OK 64 bit gurus .

[thirdlegstump]

God damnit dude, it's not "athalon," it's ATHALONG!!!

 

[Adn4n]

Originally posted by: Xatrix
Personally I am waiting for the s939 nforce4 SLI's.

And that's about the only thing worth getting an Nforce4 right now, SLI. If you're not going to get 2 cards immediately, then I'd go with Nforce 3 939.

 

[Algere]

Originally posted by: ViRGE

Originally posted by: Algere

The dual-cores are also coming out for S939, so unless AMD manages to boost performance a good deal, there's not a lot of pressure to move to S900.

More like not a lot of pressure to move to DDR2.

I can see the possibility why AMD may want to move to socket 900. To prevent "noobs" from installing a DDR1 compatible CPU into a DDR2 compatible motherboard. I can only imagine all the customer service calls AMD & motherboard manufacturers would have to face if they offered socket 939 in both DDR1 and DDR2 flavors.

The socket has to change, DDR2 has a different pin count.

The only physical relevance I am aware of between DDR2's pin count is that between the motherboard and the memory stick. I don't see why DDR2's pin count which is different from DDR1's (to prevent installation of DDR1 memory into DDR2 compatible motherboards or vice versa?) has something to do w/ a decrease in CPU pin count (939 to 900). I don't believe the socket change is due to DDR2's physical pin count. If it is, couldn't AMD just have socket 939 with 39 dead/inactive pins to be compatible with the specifications of socket 900?

If I'm wrong do you have reference saying otherwise?

 

[ViRGE]

Originally posted by: Algere

Originally posted by: ViRGE

Originally posted by: Algere

The dual-cores are also coming out for S939, so unless AMD manages to boost performance a good deal, there's not a lot of pressure to move to S900.

More like not a lot of pressure to move to DDR2.

I can see the possibility why AMD may want to move to socket 900. To prevent "noobs" from installing a DDR1 compatible CPU into a DDR2 compatible motherboard. I can only imagine all the customer service calls AMD & motherboard manufacturers would have to face if they offered socket 939 in both DDR1 and DDR2 flavors.

The socket has to change, DDR2 has a different pin count.

The only physical relevance I am aware of between DDR2's pin count is that between the motherboard and the memory stick. I don't see why DDR2's pin count which is different from DDR1's (to prevent installation of DDR1 memory into DDR2 compatible motherboards or vice versa?) has something to do w/ a decrease in CPU pin count (939 to 900). I don't believe the socket change is due to DDR2's physical pin count. If it is, couldn't AMD just have socket 939 with 39 dead/inactive pins to be compatible with the specifications of socket 900?

If I'm wrong do you have reference saying otherwise?

Looking at the DDR and DDR2 spec sheets, the first thing that immediately pops up is that DDR2 calls for 2 on-die-termination pins, although I don't know enough about ODT to know if it has to interface with the CPU. Otherwise a quick count of all the pins in Excel shows that everything else seems to be the same, with the majority of the new DDR2 pins dedicated to grounding. Still, with any memory pin change, I'm curious how it would affect routing the memory traces to the socket, and if things would need to be changed to meet the changes in DDR2.

 

[Pollock]

Originally posted by: Algere
Socket 900 (take it for FWIW)

"05/26/04 - DDR 2 and Socket 900 for Q2 2005
Some good Asian friends send us words about AMD plans regarding DDR 2 and it could come out much sooner than expected. According to well informed OEM sources, AMD is already testing a 90nm SOI DDR2 version on socket 900 code named Toledo which will be followed, within a month or two, by dual-core DDR2 90nm chips Egypt, Italy and Denmark from the K9 family. Coupled with a massive 90nm ramp, AMD is undoubtedly planning to put a tremendous pressure on Intel during the coming months..."

Source

http://www.c627627.com/AMD/Athlon64/
Near the bottom...Q4 2005. *shrugs*

 

[Algere]

Originally posted by: ViRGE

Originally posted by: Algere

Originally posted by: ViRGE

Originally posted by: Algere

The dual-cores are also coming out for S939, so unless AMD manages to boost performance a good deal, there's not a lot of pressure to move to S900.

More like not a lot of pressure to move to DDR2.

I can see the possibility why AMD may want to move to socket 900. To prevent "noobs" from installing a DDR1 compatible CPU into a DDR2 compatible motherboard. I can only imagine all the customer service calls AMD & motherboard manufacturers would have to face if they offered socket 939 in both DDR1 and DDR2 flavors.

The socket has to change, DDR2 has a different pin count.

The only physical relevance I am aware of between DDR2's pin count is that between the motherboard and the memory stick. I don't see why DDR2's pin count which is different from DDR1's (to prevent installation of DDR1 memory into DDR2 compatible motherboards or vice versa?) has something to do w/ a decrease in CPU pin count (939 to 900). I don't believe the socket change is due to DDR2's physical pin count. If it is, couldn't AMD just have socket 939 with 39 dead/inactive pins to be compatible with the specifications of socket 900?

