Printable View
I don't know why... maybe just a gut feeling, but I think the FX57/SanDiego will CERTAINLY have a "cold bug".....
:(
After seeing PCICE results from Venice, which is basically the same as San Diego... I'm sure of it. SO it may work well down into the -40C zone, but probably won't meet or exceed FX55's 3500mHz 3D stable.
So let the arrows fly, but I'm gonna' go pick up another FX55.
C
I agree, I think the FX-57s will have a problem at -90c just like the Venice did.Quote:
Originally Posted by charlie
I'm hunting for a good FX-55 myself:)
OPP
Yeah, finding the gem will be tough. My last FX55 went 3006mHz on AIR at 15C and 3704mHz++ on LN2...hope to find a BETTER one :DQuote:
Originally Posted by OPPAINTER
C
Now I know you guys aren't assuming the Venice cores are "built" the same way as the San Diegos...... ;)
I may consider do an auction of mine :D
This one :
http://pedrorocha.planetaclix.pt/Fx3899.jpg
:D What you think ???? :cool:
What you think should be the base price for the auction ? :D
3,50 and I'll take it !
C'mon Charlie - :DQuote:
Originally Posted by charlie
Venice 3500+, 3800+, etc. = E3 , 512K L2 Cache, .09, SSE3, etc.
Opteron 250, 252, FX-57, etc. = E4 , 1mb L2 Cache, .09, SSE3, etc.
That chip will climb higher in a auction then a classic corvette :cool:Quote:
What you think should be the base price for the auction ?
OMFG :slobber: Now that is a monster. May I ask what's the production date/stepping on it?Quote:
Originally Posted by Pedro Rocha
The cold bug may infact be an increase of dram drive strength as the memory controller gets chilled.
This could be sortable with a board or bios tweak.
as i said in the venice thread i dont see this being a chip issue i really think it is a bug in the bios i DONT see amd doing this to there cpus as they would loose a great deal of customers to this it may not be a huge impact but the last thing AMD wants is people to buy intel over something as stupid as this thats just my 2c tho :/
CAA2C 0441EPMWQuote:
Originally Posted by s7e9h3n
Here is a picture of it on the cascade
http://pedrorocha.planetaclix.pt/DFInaCascata1.jpg
:D
I dont no wot it is but i just love to see pics of such nice setups :)
Thanks. That seriously has got to be the highest screenshot of an fx55 I've ever seen. Also curious what the cpu was able to do on air (I'm not expecting any test done with air though ;) )and at what Vcore. Simply awesome.Quote:
Originally Posted by Pedro Rocha
Sorry for hijacking your thread Charlie.....I promise to stop asking questions after this... :D
What method is used to hunt down a "good" FX55? elephant gun!Quote:
Originally Posted by OPPAINTER
Seriously here in Spain you can legally buy things and return them if you are not satisfied within like 7 days. I have never done this but I am sure some people do it constantly. I am sure shops would get real uptight if you constantly did this.
Not an easy task, even you have a Prometeia or Vapochill - for instance mine is banal CPU on stock (much less than 3ghz benchable), rasonable on silent Promie (around 3.3Ghz benchable with 1.7v) and out of this world at -110ºC :DQuote:
Originally Posted by blueworm
My previous, FX-53 the same - nothing special on Prometeia and still have the WR with 3.600Mhz in Ricky's cascade ..
:stick: You do realize that the amount of people putting their chips under Cascade cooling is less than 100th of 1%. I don't think they are particularly worried if it doesn't work out. It works at -40 just fine and that's colder than 99% of all AMD users will ever have. It would certainly be nice if it worked, but commercial products are not, and never will be guaranteed to work at ~ -90C ever. My 2c though.Quote:
Originally Posted by Fzero
If you have a dfi mobo weaken the drive strength....see if it helps clock higherQuote:
Originally Posted by µnrealneo²
I HOPE that you optimistic guys are right :D
But I won't hold my breath :(
C
Quote:
Originally Posted by charlie
good luck :rolleyes:
But it generates a great deal of PR for AMD none the less, so don't think they don't notice this.Quote:
Originally Posted by Orthogonal
Good point - in general yes, not in xtremesystems users ..in our case the percentage is higger ;) xtremesystem forun members have the higher rate of phase-change users in planet :DQuote:
Originally Posted by Orthogonal
Where is PCIce thread about sub -40ºc bug in Venice ?
