Computer architecture : DDR3 memory technology - slot capacity scaling

1

Speaking in terms of computer architecture, what are the characteristics embedded in a DDR3 memory module or the host slot, such that a user cannot put a larger RAM than the per-slot specification of the mainboard without failing to boot his PC?

Is the limitation caused by:

  • The number of pins/contacts routed in the PCB design of a mainboard?
  • Amount of power (Watts) that the mainboard slot can serve to the module?
  • The internal design of the memory module? (e.g: memory blocks being placed in non sequential order in the module, first block 0 MB to 1 MB, second block 2048 MB to 2049 MB, etc)

F.I.V

Posted 2015-02-09T13:45:04.847

Reputation: 62

There isn't one it comes down to product differentiation. As for the reason it doesn't disregard the extra memory that comes down to not using engineering resources in order to do so. Its unlikely an adapter exists. – Ramhound – 2015-02-09T13:57:16.860

@Ramhound: Thanks, I am mainly looking forwards the reasons. What are the resources needed? Is there a power-consumption issue? Do the larger Memory Modules Architecturally differ? – F.I.V – 2015-02-09T14:02:49.780

1I was specific. It takes engineering resources, or additional man hours, to design hardware that will disregard memory the specific SKU will eventually NOT support. Memory is memory there is no difference between 4GB and 8GB except for the amount of silicon. – Ramhound – 2015-02-09T14:06:01.383

"except for the amount of silicon", That is what exactly I expected. You may remember that the old hard-disks had a "capacity-limit" jumper to discard the extra space. Same way I expected that mainboard could just discard the extra silicon and use the remaining. I could not understand the difficulty behind this which prevents its realization. – F.I.V – 2015-02-09T14:14:13.567

You are not a hardware engineering designing DD3 memory modules. DDR3 is a standard and the standard itself does not support what you are talking about. – Ramhound – 2015-02-09T14:23:39.400

This gets too close to the point. That is the reason I mentioned "Architectural" in the question. Could anyone point me to a standard, specification, etc that shows the origin of these limitations? – F.I.V – 2015-02-09T14:41:12.747

1Most of that stuff is behind pay walls. There isn't anything I can provided that didn't cost me money to access. – Ramhound – 2015-02-09T14:46:45.777

Answers

1

This has to do with internal addressing of ram chips of the module. If it's outside the chipset addressing then there's nothing one can do (more specifically - memory controller's addressing capability, which not necessarily may be part of chipset).

It's not only architectural (in the meaning you probably ask). It goes deeper: Ram chip -> Ram module -> Chipset -> CPU socket -> CPU. Whole chain...

Of course, my Integrated Circuit theory and architecture was looong time ago (Pentium III was THE SHIT back then), so I may not remember it correctly.

And no, no adapters for that are available. No reason, not cost-effective and hellishly difficult to do path routing/mapping/etc. You basically would need to give it it's own memory controller... And other stuff.

EDIT: It's somewhat similar in substance to the question: "What is the difference between 32-bit and 64-bit OS and why do I need different hardware for the latter".

AcePL

Posted 2015-02-09T13:45:04.847

Reputation: 1 571

This is also another similar substance would be : How is it possible that I install my old 32-bit OS on my new 64-bit hardware without problem (degrading to 32 bit sucessfully). – F.I.V – 2015-02-16T08:42:28.847

Aha! Trying to disprove the statement... ;) Well, it's because of the "downward compatibility" that's the god of modern computer manufacturing industry... – AcePL – 2015-02-16T09:16:18.797

Asked: 2015-02-09T13:45:04.847

Viewed: 145 times

Active: 2016-12-03T21:11:52.297