RTX 5080 vs 5060 Ti benchmarked on FLUX Krea 2: 2.1x faster, wildly asymmetric LoRA penalty
Altruistic_Print717 · reddit · 2026-10-06
The author benchmarked RTX 5080 vs RTX 5060 Ti 16GB running FLUX Krea 2 in ComfyUI on the same rig (Ryzen 7 5700X, 32GB RAM), across 1–4 MP, two samplers (euler/beta vs gauss-legendre2s/normal), with and without a LoRA, plus SeedVR ×2 upscale.
Key takeaways:
- 5080 is 2.1–2.2x faster: 9s vs 16s per image at 1 MP euler/no LoRA (0.89 vs 0.50 it/s); 39s vs 86s at 4 MP.
- The LoRA penalty is highly asymmetric: +56% on the 5080 at 1 MP (9→14s) vs only +6% on the 5060 Ti (16→17s); both converge to +13% at 4 MP.
- Sampler cost dominates GPU cost: gauss-legendre2s is 2.9–3.2x slower than euler/beta on both cards.
- SeedVR ×2 dominates at 4 MP: +157s on the 5080 (78% of 201s total), +210s on the 5060 Ti (70% of 300s); only +23–37s at 1–3 MP.
- One anomaly: at 1 MP gl2s + LoRA the 5060 Ti beat the 5080 (53s vs 58s), likely noise/caching.
Full s/it and it/s charts for every config are provided.
More from Infra
- Strata engine boosts RTX 3090 prefill 17x, reigniting the local LLM debate — Iory1998 · 2026-10-06
- Hyperscaler AI capex to jump 92% to $789B in 2026, near $1.2T/year by 2029 — luisdans · 2026-10-06
- PyTorch Conference: IBM to keynote Spyre Accelerator and distributed inference work — PyTorch · 2026-10-06
- llama.cpp v0.6.0 ships MTP speculative decoding for Qwen4Exp and more — vexatious-big · 2026-10-06
- Deep interview: why NVIDIA engineered Nemotron 3 Ultra around speed and long context — yacinelearning · 2026-10-06
- Repurposing an old X79 PC with dual GPUs to run local models and cut Claude reliance — DarkBrews · 2026-10-06