← AI Hashrate 中文

A100 80GB SXM4 vs Tesla P4 8GB for Local LLMs

The A100 80GB SXM4 has about 962% more memory bandwidth (2,039 vs 192 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 962% faster on the A100 80GB SXM4. The A100 80GB SXM4 fits more of the catalog at Q4/8K — 50 vs 16 of 69 models (80 vs 8 GB VRAM). Largest model that fits the A100 80GB SXM4 but not the Tesla P4 8GB: Qwen3.5-122B-A10B (122.0B). MSRP is $15,000 for the A100 80GB SXM4 vs $80 for the Tesla P4 8GB. MSRP is a launch list price, not live retail — Amazon prices move; click through for the current price. Not retail-shoppable: A100 80GB SXM4 (datacenter-class; cloud rental is the realistic way to use it). In the Q4/8K head-to-head below (39 shared models), the A100 80GB SXM4 posts the higher tok/s in 39 rows vs 0. Methodology.

Spec comparison

A100 80GB SXM4Tesla P4 8GB
VendorNVIDIANVIDIA
VRAM80 GB8 GB
Memory typeHBM2eGDDR5
Memory bandwidth2,039 GB/s192 GB/s
TDP400 W75 W
MSRP (list)$15,000$80
BuySearch Amazon (affiliate)

MSRP is the launch list price, not live retail. Buy links are Amazon affiliate links — the price after you click is the live price.

Q4 / 8K head-to-head — top 15 of 39 shared models

ModelParams (B)A100 80GB SXM4 tok/sA100 80GB SXM4 fitsTesla P4 8GB tok/sTesla P4 8GB fitsWinner
MiniCPM5-1B1.081201.4 est.Yes113.1 est.YesA100 80GB SXM4
Qwen3.5-2B2.0648.8 est.Yes61.1 est.YesA100 80GB SXM4
DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx2.3564.2 est.Yes53.1 est.YesA100 80GB SXM4
Qwen3.6-35B-A3B36.0432.5 est.Yes17.0 est.NoA100 80GB SXM4
Qwen3.5-35B-A3B34.7432.5 est.Yes18.2 est.NoA100 80GB SXM4
GLM-4.7-Flash31.2432.5 est.Yes21.0 est.NoA100 80GB SXM4
Qwen3-30B-A3B-Instruct-250730.5432.5 est.Yes21.9 est.NoA100 80GB SXM4
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF1630.0432.5 est.Yes24.1 est.NoA100 80GB SXM4
Ornith-1.0-35B35.0432.5 est.Yes17.0 est.NoA100 80GB SXM4
qwen35b-a3b-fable-sft-abliterated34.7432.5 est.Yes17.3 est.NoA100 80GB SXM4
Llama-3.2-3B-Instruct3.2405.5 est.Yes38.2 est.YesA100 80GB SXM4
gpt-oss-20b21.5360.4 est.Yes33.6 est.NoA100 80GB SXM4
gemma-4-26B-A4B-it25.2341.5 est.Yes23.9 est.NoA100 80GB SXM4
Qwen3-4B-Instruct-25074.0324.4 est.Yes30.5 est.YesA100 80GB SXM4
NVIDIA-Nemotron-3-Nano-4B-BF164.0324.4 est.Yes30.5 est.YesA100 80GB SXM4

est. = bandwidth-based estimate; measured = curated real benchmark that overrides the estimate. Fits = weights + KV(8K) + 1 GB overhead ≤ 95% of VRAM.

Fit coverage at Q4 / 8K

A100 80GB SXM4Tesla P4 8GB
Catalog models that fit50 of 6916 of 69
Largest exclusive fitQwen3.5-122B-A10B (122.0B)

A100 80GB SXM4 details → · Tesla P4 8GB details →