Page 1 of 2

AMD's 90nm Winchester CPU

Posted: Wed Nov 10, 2004 10:06 am
by Apoptosis
After having the 3500+ Winchester for almost two weeks I am very impressed with this CPU. It incorporates all the positives of the Newcastle it replaces, with a huge drop in operating temperatures. While some people will complain that there is virtually no performance boost over the 130nm technology, the optimists amongst us will look at an AMD processor that was 10C cooler at 2.64GHz, running without a hiccup. M system had some limiting factors that I need to iron out, but I believe this CPU has even more headroom for overclocking.

Posted: Wed Nov 10, 2004 10:48 am
by Bio-Hazard
Good review............ :shock: I was expecting to see a little higher FSB though. But the 90nm cores really do seem to rock and roll.

Posted: Wed Nov 10, 2004 11:14 am
by infinitevalence
Rich with a slam dunk of a review

Code: Select all

  o/    \  /     |
  |      o       |
  /\             |

Posted: Wed Nov 10, 2004 11:22 am
by LVCapo
Thank you guys, means alot coming from the two of you.
I too wish I had been able to hit a higher FSB, but I tried just about everything, different memory, a different PSU...... nothing really helped. My other feeling was that it was a review to look more at the thermals than overclocking or a high FSB, therefore I thought leaving it at the stock multi was more important than dropping it down and putting more pressure on the board or memory. Just my personal thought....I have seen reviews where people claim to have hit 290X9, I tried and tried after writing that, but anything over 267 borked my system totally.

Posted: Wed Nov 10, 2004 11:44 am
by infinitevalence
ehh. don’t stress it, thermals are VERY important to real overclockers, like you me and bio, yeah it would have been nice if you could have hit 290fsb but the only way people can do that is by setting a divider on the ram and by lowering the HT multi. And all of those things can have a seriously negative impact on performance if the user does not know what they are doing. if you did not try some of those things you might want to play with them and see if they do indeed give you more performance, not just fsb, because i would rather read a review that shows how to get the best performance than one that gives a ridiculous fsb even though the system performs better at stock speeds.

Posted: Wed Nov 10, 2004 9:46 pm
by Impulse
What BIOS you'd use for the A8V? 1007 or one of the betas (1008 or the new 1009)? I'm putting together a Winchester setup with the board and haven't been able to acquire much info about the ASUS BIOS'es at my usual haunts.

Think I read someone stated the 1008/1009 would allow for better voltage adjusment with Winchesters or something, not sure what's up with it on 1007. Nice to see the temps seem that the temps reported are semi-reliable though, given how many problems I've heard with temp reporting on other boards.

Edit: Forgot to mention I've got a Rev 2.0, not sure what you're using (noticed on the boards you had an earlier Rev in-house for testing a while ago, I'm guessing if you hit 240 it's a Rev 2.0 with working locks or a BIOS updated Rev. 1.x with working locks as well though).

Posted: Wed Nov 10, 2004 10:50 pm
by LVCapo
I used the 1008 BIOS, the 1007 was actually the first BIOS for ASUS that supported the 90nm Winchester....something I should have touched on more in my article. I always recommend using the newest BIOS version for any board, unless specifically told otherwise (like the 1005 BETA with the AGP/PCI locks). I did in fact use the rev 2 board, had a 1.1 and RMA'ed it after 2 weeks or so. The rev 2 boards are so much better

Posted: Wed Nov 10, 2004 11:16 pm
by Impulse
Thanks for the info, I just noticed the 1008 is now (as of today) listed under their main downloads so I guess it isn't beta anymore.

Posted: Thu Nov 11, 2004 6:00 am
by Liquid3D
Yes I really appreciated your focusing on the thermal aspects of this die shrink. It's difficult when there's not a large jump in performance, and in all fairness I can't blame the consumer for wanting to see it in every next generation. But as Overclocker's we understand all the other benefits of a die-shrink, especially when it WORKS, hehe. I sort of disagreed with all the criticism towards Intel because their Socket-478 Prescott overclocked very well for me. My 3.0E ran to 4GHz wil little Vcore.

I beleive the problems with S-478 Prescott were never really that severe where heat was concerned. Although most argue this point with me I believe the S-754 A64 ran as warm as the S-478 Prescott. I saw the trrue source of the problem was Intel's failure to implement motherboard CPU-power circuitry changes. Prescott for S-478 was launched on a motherboard which was not designed for it's power reuirements. Although the Vcore is lower on Prescott, current demands are much higher. The mosfetts on my Abit AI7 cooked at 70+ C. I wouldn't have even known this had Abit not made the AI7 their first board with uGuru. Luckily that chip read a diode monitoring power circuitry feeding the CPU. They really screwed themselves releasing Prescott onto the S-478 without beefing up the boards power circuitry.

My P4 530 (S-775) with the Abit AA8 having Mosfetts the size of small fridge's, easily clocks to 4GHz, and runs to 3600MHz without Vcore inscrease. My 3.0E S-478 Prescott would also run to 240FSB sometimes 250FSB without Vcore adjustment.

Why am I discussing alll this? because until AMD released this chip. no one had actually made SSOI technology work until now. So while Intel's chips did overclock well, they have been plagued with power circuitry problems where the mobo's are concerned. I understand the Silicon On Insulator Fab process was supposed to prvent electromigration, and junction capacitance and it seems like AMD has actually done it. When the right board come out, these chips may be the overclocking champions. And if AMD continues in this direction it seems their going to start getting some real high speeds out of them as well. Sorry about my long babbling reply here.

