The new A64 Rev-E info and release dates?????

[RichUK]

Just a quick question about the new Rev-E A64 chips, i am basically just looking to see if anyone can just give me some info on the specs and also the release dates.

All i know so far is that they are going to be using strained silcon for better performance... i have heard somewhere here on anandtech that they can clock upto 3Ghz on air, is this true? does anyone have a linky for some information, or can anyone share there information???...

thanks in advance....

 

[Megatomic]

They will have SSE3 instructions too. I don't when they'll be released though.

 

[Rich85]

i think ive heard that they will also have mem controllers that are able to run 4 banks of mem at ddr400 speeds unlike the current ones which end up running ddr333

 

[miketheidiot]

Originally posted by: Megatomic
They will have SSE3 instructions too. I don't when they'll be released though.

I believe the 2.6ghz opterons are rev E. Just released today.
Text
Text

 

[Artanis]

Athlon64 Venice core (rev.E) and the mobile Turion (rev.E core) will emerge in 2 months, as I heard...

 

[Gamingphreek]

Originally posted by: Artanis
Athlon64 Venice core (rev.E) and the mobile Turion (rev.E core) will emerge in 2 months, as I heard...

Turion is not a processor, it is like Centrino (ie More a set of components).

-Kevin

 

[gobucks]

rich85, you're correct, although i'm not sure if you can still hit 1T timings with 4 sticks. If so, I know I'm gonna be adding some more crucial ballistix in the future!!!!

 

[shira]

All of us are guessing the when (the roadmap indicates 2nd quarter 2005), and there's no way of really knowing how much Venice will overclock (the 3.0GHz figure is wishful thinking, and makes the reasonable assumption that the improved process will overclock better than rev D Athlon 64's).

Of course, it's possible that all that the new process will do is move the overall distribution of frequencies for all CPUs produced a little higher, without increasing the potential overclockability for a given CPU. That is, if enough CPUs bin at 2.4GHz, AMD can create a 3700+ designation (whereas currently, a chip binning that high would still be called a 3500+).

Some/many people on these forums (most of them a lot more knowledgable than I am, so take what I say here with a big boulder of salt) believe that all Winchesters (and therefore Venices) are really essentially identical, meaning that that, say, a 3000+ is just as likely to overclock to 2.6GHz as a 3500+. But I have to believe that AMD's frequency binning means SOMETHING.

So my own prediction is that Venices will overclock (without extreme cooling methods) a little higher - say to 2.8GHz if you're a little lucky in your purchase, whereas today we don't usually see much above 2.65GHz.

Oh, one other difference I've read about for the rev E's is the inclusion of copper between the CPU layers. That's supposed to decrease resistence and/or interference between the layers and therefore increase the potential speed of the chip.

 

[bookman]

I also understand its supposed to consume 20% less power per GHz vs the D stepping. Can anyone confirm this?

 

[RichUK]

Originally posted by: bookman
I also understand its supposed to consume 20% less power per GHz vs the D stepping. Can anyone confirm this?

if this is true which i cant confirm it shall be a very tastey platform for OCing

 

[clarkey01]

Originally posted by: shira
All of us are guessing the when (the roadmap indicates 2nd quarter 2005), and there's no way of really knowing how much Venice will overclock (the 3.0GHz figure is wishful thinking, and makes the reasonable assumption that the improved process will overclock better than rev D Athlon 64's).

Of course, it's possible that all that the new process will do is move the overall distribution of frequencies for all CPUs produced a little higher, without increasing the potential overclockability for a given CPU. That is, if enough CPUs bin at 2.4GHz, AMD can create a 3700+ designation (whereas currently, a chip binning that high would still be called a 3500+).

Some/many people on these forums (most of them a lot more knowledgable than I am, so take what I say here with a big boulder of salt) believe that all Winchesters (and therefore Venices) are really essentially identical, meaning that that, say, a 3000+ is just as likely to overclock to 2.6GHz as a 3500+. But I have to believe that AMD's frequency binning means SOMETHING.

So my own prediction is that Venices will overclock (without extreme cooling methods) a little higher - say to 2.8GHz if you're a little lucky in your purchase, whereas today we don't usually see much above 2.65GHz.

Oh, one other difference I've read about for the rev E's is the inclusion of copper between the CPU layers. That's supposed to decrease resistence and/or interference between the layers and therefore increase the potential speed of the chip.

With IBM's help ( SOI/SS and other so called transistor breakthroughs) Im pretty sure they'll hit 3 Ghz by years end/ start of Q1 06. My understanding is that yeilds with time improve via tweaks and with the help of SS (FX 55 uses it) the 90 nm offerings should hit 3 Ghz without too much trouble.

