13,471 commits

This branch

This branch

All branches

Author	SHA1	Message	Date
qazal	8ba58304f7	viz: reenable tests (#15626)	2026-04-07 07:52:44 +09:00
chenyu	2f7d085450	shared _normalize_indices for getitem (#15625) ... * shared _normalize_indices for getitem * list	2026-04-06 17:45:36 -04:00
chenyu	66ec188d50	more activations to mixin (#15624)	2026-04-06 15:41:41 -04:00
chenyu	1483f7e71c	support shift by Tensor (#15623) ... * support shift by Tensor * use mixin	2026-04-06 15:14:57 -04:00
chenyu	6e30a5f5ea	update shifts in torch backend (#15622)	2026-04-06 14:08:33 -04:00
chenyu	a444be172d	lower fuzz_symbolic_symbolic_div timeout (#15619) ... mitigate timeout crash due to high total time	2026-04-06 12:58:29 -04:00
chenyu	01b49c8647	support int operand for shifts (#15618) ... matches torch/jax, also symbolic rule to remove mask	2026-04-06 12:32:12 -04:00
nimlgen	e2700475cf	mlx: cleaner (#15617) ... * mlx: cleaner * x	2026-04-06 17:49:47 +03:00
Valtteri Valo	86c4431d74	add gpu_family detection to Metal, target MSL 4.0 on macOS 26+ (#15079) ... use supportsFamily API to detect GPU generation instead of parsing ICB debug description strings. also adds metal4.0 compiler target.	2026-04-06 06:51:38 +08:00
13Perrius	ff0c941548	remove redundant iteration and toposort in _deepwalk (#15532)	2026-04-06 06:38:45 +08:00
Andrew Cappelli	e39cfe685a	validate lr, momentum, weight_decay in optimizers (#15576)	2026-04-06 06:37:34 +08:00
nimlgen	6a334ceb27	hotfix: fix bert (#15613)	2026-04-05 23:41:21 +03:00
nimlgen	e3986a6b74	mlx: init runtime (#15612) ... * mlx: init * x * swap	2026-04-05 22:52:29 +03:00
nimlgen	e0988dbae5	hcq: support non for signal_t and compute_t (#15611) ... * hcq: support non for signal_t and compute_t * revert * x	2026-04-05 18:56:47 +03:00
nimlgen	5e134aa087	hcq: add write/poll_bit commands (#15610) ... * hcq: add write/poll_bit commands * x	2026-04-05 18:09:44 +03:00
nimlgen	604cdbf2f7	am: large allocs aligned to 2mb to use 2mb pages (#15609)	2026-04-05 18:01:31 +03:00
qazal	b2d5b29f45	assembly/amd: validate dsl keyword args (#15608) ... * assembly/amd: validate dsl keyword args * hm, this should use the SOP2 s_waits * use the sop2 s_waits	2026-04-05 23:00:24 +09:00
qazal	056fcd7758	viz: web work from rdna4 gemm (#15607) ... * add rdna4 barrier * fix realtime	2026-04-05 19:14:16 +09:00
wozeparrot	7e54992bf6	fp8 llama (#15588) ... Co-authored-by: qazal <qazal.software@gmail.com>	2026-04-04 18:24:57 -07:00
qazal	4d36366717	assembly/amd: match rdna4 hw gidx init in emulator (#15604) ... * simple rdna4 copy kernel with hw fault * the trivial fix: use ttmp instead of s * now copy kernel fails in mockgpu * rm crashing kernel	2026-04-05 02:28:18 +09:00
chenyu	2ba5a6ddc8	remove detach in selu (#15602) ... UOp does not have detach. this does not change behavior	2026-04-04 11:04:29 -04:00
qazal	f7aed180e4	viz/cli: add Other row in profiler (#15600)	2026-04-04 22:40:53 +09:00
