News for hackers. Stuff that matters.

A comment on Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL parsing benchmark. Parent comment

Re:Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL pars... by stingraycharles on Wednesday May 03, 2023 @04:37PM (#35807439)
In my totally unscientific (but consistent) benchmarks for our CI build servers, m6g.8xlarge compile our c++ codebase in about 9.5 minutes, where m6a.8xlarge takes about 11 minutes. The price difference is about 20% as well iirc, so it’s generally a good deal.

Of course the types of optimisations that a compiler may (or may not) do on aarch64 vs x86_64 are completely different and may explain the difference (we actually compile with -march=haswell for x86_64), but generally Graviton seems like a really good deal.

[ Parent | on HN ]
Re:Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL pars... by foota on Wednesday May 03, 2023 @05:36PM (#35808116)
You're probably leaving a lot of performance on the floor if you're building for haskell and running on skylakeish or newer.
[ Parent | on HN ]
Re:Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL pars... by nickpeterson on Wednesday May 03, 2023 @05:46PM (#35808216)
*haswell, in case the Haskell people come after you.
[ Parent | on HN ]
Re:Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL pars... by speed_spread on Wednesday May 03, 2023 @06:25PM (#35808629)
"don't poke the endofunctor"
[ Parent | on HN ]
Re:Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL pars... by paulddraper on Wednesday May 03, 2023 @06:06PM (#35808430)
Ah, that makes more sense.
[ Parent | on HN ]
Re:Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL pars... by stingraycharles on Thursday May 04, 2023 @01:05AM (#35811441)
Yeah we know, it’s just that we ship these binaries to customers and need a reasonably old architecture that we know “almost everyone” uses.
[ Parent | on HN ]
Re:Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL pars... by Aeolun on Wednesday May 03, 2023 @10:00PM (#35810391)
Why not use the c type instances? Is the compilation memory bound? I’d expect it to be the cpu that’s the limiting factor.
[ Parent | on HN ]
Re:Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL pars... by stingraycharles on Thursday May 04, 2023 @01:04AM (#35811431)
Heavy C++ template metaprogramming, we need up to 4GB of memory per process in some situations.
[ Parent | on HN ]