fused Metal kernel
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
fused Metal kernel has 27 facts recorded in Dontopedia across 2 references, with 3 live disagreements.
Mostly:specializes at compile time for(2), updates s and c(2), performs operation(2)
Maturity scale
raw canonical shape-checked rule-derived certifiedInbound mentions (4)
Other subjects in dontopedia point AT this entity as a value. These are inverse relationships — e.g. "X motherOf this subject" — and answer questions the forward facts can't. Grouped by predicate.
mentionedWritingOfMentioned Writing of(1)
- Xenonfun
ex:xenonfun
mentionsImplementationMentions Implementation(1)
- Message 1
ex:message-1
usesKernelUses Kernel(1)
- Decode Path Streaming
ex:decode-path-streaming
wroteWrote(1)
- It
ex:it
Other facts (26)
The long tail: predicates that appear too rarely to warrant their own section. Filter or scroll to find a specific one. Each row links to its source.
| Predicate | Value | Ref |
|---|---|---|
| Specializes at Compile Time for | M Equals 32 | [1] |
| Specializes at Compile Time for | M=32/64 | [1] |
| Updates S and C | compute output in one dispatch | [1] |
| Updates S and C | Compute Output | [1] |
| Performs Operation | Update S C | [2] |
| Performs Operation | Compute Output | [2] |
| Causes15 To17 Percent Speedup | Old Implementation | [1] |
| Enables One Dispatch | Update S C and Output | [1] |
| Has Template Params for | M 32 64 | [1] |
| Has Threads Parallelize Across | D | [1] |
| Implemented by It | It | [1] |
| Uses One Threadgroup Per | Bh | [1] |
| Shares Via | Threadgroup Memory | [1] |
| Has Thread Zero Handle | Denominator | [1] |
| Parallelizes Threads Across | D Dimension | [1] |
| Dispatches in One Call | Update S C Compute Output | [1] |
| Depends on Threadgroup Memory for Sharing | Denominator | [1] |
| Presupposes Old Implementation Exists | Old Implementation | [1] |
| References Apple Metal Framework | Metal Api | [1] |
| Supports Template Params for | M Equals 32 64 | [1] |
| Rdf:type | Code Component | [2] |
| Dispatch Mode | One Dispatch | [2] |
| Thread Group Assignment | One Threadgroup Per Bh | [2] |
| Thread Parallelization | Parallelize Across D | [2] |
| Role of Thread0 | Handles Denominator | [2] |
| Has Template Params | Compile Time Specialization | [2] |
Timeline
Timeline axis is valid_time — when each source says the fact was true in the world, not when Dontopedia learned about it. Retracted rows are kept for provenance; coloured stripes indicate the context kind.
References (2)
ctx:discord/blah/watt-activation/part-77ctx:discord/blah/watt-activation/77- full textwatt-activation-77text/plain3 KB
doc:agent/watt-activation-77/59dddcca-2e06-4d74-9254-03846e959489Show excerpt
[2026-03-07 18:47] xenonfun: It wrote a fused metal kernel, tho I think we are going to have it also do mlx path shortly: ``` Architecture Summary Training path — chunked prefix scan: - _anchor_kan_forward_chunked() processes chunks o…
See also
Keep researching
Missing something or suspicious of what's here? Kick off a research session — a Claude agent will investigate, cite its sources, and file new facts into a dedicated context you can review before accepting into the shared view.