Ideal Machine for best Asymptote Performance

[jakiestfu]

I'm looking to optimize Asymptote's performance for high-throughput. My ideal setup would involve running Asymptote on an AWS Lambda, and I want to understand the best possible hardware and software configurations for maximum efficiency.

1. Ideal Architecture – What CPU architecture is best suited for Asymptote, particularly for multi-threaded workloads?
2. Memory Requirements – How much RAM is optimal for handling large or complex graphics?
3. Disk Space – Given that Lambda functions have limited ephemeral storage, is there an ideal setup for handling intermediate files?
4. Parallelization Opportunities – If generating multiple similar graphics (e.g., 20 clock faces where only the hands differ), is there a way to parallelize only portions of the rendering process to improve performance?
5. I/O Optimization – Are there ways to bypass intermediary writes to disk and instead use stdin/stdout for handling input/output more efficiently?

Any insights or best practices would be greatly appreciated! Thanks in advance.