ai.canshu.com
中文
VRAM Calculator Advisor & TCO API Pricing Compare Models Leaderboard GPUs

Search models, GPUs, and tools

↑↓ Navigate · Enter Open · ★ Watchlist prioritized ai.canshu.com
Q4_K_M: Smooth Fit 25.03 GB / 47 GB Usable

DeepSeek-R1-Distill-Qwen-32B on NVIDIA L40S 48GB: VRAM Requirement & Speed Estimate

Yes. At Q4_K_M 4-bit quantization with a 16K context window, DeepSeek-R1-Distill-Qwen-32B requires 25.03 GB of VRAM (19.88 GB weights + 4.29 GB KV cache + 0.85 GB overhead), fitting within L40S 48G's 47 GB usable memory at ~28.2 tokens/s.

Concurrency (Batch):
Dense 32.8B·64L (d=5120)·40Q/8KV·Ctx 128K
Runnable Models →
GPUs
48 GB VRAM·864 GB/s·Safe Usable 47 GB
Smooth Fit25.03 / 47 GB~28.2 tok/s
3. Weight Quantization
4. Per-Request Context16K (16,384 tok)
Verified: 2026-11-20·Roofline bandwidth/compute estimate (Batch=1)
Smooth Fit
25.03/ 47 GB usable

Physical 48 GB · Load 53.2%

Theoretical Decode Speed
~28.2tokens/s

Actual speed varies ~±25% by runtime

Max Context at Batch=1
98K/ 128K

Enable Q8/Q4 KV quantization to scale batch concurrency

Weights: 19.88 GBKV Cache (1×16K): 4.29 GBOverhead: 0.85 GB
🏃 Smooth & Snappy~17.7s for a 500-token response
Reference Launch CLI:
OLLAMA_CONTEXT_LENGTH=16384 ollama run deepseek-r1:32b

L40S 48G Fits DeepSeek-R1-Distill-Qwen-32B · Cost & TCO

Approx. $7,500 (~$156/GB VRAM) · 350W TDP

📊 All Quantization Levels for DeepSeek-R1-Distill-Qwen-32B on L40S 48G

Quantization Total VRAM (16K) Verdict Est. Speed Max Context
FP16 / BF16 71.66 GB Partial RAM Offload (Slow) ~2.4 tok/s —
FP8 / Q8_0 40.29 GB Smooth Fit ~16.8 tok/s 41K
GGUF Q6_K 32.18 GB Smooth Fit ~21.4 tok/s 71K
GGUF Q5_K_M 28.54 GB Smooth Fit ~24.4 tok/s 84K
Q4_K_M / AWQ 25.03 GB Smooth Fit ~28.2 tok/s 98K
GGUF IQ4_XS 22.52 GB Smooth Fit ~31.8 tok/s 107K
GGUF Q3_K_M 21.1 GB Smooth Fit ~34.2 tok/s 112K
GGUF Q2_K 17.92 GB Smooth Fit ~41.3 tok/s 124K

💡 Recommended Alternative GPUs & Macs for DeepSeek-R1-Distill-Qwen-32B

Frequently Asked Questions

Can NVIDIA L40S 48GB run DeepSeek-R1-Distill-Qwen-32B locally?

Yes. At Q4_K_M 4-bit quantization with a 16K context window, DeepSeek-R1-Distill-Qwen-32B requires 25.03 GB of VRAM (19.88 GB weights + 4.29 GB KV cache + 0.85 GB overhead), fitting within L40S 48G's 47 GB usable memory at ~28.2 tokens/s.

What is the best quantization for DeepSeek-R1-Distill-Qwen-32B on L40S 48G?

The highest-quality quantization that fits cleanly on L40S 48G is FP8 / Q8_0 (99.9% Near-Lossless), using 40.29 GB VRAM at ~16.8 tokens/s with up to 41K context.

How much VRAM does DeepSeek-R1-Distill-Qwen-32B need in full FP16 precision?

In unquantized FP16/BF16 precision (16K context), DeepSeek-R1-Distill-Qwen-32B requires 71.66 GB VRAM (65.6 GB for weights alone).