Qwen3.8 27B

Run details

Source run ID

trpro-exl-qwen38-27b-5bpw-gpu1-5bpw-mtp-3-fixed-262144-allocation-failure-1-20261015

Page entry ID

trpro-exl-qwen38-27b-5bpw-gpu1-mtp-3-fixed-262144-tokens

Entry sweep

context

Raw entry label

Qwen3.8 27B EXL3 5.00bpw · MTP-3 · requested 262144-token allocation (startup OOM)

Configuration label

Single RTX3090 requested — placement unverified · EXL3 5.00bpw · MTP-3 · requested 262144-token fixed context allocation (startup OOM; actual allocation unknown)

Measured date

2026-10-15

Hardware

Not observed (outcome-only startup failure)

Runtime

exllamav3 · ExLlamaV3 1.5.0+cu128.torch2.9.0; native extension SHA-256 c1d02b410ef729618077ba287f01f05b229f31d118a7aae04a1b7e56142e3b2e

Weight placement

Unverified (startup failed)

Observed devices

None observed

Requested devices

CUDA0

Split mode

None

KV-cache precision

EXL native K8/V8 with QSA side planes and recurrent state

Prompt tokens

512

Generated tokens

256

Speculative decoding

MTP · draft max 3

Warm-ups

1

Measured repetitions

3

Timing method

ExLlamaV3 persistent worker time.perf_counter_ns with torch.cuda.synchronize at phase boundaries; fresh PP excludes prefix preparation; aggregate TG is across streams

Measurement outcomes

0 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.; 2,048 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.; 4,096 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.; 8,192 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.; 16,384 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.; 32,768 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.; 65,536 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.; 131,072 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.; 261,376 tokens · 1 stream: oom — RuntimeError: Insufficient VRAM in split for model and cache; the fixed 262144-token allocation failed during model/cache initialization. This planned prefix coordinate was not measured, so it does not establish this or any lower prefix as incapable.

Notes

Single RTX3090 requested — placement unverified · EXL3 5.00bpw · MTP-3 · requested 262144-token fixed context allocation (startup OOM; actual allocation unknown). Persistent EXL native worker: 512 fresh prompt tokens and 256 generated tokens; one warmup then three measured repetitions. PP is fresh-prompt processing after the populated prefix; TG is aggregate decode across concurrent streams. Throughput means and sample standard deviations are recomputed from the retained raw repetitions. EXL native K8/V8 cache storage includes QSA side planes and recurrent state; this is not GGUF Q8_0. Physical GPU 1 (GPU-0a2bed40-39e4-e67f-5234-868704f9704b) was selected as the single device. No successful model placement was observed for this outcome-only source. Source configuration 500bpw-context-mtp-3-c1; model revision a35e75a73baee51da709329d19294245cbeeb5d8; source manifest SHA-256 f889b720ea21d731842579e93d41eb5e4ed995a63475a7e69874a2330b529192. This separate outcome-only configuration requested the original fixed 262144-token per-stream context allocation. Actual allocation and model residency were not established because startup failed.

Source article

/benchmarks/

Raw log path

research/benchmarks/logs/trpro/2026-10-15-independent-gpus/remaining-campaign/trpro-exl-qwen38-27b-5bpw-gpu1-independent-gpus-validation-20261015/5.00bpw/context/mtp-3-c1