# NVIDIA Tesla V100

16 Hacker News commenters recommended NVIDIA Tesla V100 and 4 criticized it, November 2025 to November 2026. It was recommended in 8 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** People recommend the used V100 as cheap VRAM and FP64 compute for local AI and physics work. Caveats: no bfloat16, loud fans, high idle power.

## Why people recommend it

- Cheap VRAM per dollar (6 comments): "V100S is a very good card for some workloads. And they were cheap as well." ([source](https://news.ycombinator.com/item?id=50239204)) "You can run the same model with less context on a single £200 V100" ([source](https://news.ycombinator.com/item?id=48351187)) "I think the V100 16GBs are a better deal at $250" ([source](https://news.ycombinator.com/item?id=48900384))
- Fast local model inference (4 comments): "I have a single V100 running around 900t/s prefill and 100 t/s decode during DeepSWE runs on 27B" ([source](https://news.ycombinator.com/item?id=49933921)) "already at around 900 t/s prefill, 90-100 t/s output in Pi on coding tasks" ([source](https://news.ycombinator.com/item?id=49911612)) "Get a couple or V100S GPU. They will run it /fast/ with the right engine." ([source](https://news.ycombinator.com/item?id=50267183))
- Strong FP64 for scientific computing (3 comments): "perfectly capable of running and delivering the physics calculation optimizations" ([source](https://news.ycombinator.com/item?id=48456693)) "V100s are also quite useful for many applications that use FP64" ([source](https://news.ycombinator.com/item?id=49020662)) "the FP64 performance (7+ TFLOPS!) is fantastic given their age" ([source](https://news.ycombinator.com/item?id=48350777))
- Scales across multiple cards (3 comments): "8xV100 will outperform most of what you can buy for that price simply by way of memory and nvlink bandwidth." ([source](https://news.ycombinator.com/item?id=46222333)) "Get a couple or V100S GPU. They will run it /fast/ with the right engine." ([source](https://news.ycombinator.com/item?id=50267183)) "combined them without an issue" ([source](https://news.ycombinator.com/item?id=48351178))

## What people warn about

- Lacks bfloat16 and modern data types (2 comments): "the V100 doesn't support bfloat16" ([source](https://news.ycombinator.com/item?id=48347653)) "horribly inefficient, lacking modern data types and engines, requiring screaming server fans" ([source](https://news.ycombinator.com/item?id=49012869))
- High idle power and loud fans (2 comments): "horribly inefficient, lacking modern data types and engines, requiring screaming server fans" ([source](https://news.ycombinator.com/item?id=49012869)) "v100s idle super high! 25-35watt with nothing loaded and easily 50w when a model is loaded" ([source](https://news.ycombinator.com/item?id=48346923))

## Alternatives hackers recommend

- [NVIDIA H100](https://hackerapproved.com/p/nvidia-h100.md): 20 recommend
- [AMD Instinct MI355X](https://hackerapproved.com/p/amd-instinct.md): 14 recommend
- [Nvidia A100](https://hackerapproved.com/p/nvidia-a100-gpu.md): 10 recommend
- [Intel D270 GPU AI Accelerator](https://hackerapproved.com/p/intel-d270-gpu-ai-accelerator.md): 10 recommend
- [NVIDIA B200](https://hackerapproved.com/p/nvidia-b200.md): 9 recommend

## Where to get it

- [See price on Amazon](https://www.amazon.com/s?k=NVIDIA+Tesla+V100+GPU&tag=hackerapprove-20) (affiliate link)

## Cite as

Hacker Approved, "NVIDIA Tesla V100", https://hackerapproved.com/p/nvidia-tesla-v100/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).