15 +00:00
• 9230322266 llm: keep sequential AMD startup under a minute geohot 2026-08-02 20:04:55 +00:00
• 05bc7c6994 fix llm reasoning and Linear import (#17372) George HotzandGitHub 2026-08-02 12:31:52 -07:00
• 09dabfe05e minor pm_float_decomp cleanup [PR] (#17371) chenyuandGitHub 2026-08-02 15:17:33 -04:00
• ffc4507fb4 llm: recover packed weights in linear geohot 2026-08-02 19:06:45 +00:00
• 59df317b12 use UOp.const to create new consts [PR] (#17368) chenyuandGitHub 2026-08-02 13:31:55 -04:00
• 71869224b4 llm: share AMD quant kernel helpers geohot 2026-08-02 15:53:39 +00:00
• c64f8aeff8 Merge remote-tracking branch 'origin/master' into qwen36_27b_amd_900 geohot 2026-08-02 15:38:58 +00:00
• 82025dfe1f llm: keep 131k Qwen cache on AMD geohot 2026-08-02 15:38:04 +00:00
• 756a936be2 llm: report interrupted stream statistics geohot 2026-08-02 15:12:29 +00:00
• ce071d7f0e llm: fix recurrent server startup and reasoning routing geohot 2026-08-02 15:03:17 +00:00
• 21f91bac40 llm: reuse generic Qwen embedding geohot 2026-08-02 14:44:24 +00:00
• 73f0aa7d45 llm: add Qwen AMD acceptance benchmark geohot 2026-08-02 14:35:04 +00:00
• d1e1d587af llm: reuse standard sampling path geohot 2026-08-02 14:32:51 +00:00
• 8f3337c46f llm: checkpoint recurrent prompt state geohot 2026-08-02 14:30:06 +00:00
• f92ac85fbd llm: simplify AMD Qwen kernels geohot 2026-08-02 14:17:43 +00:00
• fb53442c17 llm: use standard RoPE table allocation geohot 2026-08-02 14:13:06 +00:00
• eba359128e llm: reuse master recurrent cache semantics geohot 2026-08-02 14:09:36 +00:00
• 3804a87b60 llm: keep Qwen warmup changes scoped geohot 2026-08-02 07:36:53 +00:00
• 9d02a3fe68 llm: remove unused Qwen AMD paths geohot 2026-08-02 07:34:56 +00:00
• 35dbe53d5d llm: trim unused AMD quant paths geohot 2026-08-02 06:22:29 +00:00
• ca29776f74 llm: fast packed Qwen3.6 inference on AMD geohot 2026-08-02 01:42:12 +00:00
• e14cadb1fb gptoss: set ASM_GEMM (#17363) wozeparrotandGitHub 2026-08-02 21:52:36 +08:00
• c9e1d82100 llm: keep Qwen warmup changes scoped qwen36_27b_amd_minimal geohot 2026-08-02 07:36:53 +00:00
• 804f5453e8 llm: remove unused Qwen AMD paths geohot 2026-08-02 07:34:56 +00:00
• 5767e5b1d3 llm: trim unused AMD quant paths geohot 2026-08-02 06:22:29 +00:00
• 5985c02005 llm: fast packed Qwen3.6 inference on AMD geohot 2026-08-02 01:42:12 +00:00
• 0258c7fefc minor argstr and alloc stack cleanup [PR] (#17361) chenyuandGitHub 2026-08-01 22:37:06 -04:00
• 475ad15f28 27b 935/47 llm_262k_minimal geohot 2026-08-02 00:36:02 +00:00
• fb607fb990 faster devectorizer with one line (#17360) George HotzandGitHub 2026-08-01 13:06:35 -07:00
• f50d40014c faster devectorizer with one line speed_one_linear geohot 2026-08-01 20:00:55 +00:00
• bc589c017c add threading to the http server http_threading geohot 2026-08-01 12:07:13 -07:00
• 15c936db01 merge devectorize + indexing (#17354) George HotzandGitHub 2026-08-01 11:40:15 -07:00
• 1ec33e2ef6 wip: test change, do not merge szdiff_bot_test geohot 2026-08-01 10:50:45 -07:00
