#1993 Community Bench: 2026-10-11-community-rx-6700-xt-qwen3-8-flash-next-iq3-xxs

本文

# Community benchmark on AMD Radeon RX 6700 XT

Measured on 2026-10-11 with `tools/strata_bench.py` (suite 1, 3 runs per test, medians with ranges). Speeds are the engine's own timings; first-token times are measured by the client over HTTP. Speed only: no correctness check was run.

## Hardware and software

- GPU and VRAM: AMD Radeon RX 6700 XT, 12 GiB VRAM
- CPU: AMD Ryzen Threadripper 3970X 32-Core Processor (32 cores, 64 threads); installed RAM: 251.5 GiB; storage: not measured
- OS: Linux 7.2.8-200.fc44.x86_64; GPU driver: not read; CUDA: not read
- Strata: project 0.1.43, commit 46894bb7; engine 0.1.43
- Background workloads and power limits: not recorded

## Model and configuration

- Model and quantization: qwen3.8-flash-next-iq3_xxs
- Context: 524288 tokens; parallel slots: 1
- Settings: the install's own configuration, unchanged; thinking off for the suite; temperature 0

## Method

Run by `python tools/strata_bench.py` against the HTTP API of a Strata server (already running); model loading is not timed. One warm-up chat and a 1K prompt are discarded first. Every prompt starts with a unique tag, so the conversation cache cannot help the fresh-prompt tests.

- short chat: the three questions of `tools/ab_engine.py`, up to 100 new tokens, 3 runs
- 4K / 16K prompts: generated word lists sized by a calibration request, 24 new tokens, 3 fresh prompts each
- agent session: system prompt and four tools, then 5 turns that each add about 1500 words of a project file and continue from the model's own earlier answers (the way agent clients resend a conversation), 64 new tokens per turn, 3 sessions; the cache numbers come from the answer's usage (`cached_tokens`) and timings (`prompt_n`)
- concurrency (when run): N requests sent at the same moment, total tokens over the wall time of the burst, latency over all requests

## Results

| Configuration | Actual prompt tokens | Reused tokens | Generated tokens | Runs | Prompt tok/s median and range | Decode tok/s median and range | TTFT seconds median and range |
| --- | ---: | ---: | ---: | ---: | --- | --- | --- |
| short chat | 46 (43-51) | 0 | 100 (100-100) | 9 | not measured | 33.1 (27.0-40.6) | 0.94 (0.86-1.02) |
| 4K prompt, fresh | 4019 (4019-4019) | 0 (0-0) | 24 cap | 3 | 350.3 (349.8-350.4) | not measured | 11.52 (11.51-11.54) |
| 16K prompt, fresh | 16058 (16058-16058) | 0 (0-0) | 24 cap | 3 | 357.7 (357.4-357.7) | not measured | 44.95 (44.94-44.99) |
| agent session, follow-up turns | 14512 (14500-14522) at the last turn | see below | up to 64 | 3 | see below | 34.5 (25.8-45.6) | 9.10 (5.17-10.41) |

Ranges are the lowest and highest of the runs; `Runs` counts requests that returned a token.

### Agent session (conversation cache)

Cache hit over the follow-up turns: 68.8 % (68.7-68.9); earlier-turn tokens read again per session: 4 (2-74) (earlier turns that the cache did not hold; the 13642 (13641-13643) new tokens of the files are read once); first turn (nothing cached) first token: 3.35 s (3.27-3.37).

| Turn (run 1) | Prompt tokens | Reused | Read | Prompt tok/s | First token (s) | Decode tok/s |
| ---: | ---: | ---: | ---: | ---: | ---: | ---: |
| 0 | 609 | 0 | 609 | 189.0 | 3.268 | 32.9 |
| 1 | 3720 | 640 | 3080 | 322.4 | 9.603 | 37.0 |
| 2 | 5253 | 3784 | 1469 | 283.0 | 5.239 | 33.9 |
| 3 | 8697 | 5316 | 3381 | 327.4 | 10.379 | 34.2 |
| 4 | 11628 | 8761 | 2867 | 317.0 | 9.1 | 39.2 |
| 5 | 14522 | 11676 | 2846 | 321.4 | 8.911 | 25.8 |

Per-request data: `results.json`. Engine log lines from the start: `engine-log-excerpt.txt` (when this run started the server).

## Correctness and limitations

Not measured: this is a speed suite. The conversation uses the model's own replies, so token counts differ a little between runs. Machine details come from the server, `nvidia-smi`/`rocm-smi` and the OS; background load and power limits are not recorded. Total time of the suite: 385 s.