Llama on NVIDIA Hopper

Can I run Llama 3.2 3B on an H100 SXM 80GB?

Yes. Llama 3.2 3B at Q4_K_M uses 3.5 GB of the 74.4 GB available on a H100 SXM 80GB, and runs at about 877 tokens per second. You can push the context to 128K.

Runs comfortably

3.5 GB of 74.4 GB · 5%
074 GB
Weights 1.8 GB
KV cache 0.9 GB
Runtime overhead 0.8 GB

Llama 3.2 3B at Q4_K_M leaves 70.9 GB spare on a H100 SXM 80GB. There is room to raise the context length or move up a quantisation level.

Generation877tok/s
Prompt processing64701tok/s
Max context128Ktokens
KV per 1K tokens0GB

Every quantisation of Llama 3.2 3B on a H100 SXM 80GB

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

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 6.0 GB 7.7 GB Runs comfortably 128K 344
INT8 / W8A8 8.50 3.1 GB 4.8 GB Runs comfortably 128K 599
Q8_0 (GGUF) 8.50 3.1 GB 4.8 GB Runs comfortably 128K 599
FP8 (E4M3) 8.00 3.0 GB 4.7 GB Runs comfortably 128K 614
Q6_K 6.56 2.5 GB 4.1 GB Runs comfortably 128K 715
AWQ 4-bit 4.25 2.1 GB 3.8 GB Runs comfortably 128K 793
GPTQ 4-bit 4.25 2.1 GB 3.8 GB Runs comfortably 128K 793
MXFP4 4.25 2.1 GB 3.8 GB Runs comfortably 128K 793
Q5_K_M 5.67 2.1 GB 3.8 GB Runs comfortably 128K 795
Q5_K_S 5.52 2.1 GB 3.7 GB Runs comfortably 128K 811
Q4_K_M 4.85 1.8 GB 3.5 GB Runs comfortably 128K 877
Q4_K_S 4.58 1.8 GB 3.4 GB Runs comfortably 128K 907
Q4_0 4.55 1.7 GB 3.4 GB Runs comfortably 128K 911
IQ4_XS 4.25 1.6 GB 3.3 GB Runs comfortably 128K 947
Q3_K_M 3.91 1.5 GB 3.2 GB Runs comfortably 128K 992
IQ3_M 3.70 1.5 GB 3.1 GB Runs comfortably 128K 1021
IQ3_XXS 3.06 1.3 GB 2.9 GB Runs comfortably 128K 1124
Q2_K 2.63 1.1 GB 2.8 GB Runs comfortably 128K 1206
IQ2_XXS 2.06 0.9 GB 2.6 GB Runs comfortably 128K 1334
IQ1_M 1.75 0.8 GB 2.5 GB Runs comfortably 128K 1415

Also worth checking

Llama 3.2 3B on every GPU
Full hardware tableEverything that runs on a H100 SXM 80GB
Full model tableWhy the KV cache is the part that bites
Guide