If I'm wrong do you have reference saying otherwise?

Looking at the DDR and DDR2 spec sheets, the first thing that immediately pops up is that DDR2 calls for 2 on-die-termination pins, although I don't know enough about ODT to know if it has to interface with the CPU. Otherwise a quick count of all the pins in Excel shows that everything else seems to be the same, with the majority of the new DDR2 pins dedicated to grounding. Still, with any memory pin change, I'm curious how it would affect routing the memory traces to the socket, and if things would need to be changed to meet the changes in DDR2.

A far cry from knowing/stating as definite that the pin count on DDR2 is relevant to the changed pin count from socket 939 to 900, assuming the socket change is true and not rumor. At this time it's all speculation & for all we know the decrease in pin count (other than what was already said) could be due to enhancements to socket 939 itself not needing as many pins as before or as you've said DDR2, but not for the reasons you've stated. Perhaps the lower voltage requirements of DDR2 (in contrast to DDR1) allows for a lower pin count or something else altogether.

 

[Algere]

Originally posted by: Pollock

Originally posted by: Algere
Socket 900 (take it for FWIW)

"05/26/04 - DDR 2 and Socket 900 for Q2 2005
Some good Asian friends send us words about AMD plans regarding DDR 2 and it could come out much sooner than expected. According to well informed OEM sources, AMD is already testing a 90nm SOI DDR2 version on socket 900 code named Toledo which will be followed, within a month or two, by dual-core DDR2 90nm chips Egypt, Italy and Denmark from the K9 family. Coupled with a massive 90nm ramp, AMD is undoubtedly planning to put a tremendous pressure on Intel during the coming months..."

Source

http://www.c627627.com/AMD/Athlon64/
Near the bottom...Q4 2005. *shrugs*

At this point it's all rumor to me. Looks more plausible than Q2 nevertheless.

 

[PumpActionWalrus]

Originally posted by: Algere
Socket 900 (take it for FWIW)

"05/26/04 - DDR 2 and Socket 900 for Q2 2005
Some good Asian friends send us words about AMD plans regarding DDR 2 and it could come out much sooner than expected. According to well informed OEM sources, AMD is already testing a 90nm SOI DDR2 version on socket 900 code named Toledo which will be followed, within a month or two, by dual-core DDR2 90nm chips Egypt, Italy and Denmark from the K9 family. Coupled with a massive 90nm ramp, AMD is undoubtedly planning to put a tremendous pressure on Intel during the coming months..."

Source

No way it is true, using DDRII will require a new memory controller, also increasing the pincount of current chips, unless of course they decide to go with 64bit memory channel which is highly unlikely, I doubt this is even remotely true seeing that dual channel is one of the main things the Athlon 64 FX has going for it.. Just wait for that San Diego to come out I believe it will be capable of 3.2 easy

 

[PumpActionWalrus]

Originally posted by: deathkoba
God damnit dude, it's not "athalon," it's ATHALONG!!!

Athlon, or as you put it ATHLONG!!!

 

[Algere]

Originally posted by: PumpActionWalrus

Originally posted by: Algere
Socket 900 (take it for FWIW)

"05/26/04 - DDR 2 and Socket 900 for Q2 2005
Some good Asian friends send us words about AMD plans regarding DDR 2 and it could come out much sooner than expected. According to well informed OEM sources, AMD is already testing a 90nm SOI DDR2 version on socket 900 code named Toledo which will be followed, within a month or two, by dual-core DDR2 90nm chips Egypt, Italy and Denmark from the K9 family. Coupled with a massive 90nm ramp, AMD is undoubtedly planning to put a tremendous pressure on Intel during the coming months..."

Source

No way it is true, using DDRII will require a new memory controller, also increasing the pincount of current chips, unless of course they decide to go with 64bit memory channel which is highly unlikely, I doubt this is even remotely true seeing that dual channel is one of the main things the Athlon 64 FX has going for it.. Just wait for that San Diego to come out I believe it will be capable of 3.2 easy

Uhm... Welcome to socket 939?

Socket 900 is only 39 pins less. You think it's out of the realm of possibility for AMD to "modify" socket 939 and make it more efficient/workable with 39 pins less? & AFAIK there hasn't been much talk (if any) of how many pins more or less a DDR2 compatible memory controller on a A64 will take in comparison to the existing memory controller that's on the A64 / FX today.

P.S. "AMD recently said that its desktop AMD64 chips with DDR2 support will also have 939-pin packaging, though, it is not clear whether such processors will be able to work with Socket 939 mainboards made in 2004." - X-bit

The news bit is pre socket 900 so the rumor still stands unless someone can find a source stating otherwise.