Any Sandiego sub-zero testing yet?
Pedro, it's the sticky on this page
http://www.xtremesystems.org/forums/...ad.php?t=57291
He ran it at -40C fine, but it wouldn't even post at -90C :(
Anyway, no news on San Diego sub-zero testing yet, it appears we'll have to wait a little longer, no one has one yet.
suscribed
Its going to be an Issue with the SOI manuf process at 90nm Im sure of it.. That and that the ddr controller is on the chip.
IMO its unfixable
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!
:(
that sucks...
Both the chips were from me and they weren't engineering samples. You'll see the exact same cpu's when they become availble to the public. What you saw is what you'll get. ;)Quote:
Originally Posted by Orthogonal
I would not buy any fx55 and not jump to conclusions to assume fx57 will have the cold bug and even if it did, -40c is still plenty and diego is faster than claw clock for clock too
:D So what will be on single phase below -40°c ??? Many poeple don't use cascades to reach the -100°C range but have phase changers which go down to -50°C / -60°C.
Think we all have to wait till we will se the first results on single phase with the FX-57.
But who will be the first to buy and test it ???
:toast:
0510cpmw that cant do 2.8ghz@1.6v??? thats gotta be the worst fx55 I have ever seen..
do me a fav look at the chips PCB. There is a letter in gold on one of the corners on the TOP is it an "N" or a "T"?
Just curious
Yes, that was a super FX-53 you had then. I "only" done 3553 with FX-53.Quote:
Originally Posted by Pedro Rocha
I agree on the subjekt. God OC on air is not the same as god OC on Cascade.
And visa versa.
The same with volt, some FX can do great OC with low volt, and no more at high volt. Some like volt, and do better and better with high volt, if the Mobo and cooling can handle it.
I have also seen some Pifast close to 33 sec. I'll soon be ready, (and have some time) to race you again. :D
Well im working with a few companies on getting the boards clocking higher while chilled, this was one of the findings.Quote:
Originally Posted by matt9669
Frank, those temps on display aren't real temps ;)Quote:
Originally Posted by HARDCORECLOCKER
Die temps should be "somewhat" higher :D So -40°C temps (loaded) won't be touched by commercial units at all...
Don't give up on women. Sure they don't offer as much bang for the buck but ... (insert rimshot sound effect here)Quote:
Originally Posted by BLHealthy4life
Seriously, great thread started by charlie's gut. What else is it telling you? :D Whats it say about 4x512 1T on the Venice?
dont think thats coming true anytime soon. :)Quote:
What's it say about 4x512 1T on the Venice?
EDIT: chill
-saaya
Maybe I should RMA it for another. It will run at 2.8 doing regular things, but when I stress test it with S&M, it will crash when it gets to the FPU part.Quote:
0510cpmw that cant do 2.8ghz@1.6v??? thats gotta be the worst fx55 I have ever seen..
Yeah even the new cores are great air/water clockers.. Ive seen 0507 weeks do 3ghz on air at 1.6v and 3060mhz on reg water at 1.63v so for you to not hit 2.8ghz Is bad.
Rem Macci's? ;) 2750mhz 1.75V under XP-120 iirc :slobber:Quote:
Originally Posted by sandybeach
lol, yeah second worst thenQuote:
Rem Macci's? 2750mhz 1.75V under XP-120 iirc
I know where you get your money from :D My mother just paid 22,000 DKK for some complex procedure with her teeth. That's enough for a decent PC ;)Quote:
Originally Posted by Tom Holck
Regarding 100th of 1% of people using cascades, a similair amount of people buy FX chips. Your average outback jack will get a whooping 2,66ghz CeleronD with awesome 2*128mb ram and perhaps onboard GFX.
