PDA

View Full Version : Record Breaker


AMDCam
09-19-2005, 07:42 AM
Hey, I just tested my 100% stable computer, after months and months of instability problems, and I broke a record on 3Dmark 03 compared to EVERYONE else who owns my graphics card and any processor in my field (Athlon XP).

So, obviously I'm pretty happy. Has anyone else had an experience like this on 3dmark or benchmark program? What are the specs of yours if you have?

wazzledoozle
09-19-2005, 09:26 AM
Im #2 Athlon XP on PCMark05 at 2.6 GHz, and I had a 9550 when I scored that.

What vcore do you run your AXP at? Also what are your average load/idle temps? (dont give me the numbers from when your blowing chilled air into your case or something)

I have to run 1.8 vcore to get to 2.6 GHz and its pretty hot. Ill run 220x11.5 once I get an SI-97A with 92mm Panaflow :D


I noticed you are running an insane FSB of 255, how did you do that? (which custom bios?)

Velocity
09-19-2005, 02:47 PM
lol nice... thats a crazy oc...

AMDCam
09-19-2005, 03:30 PM
My Vcore is actually only 1.7. The thing is with that, I still haven't done that shortage trick to unlock my multiplier (because I really don't want to take apart everything I've spent so long on getting to work right), so actually what I used on 3DMark03 is 2.4+ghz straight overclock, which is pretty much yours (219mhz but I've had it up to 224 without a problem). I'll go ahead and edit my sig until I actually do the trick, sorry about that. When I first wrote that sig was when I first came here and had a lot of plans to hardmod my stuff. Although what I got for the graphics card is damn close, I can get 453 core and 400 (800) RAM. Again, sorry about that, it really was something I was going to do, and still is, but I'll wait until I actually do it before I brag in the sig.

Even with that OC I really did break that record though.

DanTheBanjoman
09-19-2005, 03:46 PM
So, what is your score?

Solaris17
09-19-2005, 05:04 PM
that pretty sweet :)...and please anyone i seriously whant to oc even more i head something bout custom/hacked bios's can i get one will it work i whant to adjust my vcore(for stability) but i dont whant to ground jmpers please !!! :( :o :cool: :) ;) (i also whant to just oc more like multipliers etc

AMDCam
09-20-2005, 02:28 AM
6748 with a Radeon 9800 Pro and overclocked Athlon XP (both were overclocked). That's my class.

turbopsi
09-20-2005, 09:40 PM
I score 6867 on 3dmark03 @ 420mhz core and 2.4ghz cpu..

I've scored 7k though at 430mhz, and 2.6ghz cpu.

bim27142
09-21-2005, 06:45 AM
does a score of 4k on 3dmark03 seems far away from your OC'ed 9800's? i scored 4k with my card (on my sig)...............

or is this score good enough already for my card?

AMDCam
09-21-2005, 06:49 AM
for a 9550 that seems good. See, I'm just proud because I beat everyone in my class until "now" with turbopsi. I don't know how he did it, but I'll try 3x in a row tests and I'll unlock the multiplier, try it with 2.8ghz (12.5x multiplier, 442mhz fsb), and then see what I get. But turbo, how does that work out? Did you do 3x in a row or something? Or is it the XT flash? I mean I overclocked mine higher, so I would assume we're pretty much even, but how do you do that? Thank you

turbopsi
09-21-2005, 04:04 PM
for a 9550 that seems good. See, I'm just proud because I beat everyone in my class until "now" with turbopsi. I don't know how he did it, but I'll try 3x in a row tests and I'll unlock the multiplier, try it with 2.8ghz (12.5x multiplier, 442mhz fsb), and then see what I get. But turbo, how does that work out? Did you do 3x in a row or something? Or is it the XT flash? I mean I overclocked mine higher, so I would assume we're pretty much even, but how do you do that? Thank you

All I did was run the benchmark at default, with those clocks.

Also, how do you unlock the multiplier? Barton core processors are locked after 0338 unless you have a mobile.

AMDCam
09-22-2005, 01:51 AM
I haven't yet. Someone I know told me a trick to unlock the processor, a hardmod. You short out 2 pins by connecting them and there you go, the clock's unlock. It's on a website and shows you step-by-step.

Dippyskoodlez
09-22-2005, 04:43 AM
that pretty sweet :)...and please anyone i seriously whant to oc even more i head something bout custom/hacked bios's can i get one will it work i whant to adjust my vcore(for stability) but i dont whant to ground jmpers please !!! :( :o :cool: :) ;) (i also whant to just oc more like multipliers etc

You cannot unlock locked bartons. you can enable powernow! but that is useless on Nforce II motherboards, as they are incapable of utilizing this feature.

