Benchmarking MiniMax H3 on RTX 5090: Efficiency vs. Pixel Workload
Forward-Parsley-148 · reddit · 2026-08-05
The test introduces MP·s (Megapixel-seconds) as a metric to evaluate hardware efficiency when running the MiniMax H3 model for video generation.
- Efficient Linear Range: Below a certain workload threshold, generation time scales linearly with MP·s. Beyond this, generation time increases noticeably.
- Hardware Performance: The RTX 5070 Ti maintains steady performance (approx. 70s/MP·s) up to 2.8 MP·s, while the RTX 5090 extends this threshold to 4.0 MP·s (approx. 37.6s/MP·s).
- Extreme Optimization: By combining SageAttention and TeaCache, the RTX 5090's threshold jumps to 10.4 MP·s, with efficiency improving to roughly 22.9s/MP·s.
The conclusion suggests keeping workloads below the hardware's linear threshold for optimal generation efficiency.
Related event: MiniMax H3 Local Tests: Performance and Optimization Across GPUs(13 posts)→
More from Infra
- Trading 2 RTX 5090s for 4 R9700s: does doubled VRAM pay off? — poofph · 2026-10-16
- Running PolyStrata + Cline coding agent on a $0-fancy laptop at ~90 tok/s — rpwoerk · 2026-10-16
- TensorFold 1.0.12 runs 120B vision model on 1-2 DGX Sparks, adds Mac image input — HankYeomans · 2026-10-16
- vLLM brings frontier model local deployment to NVIDIA RTX Spark PCs — vllm_project · 2026-10-16
- NVIDIA signals $2B in new European AI deals at GTC Berlin, totaling ~$10B in a year — CurieuxExplorer · 2026-10-16
- Making Your Data Ready for Agentic AI: trust first, then semantics, then writes — bibryam · 2026-10-16