The people who dish out $900 for a FX would like to see it run under their expensive cascades/LN2 - if it cannot, and another chip (e.g. Intel 660 or 3,46) can give better performance under cascade than FX under water, then of course they will get the alternate.
If you are serious into WR, then you need the best chips out there with the best cooling out there. THere is no compromise in this. If the FX-57 is not capable of accomodating -40+ cooling, then the sales will be minimal, and the fault will be on AMD's hands.
I pray that at least San Diego, if not Venice, will work sub zero :)
I agree with you Juliette, if AMD doesnt fix the sub zero temps on their FX-57, it would be a bad move for them. Whenever anyone here breaks the WR, its usually with extreme subzero cooling and most of the record are with AMD FX cpus. Free publicity for them and thus increasing sales.
Correction :)Quote:
Originally Posted by neokenzo
Always :D
Don't you mean -40- :DQuote:
Originally Posted by Juliette
Anyway, It appears that something has obviously changed with the thermal workings of the San Diego, hence the need for a BIOS update. I would be willing to bet they purposely made the San Diegos not have the Cold Bug and the Venices sensetive to it so ppl can't get away with buying a 3000+ and OCing it to well above FX performance at a fraction of the cost.
Yeah I was thinking what to write :D
Considering the "-" sign is before the digits, I found it best to make a "+" sign to indicate the increasing numerals.
PC ICE makes affordable -100C cascades, so a FX-57 that doesnt like to cold is a CPU for a non-existant market.
For real??Quote:
Originally Posted by Juliette
I'll have to contact and see what his prices are.
Way off topic here briefly:
Is it possible to make a dual cascade with the compressors "stacked" in an enclosure with the same length and width deminsions of the prommie? ...of course height would be different....
Thanks
i wouldnt care if it was just 1 person its still a sale most reviewers have cascades or PS units and like to see how far they can go with a chip, what is the point in spending over one thousand bucks on a chip if you cant fully use it and you would be surprized how many people actually us ps i mean you say it is only 100th of 1% but do you have documentation to back that up?Quote:
Originally Posted by Orthogonal
Of course I can't back that up, I pulled it out of my :bananal: , I was just making the point that when a company has to decide on making an overall solid product for the masses, or cater to a niche market at the same time, the masses usually win.Quote:
Originally Posted by Fzero
You think the FX is made for the masses? I beg to differ, I think they cater the niche market. Not everyone do get the FX series and most that do use phase change with it.
I never said the FX is made for the masses. I was talking about their processors in general. They are far more concerned with getting the garden variety processors out with no bugs and solid performance, than tweaking the FX line to have no cold issues.
well like i said befor and many others think to its just gana be something as silliy as a bug in the bios or a couple pins bent on the bottom
Good to know you are back Danish OC Champ.. :toast:Quote:
Originally Posted by Tom Holck
This weekend we a few mobo to test, new graphic card, between Big4 try runs and 3Dmakr runs I promise I will try to do a good PIFast run..
The prob with my chip may be multipliers our 1Giga BH5 max at around 277Mhz 1.5-2-2, using the 13x can only go up to just 3.600Mhz .. :rolleyes:
If my chip will do 3878 PIfast run (multiplier 14x) I promise no to seel it even for $5.000 :D :D
I hope the new DFI bios will help :cool:
Back on topic : After all, in a few months can be a main role for the 0.13 chips in OC comunity .going under cascades and staying there :D
Hey Chilly1, PCIce and Ricky there is a good business oportunity for you guys :D
Any ideia what was the best OC of an 0.09 AMD chip yet??
"I would be willing to bet they purposely made the San Diegos not have the Cold Bug and the Venices sensetive to it so ppl can't get away with buying a 3000+ and OCing it to well above FX performance at a fraction of the cost."