Posted: Thu Nov 11, 2004 10:31 am
by LVCapo
Please read the whole article!!!!!!:axe:
I touched upon all of that in "Rich's Thoughts".
I have personally owned three Prescott CPUs, a 2.8 and a 3.2GHz in the socket 478 and an unlocked testing sample in the socket T which I run at 3.6 GHz (This will be the subject of a future article); both socket 478 CPUs overclocked extremely well in my opinion. I ran the 2.8E 24/7 at 3.5GHz (until its untimely demise at my hands while trying to remove the IHS), I have also had the 3.2E clocked over 4GHz and run it at 3.71 GHz 24/7 without issue. In my opinion Prescott CPUs overclock extremely well and handle multitasking better than any AMD CPU I have ever used.

Posted: Thu Nov 11, 2004 11:14 am
by Liquid3D
I didn't mean to say I disagreed with "your" criticism, just THE criticism about Prescott. I was referring to criticism on the whole about Prescott. Were in total agreement in "thoughts."

Posted: Thu Nov 11, 2004 4:20 pm
by LVCapo
I think something you have to understand is that when the Pressie first arrived, people were disappointed in the fact that compared to the Northwood it ran extremely hot and showed almost no performance boost at stock speeds, but it seemed to scale better than the NW. I think people were expecting a boost worthy of "P5", and when they didn't get it, the negative feedback kinda wiped out any and all momentum. I also think, and this is just my opinion, that this is part of the reason both Intel and AMD are a little subdued in their recent releases, or even to an extent underselling their products. Also, another point is that i have on more than one occasion, linked your article regarding the Prescott and its effects on motherboard circuitry..... a very nice article.

Posted: Thu Nov 11, 2004 4:32 pm
by Liquid3D
Thank you, I take that as a huge compliment coming from you. Although I go back over my articles and say to myself "I could have wrote that better."

Yes I do agree, and once the Prescott gets going, like 3.5GHz, the 1MB cache kicks in, and you no longer have to be a spulunker to accept the pipeline depth.

Posted: Thu Nov 11, 2004 5:18 pm
by infinitevalence
Just a little note, its actualy IBM that got SOI and SSOI to work, while AMD is seeing the majority of the production of this stye chip the power5 and G5 have been using 90nm and SOI for a while now. You may even remember that one of the deals that AMD cooked up was to "license" hypertransport to IBM in exchange for help to fix SOI production. Lots of people seem to forget that the first A64's in the retail channel were a late stepping. Its my understanding that AMD went through about 5 steppings before releasing the A64. Not that i have a problem with this but its nice to know that EVERYONE has had problems transitoning to 90nm and to SOI. The problem is nither Intel or AMD, but phisics.

Posted: Mon Dec 27, 2004 6:54 pm
by FZ1
Hey, that was a great review. I've had my eye on the 3500+ CPU for a week or so and I have seen a price difference of anywhere from $50-$100. Some retailer sites don't even give the specs to differentiate. :x After reading this informative review, I'll definitely shell out the extra cash for the 90nm version.

BTW - I'm planning on building a PC for the first time. I'm looking at the new ABIT AX8 mobo as it supports PCIE to go with this CPU. I haven't seen a review on it yet but it just hit the market as I understand it.

Posted: Mon Dec 27, 2004 8:04 pm
by grunt
I too will be building a 90nm system but so far I havent seen enough gain to go with pcie. Of course I am a gamer and like to play newer titles so therefore I look more to the newer video cards which seem either hard to find or hard to fork over that much cash for being as the gains on benchmarking tests dont impress me (just my uneducated oppinion). Just today I noticed a 6600gt fairly cheap but even though its ddr3 memory I think its 128bit. Anyhow, just my 2 food stamps worth as we all know Im the nOOb here...

Posted: Mon Dec 27, 2004 8:35 pm
by FZ1
I guess I'm the noob here now :) I wanted the PCIE for the future as they'll prolly make the new cards for PCIE only and then I'll have to upgrade the mobo again. Although I'll prolly want to go the SLI route anyway and face the same predicament. :cry:

Posted: Tue Dec 28, 2004 2:26 am
by infinitevalence
For the cost of an SLI mobo you can just upgrade to a top of the line vid card. figure two 6600gt @ 200 each, mobo at 200 and your at 600 flat, now for a regular mobo 125 with top end vid 500 your at 625, in most cases the top end video card will preform just as well as two SLI 6600gt's and much better in some games that cannot take advantage of the SLI like everquest.

Posted: Tue Dec 28, 2004 4:34 am
by grunt
infinitevalence wrote:For the cost of an SLI mobo you can just upgrade to a top of the line vid card. figure two 6600gt @ 200 each, mobo at 200 and your at 600 flat, now for a regular mobo 125 with top end vid 500 your at 625, in most cases the top end video card will preform just as well as two SLI 6600gt's and much better in some games that cannot take advantage of the SLI like everquest.

Posted: Tue Dec 28, 2004 6:50 am
by FZ1
infinitevalence wrote:For the cost of an SLI mobo you can just upgrade to a top of the line vid card. figure two 6600gt @ 200 each, mobo at 200 and your at 600 flat, now for a regular mobo 125 with top end vid 500 your at 625, in most cases the top end video card will preform just as well as two SLI 6600gt's and much better in some games that cannot take advantage of the SLI like everquest.
That holds true for now which is why I'm not going that route. However, in a year or 2 I think the SLI will really start living up to it's potential. I really don't look forward to buying 2 $600 vid cards though.

I am also rethinking the AX8 mobo and may just go with the AV8. The AX8 is not readily available (I have been unable to find it in stock by a "reputable" retailer) and per my earlier comments, by the time the AGP is really not available on the new cards, it'll be time for SLI anyway so either way I'll "need" to upgrade if I want to stay current. Talking it out really does help! :mrgreen: