0
$\begingroup$

I am wondering why modern consumer cpu usually has 128GB memory limit when server cpu supports terabytes. 128GB is really not that much.

Do they really can't handle more RAM? How wide is their address bus?

Is it possible to put more RAM if motherboard handles it?

-- Edit --

I think that my question is simple, but let me clarify it.

As we know theoretical limit of address space for 64bit is very large. I know that currently no one supports it (both on software and hardware). Windows x64 supports at least 2TiB, but it works with virtual memory so it doesn't have to be RAM only. On the other hand wikipedia states that virtual and physical address space uses 48 bits which is 256TiB of RAM.

Looking at the numbers MMU is able to translate at least 2TiB from virtual to physical space so my question is: "Why Intel/AMD states in the spacification that processor supports only 128GB of RAM?" Is it limited by design in hardware or artificialy to force people to buy more expensive cpus with larger limits?

Improve this question
$\endgroup$
8
  • 1
    $\begingroup$ a CPU definitely does not have that much memory... Maybe you meant hard disk (or SSD)? $\endgroup$
    – nir shahar
    Jan 21, 2022 at 18:21
  • 1
    $\begingroup$ Questions about consumer hardware are off-topic here. $\endgroup$
    – Yuval Filmus
    Jan 21, 2022 at 19:42
  • 1
    $\begingroup$ See: Fan out $\endgroup$
    – Guy Coder
    Jan 21, 2022 at 19:44
  • 1
    $\begingroup$ Around 50,000 dollars will buy you a Dell computer or a Mac with 1.5TB of RAM. For storage, you can buy 60TB SSD drives, plug 10 of them into a USB hub, plug 10 of those into another 10 port USB hub, and plug probably four of those hubs into your $50,000 computer. Can’t see a reason why that wouldn’t work. $\endgroup$
    – gnasher729
    Jan 22, 2022 at 16:20
  • $\begingroup$ Where is it specified in Intel/AMD manuals that there is a limit of 128GB of RAM? $\endgroup$
    – user123
    Jan 24, 2022 at 14:54

1 Answer 1

Reset to default
1
$\begingroup$

There is no fundamental reason. It's simply an engineering decision, an implementation detail. If you want 1TB of RAM, you buy a server system that can handle 1TB of RAM, not a desktop system that can only handle 32GB.

The limitation is more likely to be in the memory controllers which generate the electrical signals to talk to the memory chips, not in the memory management unit. More chips means more signals, more complex decoding, and also stronger drive current needed to make the signal reach all the chips. Servers often use "registered ECC" where each RAM stick has a "register" which acts as a signal booster.

The same memory management unit (or a very similar unit) is used on the 1TB systems and the 32GB systems - so the same software can run on both units

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.