Skip to main content

Benchmark Planning

Plan the matrix before creating workloads and scenarios. The goal is to make every result explainable and reproducible.

Model And Modality

Record both the model identity and the serving shape:

DimensionExamplesWhy it matters
Modeldeepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B, Qwen/Qwen3-32BDetermines memory, tokenizer, prompt format, and license.
ModalityLLM, VLM, ASR, image, KYAISelects benchmark tool, scenario fields, and report interpretation.
PrecisionBF16, FP16, FP8, INT8, AWQ, GPTQChanges memory, speed, and quality.
Context4k, 32k, 128kDetermines KV cache pressure and scheduling behavior.
Endpointlocal vLLM, Dynamo, OpenAI-compatible API, OpenRouterChanges auth, latency path, and command templates.

Traffic Shape

For LLM serving, define at least:

  • input sequence length (ISL);
  • output sequence length (OSL);
  • concurrency;
  • total requests;
  • streaming or non-streaming;
  • prompt source and sampling parameters.

Common sweep grid:

ISLOSLConcurrency
12812832, 256, 512, 1024
128102432, 256, 512, 1024
1024102432, 256, 512, 1024
102412832, 256, 512, 1024

Create one scenario per grid point. Keep scenario codes machine-sortable, such as isl128-osl1024-c256.

Tool Selection

Use the tool that matches the question:

ToolUse when
metrumbench-llmOpenAI-compatible text generation throughput and latency.
GenAI-PerfNVIDIA-oriented performance tests and comparison with existing GenAI-Perf runs.
KYAIQuality, instruction-following, and judge-based evaluation.
metrumbench-imagegenImage or multimodal workflows covered by metrumbench-imagegen runbooks.
metrumbench-vlmPrompt-set based custom workloads.

Success Criteria

Pick metrics before the run:

  • throughput: tokens/sec, requests/sec;
  • latency: TTFT, TPOT, p50, p95, p99;
  • quality: KYAI pass rate, judge score, failure modes;
  • utilization: GPU utilization, memory, power, CPU, network;
  • cost: hourly cost, cost per 1M tokens, tokens per dollar;
  • reliability: error rate, timeout rate, restart count.

Minimal Planning Record

Question: Can DeepSeek R1 Distill Qwen 1.5B meet 256-concurrency chat traffic on 1x RTX PRO 6000?
Project: deepseek-qwen15b-rtxpro6000-smoke
Model: deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
Framework: vLLM latest, OpenAI-compatible server
Precision: auto/BF16 baseline, FP8 follow-up
Hardware: specific 1x RTX PRO 6000 instance with recorded CPU/RAM/disk/NIC
Scenarios: ISL/OSL 128/128, 128/1024, 1024/1024, 1024/128 at c32/c256/c512/c1024
Success: no job failures, TTFT and p95 under target, GPU memory below limit, report package complete