Probably true and I also heard amd implemented some sort of OVP in Venice, the few ive seen were not even overvolted cause it didnt seem to even make a difference, perhaps the cpu ignores any higher volt and still runs internally at 1.4v? This would further the gap between Venice and Diego
Well seems week 0509's are awsome. This guy over at this website has just started posting some results.
http://forums.amd.com/index.php?showtopic=43824
Yea I was just talking to MM. HIs chip is friggin nice! :slobber:
It has a N . Sending it back too.Quote:
0510cpmw that cant do 2.8ghz@1.6v??? thats gotta be the worst fx55 I have ever seen..
do me a fav look at the chips PCB. There is a letter in gold on one of the corners on the TOP is it an "N" or a "T"?
Sorry, I said T before, my bad.
T for terriable :( try to get a refund, if not, ebay it. I would save for a Diego at this point :/
Now where did you get your info about the "OVP" thing? If it's soley based on the 3500+ PCIce tested, then you can't make this conclusion. I have a feeling there may have been a problem with Ice's harwdware. The 3800+ scaled beyond 1.4 and stopped at ~1.65. And, if for some reason, you're right about Venice's OVP, then how do you know that the San Diego doesn't have it as well? ;)Quote:
Originally Posted by Geforce4ti4200
Venices (E3) will not get a reworked temp diode. So far they have been proven to be as resistant to cold as Winchesters.
San Diegos (E4) will, on the other hand, receive a reworked temp diode, suggesting the AMD might solve the cold bug for the Diegos ONLY.
It seems to me that AMD is still catering to the high-end enthusiast segment (however small it is) with their new CPUs. Like some people have said, it's a real money-maker for AMD to spend a little more time on a CPU to get some kick-ass overclocking publicity a little later so people will want the CPU.
Imagine a conversation between two people:
person 1: "OMG WTF! I saw this dude at XS.org get a FX-57 to 4GHz with a three-stage cascade! It was frickin' awesome!"
person 2: "Dude! I am so buying one of those CPUs, and even though I can't possibly imagine running that thing at even close to below ambient temps, the allure of that overclock is simply too much for my feeble mind to resist!"
Ding! Another sale for AMD. Check please.
Ehhh.... I'll try a San Diego just 4 fun, but I'm not expecting BIG things...
C
http://community.allhiphop.com/images/smilies/rofl.gif http://community.allhiphop.com/images/smilies/rofl.gifQuote:
Originally Posted by mcnbns
Quote:
Originally Posted by charlie
If you are in doubt, you can wait and see what others get with their fx57 first, if it looks good enough, you can base your decision on that :)
We will have to see to what extend amd OVP their Venice, I doubt they will do any extent to the diegos. We dont yet know how cold tolerant Diego will be and we havent had a large sample of Venice to test its cold tolerance. It wont look too good for amd unless at least some diegos can handle down to -100c
No, I'd buy one anyways....Quote:
Originally Posted by Geforce4ti4200
I've been an EARLY adopter of FX51, FX53 and FX55... wouldn't want to break my routine ;)
I'll haave one and it'll work well, or it'll be eBay bait....
C
Has anyone hinted to you that perhaps you have too much money :DQuote:
Originally Posted by charlie
Sad but true.Quote:
Originally Posted by mcnbns
Our current Venice test subjects have been running fine at -40C, just not at -90C, no Winchester I ever played with liked anything below 0C so that's a step in the right direction . . .
As for the San Diego, it would make sense that AMD would spend time reworking the temp diode to be sure it plays nice at xtremely low temps, so let's hope this is the case - if I have to plunk down the change for an FX-57 I will :D
same here agree alot of winnys all hated cold or below zero.Quote:
no Winchester I ever played with liked anything below 0C so that's a step in the right direction . . .
not diode its the onchip ddr controller or the process not being able to take the cold temps IMO.. :)Quote:
As for the San Diego, it would make sense that AMD would spend time reworking the temp diode to be sure it plays nice at xtremely low temps, so let's hope this is the case - if I have to plunk down the change for an FX-57 I will
lol... I actually recycle a lot of hardware, so it doesn't cost that much :DQuote:
Originally Posted by Juliette
C
Imo, forget the venices....FX57 is almost here...till then we cant say that it wont be able to run with cascade @ -90....