← AI Hashrate 中文

Can Titan Xp 12GB run DeepSeek-R1-Distill-Qwen-32B?

NVIDIA · 12 GB GDDR5X · 548 GB/s bandwidth · DeepSeek · 32.8B · model ctx up to 128K

FF · No FitComposite score: VRAM headroom × 0.5 + tok/s speed × 0.5. Q4 @ 8K context.
CUDA

No — DeepSeek-R1-Distill-Qwen-32B at Q4 needs ≈ 22.32 GB even at 4K context, beyond the Titan Xp 12GB's 12 GB GDDR5X. It would only run with heavy CPU/disk offload. Decode at Q4/8K: ≈ 6.6 tok/s (estimated, batch 1). FP16 needs ≈ 73.16 GB — does not fit on this card. Estimates come from memory-bandwidth math; rows tagged measured override estimates. Relative ranking is more reliable than absolute tok/s. Methodology.

Cheapest GPU that can: H100 80GB SXM5 See fit details →or☁️ Rent on RunPod from ~$0.50/hr

Fit & speed by quant and context

QuantContextVRAM neededFits?tok/s (decode)
Q44K22.32 GBNo8.7 ± est.
Q48K default25.6 GBNo6.6 ± est.
Q432K45.28 GBNo2.1 ± est.
FP164K69.88 GBNo
FP168K default73.16 GBNo
FP1632K92.84 GBNo

Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s means the model is far beyond this card (offload-only territory).

VRAM breakdown at 8K context

QuantWeightsKV cacheOverheadTotal neededTitan Xp 12GB VRAM
Q418.04 GB6.56 GB1 GB25.6 GB12 GB
FP1665.6 GB6.56 GB1 GB73.16 GB12 GB

Other GPUs that run DeepSeek-R1-Distill-Qwen-32B

All GPUs for DeepSeek-R1-Distill-Qwen-32B →

Other models for the Titan Xp 12GB

All models on the Titan Xp 12GB →