If we go back to 2003, one of the reasons AMD was over a year late with the Hammer was due to poor yeilds at the planned 2 Ghz clock speed for the "3200+", they were pulling thier hair out and its alot harder to scale up a high IPC design like the K8 then any netburst chip out there. 2 years on and AMD has achieved 2.6 Ghz on the 130 nm process and it goes alot higher as seen with the overclocks ppl are getting, two years later and with quite a bit of tweaking they have gotten another 600 Mhz + more out of the design/tech.

And with Intel only adding cache to future single core designs they have plenty of time to ramp up.

 

[RichUK]

i surpose by what clarkey01 is saying, that the 90nm has just started off and has some time yet before it matures..

i have a question is this going to be the last rev before we see dual core? or will they still plod on with single cores aswell as the dual core?

 

[clarkey01]

AMD are not exactly rushing, well, to be honest theres no need to rush, they'll just see what Intel throws out and respond with there next speed bump.

But if it does comes with SS I will walk to Dresden to FAB 30 and hug the all the FAB workers

 

[shira]

Originally posted by: clarkey01
[With IBM's help ( SOI/SS and other so called transistor breakthroughs) Im pretty sure they'll hit 3 Ghz by years end/ start of Q1 06. My understanding is that yeilds with time improve via tweaks and with the help of SS (FX 55 uses it) the 90 nm offerings should hit 3 Ghz without too much trouble.

I was interpreting the OP's question to be asking how high the initial Rev E's will overclock.

I agree with you that ultimately we'll see Venice chips reach 3.0 GHz, and they highest-binned ones will be labeled as such, "Venice 4500+". Initially, however, we'll see the Venice 3700+ (2.4GHz), and we might see the better overclockers of the 3700+ and lower Venices reach 2.8.

 

[LrngToFly]

This site now lists Venice as being released in April:

http://www.c627627.com/AMD/Athlon64/

If anyone knows the actual date in April, please let me know! I'm itching to build myself a new computer- if this processor is being released in early April I might wait on it. If it's late April I'm going to order parts soon...

A couple of other interesting roadmap sites:

http://endian.net/details.asp?ItemNo=4022

http://www.a1-electronics.net/AMD_Section/CPUs/AMD_Roadmap.shtml

 

[bradley]

Looks like they're here.... that is, if *here* is in Switzerland. Check out these links.

http://www.stegpc.com/browse.asp?cat=123
http://www.stegpc.com/details.asp?prodid=amd-64-35v (look at bottom of page)

 

[LrngToFly]

If you're a lurker and interested in this, here's a more active thread on the waiting game for Venice:

http://www.hardforum.com/showthread.php?t=859000&page=9&pp=20

 

[LrngToFly]

The steg pc site is taking pre-orders for March 8th in Germany- let's hope it's not a paper launch! And that there's plenty of stock!

 

[LrngToFly]

http://www.stegpc.com/details.asp?prodid=amd-64-35v

By the way, even Winchester cores are built on a 90 nm process with SOI. Venice has SS - strained silicon- in addition to SOI. I suppose there's a chance that the stegpc site is "confused" - although Wincester obviously doesn't have SSE3 or the newer memory controller.

 

[bradley]

Originally posted by: LrngToFly
The steg pc site is taking pre-orders for March 8th in Germany- let's hope it's not a paper launch! And that there's plenty of stock!

False Alarm. Turns out these are actually Newcastles; and they only advertised them as Venice cores based on what was mistakenly written on the invoice.

 

[Avalon]

Originally posted by: LrngToFly
http://www.stegpc.com/details.asp?prodid=amd-64-35v

By the way, even Winchester cores are built on a 90 nm process with SOI. Venice has SS - strained silicon- in addition to SOI. I suppose there's a chance that the stegpc site is "confused" - although Wincester obviously doesn't have SSE3 or the newer memory controller.

Isn't it the other way around? Winchesters are using SS right now, and will be implementing SOI with their E revision?

I suspect we'll see rev E cores in a couple months, and I do think that, on the whole, they will reach a little higher than the D0 cores. However, if you've already got a good winchester sitting pretty at 2.6, there shouldn't be much incentive to upgrade, unless people are indeed hitting 3.0 on air.

 

[LrngToFly]

This is the 1 MB cache San Diego:

http://www.theinquirer.net/?article=21646

Venice has 512 KB- but let's hope a sighting of a San Diego means that they're close!

 

[clarkey01]

Originally posted by: LrngToFly
This is the 1 MB cache San Diego:

http://www.theinquirer.net/?article=21646

Venice has 512 KB- but let's hope a sighting of a San Diego means that they're close!

"The 4200+ has 1MB of cache built on 90 nanometre marchitecture and works with 1.4 volts. "

#
Venice is the athlon 64 and the FX core is San Diago, correct me if I am wrong.

Venice is however the replacment for Clawhammer which would indecate thats it ment for the FX's. They must just be using the FX core for the 4200+ as a one off.

 

[Hyperlite]

if Venice is the replacement for clawhammer, why does it have 512k L2?

 

[gate1975mlm]

http://www.c627627.com/AMD/Athlon64/

Next Month!