is DDR2 backwards compatible?

H

HallaDuke

Limp Gawd
Joined
Jul 9, 2005
Messages
154
I just finished reading an article that suggested buying newer DDR2 ram modules when they are more plentiful, as opposed to buying older DDR chips which are averaging more costly. Is DDR2 backwards compatible? Or am I too lazy to research it myself...? :eek: :eek: ;)
 
(cf)Eclipse said:
it would be like putting a s939 a64 into a s754 socket :p
Click to expand...
im sure if you pushed hard enough you could get it to work!!! elbow greese and headlight fluid work well ;)


soulsaver_8229
 
i wished that amd set it up in such a way that you can use a s939 chip as single channel in s754 or opposite with s754 as a single channel cpu in a dual channel capable board, just not use the ram slots for the 2nd channel. would make things sooo much easier in some respects
 
what they need to do is

Make the next CPU 1700 Pins, and have many of them do nothing.... and when a future revison comes out, they can use some of them pins for some of the newer features... thus, keeping the same socket for liek 10 years
 
USMC2Hard4U said:
what they need to do is

Make the next CPU 1700 Pins, and have many of them do nothing.... and when a future revison comes out, they can use some of them pins for some of the newer features... thus, keeping the same socket for liek 10 years
Click to expand...

rofl
I dont know if you're joking, but if you aren't......well, too bad.
 
im prob gunna screw this up and i forget where i read it but there is an adapter that allows DDR1 memory to fit into a DDR2 motherboard.
 
USMC2Hard4U said:
what they need to do is

Make the next CPU 1700 Pins, and have many of them do nothing.... and when a future revison comes out, they can use some of them pins for some of the newer features... thus, keeping the same socket for liek 10 years
Click to expand...
i'm hoping that is what M2 will end up being like. remember how long socetA stuck around? :eek:
 
EquaLiZr said:
im prob gunna screw this up and i forget where i read it but there is an adapter that allows DDR1 memory to fit into a DDR2 motherboard.
Click to expand...
yep, DDR 2 is nothing more than a socket change and a diffrent way of working
 
and that adaptor has chances of killing oc'ability and maybe latency. i'll have to see how it works.
 
ScHpAnKy said:
Ten years is a stretch, but the concept is quite feasible.
Click to expand...

First off, I thought it might be as well. And then I thought again.
What about, when a new processor comes out, that invariably supports the new type of memory, or a new chipset, or something like that? I mean, if the processor was to support the old technologies as well as the new ones (depending on what was on the board), that would involve building into the processor backwards compatibility, which inevitably raises costs as well as die size. Think of the new generation of video cards that don't support hardware DX7.0 or 8.0, all of those functions are done through the shaders, so as to reduce costs and advance the new features.
It's really the same sort of thing with backward-compatible CPUs you're suggesting. Creating redundant pins/connectors that link to nothing would be pointless, because if a new technology that came along and replaced the old one (eg DDR3 over DDR2), the processor would have to A)use it's current pins for DDR2, and B) new pins for DDR3. This would eat up the available redundant pins like Winnie the Pooh in a honey shop, and with the way things move these days, that's not altogether a good thing.
And in the case of say, the Athlon64, with it's onboard memory controller, would that mean that in a case of new memory standard, would both types of memory controller have to be integrated as well? Or would R&D be put into developing a dual-access controller?
As you said, the concept is quite feasible, but practice really doesnt seem to be, unless you're keen on increasing costs for consumers and decreasing the speed at which things move by holding onto the past.
 
banGerprawN said:
As you said, the concept is quite feasible, but practice really doesnt seem to be, unless you're keen on increasing costs for consumers and decreasing the speed at which things move by holding onto the past.
Click to expand...
Actually even the concept is quite unfeasible...

The North and South Bridge Chipsets are Hardware, and in most aspects, umprogramable. Bios Flashing does alter some of the specs, but with a very low range of choices...

If you wanted to create something as described above, you would have make changes in the Motherboard (not Processor or Memory or etc...) and the changes would mean making a more programable Chipset.

Now how could you possibly predict what options to insert into the programming eviroment so as to make the Chipset accept future advancements...?

It would be like predicting the future... And quite unfeasible...
 
(cf)Eclipse said:
i'm hoping that is what M2 will end up being like. remember how long socetA stuck around? :eek:
Click to expand...
socket M2 will not be much of an improvement over 939 & 754, is it? i think that only something more revolutionary will be able to do that.
 
erorr404 said:
socket M2 will not be much of an improvement over 939 & 754, is it? i think that only something more revolutionary will be able to do that.
Click to expand...
not a large jump. it'll definitly help out dual core more than single core though.
 
USMC2Hard4U said:
what they need to do is

Make the next CPU 1700 Pins, and have many of them do nothing.... and when a future revison comes out, they can use some of them pins for some of the newer features... thus, keeping the same socket for liek 10 years
Click to expand...
1700 pins is a lottt

I liked your idea of having many of the pins do nothing. Not the part about using pins for newer features though, it's funnier to have pointless pins.
 
ChingChang said:
I liked your idea of having many of the pins do nothing. Not the part about using pins for newer features though, it's funnier to have pointless pins.
Click to expand...

We all know AMD is in the business to give us a good chuckle :p. Seriously though, AMD should find some way to keep the socket changes reasonable, I'd like to see another socket A ;) (I'm sure it's going to be much more difficult after integrating the memory controller into the processor though).
 
Pin function need typically does not drive socket change, speed and power capabilities do. I think the on-die controller was the first time in microprocessor history where a new function drove the new socket with more pins.

-tReP
 
You must log in or register to reply here.
Back
Top