• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

4000+ 939 with 1MB of L2 cache?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

OC Detective

Member
Joined
Jul 30, 2001
Location
Mauritius
According to the link here..
http://www.aceshardware.com/forum?read=115098351
Whether it is true or not is an entirely different matter (maybe Petr can come by and throw cold water on it)! The letter "I" within the code DEI5AS currently does not exist for A64's and would indicate perhaps a lower than normal maximum T Die temperature if it were true!
Also states FX55 will be 130nm - nothing new there and also 939's at 90nm will initially be 3000+ and 3200+. Hope this turn outs to be true, I am definitely interested in a 3200+ if it is genuine!
 
Last edited:
I wonder if AMD will just do a soft launch of the 90nm 939's, (especially if its just lower frequency ones such as the 3000 and 3200+, with nobody actually knowing until they appear in the channels? Might even appear this month!
 
fx-55 on 130nm and 90nm starting at 3000-3200+ sounds to me like they are having the same troubles with 90nm as intel and ibm are.

why make the fx-55 130nm if you have no problems making a 2.6ghz 90nm chip?
 
My $ is on Athlon64 4000+ being a Q4 2004 130 nm part.

Also, there'll probably be

no 130nm 3000+/3200+ Athlon64 parts,
yes 90nm 3000+/3200+ Athlon64s which we'll see by the end of the month.

There will be no such thing as Socekt 939 4000+ 939 with 1MB of L2 cache. When it gets eventualy released in 2005, it will have 512 KB L2 cache.

130nm Athlon 64 FX-55 at 2.6 GHz was always rumored to be coming in October 2004. Even AMD official CPU roadmap has only 130 nm Athlon 64 FX parts in 2004.

90nm Athlon 64 FX was never rumored to be here before 2005.

Today is Sep 09, 2004

EDIT:
c627627 said:
There will be no such thing as Socket 939 4000+ 939 with 1MB of L2 cache. When it gets eventualy released in 2005, it will have 512 KB L2 cache.

The 90nm 4000+ will be released in 2005 but that doesn't mean that there won't be a 130nm Newcastle 4000+ in 2004.

What I meant to say is there'll be a SH7-CG 2.6 GHz / 512kB 130nm 4000+ which may have 1 MB cache but 1/2 will be disabled so it'll effectively have 512 KB cache.
 
Last edited:
c627627 said:
There will be no such thing as Socekt 939 4000+ 939 with 1MB of L2 cache.

And then it would be right again, according to AMD they only want one CPU with the full meg of ²Cache (on the normal sockets, of course Opteron will always have 1MB ²Cache, maybe even more in the future, but unlikly I think)
 
Dual Core Opterons will have 2 MB of cache in 2006.
...and so will desktop Toledo.
 
Last edited:
c627627 said:
Dual Core Opterons will have 2 MB of cache in 2006.

Uh, I didn't know that- thank you! (I'm not really into Opterons, but thanks for that correction again)

I actually do not think AMD will wait until 2005 to release 90nm, maybe they are closer to ship them than we all think. I mean, I doubt that they will make that much commercial about it as intel did, maybe they are nearly out- however, I'm looking forward to it.

But first we'll have the "e- stepping" of A64, and I want to see if it has strained SOI
 
Revisions:
CG = 130nm
D0 = 90nm
E = dual core

FX55 on SH7-CG is sure thing, we knew that for months. The SH8-D0 core is now available only for Opterons, I really don't know why not for FXs.

The 4000+... well, originaly it was planed as DH8-D0 core (Winchester) at 2.6 GHz but CPU-Z screenshot posted on Xtremesystems shows it as 2.4 GHz / 1MB L2 SH7-CG. Considering BrandID matches on that screenshot to A64 (and not FX) this is possible. But still it may be a prototype, originally there was a plan for 3600+ s939 at 2.2 GHz / 1MB that was later changed to 2.4 GHz / 512kB and renamed to 3700+ and then later renamed to 3800+. As far as I know AMD plans were for 1MB cache only for FXs and all A64s were supposed to have 512kB only. Such a change must mean it is an early prototype or AMD has problems with 90nm reaching high clock speeds. In fact all AMD people I spoke with about this were surprised.
 
There wont be socket 754 90nm chips and there won't be any Socket A chips beyond Athlon XP 3200+.

90nm Athlon 64 3000+ and Athlon 64 3200+ Socket 939 Winchesters are coming this month, and then even the Semprons will switch over to Socket 939.

They're going to eventually figure out how to scale Winchesters beyond 4000+. This will happen in 2005 and then it'll be on to Socket 900 dual core DDR II 2MB cache Toledo in 2006.
 
DFI definatly fubared it then! lol but i take it that the 3000+ and 3200+ will be reasonabley priced? if so, then its time to go s939, well for me, time to change my plans. lol
 
Where did you hear Sempron will switch over to Socket 939? I do not keep up with this stuff that well and wondered where you get your information.

From what I understand, is that Sempron will be Socket 462 as a Thornton, then Socket 754 as Paris with 256KL2 as well, but the 90nm Sempron will still be Socket 754, but may have 512K of L2, Palermo.

But...it would be cheaper for AMD to standardize everything into 939, then they could give Sempron dual channel memory, but instead of 512K, only 256K L2. Or if needed, they could make it a 939 CPU, put only have a single channel core and the extra pins do nothing. It is a theory....

EDIT: Here it is:
http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_11599_11603,00.html
AMD Sempron™ Processor Overview
The AMD Sempron™ processor performs at the top of its class when running the home and business applications that you need most. The AMD Sempron™ processor’s full-featured capabilities can include HyperTransport™ technology, up to 512k total high-performance cache, advanced 333Mhz front-side bus, 3DNow! Professional technology, and an integrated DDR memory controller.

Since Thornton and Paris are 256K, the only one left is the upcoming 90nm Palermo that could possibly have 512K.

Now..if it was 939 and dual channel...and had 512k....the only difference it would have from the 90nm version of Newcastle would be that it does not have any 64bit support.
 
yeah, but the new DFI boards are supposed to support mobiles so we might still get to use 90nm Socket 754 chips on a desktop
 
Back