Mobiles, however, are unlocked..

To use over 13x+ on a mobile you must have the pin mod for setting a base multi of 13x or better, or a mobo (DFI, ABIT, Epox (I believe), all use the extra VID range required to utilize all multiples.)

yeah.. I know... socket a nut. :D

AMDCam
09-22-2005, 06:16 AM
Okay, well it's funny that there is a BARTON walkthrough for 3000+ and 3200+'s, where you short out 2 pins. But I guess you know better than me, and I guess I'm imagining the walkthrough. I've got a lot of e-mails from a guy that unlocked his, and as soon as I find the one with the link to the walkthrough in it, I'll show you.

turbopsi
09-22-2005, 04:53 PM
Yeah, are you talking about the trick where you turn the barton into a mobile? It only works on VIA chipsets though right?

The only bad thing about that trick is you have to change the multiplier with a utility from within windows... each time you reboot, just like a laptop. That is the only way to unlock a superlocked barton cpu.

I forgot the actual bridges you have to connect.

Mobiles are way better :)

AMDCam
09-22-2005, 11:19 PM
Oh wow. Maybe that is the one, although I saw screenshots of the BIOS showing an screen where the multiplier is unlocked. It's been a few months since I saw that walkthrough, but it might be the one you're talking about.

Dippyskoodlez
10-02-2005, 09:30 PM
Oh wow. Maybe that is the one, although I saw screenshots of the BIOS showing an screen where the multiplier is unlocked. It's been a few months since I saw that walkthrough, but it might be the one you're talking about.

the "powernow!" "unlock" is from windows only, and is not technicly an unlock. its enabling a feature, which happens to involve multiplier changing.. ;) its from windows only though.. so its of very little use.

turbopsi
10-02-2005, 11:49 PM
the "powernow!" "unlock" is from windows only, and is not technicly an unlock. its enabling a feature, which happens to involve multiplier changing.. ;) its from windows only though.. so its of very little use.

Not only that, it's only for VIA chipsets... which suck..

AMDCam
10-03-2005, 07:47 AM
Well mine is a hardware thing, and I don't think it's the Powernow one. I'll ask the guy that told me about it to send the link again. I hope it's not the Powernow, I mean I have that on my laptop and it's kind of a sketchy feature to use, especially with a 3rd party program.

DanTheBanjoman
10-03-2005, 09:42 AM
As far as I know newer Bartons can't be unlocked. (3000+) You can however make MP's from them. MP means multiprocessor here, not multiplier.

AMDCam
10-03-2005, 03:39 PM
Woah, awesome. But does that mean like you can hook up 2 of those processors and make them run like a server? Or does that mean AMD's version of hyper-threading? I assume it means the first.

turbopsi
10-03-2005, 03:58 PM
Well mine is a hardware thing, and I don't think it's the Powernow one. I'll ask the guy that told me about it to send the link again. I hope it's not the Powernow, I mean I have that on my laptop and it's kind of a sketchy feature to use, especially with a 3rd party program.

There's no way to do it, besides the way we said above. And it does work, only with a certain Via chipset, and only by using powernow.

EDIT: And making it into an MP processor only allows you to use them in dual cpu mobos.

DanTheBanjoman
10-03-2005, 04:15 PM
Woah, awesome. But does that mean like you can hook up 2 of those processors and make them run like a server? Or does that mean AMD's version of hyper-threading? I assume it means the first.

It means you will need an MP or MPX board, they're dual socket A boards. Then again those chipsets are pretty crappy. They're one of the reasons the Opterons have/had (in case anyone plans to start a discussion about their share) problems taking market share, they gave AMD a bad name in the server/workstation market.

AMDCam
10-04-2005, 03:47 PM
Okay, for right now I'll believe you until I find my friend's e-mail he sent. I mean I even remember screenshots for it. But you know guys, other people would give up trying from what you say, I mean you make yourself sound so sure when you say "there's no way to do it". Like I said, I believe you for now but if I really do find the walkthrough that guy sent me, and it IS a multiplier unlock, that's not the greatest thing to say.

DanTheBanjoman
10-04-2005, 04:09 PM
If everybody gave up right away such nice mods wouldn't exist. It can be good to be stubborn.

