#2044 Community Bench: 2026-10-12-community-2x-2080ti-22gb-qwen3-8-flash-next-gsq-rco-iq3 (Proxmox 9 VM)

Descripción

[engine-log-excerpt.txt](https://github.com/user-attachments/files/33320588/engine-log-excerpt.txt)
[README.md](https://github.com/user-attachments/files/33320589/README.md)
[results.json](https://github.com/user-attachments/files/33320587/results.json)

## Hardware and software

- GPU and VRAM: 2x NVIDIA GeForce RTX 2080 Ti 22GiB

, 44 GiB VRAM
- CPU: AMD EPYC 7F52 16-Core Processor (32 cores, 32 threads); installed RAM: 94.3 GiB; storage: not measured
- OS: Linux 6.12.0-211.64.1.el10_2.x86_64; GPU driver: 615.71.09; CUDA: 13.4
- Strata: project 0.1.43, commit 7695859a; engine 0.1.43
- Background workloads and power limits: not recorded

## Model and configuration

- Model and quantization: qwen3.8-flash-next-iq3_s
- Context: 204800 tokens; parallel slots: 1
- Settings: the install's own configuration, unchanged; thinking off for the suite; temperature 0

```text
strata --pack iq3_s --native Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf --ple-gguf Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf --expert-profile expert-profile.bin --expert-cache auto --prefill auto --spec 4 --mtp rt --max-context 204800 --kv int8 --kv-resident 32768 --spec-min-p 0.5
```

## Method

Run by `python tools/strata_bench.py` against the HTTP API of a Strata server (started by the benchmark from the install configuration); 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 | 72.5 (63.7-94.1) | 0.39 (0.35-0.43) |
| 4K prompt, fresh | 4019 (4019-4019) | 0 (0-0) | 24 cap | 3 | 955.0 (901.4-956.1) | not measured | 4.24 (4.23-4.49) |
| 16K prompt, fresh | 16058 (16058-16058) | 0 (0-0) | 24 cap | 3 | 1375.1 (1354.1-1376.7) | not measured | 11.71 (11.70-11.90) |
| agent session, follow-up turns | 14465 (14456-14467) at the last turn | see below | up to 64 | 3 | see below | 68.9 (53.5-88.4) | 3.79 (2.52-4.54) |

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 % (68.6-68.8); earlier-turn tokens read again per session: 4 (4-74) (earlier turns that the cache did not hold; the 13634 (13634-13635) new tokens of the files are read once); first turn (nothing cached) first token: 1.69 s (1.68-1.77).

| Turn (run 1) | Prompt tokens | Reused | Read | Prompt tok/s | First token (s) | Decode tok/s |
| ---: | ---: | ---: | ---: | ---: | ---: | ---: |
| 0 | 609 | 0 | 609 | 348.9 | 1.772 | 63.1 |
| 1 | 3713 | 632 | 3081 | 723.7 | 4.295 | 83.8 |
| 2 | 5246 | 3776 | 1470 | 538.2 | 2.772 | 68.4 |
| 3 | 8690 | 5309 | 3381 | 750.1 | 4.544 | 59.9 |
| 4 | 11595 | 8735 | 2860 | 709.7 | 4.07 | 70.6 |
| 5 | 14465 | 11619 | 2846 | 712.8 | 4.037 | 70.7 |

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