George Hotz 838cd078bc use atomics for embedding backward (#14400) ...

* embedding is slow

* failing

* float is fine

* null

* it fails

* simplify embedding with broadcasting

* ATOMIC_ADD incoming

* min change

* simpler test

* better test

* fix test

* real test

* simpler

* cleanups

* types and names

* _zero_kernel

* grad multi

* hack

* none

* multi unshard

* more for call

* don't tag in call

* good

* call_multi

* call_multi wow claude is useless

* embedding backward mutli test

* test passes

* fix as_param

* shape_to_shape_arg

* add clip

* before cast

* fix spec=2, use atomics

2026-01-30 18:10:59 +08:00

..

device

assembly/amd: bug fixes for PYTHON_REMU (#14347)

2026-01-27 00:48:22 +08:00

external

test for am recovery and faults handling (#14421)

2026-01-29 17:11:24 +03:00

mockgpu

_MockMemoryviewMeta for mockgpu (#14405)

2026-01-28 11:59:00 -05:00

models

Whisper audio helpers (mel filters in tinygrad) (#13478)

2026-01-20 10:50:02 -05:00

opt

amd tc 1616128 (#13439)

2026-01-02 09:01:05 -05:00

speed

ops_gpu -> ops_cl (#12103)

2025-09-10 15:15:48 -04:00

testextra

viz: sqtt work from cdna gemm (#14434)

2026-01-30 14:00:56 +09:00

unit

use atomics for embedding backward (#14400)

2026-01-30 18:10:59 +08:00

web

remove stale / untested viz related files (#13785)

2025-12-21 16:42:48 +08:00

__init__.py

All devices are equal! (#196)

2020-12-15 23:44:08 -08:00

Dockerfile

new cloud is cloudy [pr] (#7631)

2024-11-11 20:18:04 +08:00

helpers.py

remove REAL_DEV [pr] (#14337)

2026-01-26 10:08:16 -05:00

test_arange.py

use atomics for embedding backward (#14400)

2026-01-30 18:10:59 +08:00

test_compile_failures.py

remove ScheduleItem and merge it with ExecItem (#13759)

2025-12-19 17:04:24 -04:00

test_const_folding.py

remove dtypes.index_like and dtypes.fields [pr] (#14207)

2026-01-18 11:49:01 -05:00

test_custom_kernel.py

multi custom kernel: support input mixed with copy and shard (#13748)

2025-12-29 12:54:27 -05:00

test_dtype_alu.py

decompose long to ints where unsupported, try 2 (#14383)

2026-01-27 23:20:43 -05:00

test_dtype.py

decompose long to ints where unsupported, try 2 (#14383)

2026-01-27 23:20:43 -05:00

test_edgecases.py

decompose long to ints where unsupported, try 2 (#14383)

2026-01-27 23:20:43 -05:00

test_fusion_op.py

remove ScheduleItem and merge it with ExecItem (#13759)

2025-12-19 17:04:24 -04:00

test_gc.py

fix a few tests [pr] (#13498)

2025-11-29 13:43:45 -08:00

test_graph.py

amd: copies w/o sdma (#14036)

2026-01-06 21:15:58 +03:00

test_image_dtype.py

remove REAL_DEV [pr] (#14337)

2026-01-26 10:08:16 -05:00

test_interop.py

hotfix: interop example (#9237)

2025-02-25 10:32:00 +03:00

test_jit_cases.py

enumerate cases of Tensors in the JIT (#10548)

2025-05-28 11:51:27 -07:00

test_jit_footguns.py

raise if Tensor._buffer is called during jit (#14114)

2026-01-22 17:30:18 -05:00

test_jit.py

remove REAL_DEV [pr] (#14337)

2026-01-26 10:08:16 -05:00

test_kernel_cache.py

make method_cache account for compiler (#12156)

2025-09-13 17:00:11 +03:00

test_linearizer_dumb.py

add loads at the end (#12988)

2025-10-30 10:42:19 +08:00

test_linearizer_failures.py

require renderer argument in get_program, removes device opening in process replay [pr] (#13524)