Anarion
10-04-2005, 05:14 PM
It means you will need an MP or MPX board, they're dual socket A boards. Then again those chipsets are pretty crappy. They're one of the reasons the Opterons have/had (in case anyone plans to start a discussion about their share) problems taking market share, they gave AMD a bad name in the server/workstation market.


maybe, but they were price breaker...

turbopsi
10-04-2005, 06:21 PM
Okay, for right now I'll believe you until I find my friend's e-mail he sent. I mean I even remember screenshots for it. But you know guys, other people would give up trying from what you say, I mean you make yourself sound so sure when you say "there's no way to do it". Like I said, I believe you for now but if I really do find the walkthrough that guy sent me, and it IS a multiplier unlock, that's not the greatest thing to say.

Funny how all the other overclockers on the web can't seem to do it. I've did countless hours of research on this subject.

I'd like to see you unlock it, I'll bet 5bucks you can't. It's superlocked.

I'd also like to see you hit 2.8ghz with that 3200+.

AMDCam
10-04-2005, 11:04 PM
Dang man, I guess a guy can't dream on this site anymore. It's always a possibility you can hit 2.8ghz with the unlocked clock. Plus, if you weren't all down about it, I'd take you seriously with your research, I mean if you've done it then I should believe you, which I do, but no need to be depressing about it. What if I get phase-change cooling in there or something? Everything is possible, no need to be a downer.

turbopsi
10-05-2005, 12:28 AM
Not trying to be a downer man, just trying to be realistic. BTW, I think you mean unlocked multiplier, not clock.

I have seen people hit over 3ghz with chilled air or phase, you don't even need to unlock it for that.

AMDCam
10-05-2005, 01:54 AM
Well yeah I do mean multiplier, clock is just easier to say. Oh yeah, does anyone know if it might be a good idea to REFRIDGERATE your computer? Not freeze, because condensation would happen from being frozen, and I couldn't take it out once it was in there (or else it would turn from ice to water everywhere around the components), and the fans would stop running. But refridgerating at a considerably warmer temperature, none of those drawbacks would be there. The only downside I can think of is- it's not pretty (you can't see the case in the fridge) and the wires would hang down off the door since I couldn't really make holes. But I'm also thinking, since a computer's speed fluctuates, hopefully the fridge could keep up cooling it, but I don't think the fridge would get hurt, right? Just it wouldn't be as cold as a normal fridge because the computer would suck it out in pretty quick. So practical idea? Try not to be depressing, thank you

Dippyskoodlez
10-05-2005, 02:16 AM
Dang man, I guess a guy can't dream on this site anymore. It's always a possibility you can hit 2.8ghz with the unlocked clock. Plus, if you weren't all down about it, I'd take you seriously with your research, I mean if you've done it then I should believe you, which I do, but no need to be depressing about it. What if I get phase-change cooling in there or something? Everything is possible, no need to be a downer.

theres a big difference between dreaming, and spreading rumors about a way to do something that has been proved by many many EXTREMELY knowledgable, PROVED, to be impossible. (I have done 3ghz on a mobile(air cooling)... im no noob to this stuff..

Its the same with Athlon 64's. Even moreso impossible, actually. If you would like a detailed explination, i would be glad to PM it to you. I know a lot about this subject.. several school projects on the stuff... ;)

Well yeah I do mean multiplier, clock is just easier to say. Oh yeah, does anyone know if it might be a good idea to REFRIDGERATE your computer? Not freeze, because condensation would happen from being frozen, and I couldn't take it out once it was in there (or else it would turn from ice to water everywhere around the components), and the fans would stop running. But refridgerating at a considerably warmer temperature, none of those drawbacks would be there. The only downside I can think of is- it's not pretty (you can't see the case in the fridge) and the wires would hang down off the door since I couldn't really make holes. But I'm also thinking, since a computer's speed fluctuates, hopefully the fridge could keep up cooling it, but I don't think the fridge would get hurt, right? Just it wouldn't be as cold as a normal fridge because the computer would suck it out in pretty quick. So practical idea? Try not to be depressing, thank you


Using a fridge (i have done it) is a very very bad way to do things, for several reasons- fridges arent made to handle a 1000w/HR load, rather a 1000w load over 12+ hours.

A computer continiously created heat, unlike pop, which sits and gets colder constantly.

The condensor of the fridge, which is in the back, and is designed to be passive. Using a large fan, and most likely you would need to use a special, much larger condensor, it could very well be done, but if your going to do that, might aswell just make the whole phase system, and save yourself some peace of mine along with betting much colder temperatures.

Along with condensation from opening the door... if you store food and stuff ;)

AMDCam
10-05-2005, 02:26 AM
Well okay, so a big fan blowing into the condensor will work? I'm just saying, $700-$1000 for phase-change is not in my pricerange, but at wal-mart I can get a $100 fridge and mess with it. How could I find a bigger condensor?

