Howdy, Stranger!

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


In this Discussion

How many vCPUs per physical core?
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.

How many vCPUs per physical core?

bruh21bruh21 Member, Host Rep
edited July 2021 in General

Just wondering, how many vCores are assigned per actual core? I know providers say fair share, but how many neighbors would you normally have on each core?

Comments

  • jlayjlay Member
    edited July 2021

    It varies, a couple dozen isn't that unusual. It is on the high end though

    Anywhere from one to that, I'd guess.

    High frequency CPUs with a lot of cache can handle a lot more per-core. It comes down to a certain amount of overlap/Tetris.

    How much monitoring/QoS do they want? Can their customers appreciate something consistent, but measurably slower?

    The host might not even bother to pin - leaving it up to the kernel

    edit: It gets really complicated when you consider things like NUMA

    edit2: it gets even more complicated when you consider what 'on a core' really means.

    They tend to share pools of cache, activities on another core impact the performance of yours.

    The CPU core and 'hyper thread' are often treated as separate and I don't think they should be

    Thanked by 1bruh21
  • bruh21bruh21 Member, Host Rep

    @jlay said:
    It varies, a couple dozen isn't that unusual. It is on the high end though

    Anywhere from one to that, I'd guess.

    High frequency CPUs with a lot of cache can handle a lot more per-core. It comes down to a certain amount of overlap/Tetris.

    How much monitoring/QoS do they want? Can their customers appreciate something consistent, but measurably slower?

    The host might not even bother to pin - leaving it up to the kernel

    edit: It gets really complicated when you consider things like NUMA

    edit2: it gets even more complicated when you consider what 'on a core' really means.

    They tend to share pools of cache, activities on another core impact the performance of yours.

    The CPU core and 'hyper thread' are often treated as separate and I don't think they should be

    thanks for the answer
    i wonder how many are assigned from top providers like racknerd and nexusbytes

Sign In or Register to comment.