Speed sweep

qwen3-6-35b-a3b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1-sweep

qwen3-6-35b-a3b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1-sweep

Record

Recipe
qwen3-6-35b-a3b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1
Measured
2026-06-29T18:23:54.994Z
Points
1

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatus
18,192894.275.98,653observed

Remaining fields

Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.

id
qwen3-6-35b-a3b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1-sweep
measured at
2026-06-29T18:23:54.994Z
recipe id
qwen3-6-35b-a3b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1
schema version
local-ai-registry/v1

metrics

concurrency
1
inference engine version
toolbox-rocm-7.2.4-post-gtt
latest point at
2026-06-29T18:23:54.994Z
max context tokens
8,192
peak generation tps
75.89
peak prompt tps
894.22
point count
1

rows

concurrencycontext tokensdecode tok sdecode tok s per streamoutput tokenspeak vram gbprefill tok ssamples
18,19275.8975.89128Unknown894.221
Provenance & metadata (1)