Hacker News Reader

Comment by jeffbee | Hacker News Reader

By the way for anyone curious how the numbers in the article compare to a modern desktop computer, a 13th-generation Intel Core does 169ns/url, twice as fast as either of these server rigs. If I cap mine to 3GHz to reflect more realistic server scenarios, 310ns/url, a bit quicker than the Graviton 3. Even the "efficiency cores" on a desktop are faster than these server parts, at 286ns/url.

View on HN

Cloud servers have many, many more cores, cache and memory channels than desktop chips. You can’t clock them that high, nor would hyperscalers want to. They would cost more in power and cooling. Cloud is about multi factor optimization.

Just to note the Graviton 3 is on 5nm. The 13th Gen is on Intel 7. The Core is also Real Core compared to a VPC which is a thread.

I tried this benchmark on both my desktop (Ryzen 5800X3D) and phone (Sony Xperia 1 IV)

    Windows => time/url=364.626ns
    WSL2    => time/url=234.915ns
    Android => time/url=252.276ns

This benchmark feels flawed to me, but I'm not qualified to speculate why.

Another datapoint on a modern desktop CPU, this time ARM (M1 Max laptop, so the same TSMC 5nm process as Graviton 3): 175ns/url

Well the Graviton 3 runs at 2.6 GHz so not sure that really shows 13th-gen Core in a great light.

How about running as many threads as your cpu has SMT threads? Those are what AWS sells as vCPUs.

Reply on news.ycombinator.com