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

Winchester 90nm SS CPU Thread.. (Good Read)

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

OC-Master

Member
Joined
Jul 14, 2001
Location
Edmonton, Alberta
Ok, I've spent all weekend researching literly hundreds of sources on the web finding information on Strain Silicon Based Winchester Cores.

The reason for my research is simple, I'm looking on the market for a S939 Winchester CPU that can overclock like mad.

I have came to a very interesting conclusion and have figured out the mystery to which CPUs are SS garanteed and which CPUs are not SS garanteed.

Ok, lets fill those of you in on strained silicon and what exactly it accomplishes. Strained silicon is one of those rare new technologies that enables a fairly dramatic increase in performance (MHz) with a relatively simple change in starting materials. Proof that transistors fabricated with strained silicon were faster due to increased electron mobility and velocity was first demonstrated in the mid-1980s. Basicly, strained silicon helps overall CPU stability which in turn allows higher clock rates.

Let me explain to you which CPUs are available that are strained silicon process. Currently, only 130nm 4000+ and FX55 Athlon64 CPUs from AMD's line of processors carry this technology. To date, all of Intel's Pentium4's carry this process but that's a different story. Lately, many people have been wondering about the 90nm Winchesters and when they would recieve the same technological advancement boost. The anwser is, first week of november production.

So, any Winchester CPU made before week 45 is made with standard silicon process technology. Winchester cores created after week 44 are all produced with strained silicon process technology.

Their is several advantages to strained silicon CPUs. First being, its gonna be cheaper to build them in the long run because AMD is currently switching all FAB process'es to strained silicon with SOI which means they will eventually only need one method of production.
Strained silicon CPUs allow for higher yeilds which in return allow for more production of product.
And of course, strained silicon CPUs are supposed to clock better giving overclockers the edge.

However, to my studies, it seems that a few rare Winchesters floating about that were made before week 45 are carrying this advancement. I could easily see this possibilty simply because AMD's 4000+ and FX55 were in all in production on Week 32. I estimate at least 5% of Winchesters made between week 32 and week 45 have this process.

How to tell if your CPU has strained silicon process;
-It's extremely difficult to tell. Really, the only real way to find out is to see how far the chip can be pushed on air cooling without exceeding high temps or loosing stability. A good SS Winchester should be capable of clocking as high as 2.60GHz with stock or 1.35~1.40V core voltage.
-SS Winchesters when pumped with voltage and are well cooled should be able to push 2.7/2.8GHz as a maximum overclock.

A rumor behind this technology being implemented in Winchester's is for the reason, AMD wants to replace 1MB 2.4GHz 4000+ (FX53) with the appropriate line being 512KB 2.60GHz models. AMD's yeilds on 2.60GHz Winchester's just were not sutiable enough to sell so AMD simply renamed their FX53 to 4000+ and called it a day.

As soon as people can begin to buy those week 44 and week 45 CPUs, I wanna see some overclocks of both and see just how much of a difference their really is. We need some guinepigs.

I hoped to have educated some of you with this new process technology and how its going to affect AMD's new line of processor's in the near feature.

Please drop your feedback on what I've learned and most certainly post your overclocks if you think your CPU is Strained Silicon Processed. I would like to dig in more to see if its possible to identify a SS chip with software or by stepping somehow.


OC-Master
 
9mmCensor said:
So all winchester Chips after week 44 are SS?

Is this for all the winchesters?

Yes. AMD's production plants are now only producing Winchesters with strained silicon process. Only stock (week 44 or older) is most likely not made with the new process. Also give or take that 5% rule. This 5% could be higher but we'll stick with 5% so not everyone is gonna think they have SS Winchesters at the moment.

Also note that the 5% of people who have SS Winchesters will most likely have just gotten their CPUs and will be made most likely between week 32-44.