2025-12-03 02:05:31 +08:00

test_linearizer.py

assembly/amd: emu speed (#14344)

2026-01-26 22:21:34 +08:00

test_memory_planner.py

memplan: fix large buffers planning (#11878)

2025-08-27 23:54:27 +03:00

test_method_cache.py

make method_cache account for compiler (#12156)

2025-09-13 17:00:11 +03:00

test_multitensor.py

use atomics for embedding backward (#14400)

2026-01-30 18:10:59 +08:00

test_nn.py

mark slow tests as slow instead of as CI (#13736)

2025-12-17 10:29:57 -04:00

test_ops.py

fix webgpu runtime types (#14402)

2026-01-28 10:37:25 -05:00

test_opt_gemm.py

remove ScheduleItem and merge it with ExecItem (#13759)

2025-12-19 17:04:24 -04:00

test_optim.py

mark slow tests as slow instead of as CI (#13736)

2025-12-17 10:29:57 -04:00

test_opts.py

require renderer argument in get_program, removes device opening in process replay [pr] (#13524)

2025-12-03 02:05:31 +08:00

test_outerworld_range.py

remove dtype from range, it will be dtypes.index soon [pr] (#11914)

2025-08-29 09:52:07 -07:00

test_outerworld.py

vmap on full model (#13340)

2025-11-18 16:06:06 -08:00

test_pickle.py

no sym in pm_reduce [pr] (#13398)

2025-11-20 16:49:09 -05:00

test_profiler.py

fix test_dev_jitter_matrix (#14255)

2026-01-20 20:07:51 +03:00

test_quantize_onnx.py

remove ScheduleItem and merge it with ExecItem (#13759)

2025-12-19 17:04:24 -04:00

test_randomness.py

assembly/amd: bug fixes for PYTHON_REMU (#14347)

2026-01-27 00:48:22 +08:00

test_rangeify.py

add device to local, fix PCONTIG=2 (#14266)

2026-01-21 22:12:18 +09:00

test_renderer_failures.py

no full_rewrite [pr] (#13809)

2025-12-22 23:20:01 -05:00

test_schedule.py

also return index in Tensor.cummax (#14117)

2026-01-12 22:42:10 -05:00

test_setitem.py

IGNORE_OOB -> CHECK_OOB [pr] (#14374)

2026-01-27 12:20:59 -05:00

test_softmax_fusion.py

remove ScheduleItem and merge it with ExecItem (#13759)

2025-12-19 17:04:24 -04:00

test_stunning.py

remove FUSE_ARANGE (#12511)

2025-10-08 04:54:07 -04:00

test_subbuffer.py

remove REAL_DEV [pr] (#14337)

2026-01-26 10:08:16 -05:00

test_symbolic_jit.py

don't allow jit input to be const (#14045)

2026-01-06 18:15:22 -05:00

test_symbolic_ops.py

minor type cleanups [pr] (#14408)

2026-01-28 14:11:50 -05:00

test_tensor_variable.py

fix test_symbolic_arange_sym_step (#13952)

2026-01-01 09:41:07 -05:00

test_tensor.py

remove dtypes.index_like and dtypes.fields [pr] (#14207)

2026-01-18 11:49:01 -05:00

test_tiny.py

beautiful SQTT_PARSE=1 with color (#13428)

2025-11-23 01:05:14 -08:00

test_to_numpy.py

Apply ruff linting rules to tests (#2473)

2023-11-27 21:24:06 -08:00

test_transcendental.py

update tests, WEBGPU=1 pytest . passes (#14089)

2026-01-10 00:03:02 -05:00

test_uop_graph.py

add wrapper class for the -0.0 != 0.0 issue (#14339)

2026-01-26 16:52:37 +08:00

test_uops_stats.py

no full_rewrite [pr] (#13809)

2025-12-22 23:20:01 -05:00

test_uops.py

DISABLE_FAST_IDIV actually works as a ContextVar (#14378)

2026-01-27 16:12:42 -05:00

test_zero_copy.py

unify cpu and llvm (#11982)

2025-09-09 13:54:44 +03:00