The cr.yp.to microblog: 2022.06.16 03:31:57

2022.06.16 03:31:57 (1537246569216454656) from Daniel J. Bernstein, replying to "Ruben Kelevra (@RubenKelevra)" (1537196896808214529):

I agree that the user doesn't want to wait for the computer. I don't agree that spinning up threads on all cores needs a user-perceptible slowdown. I'm not surprised at the limited attention to super-fast browser startup: don't users normally have a browser running continuously?

Context

2022.06.15 23:33:54 (1537186663356506112) from Daniel J. Bernstein:

Given how many major applications that users care about are already multithreaded and vectorized, it's wrong to cherry-pick unoptimized single-threaded applications as pictures of total system performance. This error will increase as more and more applications add optimizations.

2022.06.16 00:10:49 (1537195950682382337) from "Ruben Kelevra (@RubenKelevra)":

Yeah, I think we just have a different view on "total system performance". I see peak workloads with a low base load as primary focus for desktops. Low latency is key here to provide a good user experience. If a CPU can "snap" a chunk of work as a whole in a shorter...

2022.06.16 00:12:53 (1537196473380704258) from "Ruben Kelevra (@RubenKelevra)", replying to "Ruben Kelevra (@RubenKelevra)" (1537195950682382337):

... amount of time on 2-3 threads with nearly twice the processing power the user experience will be better, as there's lower latency. For 5 second workloads, like starting Firefox I'm really not sure if there's enough time to make use of more than 2-3 threads.

2022.06.16 00:14:34 (1537196896808214529) from "Ruben Kelevra (@RubenKelevra)", replying to "Ruben Kelevra (@RubenKelevra)" (1537196473380704258):

A lot of memory has to be claimed, copied, shared etc. My Firefox is actually configured to launch up to 8 threads to process webcontent - with the drawback of a bit more memory consumption (states the manual).