ATOM Mesh PD Benchmark
— SLURM nightly
Loading…
☼ Theme
Source
Model
Backend
Config
Precision
ISL
OSL
Date
Reset
Pareto (TPOT vs Throughput)
TTFT vs Concurrency
Time Series
Table
X
Interactivity (tok/s/user)
Per-user output throughput (tok/s/user)
Total token throughput (tok/s)
Output throughput (tok/s)
Concurrency
Y
Token Throughput per GPU (tok/s/gpu)
Mean TPOT (ms)
P99 TPOT (ms)
Mean TTFT (ms)
P99 TTFT (ms)
Mean E2E latency (ms)
Y
Mean TTFT (ms)
P99 TTFT (ms)
Mean TPOT (ms)
Output throughput (tok/s)
ISL
OSL
Concurrency
Ratio
Metric
Mean TPOT (ms)
Mean TTFT (ms)
Output throughput (tok/s)
Total token throughput (tok/s)
Mean E2E latency (ms)
GSM8K accuracy
Date
Config
ISL
OSL
Conc
Ratio
TTFT (ms)
TTFT P99
TPOT (ms)
TPOT P99
Output tok/s
Total tok/s
tok/s/user
GSM8K