GPUs for scientific computing and simulation
Simulation codes such as molecular dynamics, computational fluid dynamics, weather or quantum chemistry are often limited by memory bandwidth and, for many of them, by double-precision (FP64) arithmetic. Data center GPUs with HBM memory are built for both.
GPUs that fit scientific computing
Swipe sideways to see every column.
| GPU | Memory | Memory type | Bandwidth | Interconnect | Price/h |
|---|---|---|---|---|---|
| Tesla V100Volta | 32 GB | HBM2 | 900 GB/s | Not published | from $0.10 |
| A100 PCIeAmpere | 80 GB | HBM2e | 1,935 GB/s | NVLink bridge (600 GB/s) · PCIe 4.0 | from $1.79 |
| A100 SXMAmpere | 80 GB | HBM2e | 2,039 GB/s | NVLink 3 (600 GB/s) · PCIe 4.0 | from $1.79 |
| H100 PCIeHopper | 80 GB | HBM2e | 2,000 GB/s | NVLink bridge (600 GB/s) · PCIe 5.0 | from $2.89 |
| H100 SXMHopper | 80 GB | HBM3 | 3,350 GB/s | NVLink 4 (900 GB/s) · PCIe 5.0 | from $2.88 |
| GH200Grace Hopper | 96 GB | HBM3 | 4,000 GB/s | NVLink-C2C (900 GB/s) | from $6.81 |
| H200Hopper | 141 GB | HBM3e | 4,800 GB/s | NVLink 4 (900 GB/s) · PCIe 5.0 | from $4.19 |
| MI300XCDNA 3 | 192 GB | HBM3 | 5,300 GB/s | AMD Infinity Fabric (128 GB/s per link) · PCIe 5.0 | from $4.16 |
| MI325XCDNA 3 | 256 GB | HBM3e | 6,000 GB/s | AMD Infinity Fabric (128 GB/s per link) · PCIe 5.0 | from $4.82 |
Manufacturer specifications, per GPU; tensor throughput without sparsity. Model sizes are cautious estimates in billions of parameters, counting 80% of memory; context length, batch size and framework change the real limit. Prices are the best live offer for one GPU per hour, excluding tax.
All GPU modelsWhat matters for scientific computing
Double precision
Many solvers need FP64. Data center GPUs such as the V100, A100, H100 and AMD Instinct run FP64 at a high rate, while GeForce and most workstation cards run it at a small fraction of their FP32 rate. Check your code: some packages run mostly in single or mixed precision.
Memory bandwidth
Stencil codes, sparse linear algebra and particle methods move more data than they compute. HBM memory, from about 0.9 TB/s on the V100 to more than 4 TB/s on the H200 and MI300X, is what makes these codes fast.
Scaling across GPUs
Domain decomposition splits a simulation across GPUs that exchange boundary data at each step. NVLink or Infinity Fabric between GPUs keeps that exchange from becoming the bottleneck.
Compare GPUs for scientific computing
GPUs for scientific computing and simulation: frequently asked questions
Why are FP64 figures not shown?
This site only shows values published for each exact model in a comparable form. FP64 rates differ between board versions (SXM, PCIe), so check the manufacturer data sheet for the board you plan to rent.
Which GPU for molecular dynamics?
Packages such as GROMACS, AMBER and NAMD run largely in single or mixed precision and do well on many GPUs, including workstation cards; codes that need FP64 throughout are better served by A100, H100 or MI300X GPUs.
Does my code run on AMD GPUs?
AMD Instinct GPUs run code written for ROCm and HIP; many HPC packages ship ROCm builds. Code written only for CUDA needs porting, often with the HIPIFY tools, before it runs on AMD.
How do I rent a GPU for scientific computing on EverXCloud?
Create an account, top up your prepaid balance in euros, US dollars or pounds, then buy GPU hours on the spot exchange at the best offer, or place a request at the price you are willing to pay. Rentals are billed by the hour. Prices are shown excluding tax and EverXCloud is the merchant of record for every rental.
Plan your GPU rental
- GPU rental: how to rent a cloud GPU by the hour
- How to rent your first GPU
- How to choose a GPU for LLM inference
- How to estimate a GPU rental session
- H100 price per hour: what sets the rate
- How a GPU spot exchange works
- H200 price per hour: cloud pricing explained
- B200 price per hour: B200, B300 and GB200
- A100 price per hour: rent an NVIDIA A100
- RTX 4090 price per hour: rental guide
- RTX 5090 price per hour: rental guide
- L40S price per hour: rental guide
- MI300X price per hour: rental guide
- How much VRAM do you need for an LLM?
- Best GPU for Stable Diffusion and Flux
- Cheapest way to rent a GPU for AI
- Best GPU for scientific computing (FP64)
Check the workload, complete configuration and billing conditions before choosing a live offer.
Be ready on launch day
Create your account now: you can top up your balance and rent the GPU that fits your workload as soon as the market opens.
Have GPU servers?
Businesses can join the EverXCloud Network and sell hours of their GPU servers on the spot exchange.
Other workloads
- GPUs for training large language models
Pre-training and full fine-tuning on data center GPUs with HBM memory.
- GPUs for fine-tuning open models
LoRA and QLoRA adapters on a single GPU, from 24 GB up.
- GPUs for LLM inference
Serve open models: what fits in memory, and how fast tokens come.
- GPUs for image and video generation
Diffusion models for images and video, from 24 GB cards to 80 GB GPUs.
- GPUs for 3D rendering and visual effects
Ray-traced rendering, visual effects and real-time engines.