• fd62ad3034 Merge pull request 'szdiff: make workflow work on both github and gitea (supersedes #8)' (#9) from szdiff_portable into master geohot 2026-08-01 10:49:44 -07:00
• 60996454a3 szdiff: make comment steps work on both github and gitea geohot 2026-08-01 10:44:41 -07:00
• a54bb3b795 fix szdiff workflow for Gitea: use local master and REST API comments geohot 2026-07-30 09:00:54 -07:00
• 98b700bad1 gptoss optim fixes (#17356) wozeparrotandGitHub 2026-08-02 00:10:06 +08:00
• 8e524ca467 CONST related cleanups [pr] (#17352) chenyuandGitHub 2026-08-01 09:54:28 -04:00
• 665822ab34 hcq2: faster replace (#17353) nimlgenandGitHub 2026-08-01 16:32:04 +03:00
• 7727698486 test tiny passes geohot 2026-07-31 23:59:17 -07:00
• 6c0ec39279 UOp.is_invalid [PR] (#17351) chenyuandGitHub 2026-08-01 02:36:32 -04:00
• b502fc1367 more const arg -> val (#17350) chenyuandGitHub 2026-08-01 02:18:10 -04:00
• 161783d8f7 add _device_num back to ast.variables (kimi) (#17327) qazalandGitHub 2026-08-01 13:56:50 +08:00
• 5a1c641f79 more arg -> val (#17349) George HotzandGitHub 2026-07-31 22:44:24 -07:00
• 20b8ecff50 support new USB vendor ID (#17348) Adeeb ShihadehandGitHub 2026-07-31 21:45:06 -07:00
• 9082ecef5d use .val to access the value of Ops.CONST (#17347) George HotzandGitHub 2026-07-31 19:51:42 -07:00
• 099d69ff7d ci: split macos unit test into metal and mock runners (#17346) George HotzandGitHub 2026-07-31 19:41:18 -07:00
• a88f832f0c remove UOp.val (#17345) George HotzandGitHub 2026-07-31 18:58:38 -07:00
• 850989115d __int__ and __float__ work for weak (#17342) chenyuandGitHub 2026-07-31 19:56:37 -04:00
• 15d515299e heuristics: try multiple TC axes (#17341) sirhcmandGitHub 2026-07-31 19:25:18 -04:00
• 6d526f0252 /raid/models/Qwen3.6-27B-IQ4_XS.gguf works at 704/40 geohot 2026-07-31 21:25:09 +00:00
• ceb2c0e673 contig is copy geohot 2026-07-31 14:22:26 -07:00
• 85ced44db6 tc: don't allow reduce over output dims (#17340) sirhcmandGitHub 2026-07-31 16:48:51 -04:00
• 165ede2471 extra geohot 2026-07-31 13:35:49 -07:00
• c22d91760e fix cat geohot 2026-07-31 13:28:52 -07:00
• 2c97b39529 pad ish geohot 2026-07-31 13:23:33 -07:00
• 277433259e fix sym_infer for CAST (#17338) chenyuandGitHub 2026-07-31 14:38:38 -04:00
• 8dc225e28e dtype_from_uop(INS) is None [PR] (#17337) chenyuandGitHub 2026-07-31 14:33:15 -04:00
• 85a1e25004 pad issues geohot 2026-07-31 11:30:44 -07:00
• 394943eeaf simple rangeify is simple geohot 2026-07-31 10:29:55 -07:00
• ad24750487 const(value, dtype) -> const(value).cast(dtype) in tests (#17335) chenyuandGitHub 2026-07-31 13:24:31 -04:00
• a11ee26bb8 viz: prep for faster cli DEBUG=3 (#17334) qazalandGitHub 2026-08-01 01:16:30 +08:00
• b95bd5b2a5 don't reset chestnut in benchmark (#17333) sirhcmandGitHub 2026-07-31 12:39:04 -04:00
• 9540700bbc direct lds hk_gemm geohot 2026-07-31 09:03:32 -07:00
• 1095bbe409 hcq2: fix ib reuse (#17330) nimlgenandGitHub 2026-07-31 18:43:23 +03:00
• 7f4dbb8090 remove shape= from UOp.const [PR] (#17331) chenyuandGitHub 2026-07-31 11:18:59 -04:00
• 8e86cfa60e /raid/models/Qwen3.6-35B-A3B-UD-IQ4_XS.gguf geohot 2026-07-31 14:33:37 +00:00