Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


One core vs multiple cores
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

One core vs multiple cores

nfnnfn Veteran
edited February 2013 in General

Hi,

For a LEMP VPS, what disavantage or performance lost we have with only one core versus multiple cores either in OpenVZ, KVM and XEN.

Any Reading about this?

Thanks.

Comments

  • MaouniqueMaounique Host Rep, Veteran
    edited February 2013

    I think more cores arrange work better, it is best with multithreaded apps.
    About how much better is to have more cores, I guess the answer depends on how loaded is the cpu. If you already max it out, even with same power divided into 2 cores, will be some performance gain. I would expect at max 20% or so.
    If you cpu idles much of the time, there is no real difference.

  • Additional cores allows you to balance out the load. It all depends on what services you're running though, but from what I've always seen the more cores the better! ^TG

  • It also depends if it's 1 very fast core vs multiple slower cores.

  • Which is better for LEMP stack, 1 2Ghz core or 4 500Mhz cores?

  • MaouniqueMaounique Host Rep, Veteran
    edited February 2013

    4x500 However I dont see a 500 core in standard processors so I believe that is a heavily throttled core suggesting a saturated node which must be kept alive with such tricks or a lot of abusers which are not kicked.

  • @Maounique said: 4x500 However I dont see a 500 core in standard processors so I believe that is a heavily throttled core suggesting a saturated node which must be kept alive with such tricks or a lot of abusers which are not kicked.

    My Urpad CyberMonday I think is throttled, that's why I ask =) Perhaps just preventing abusers thats why.

  • At some point, too many VCPUs per container/VM can generate enough context switching to cause problems of its own.

  • MaouniqueMaounique Host Rep, Veteran
    edited February 2013

    I do that too from time to time to people which have been warned but do not take any action and the problems happen again while also not being bad enough to warrant a suspension (node being able to take the load, just not fair to eat out our margin of 60-70% idle time by maxing out their core all the time). I have some 5-7 of these people that either didnt notice it or dont care, just dont want to be bothered and are better to maxing out the available cpu instead of taking extra steps to stay in line.

    @Microlinux said: At some point, too many VCPUs per container/VM can generate enough context switching to cause problems of its own.

    Yes, this is why we dont give out too many, once because we have enough idle time on all cores, second because if you need that much CPU you need a special plan with the other resources as well, unless you intend to run some kind of seti@home or bitcoin.
    4 is not too much, tho, but I would rather have 2x1000 than 4x500. 4x500 is still much better than 1x2000 in identical CPUs.

  • nfnnfn Veteran

    I'm comparing a Xen with 4 cores raid10 with a cloud provider with One core + raid 10 San with the same RAM. This is for a few WordPress sites (nginx + php + mariadb)

    I don't know the cpu used in the cloud provider at this time.

Sign In or Register to comment.