Llama on NVIDIA Ada

Can I run Llama 3.1 70B on an RTX 4080 Super?

Not at Q4_K_M — it needs 43.4 GB against 14.2 GB available. You would need 4 of these cards.

Won't fit

43.4 GB of 14.2 GB · 305%
045 GB
Weights 40.0 GB
KV cache 2.5 GB
Runtime overhead 0.9 GB
Over the limit 29.1 GB

Short by 29.1 GB. You can run it with 21 of 80 layers on the RTX 4080 Super and the rest in system RAM, at roughly 1.22 tok/s — usable for batch work, painful for chat. A smaller quantisation or a shorter context is usually the better trade.

Generation1.22tok/s
Prompt processing310tok/s
Max context0tokens
KV per 1K tokens0GB

Every quantisation of Llama 3.1 70B on a RTX 4080 Super

Highlighted row is the highest quality that still fits at 8K context.

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 131.4 GB 134.8 GB Won't fit — 0.31
INT8 / W8A8 8.50 69.3 GB 72.7 GB Won't fit — 0.63
Q8_0 (GGUF) 8.50 69.3 GB 72.7 GB Won't fit — 0.63
FP8 (E4M3) 8.00 65.7 GB 69.1 GB Won't fit — 0.67
Q6_K 6.56 53.9 GB 57.3 GB Won't fit — 0.85
Q5_K_M 5.67 46.6 GB 50.0 GB Won't fit — 1.01
Q5_K_S 5.52 45.3 GB 48.7 GB Won't fit — 1.05
Q4_K_M 4.85 40.0 GB 43.4 GB Won't fit — 1.22
Q4_K_S 4.58 37.8 GB 41.2 GB Won't fit — 1.31
AWQ 4-bit 4.25 37.8 GB 41.2 GB Won't fit — 1.31
GPTQ 4-bit 4.25 37.8 GB 41.2 GB Won't fit — 1.31
MXFP4 4.25 37.8 GB 41.2 GB Won't fit — 1.31
Q4_0 4.55 37.6 GB 41.0 GB Won't fit — 1.34
IQ4_XS 4.25 35.2 GB 38.6 GB Won't fit — 1.44
Q3_K_M 3.91 32.5 GB 35.9 GB Won't fit — 1.61
IQ3_M 3.70 30.8 GB 34.2 GB Won't fit — 1.75
IQ3_XXS 3.06 25.7 GB 29.1 GB Won't fit — 2.27
Q2_K 2.63 22.3 GB 25.7 GB Won't fit — 2.87
IQ2_XXS 2.06 17.8 GB 21.2 GB Won't fit — 4.44
IQ1_M 1.75 15.3 GB 18.7 GB Won't fit — 6.40

Also worth checking

Llama 3.1 70B on every GPU
Full hardware tableEverything that runs on a RTX 4080 Super
Full model tableWhy the KV cache is the part that bites
Guide