OC-Master
 
Last edited:
I think mine could be SS, just ask sen. I have the CBBGD core. Made in week 39 too. almost everyone else has week 37 and the older "F" core. Me and sen has agreed its a new stepping at that. Might have SS, I will need to see how high I can hit at 1.4 volts though. Charlie hit 3GHz for a screenshot and 2.85GHz superpi stable so I am guessing his was SS but was also a 3500+ He sold his for a nice profit cause he wanted to go fx55 and also cause winchesters dont like temps below 0 so he couldnt use his dry ice :( Funny thing is my mobo might run out of fsb before I max the cpu, id need 333fsb for 3GHz so if my cpu ocs like a gem, ill assume its SS and I will sell it, ask for $200 then buy a 3200+ SS for $200 a month later and run 290-300 fsb times 10.
 
f012t12 said:
OC-master, where'd you get your nforce4?

Btw, interesting post.

I dont have it yet. Thats my RIG that's gonna be completed by January. I want the Abit AN8-SLI but Abit says it wont be out untill the 2nd week of January. I also refuse to purchase an Athlon64 Winchester made before week 45 so I'm forced to wait.

Most of that stuff I got, I just gotta get the CPU, Board and graphics card..


OC-Master
 
Last edited:
The Winchesters sound good. I'll probably pick up one when the nForce4 boards come out, though I might end up with an ATi board when they come out. I'm not going to upgrade again until I can have a PCI-E slot, and that means getting a new video card at the same time...it'll be one hell of an investment.
 
Overclocker550 said:
I think mine could be SS, just ask sen. I have the CBBGD core. Made in week 39 too. almost everyone else has week 37 and the older "F" core. Me and sen has agreed its a new stepping at that. Might have SS, I will need to see how high I can hit at 1.4 volts though. Charlie hit 3GHz for a screenshot and 2.85GHz superpi stable so I am guessing his was SS but was also a 3500+ He sold his for a nice profit cause he wanted to go fx55 and also cause winchesters dont like temps below 0 so he couldnt use his dry ice :( Funny thing is my mobo might run out of fsb before I max the cpu, id need 333fsb for 3GHz so if my cpu ocs like a gem, ill assume its SS and I will sell it, ask for $200 then buy a 3200+ SS for $200 a month later and run 290-300 fsb times 10.

I'm almost positive that the strained silicon cores will have a different batch name from non-strained silicon cores.

Run your CPU at 1.4 volts and see how far you can push it without loosing stability. AMD will most likely be using 1.35V or 1.4V in their 4000+ Winchester CPU to reach 2.6 stable.

Can you stabilize at 2.6 with 1.4 volts or less? Can you prime/superpi and 3dmark without any problems?


OC-Master
 
I would like to know if a 3000+ Winchester at 2.6Ghz performs as well as a FX-55 at stock benchmark-wise.

If not, I wonder if it's just the extra cache.
 
f012t12 said:
I would like to know if a 3000+ Winchester at 2.6Ghz performs as well as a FX-55 at stock benchmark-wise.

If not, I wonder if it's just the extra cache.

The FX55 will pull ahead in almost everything beause of the extra cache. If you compare benchmarks and AMD's Performance Rating sheme, the extra 512KB of cache works out to 200MHz difference.

So, you would need a Winchester to run at 2.80GHz to match the speed of a stock FX55 on 95% of benchmarks. A Winchester at 2.6 delivers roughly the same performance as an FX53. Remember, currently, AMD renamed their FX53 to 4000+ so if you want to bench how much faster cache makes an Athlon64, look at the benchmarks of the 4000+ vs 3800+. Both CPUs run at 2.4GHz.

This will stay true untill AMD can get their yeilds up on the Winchesters.


OC-Master
 
Last edited:
OC-Master said:
I want the Abit AN8-SLI but Abit says it wont be out untill the 2nd week of January. I also refuse to purchase an Athlon64 Winchester made before week 45 so I'm forced to wait.

Sounds very similar to what I'm waiting on. I don't feel so bad waiting on the mobo now since it gives time for the CPU's to rotate stock to the SS. That and hopefully the quanities of PCIE 6800 GT's go up by then.

Good info, thanks for the read.
 
