Back to search results
Llama-3.1-8B-Instruct · IQ3_S
NVIDIA RTX 509032GBStrata v0.22L2 Cross-framework verified
This page aggregates 1 real-world runs of Llama-3.1-8B-Instruct (IQ3_S) on NVIDIA RTX 5090 with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.
280 tok/s
Decode
Decode speed
1,358.66 tok/s
Prefill
Prefill speed
1.41 s
TTFT
Time to first token
13.19 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
16 days ago
Last verified
Performance
- Ollama · Q5_K_S Decode 220.5 · Prefill 854 ·
- Strata · IQ3_S (current)Decode 280 · Prefill 1358.66 ·
- Strata · Q5_K_M Decode 297.96 · Prefill 1229.8 ·
- vLLM · IQ4_XS Decode 310.8 · Prefill 1667.93 ·
- Ollama · IQ1_S Decode 316.88 · Prefill 1203.93 ·
- llama.cpp · Q8_0 Decode 162.5 · Prefill 1218.8 ·
- llama.cpp · Q5_K_M Decode 236.9 · Prefill 1776.8 ·
- llama.cpp · Q4_K_M Decode 285.3 · Prefill 2139.8 ·
Core figures
- Decode (avg)
- 280 tok/s
- Prefill (avg)
- 1,358.66 tok/s
- TTFT (avg)
- 1.41 s
- VRAM (avg)
- 13.19 GB
- MTP acceptance rate
- 0.56
- TTFB
- 2.24 GB
- Power draw
- 429.1 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Llama-3.1-8B-Instruct
- Quantization
- IQ3_S
- Framework
- Strata
- Version
- v0.22
- Context length
- 8192 tokens
- Batch size
- 1024
- GPU layers
- 15
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA RTX 5090
- Nominal VRAM
- 32 GB
- Measured VRAM (avg)
- 13.19 GB
- OS
- Ubuntu 24.04
- Driver
- 572.83
- CUDA
- 12.6
- Power draw
- 429.1 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.22 · Ubuntu 24.04 · CUDA 12.6 · 8192 ctx
280 tok/s
Decode
1,358.66 tok/s
Prefill
1.41 s
TTFT
13.19 GB
VRAM
0.56
MTP
429.1 W
Power draw
L2 batch seed 2026-10-19: Strata on hw_id=13 running model_id=68 at IQ3_S