De-Bloated Windows 11 Build Runs on 2GB of RAM

Re:De-Bloated Windows 11 Build Runs on 2GB of RAM by wizofaus on Sunday February 05, 2023 @04:48PM (#34669857)

I'm struggling to believe any single server could usefully service 2100k (well over 2 million!) requests per second. Even Google, with their vast farm of servers, reportedly only process less than 100k requests per second globally. I've certainly read of servers capable of handling in the order of 1000k requests per second as a benchmark, but the requests are usually pretty trivial (the one I saw literally did no input processing at all, and just returned a single fixed byte! But was written in Java, surprisingly.) At any rate, I would think a tiny % of real-life systems actually need to be able to support that sort of load, and bringing in somebody to do the scalability work once it's clear it's needed seems like exactly the right strategy to me.

[ Parent | on HN ]

Re:De-Bloated Windows 11 Build Runs on 2GB of RAM by Genbox on Tuesday February 07, 2023 @05:58PM (#34701128)

Not serving, but handling 2100k requests. Your skepticism is rightly placed, as the HTTP protocol is yet an example of an inefficient protocol that nonetheless is used as the primary protocol on the internet. Some webservers[1] can serve millions of requests pr. second, but I'd never use HTTP in code where efficiency is key.

No, I'm talking about handling requests. In this particular case, requests (32 to 64 bytes) were flowing through several services (on the same computer). I replaced the processing chain with a single application to remove the overhead of serialization between processes. Requests were filtered early in the pipeline, which made a ~55% reduction in the work needed.

Requests were then batched into succinct data structures and processed via SIMD. Output used to be JSON, but I instead wrote a custom memory allocator and just memcpy the entire blob on to the wire.

Before: No pre-filtering, off-the-shelf databases (PSQL), queue system for I/O, named pipes and TCP/IP for local data transfer. Lots of concurrency issues, thread starvation and I/O bound work.

After: Agnessive pre-filtering, succinct data structures for cache coherence, no serialization overhead, SIMD processing. Can saturate a 32 core CPU with almost no overhead.

[1] https://www.techempower.com/benchmarks/#section=data-r13&hw=...

[ Parent | on HN ]