Training at a defined batch size
Check whether weights, activations and optimizer state fit in 40 GB HBM2.
A100 SXM 40 GB rental
Ampere in the SXM form factor with 40 GB HBM2 for AI and scientific computing.
$1.1888/ hour
Workloads for this GPU configuration
Inference within a tighter budget
Tune quantization and context length around the actual memory limit.
Scientific computing
Test your CUDA application on a representative problem at the required precision.
Software compatibility
- The GPU has 40 GB of HBM2. Include runtime memory and working buffers in the total.
- This configuration is A100 SXM with 40 GB HBM2. The 80 GB version is listed separately.
- Estimate training memory from the method and batch size, not just the weights.
Rent A100 SXM 40 GB
Provide your model, software environment and rental period. We will confirm the rate, availability and connection method before the rental starts.
$1.1888/ hour
About this configuration
How much memory does A100 SXM 40 GB have?
The GPU has 40 GB of HBM2. Include runtime memory and working buffers in the total.
When should I choose A100 80 GB instead of 40 GB?
Choose 80 GB if model weights, context and working data do not fit in 40 GB, or if you need a longer context or larger batch size. The 80 GB version uses HBM2e. Compare performance on your workload.
How do I request this configuration?
This configuration is open for rental requests. Share your workload, preferred dates and duration. Hardware availability, pricing and access dates are confirmed individually; a request does not create a confirmed booking.
