Multi-GPU Full VRAM Deployment of DeepSeek V4 Yields Only 600 t/s PP
fragment_me · reddit · 2026-08-02
A developer loaded the IQ3XXS quantized version of DeepSeek V4 entirely into VRAM across 5 consumer GPUs (2x 3080, 2x 3090, 5090), but found the Prompt Processing (PP) speed to be a sluggish 600 t/s, making it unusable as a daily driver.
The author verified the model was fully in VRAM and updated CUDA/NCCL/Llamacpp, and is now asking the community for baseline PP numbers to troubleshoot the bottleneck.
Related event: DeepSeek V4 Flash Sparks a Wave of Local Deployment Tests(26 posts)→
More from Infra
- NVIDIA quietly delists the entry-level RTX 5050 from its marketplace, following the RTX 5090 — chemist_slime · 2026-10-16
- Grok claims 816B MoE model runs locally on dual M5 Ultra Mac Studios via MLX — grok · 2026-10-16
- Same 1,000 Cloudflare Durable Objects: $4,050/month cut to $22 with one trick — threepointone · 2026-10-16
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- Baidu's Virtual Head Attention tames the KV cache, FLOPs, quality triangle in long-context LLMs — baidu · 2026-10-16
- DuoDiscover agents auto-build fast eval proxies, scoring 2.68x TPU kernel speedup — berkeley_ai · 2026-10-16