The best datacenter gpus, according to hackers
11 datacenter gpus recommended in Hacker News comments over the past year, ranked by how many separate people recommend each one.
- Most recommendedNVIDIA H10020 people recommend it
- Least regretNVIDIA Tesla V10016 recommend, 4 criticize
All datacenter gpus
- NVIDIA H100People recommend it for top AI training and cheap rental access, though critics cite high purchase cost, power draw and failure rates.20recommend
- NVIDIA Tesla V100People recommend the used V100 as cheap VRAM and FP64 compute for local AI and physics work. Caveats: no bfloat16, loud fans, high idle power.16recommend
- AMD Instinct MI355XStrong memory and performance per dollar for large AI inference and training; critics say its software stack lags and it trails Nvidia's B300 on some benchmarks.14recommend
- Nvidia A100Long-lived datacenter GPU that people still run for AI and compute work years after release; the most common caveat is that it can be unreliable under long runs.10recommend
- Intel D270 GPU AI AcceleratorRecommended for holding many resident AI models in 480 GB of LPDDR5X memory. Critics say its low memory bandwidth makes it slow, and it is very expensive for individuals.10recommend
- NVIDIA B200Recommended for top raw AI throughput, a maturing software stack and low cost per token. Main caveat: rented B200 instances are scarce or pricey.9recommend
- NVIDIA GB300People recommend it for high-end compute such as CGI rendering and say it is a decade ahead of competitors. Critics say it is expensive, around $100k, and hard to deploy.9recommend
- AMD Instinct MI50People recommend it for its 32GB of fast HBM2 VRAM, about 1TB/s bandwidth, for running large local AI models. Critics note weak compute, manual recompilation and AMD deprecating cards.8recommend
- NVIDIA H200Datacenter GPU that people recommend for running large open-source models unquantized, mostly by renting cloud instances by the hour. Critics call it overpriced, with limited concurrency and throughput.8recommend
- AMD Instinct MI300XRecommended for large VRAM, competitive throughput and easy availability. Common caveat: the software stack lags Nvidia, missing features like int4 and fp8 quantization.8recommend
- NVIDIA B300People recommend the B300 for price-performance and for running models far below API costs. One critic says a single card cannot run much on its own.7recommend