Speed sweep

pg-019f3295-e1c8-7d72-b353-aa5f35d82692-sweep

pg-019f3295-e1c8-7d72-b353-aa5f35d82692-sweep

Record

Recipe
pg-mlx-community-nvidia-nemotron-3-nano-30b-a3b-mlx-8bit-8bit-apple-m3-ultra-96gb-80c-mlx-0742bd2b5c
Measured
2026-07-05T14:01:51.299Z
Points
4

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatus
18,25695.2observed
132,83287.5observed
1258,11152.1observed
132,76817,498.5observed

Remaining fields

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

id
pg-019f3295-e1c8-7d72-b353-aa5f35d82692-sweep
measured at
2026-07-05T14:01:51.299Z
recipe id
pg-mlx-community-nvidia-nemotron-3-nano-30b-a3b-mlx-8bit-8bit-apple-m3-ultra-96gb-80c-mlx-0742bd2b5c
schema version
local-ai-registry/v1

metrics

base memory bytes
34158707976
base memory context tokens
192
concurrency
1
decode32k context tokens
32,832
decode32k tps
87.531
decode8k context tokens
8,256
decode8k tps
95.183
decode max context tokens
258,111
decode max context tps
52.079
decode mode
non-mtp
inference engine version
unknown
latest point at
2026-07-05T14:01:51.299Z
max context tokens
258,111
max prompt tokens
258,047
memory8k bytes
37118912378
memory8k context tokens
8,256
memory max context bytes
40347049862
memory max context tokens
258,111
peak generation tps
87.531
peak memory bytes
40347049862
peak prompt tps
1,872.62
point count
223
ttft32k cached prompt tokens
28,736
ttft32k context tokens
32,768
ttft32k seconds
17.498

rows

concurrencycontext tokensdecode tok sdecode tok s per streamoutput tokenspeak vram gbprefill tok ssamples
18,25695.18395.183Unknown34.57Unknown1
132,83287.53187.531Unknown37.576Unknown1
1258,11152.07952.079Unknown37.576Unknown1
132,768UnknownUnknownUnknownUnknownUnknown1
Provenance & metadata (1)

source

paths
publication:pg-20260827T060320709Z, run:019f3295-e1c8-7d72-b353-aa5f35d82692
repository
local.ai