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 23.96 GB / 61.4 GB Usable

Qwen3.6-35B-A3B (MoE) on 海光深算 DCU K100-AI 64GB (HBM2e): VRAM Requirement & Speed Estimate

Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.6-35B-A3B (MoE) requires 23.96 GB of VRAM (21.83 GB weights + 0.4 GB KV cache + 1.74 GB overhead), fitting within 海光 DCU K100-AI's 61.4 GB usable memory at ~319.4 tokens/s.

Concurrency (Batch):
MoE 36B / A3B·40L (d=2048)·16Q/2KV·Ctx 256K·Text+Image+Video
Runnable Models →
GPUs
64 GB VRAM·896 GB/s·Safe Usable 61.4 GB
Smooth Fit23.96 / 61.4 GB~319.4 tok/s
3. Weight Quantization
4. Per-Request Context16K (16,384 tok)
Verified: 2026-11-20·Roofline bandwidth/compute estimate (Batch=1)
Smooth Fit
23.96/ 61.4 GB usable

Physical 64 GB · Load 39%

Theoretical Decode Speed
~319.4tokens/s

MoE: reads only 1.82 GB active weights/step

Max Context at Batch=1
256K/ 256K

Enable Q8/Q4 KV quantization to scale batch concurrency

Weights: 21.83 GBKV Cache (1×16K): 0.4 GBOverhead: 1.74 GB
🚀 Blazing Fast (Instant CoT)~1.6s for a 500-token response
Reference Launch CLI:
OLLAMA_CONTEXT_LENGTH=16384 ollama run qwen3-6-35b-a3b

海光 DCU K100-AI Fits Qwen3.6-35B-A3B (MoE) · Cost & TCO

Approx. $8,000 (~$125/GB VRAM) · 350W TDP

📊 All Quantization Levels for Qwen3.6-35B-A3B (MoE) on 海光 DCU K100-AI

Quantization Total VRAM (16K) Verdict Est. Speed Max Context
FP16 / BF16 75.14 GB Partial RAM Offload (Slow) ~40.8 tok/s —
FP8 / Q8_0 40.72 GB Smooth Fit ~190.4 tok/s 256K
GGUF Q6_K 31.81 GB Smooth Fit ~242.5 tok/s 256K
GGUF Q5_K_M 27.82 GB Smooth Fit ~276.3 tok/s 256K
Q4_K_M / AWQ 23.96 GB Smooth Fit ~319.4 tok/s 256K
GGUF IQ4_XS 21.21 GB Smooth Fit ~359.5 tok/s 256K
GGUF Q3_K_M 19.65 GB Smooth Fit ~387 tok/s 256K
GGUF Q2_K 16.16 GB Smooth Fit ~420 tok/s 256K

💡 Recommended Alternative GPUs & Macs for Qwen3.6-35B-A3B (MoE)

Frequently Asked Questions

Can 海光深算 DCU K100-AI 64GB (HBM2e) run Qwen3.6-35B-A3B (MoE) locally?

Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.6-35B-A3B (MoE) requires 23.96 GB of VRAM (21.83 GB weights + 0.4 GB KV cache + 1.74 GB overhead), fitting within 海光 DCU K100-AI's 61.4 GB usable memory at ~319.4 tokens/s.

What is the best quantization for Qwen3.6-35B-A3B (MoE) on 海光 DCU K100-AI?

The highest-quality quantization that fits cleanly on 海光 DCU K100-AI is FP8 / Q8_0 (99.9% Near-Lossless), using 40.72 GB VRAM at ~190.4 tokens/s with up to 256K context.

How much VRAM does Qwen3.6-35B-A3B (MoE) need in full FP16 precision?

In unquantized FP16/BF16 precision (16K context), Qwen3.6-35B-A3B (MoE) requires 75.14 GB VRAM (72 GB for weights alone). Because it uses a Mixture-of-Experts (MoE) architecture, all 36B parameters must sit in memory, while only 3B active parameters are read per token.