jvd said:my understandnig is they wont move to ddr2 till late 2005 .
Dunno if its valid or not.
They will most likely just move to ddr 500 as its becoming more avalible .
Moffell said:Well if you buy a socket 939 mobo today you won't be able to use DDR2 with it. The same goes for if you get a s754 board so in that aspect the upgradability is the same for the different sockets. S939 may or may not get dual core eventually but such a cpu might be handicapped for the lack of DDR2 bandwidth goodness. On the other hand it might not, s754 is very close in performance to s939 despite half the bandwidth. Any more question marks that need to straightened out?
www.aceshardware.com/read.jsp?id=65000314
I can tell you what I'm going to do if that's of any interest to you. I'm waiting for a cheaper s939 cpu, preferably on 90 nm process. Put that on a nice and pricey mobo (with DDR) and hope that the mobo manufacturor made a good job in sticking to AMD's spec so that it can handle a dualcore in the future. I'm thinking November-December timeline here. My AXP 1800+ comp still feels "spiffy", but I could use a couple of SATA ports on it.
If I were buying something today I'd get a s754 3200+ with a nice mobo, they have really come down in price lately and Semprons will likely fit the mobo the next couple of years. I sure as hell ain't no Oracle though, take heed.
I believe I read on overlclockers.com that there is a socket change.3dilettante said:How does AMD hope to keep the exact same socket through the transition?
They'll have to change something in the socket, or have implemented some hidden ddr/ddr2 logic in their controllers, otherwise people are going to plug a ddr-only chip into a ddr2 board, or vice-versa.
The next question is what pins on a DIMM must be routed to the controller. I'd hate to see what happens if a power or ground pin gets shunted to a data pin, potentially killing a cpu or a DIMM.
You really shouldn't believe everything you read at overclockers.com. They also claimed that socket939 would use a 250Mhz "fsb" (instead of 200Mhz) when it was obvious (to me at least) that AMD would simply use a 5x multiplier (instead of 4x). They said this would prevent you from overclocking, since there would be no headroom left. They kept repeating that fud as fact for a long time, then when they were proved wrong about it they never mentioned it again. They've also said lots and lots of other shit that has turned out to be false. I really wouldn't trust them further than I could throw them.radeonic2 said:I believe I read on overlclockers.com that there is a socket change.
radeonic2 said:I believe I read on overlclockers.com that there is a socket change.
3dilettante said:How does AMD hope to keep the exact same socket through the transition?
They'll have to change something in the socket, or have implemented some hidden ddr/ddr2 logic in their controllers, otherwise people are going to plug a ddr-only chip into a ddr2 board, or vice-versa.
The next question is what pins on a DIMM must be routed to the controller. I'd hate to see what happens if a power or ground pin gets shunted to a data pin, potentially killing a cpu or a DIMM.
Tim said:I hope you don't take anything that idiot Ed writes serious.
Tim said:radeonic2 said:I believe I read on overlclockers.com that there is a socket change.
I hope you don't take anything that idiot Ed writes serious.
Guden Oden said:Tim said:I hope you don't take anything that idiot Ed writes serious.
Ed thinks someone died and made him king shit of everything. It seems there's not a single topic in the entire universe he does not know all there is to know about, at least according to he himself...