So according to the posts it is being hypothesized that the FX57 will be fine on a single stage cooler such as the prommie/vapochill but no good on anything colder than that?
Is this correct?
Printable View
So according to the posts it is being hypothesized that the FX57 will be fine on a single stage cooler such as the prommie/vapochill but no good on anything colder than that?
Is this correct?
yes
According to the 2 samples tested so far, that is correct. It remains to be seen if a simple BIOS update or tweak can fix it, but it seems to atleast be a problem. It will be interesting to see if this is just an isolated incident or general throughout the market. I don't know if the chips they got are simply engineering samples, or leaked versions of the final product.Quote:
Originally Posted by BLHealthy4life
But I thought it would not need to be any colder because of the core improvements?? :shrug:
Also I have 4 55's 1 0441, 1 0503, and 2 0435's.. I have reached my best results using one of the week 35's.
FOUR FX55s?! Wow..must be nice.Quote:
Originally Posted by FROG
One is plenty for me.
Would you please be kind enough to briefly explain the method for dating a CPU.
Thanks.
04 41--week
|
|
year
On the heatspreader. The second line. Will say something like 0430 or 0505 etc in the line.
it's year then weekQuote:
Originally Posted by BLHealthy4life
e.g. 0435 is the 35th week of 2004
ya that's a good point, BUT..you know what?
I believe that all the fx55 from year 2005 are really really bad.
Unless you can find some year 2004.:)
My FX-55 it seems to have a "cold bug" on single phase change. More than 215MHz was hopelessly unstable. No matter what I do (Tried 3 MBs, 2 PSUs, 3 different types of RAM), I couldnt stabilize it, that's why my 3Dmark scores are 200HTT. I've lousy mc thought.
However, now on air cooling (XP-120) and DFI LanParty NF4 Ultra-D, I can run 270 2-2-2-6 with 2x512MB BH-5(3.5V), and 333 HT with TCCD(2.9V) T1 (or CPC enabled).
Stepping: 0444 EPMW.
From 2.6GHz to: 3.0GHz(1.525V) - 3.2GHz(1.70V) - 3.3GHz(1.85V) Rock Stable.
I can't afford a FX-57 upgrade :)
I'll stay on my FX-55 until Q4 05 (FX-59 ??)
This certainly does not mean that peeps won't want to run a San Diego at -110C :DQuote:
Originally Posted by FROG
What percentage of the market buys FX CPUs to begin with? I doubt that AMD has no concern about Xtremist (hehe) users and their ability to clock the chip well, after all they demonstrate what is possible and can influence like-minded buyers to buy an FX over other chips. Both San Diego and Venice can be considered Revision E A64's, in Intel's system these would both be "E0 revision" but AMD chose to stick with E3 and E4, the chips themselves share everything but half a meg of cache. There's reason to believe that since the chips are so physically similar, they will both respond similarly to low temps, but there is AMD incentive to be sure the FX-57 has no issues even at the lowest temps.
As for the elusive cold bug - we have zero evidence that it is inherent to the 90nm process, strained silcon or SOI. Intel's 90nm strained silicon chips have not been shown to have any cold-related issues beyond that of the 130nm variants . . .
:).Quote:
Originally Posted by BLHealthy4life
Dont try and compare intel to amd Matt..Quote:
strained silcon or SOI. Intel's 90nm strained silicon chips have not been shown to have any cold-related issues beyond that of the 130nm variants
AMD has the ddr controller on the damm chip, lol..
your way off base with that comment, hehe
Tony - what gives you this impression, and why would this only affect AMD's 90nm processors?Quote:
Originally Posted by bigtoe
I have been of the opinion since Winchester that it was an issue with the thermal diode reporting, as I understood it both Winchester and Venice had changes made to the thermal diode, and you'll notice that issues do not seem to occur until much lower temps with Venice chips which would be in accordance with this theory . . .
Way off base? People are suggesting the issue is inherent to the 90nm process, if so why haven't Intel chips seen any cold related issues at all at 90nm? It would only be logical if the issue were directly related to the feature size . . .Quote:
Originally Posted by sandybeach
See people might be relating the issues to the 90nm process in general, but its the AMD 90nm process not Intels. I can see how alot get draw to the fact that its a 90nm manuf. process thats the casue, but your right thats not true its a specific 90nm process only and that in which AMD is using for there new chips not throughout the industry.
Quote:
Originally Posted by matt9669
Isn't that what he was saying, The memcontroller could very well be the reason it isn't doing it.
AMD is alot more complex now than Intel and the shrinking of the DIE just means they have to crap the same stuff they had before in a smaller space and make it more efficant at the same time. This causes any big change in resistance "which occurs at sub freezing levels" to throw things way out of spec.
Rem alot of the VID steppings the chip goes through in the boot and idle/working process are set VIA Resistance changes through traces and bridges in the chip itself.. Cold temps will casue these to Fluct. bigtime which isnt good.
Yes, it could be the memory controller - interesting, must run to dinner but I hope this debate continues . . .
Frankly I think we should focus on plausible theories which seem fixable regardless of what the issue actually is, which I why I like Tony's theory - give me more :D
yeah, I wish we never saw any cold issues. who would have ever thought we would be walled by temps being too cold, lol
The tooth fairy might be able to fix it... :demon:
I have found local 0508 and 0509, any results on them??
C
From my understanding, the cold bug is mainly caused by the prcessor temperatur diode setup. Now it was changed for the winchester and people started getting the no boot issues at extreme cold. I remember reading somewhere the Venice still has the same setup, while the San Diego has a changed diode setup. That is also the reason why Venice run on Mobos with old Bios, but San Diego requires a new Bios. *Looks in to crystal ball* I predict, the San Diego will be the almighty cold chip and crush all world records... :banana:
I just got a FX-55 from ZipZoomFly. Stepping is CAA2C 0510CPMW
Doesn't overclock worth a :banana::banana::banana::banana:!
Keep in mind I'm using stock cooler, but it won't even do 2800 at 1.6v
Also, my Fatal1ty board either overvolts or the probe is wrong so I don't know if it's 1.6v like I set in bios or 1.64 like it shows in monitoring software.
Maybe I need better cooling to get better overclocks? I have a Zalman AlCu 7000? :mad:
I thought the CAA2C stepping was supposed to OC pretty well. that's a bummer your FX55 isn't on par with the rest of the CAA2Cs out there.Quote:
Originally Posted by Gitsum
Keep us updated.
Thanks for the CPU date explaination guys...greatly appreciated!