And yes, I would love a private message to see how you ran an air-cooled Mobile at 3ghz, and what I could do with an Athlon 64 when I finally got one. And your use of words (spreading rumors) is definitely not what I was doing, I was just saying that I knew of a guy who sent me a link to a way to unlock a Barton XP multiplier. Until I find it I'll keep my mouth shut, but I'm not spreading rumors.

Velocity
10-05-2005, 03:46 AM
a fridge definatly is not the way to go lol... the main reason is that it will eventualy condense water on the parts. (which is bad) and rack up your powerbill enough to make phasechange a whorthwile purchase.

but if you have the money for it a (and dont care about buying a new mobo/cpu you could try getting a custom made cpu block that lets you put liquid nitro in it (now this is an extremety that not many people can afford (i have done it before with a p4 and got it to 4.6ghz before it crashed) but its spendy and dont plan on using the parts again...

Also you could have a custom hsf made out of copper ( i custom made one for a p3... its not that hard nor expensive)

wazzledoozle
10-05-2005, 04:37 AM
I call bullshit on the 3 GHz air cooled mobile. Thats a record overclock overall, let alone on air.

Start thinking LN2/"Prometeia" (what most people seem to use around 3 GHz)
http://www.vr-zone.com/guides/AMD/Barton/

Prometeia info-
http://www.tomshardware.com/howto/20021230/chip_con-01.html

Dippyskoodlez
10-05-2005, 12:15 PM
I call bullshit on the 3 GHz air cooled mobile. Thats a record overclock overall, let alone on air.

Start thinking LN2/"Prometeia" (what most people seem to use around 3 GHz)
http://www.vr-zone.com/guides/AMD/Barton/

Prometeia info-
http://www.tomshardware.com/howto/20021230/chip_con-01.html

Ive got a cpu-z for 2923mhz, and a bios of 3ghz.. ;) back when 1.27 was latest, it was verified :)

I hold #2, but havent sent it in... 3016 is #1.. (#2 is now 2923mhz )

AMDCam
10-09-2005, 06:34 PM
dippy, why don't you send in the record you got?

wazzledoozle
10-10-2005, 12:01 AM
Ive got a cpu-z for 2923mhz, and a bios of 3ghz.. ;) back when 1.27 was latest, it was verified :)

I hold #2, but havent sent it in... 3016 is #1.. (#2 is now 2923mhz )
Your sig shows Athlon 64...

Dippyskoodlez
10-10-2005, 12:02 AM
dippy, why don't you send in the record you got?

atually, I had never thought of that.. I havent stumbled upon the page that had the records(again), but the page was in japanese, and i barely speak french/spanish.. not japanese.. :laugh:

if anyone knows of the e-mail address, id gladly send it in :)

The record for extreme cooling of an Athlon XP is somewhere around 3616mhz. Memesama got that using I think a desktop 3200+ barton.

3016mhz I know is the air cooled record.. but most definatly not extreme cooling.

Your sig shows Athlon 64...

I did this back.. around december/january when I was using an Abit NF7-S, and my trusty 2600+M.

0352 RPMW I think it was. There have been numerous records set by this exact stepping.. most notable being 1M superpi by S*A*E over at dfi-street at around 3.2ghz I think. Phase change cooling.

Dippyskoodlez
10-10-2005, 12:18 AM
Well okay, so a big fan blowing into the condensor will work? I'm just saying, $700-$1000 for phase-change is not in my pricerange, but at wal-mart I can get a $100 fridge and mess with it. How could I find a bigger condensor?

And yes, I would love a private message to see how you ran an air-cooled Mobile at 3ghz, and what I could do with an Athlon 64 when I finally got one. And your use of words (spreading rumors) is definitely not what I was doing, I was just saying that I knew of a guy who sent me a link to a way to unlock a Barton XP multiplier. Until I find it I'll keep my mouth shut, but I'm not spreading rumors.

How I did it, was....

I was sitting in my room one cold wintery night.. and happen to look at the temperature.. -7F... im sitting there thinking.. damn... I wish I had a prommie or something so I could have temps like that with thus cpu that was doing 2700mhz inside using my SLK-947U+tornado combo.

Then I had the bright idea to take the sucker outside :D

We had almost 3 feet of snow.. but I took it out to the garage, carrying everything myself at 10:00 at night, and standing outside for an hour and a half freezing my arse off trying to get this sucker to do well.. booted at 2700mhz easy, 2800.... then 2900... tried to keep going, but 3ghz wouldnt boot either because of temps being to high (which I think it was), or my antec 350w not being able to handle the load. I did manage to grab the camera and snap a few shots of the monitor displaying 3ghz. 2923 was best I could get windows at.. and my onboard sensor was saying my cpu was at around 43C.. at 1.9v :eek: I ended up trying 1.9 all the way to 2.3v and yielded no success, but the thing im most proud of, is it doing 2800mhz at 1.85v :respect: Once I get a phase change unit finished, I plan on getting that CPU back and building a socket A system, just for the heck of it.. :)


As for your condensor problem, you must first understand how phase change works. It compresses a *gas* to raise and lower the boiling points, and thus making one end freezing cold while the other gets hot.