Sounds very nice to see A64 running 1.4V and reaching 2.6GHz, but switching from Newcastle to Winchester already got me from 2.4GHz to 2.6GHz. Even though I'm really jealous, I don't think I can afford to buy another A64. Plus, it'll be some endless change: 90nm-Strained Sillicon-SSE3 Instructions-New Steppings, etc.
I don't think mine is a SS either. Right now priming at 237x11 (2.61GHz) @ 1.6V
 
"Run your CPU at 1.4 volts and see how far you can push it without loosing stability. AMD will most likely be using 1.35V or 1.4V in their 4000+ Winchester CPU to reach 2.6 stable.

Can you stabilize at 2.6 with 1.4 volts or less? Can you prime/superpi and 3dmark without any problems?"


uh even if mine was SS I highly doubt it because amd isnt that dumb to mark mine down as a 3000+ when they could be selling it as a 4000+ its speed binned. I think all the SS chips before week 45 are amd's first trys with SS that failed to pass as 2.6GHz on 1.4v so I recon I can hit 2.6GHz at 1.5 to 1.6 volts and 2.7GHz at about 1.7v but I doubt ill push the volts this high. I think Osirus may be a speed binned SS too. A weak SS cpu may not oc any higher than a good one without SS
 
Overclocker550 said:
"Run your CPU at 1.4 volts and see how far you can push it without loosing stability. AMD will most likely be using 1.35V or 1.4V in their 4000+ Winchester CPU to reach 2.6 stable.

Can you stabilize at 2.6 with 1.4 volts or less? Can you prime/superpi and 3dmark without any problems?"


uh even if mine was SS I highly doubt it because amd isnt that dumb to mark mine down as a 3000+ when they could be selling it as a 4000+ its speed binned. I think all the SS chips before week 45 are amd's first trys with SS that failed to pass as 2.6GHz on 1.4v so I recon I can hit 2.6GHz at 1.5 to 1.6 volts and 2.7GHz at about 1.7v but I doubt ill push the volts this high. I think Osirus may be a speed binned SS too. A weak SS cpu may not oc any higher than a good one without SS

Your theory sounds dead on the nose.

According to http://news.com.com/, AMD's yeilds are finally looking decent and this will start looking good for the overclocking community.

With the newer made CPUs, dont forget about supply and demand. Once yeilds start picking up with the process, even the poorest of the batches could still be stamped as a 3000+ model over a 4000+ model. If the demand is very high, we'll be seeing alot of 4000+ CPUs stamped as only 3000+ models with very high overclocking potential and sold at dirt cheap prices.

I read in your other thread that you were thinking of swapping for a SS chip eventually, I would recommend selling it to someone who does overclock, and you letting them know exactly what the chip is capable of. Your chip will have more resell value if you state its capabilities.


OC-Master
 
If mine doesnt overclock well, its going to ebay once the SS comes out. If mine does well, ill either keep it or sell it now and buy a 3200+ for its 10x multi so I can run 290-300 with ram at 5:4 instead of having to run 3:2 ram and 9x multi. funny thing is I may run out of fsb since id need 333x9 for 3GHz LOL. the 3000+ might actually be limited by mobo or ram before the cpu so it would be wiser to get a 3200+
 
Sounds very similar to what I'm waiting on. I don't feel so bad waiting on the mobo now since it gives time for the CPU's to rotate stock to the SS. That and hopefully the quanities of PCIE 6800 GT's go up by then.

Thats exactly what i'm waiting for. Are we suppose to be seing nf4 ultra boards by the end of this month? And will they be up to par as far as overclocking FSB? I saw somewhere that the first batch of nf4 boards were going to have a weaker nf4 chipset, but nvidia claims they fixed the problem with a new revision of nf4 chipset.
 
Overclocker550 said:
If mine doesnt overclock well, its going to ebay once the SS comes out. If mine does well, ill either keep it or sell it now and buy a 3200+ for its 10x multi so I can run 290-300 with ram at 5:4 instead of having to run 3:2 ram and 9x multi. funny thing is I may run out of fsb since id need 333x9 for 3GHz LOL. the 3000+ might actually be limited by mobo or ram before the cpu so it would be wiser to get a 3200+



Thats what happens to the people who rush ahead and buy things, u just got the cpu and now if its a lemon youre ganan sell it cuss SS just now came out a week later hehe.

Well i hope youre cpu is a good one
 
Back