HardOCP Overclocks A Athlon64 3200+ 939-Pin 90nm (Winchester)

[jpeyton]

Text

HardOCP's sample hit 2.32GHz at stock Vcore, stock cooling. It was a week 31 chip; not the newest (I've seen some people with week 40+ chips).

Notably, Anandtech's review of the same 90nm chips used a better heatsink and used small Vcore increases (in the case of the 3000+) to reach 2.6GHz.

 

[stelleg151]

Yeah, I think 2.327 isnt bad at all for stock everything. I bet that chip could be brought up to 2.5 at least. Also notice that the winchester @ 2.327 = newcastle @ 2.4. Winchester is sweeet. That means 3000+ @ 2.6 = FX-55.

 

[jpeyton]

I would argue a 3000+ at 2.6GHz would be faster than an FX-55, because of the 90nm optimizations and more importantly because of the increase HTT and Memory Clock speeds needed to reach that overclock on a 9x multiplier.

 

[Chadder007]

Isn't it strange how Intels CPU's when moving down in die size have increased in heat....and AMDs have lowered in heat production?

 

[fkloster]

interesting side note.... my FX-55 gets to desktop @ 2.8 ghz and comes stock with a copper heat pipe cooler in the retail package...

 

[jpeyton]

Originally posted by: fkloster
interesting side note.... my FX-55 gets to desktop @ 2.8 ghz and comes stock with a copper heat pipe cooler in the retail package...

FX-55s are the $100 steaks of the processor world; the finest cut of silicon from the wafer. Does the stock cooler keep it at decent temps; would the cooler be comparable to the quality of something like a Hyper6?

 

[gobucks]

I've seen 2.6GHz benchmarks of the 3000+, and it's pretty much dead even with the FX-55. Winchester is faster clock-for-clock, but remember that the FX has 1MB L2, versus 512KB L2 for winchester.

As for the 90nm transition, Intel had problems because they tried to add too much new stuff to the prescott core. Didn't they add like 50 million transistors more than northwood or something? The problem was that northwood was pretty much at its clockspeed limit because of its pipeline, so they had to completely re-engineer the pipeline to allow faster speeds, which turned out to be a big waste due to other problems. Adding 50 million transistors and fitting them into a smaller area is a pretty big task, especially when trying to pump 3+ GHz frequency signals through them, and I'm not at all surprised that current leakage and heat problems resulted. Had they just made a straight transition to 90nm with northwood, with added cache, performance would have gone up, and while they would not have had the clock-ramping potential, it's not like prescott has it either, so at least they would have a decent processor with decent power usage and low manufacturing costs. After all, AMD underwent pretty much just a simple die shrink, and it went great. Basically, intel forgot the all important rule of Keep It Simple, Stupid.

 

[gobucks]

Oh, yeah, the reason the FX can hit 2.8GHz is because of the use of strained silicon. I imagine this tech will work its way down into the 90nm parts eventually, and will probably buy an extra 200MHz or so.

 

[jpeyton]

Originally posted by: gobucks
Oh, yeah, the reason the FX can hit 2.8GHz is because of the use of strained silicon. I imagine this tech will work its way down into the 90nm parts eventually, and will probably buy an extra 200MHz or so.

I believe the 90nm parts already use strained silicon; right now, only the high end 130nm also use it, but it might make its way down the ladder.

 

[Zero Plasma]

The 90 nm's do use strained siilicon.I just order parts for a new build and i'm using the 3500+ 90nm.
Can't wait for them to get here.:0

 

[XBoxLPU]

http://www.anandtech.com/cpuch...ts/showdoc.aspx?i=2249

The FX-55 uses a type of strained silicon developed with one of AMD's partners, but unfortunately at this point AMD is not releasing much information on their implementation of strained silicon. IBM has been demonstrating strained silicon for years now so it is not too much of a surprise that AMD would have access to this technology for use in their CPUs. Intel first introduced strained silicon to desktop CPUs with their 90nm Prescott chips.


The big announcement will be whenever AMD brings strained silicon technology down to their 90nm chips, since 130nm advancements won't mean much going forward. It does appear that AMD's manufacturing partnerships are definitely paying off though, which has helped them address manufacturing as a serious weakness in years past.

 

[fkloster]

the cooler is very high quality... I did not really study it to much but 160% better than AMD used to use... I will have more benches this weekend once the kids are in bed...

 

[Markfw]

Why do I have such trouble ? I can;t get mine to work good at 2.3 ! at 1.6v !