Datasets:
metadata
license: mit
task_categories:
- text-generation
tags:
- benchmark
- inference
- llm
- nvidia
- rtx-5090
- llama-cpp
- vllm
- speed
- moe
size_categories:
- n<1K
configs:
- config_name: default
data_files:
- split: train
path: benchmarks.csv
RTX 5090 LLM Benchmarks
Speed benchmarks for quantized LLMs on NVIDIA RTX 5090 32GB, measured with llm-bench-rig.
What's measured
- Prompt processing (pp): parallel batched token throughput at context lengths 128, 512, 2048, 4096, 8192, 16384
- Text generation (tg): sequential autoregressive token throughput at 128 tokens
- All models fully GPU-offloaded (ngl=99)
Hardware
| Component | Spec |
|---|---|
| GPU | NVIDIA GeForce RTX 5090 32GB (Blackwell, sm_120a) |
| CPU | AMD Ryzen 5 9600 (6c/12t) |
| RAM | 64GB DDR5-5600 |
| OS | Ubuntu 26.04 LTS |
| CUDA | 12.8 (patched for glibc 2.41) |
Schema
| Column | Description |
|---|---|
model |
Model name |
architecture |
Dense or MoE (with active param count) |
params_b |
Total parameters in billions |
quant |
Quantization method |
size_gib |
File size in GiB |
engine |
Inference engine (llama.cpp or vLLM) |
backend |
Compute backend (CUDA) |
gpu |
GPU model |
vram_gb |
VRAM in GB |
test |
Benchmark test (pp128, pp512, ..., tg128) |
tokens_per_sec |
Throughput in tokens/second |
stddev |
Standard deviation |
date |
Benchmark date |
Key findings
MoE (3B active) vs Dense (27B) on same-family Qwen3.6 models:
- Prompt processing: 2.4x faster across all context lengths
- Text generation: 3.5x faster (271 vs 77 t/s)
- Both degrade ~17% at 16K context (attention + VRAM, not parameter count)
Tooling
Benchmarks generated with llm-bench-rig — open-source benchmark pipeline for GGUF and safetensors models.