Dontopedia

Gpu Utilization Dips

From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)

Gpu Utilization Dips has 4 facts recorded in Dontopedia across 2 references, with 1 live disagreement.

4 facts·3 predicates·2 sources·1 in dispute
Maturity scale raw canonical shape-checked rule-derived certified

Other facts (4)

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.

4 facts
PredicateValueRef
Caused byCpu Gpu Sync[1]
Caused byRouter Probs Operation[2]
Correspond toRouter Probs to Vec1[1]
Rdf:typePerformance Observation[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.

4 undated
—
causedByblah/safiersemantics/part-74
ex:cpu-gpu-sync
—
correspondToblah/safiersemantics/part-74
ex:router-probs-to-vec1
—
typeblah/safiersemantics/72
ex:PerformanceObservation
—
causedByblah/safiersemantics/72
ex:router-probs-operation

References (2)

2 references
  1. [1]Part 742 facts
    ctx:discord/blah/safiersemantics/part-74
  2. [2]722 facts
    ctx:discord/blah/safiersemantics/72
    • full textsafiersemantics-72
      text/plain2 KBdoc:agent/safiersemantics-72/ed67a7a2-21de-4492-a760-749e1ee367d2
      Show excerpt
      [2026-02-21 13:58] xenonfun: ``` 1. Persistent fp16 (not cast-per-matmul) Our fp16 failed because mm() does 3 dtype casts per matmul × ~120 matmuls = 360 extra GPU kernels. Instead: cast once at the start of each layer, keep intermediat…

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.

↑