Christopher Milan	74ecf6d3e6	opaque structs are also c.Struct (#15596)	2026-04-03 19:40:43 -04:00
Christopher Milan	645d45d968	DEV has arch (#15577) ... Co-authored-by: Comma Device <device@comma.ai>	2026-04-03 19:17:19 -04:00
nimlgen	902edc3781	hcq: hcqbuf in copy (#15595)	2026-04-03 22:47:36 +03:00
nimlgen	2c4271209e	hcq: peer groups for remote (#15594) ... * hcq: set real peer group * x * x * x	2026-04-03 19:03:07 +03:00
chenyu	8fdef2d3e4	mean/std/var to mixin (#15593)	2026-04-03 10:42:41 -04:00
qazal	9920b42b5e	hotfix: renderer.target.arch in disasm (#15592)	2026-04-03 22:23:51 +09:00
nimlgen	237084b276	remote: support several hosts (#15585) ... * remote: support several hossts * f	2026-04-03 11:22:15 +03:00
Christopher Milan	0ed8d9271d	Renderers accept Target or nothing (#15590)	2026-04-03 01:09:41 -04:00
wozeparrot	3a26920141	feat: framework ci (#15589)	2026-04-02 22:03:51 -07:00
Christopher Milan	736fea8412	select_first_inited cleanup and better errors (#15587)	2026-04-02 19:27:58 -04:00
Christopher Milan	8c50da800d	[pr] cleanup unused ctx's in codegen (#15586)	2026-04-02 19:06:58 -04:00
nimlgen	694dc5a717	install script in benchmark (#15584)	2026-04-02 18:15:58 +03:00
nimlgen	046c3f1240	mlx: add loopback with send/recv (#15583)	2026-04-02 18:15:46 +03:00
chenyu	c64226e97c	fix CreationMixin doc (#15582)	2026-04-02 09:46:28 -04:00
qazal	fefb0ebc2a	gemm/asm: fp8 cleanups (#15580) ... * normal gemm here * s/dtypes.fp8e4m3/FP8_DTYPE * gemm_bw * device UOp stays NULL	2026-04-02 19:02:38 +09:00
chenyu	61bc91aa8c	Tensor cumalu cleanups (#15579) ... * Tensor cumalu cleanups * happy	2026-04-02 05:23:22 -04:00
chenyu	1aa04eab08	simple CreationMixin (#15567) ... start with full_like, zeros_like, ones_like	2026-04-01 23:00:56 -04:00
wozeparrot	5b2a3251c4	mlperf system json for mi350 (#15575)	2026-04-01 15:30:33 -07:00
Christopher Milan	6c67bd4c14	better error message when invalid renderer is specified (#15573)	2026-04-01 17:12:55 -04:00
Christopher Milan	0d6fbc2355	remove flaky and redundant image test (#15574)	2026-04-01 16:33:13 -04:00
Christopher Milan	20f7f0be8e	nir renderers use arch (#15556) ... * nir renderers use arch * fix * fix null	2026-04-01 16:32:51 -04:00
nimlgen	148ad09559	am: do not use dbell for ih (#15571)	2026-04-01 21:34:21 +03:00
nimlgen	93a85c7348	am: raise when using more sdma engines (#15569)	2026-04-01 21:33:42 +03:00
nimlgen	da12c2ea16	better install msg (#15570)	2026-04-01 20:09:37 +03:00
b1tg	20497f2840	fold BIND to CONST when min==max (#15568)	2026-04-01 11:19:04 -04:00
qazal	9275f283e5	viz: update flag and display names (#15566) ... * rename to occ, other_simd * se pkts * match viz cli tool in names	2026-04-01 21:48:37 +09:00
chenyu	f5c0794df2	fix Tensor.const_like (#15565) ... used to always return a 0-d tensor, now returns an expanded Tensor based on self.shape and matches UOp	2026-04-01 08:35:19 -04:00
qazal	09f60d80fd	llama: fix FP8=1 FAKEDATA=1 (#15564)	2026-04-01 20:53:03 +09:00