The problem with just "finding a bigger condensor" is that you would have to vaccuum the gasses (most likely R134a in a fridge) out, cut off the already built in condensor, and then braze in the new one, and leak testing the system.

After that, you will have to deal with cooling the condensor, which is simple enough, but also the compressor size.

If its a small minifridge like the one I was using it will probably be something around 1/8hp, which is about right for R134a and the tamer gasses.

But its not nearly cold enough to justify the hassle of brazing a new condensor into the unit and regassing it with a system that will most likely not even yield sub zero temperatures while loaded with an athlon 64 system, letalone a preshott.

This brings up the two much more rewarding solutions- a water chiller and a standard phase change.

Standard phase change can be had for ~$600 from chilly1 over at XS and will perform extremely well, and just leave vapochills in the dust :)

The second option, which would be much more justifiable in buying and/or finding the stuff necessary to do a condensor swap, would be a chiller. Its just making a copper coil in place of the evaporator, and sticking it in a resevoir for water cooling. This way, 40C liquid can be achieved (requires a hefty mix of antifreeze/denatured alcohol to keep h2o from freezing up tho, but at those temperatures, it doesnt matter as much).

The work needed to convert a fridge is just overhwhelming compared to the alternative solutions, and rewards from those. :)

The temperature inside my minifridge while running my AXP at ~2ghz passivly cooled, at 1.3v was around 13C. definatly not very cold.

AMDCam
10-10-2005, 12:59 AM
Well damn, that explained a lot, seriously. It's a wonder how colder environmental air (like 10F in a bedroom) can affect overclocking so much. But okay, for christmas I might get one of these things. Do you mean -40C water? That is crazy, crazy, crazy, plus it's simple and fixable because I probably couldn't repair a phase-change system. How much would a water chiller cost? Thanks man, you may have catapulted my Barton 3200+ into a new world of speed (3.6ghz?!?!?! UNBELIEVABLE!!!!!!). I was wondering about that though, with water and phase-change you can add connectors to the GPU, RAM, GPU RAM, North Bridge and South Bridge too right (although I have no idea why I've never seen any record-breaking extreme overclocked GPU's)? Plus, according to 3dmark 03 (hardware analyzations) there's a ceiling of 3ghz on the 3200+, but I don't know why it would have one. And did he have an unlocked multiplier, because I don't have one. Thanks dude.

And before I do that, do you guys know of portable air conditioners? I hear wal-mart sells them in the summer, but I can't find them anywhere. That might help performance if I just let it blow into my case. Thanks guys.

wtf8269
10-10-2005, 02:29 AM
(although I have no idea why I've never seen any record-breaking extreme overclocked GPU's)
Ever hear of a guy named ViperJohn?

Dippyskoodlez
10-10-2005, 04:10 AM
Well damn, that explained a lot, seriously. It's a wonder how colder environmental air (like 10F in a bedroom) can affect overclocking so much. But okay, for christmas I might get one of these things. Do you mean -40C water? That is crazy, crazy, crazy, plus it's simple and fixable because I probably couldn't repair a phase-change system. How much would a water chiller cost? Thanks man, you may have catapulted my Barton 3200+ into a new world of speed (3.6ghz?!?!?! UNBELIEVABLE!!!!!!). I was wondering about that though, with water and phase-change you can add connectors to the GPU, RAM, GPU RAM, North Bridge and South Bridge too right (although I have no idea why I've never seen any record-breaking extreme overclocked GPU's)? Plus, according to 3dmark 03 (hardware analyzations) there's a ceiling of 3ghz on the 3200+, but I don't know why it would have one. And did he have an unlocked multiplier, because I don't have one. Thanks dude.

And before I do that, do you guys know of portable air conditioners? I hear wal-mart sells them in the summer, but I can't find them anywhere. That might help performance if I just let it blow into my case. Thanks guys.

3.6ghz on an extreme cooled AXP (ln2 it was) takes not only a lot of skill, but also a very good CPU to begin with.

My 2600+M would do 2.8ghz suicide easy on air, which results in high operating temps overall. As long as that CPU follows the fact that the laws of physics says, and how most CPUs responded to cold at that time, I'm very confident a cascade would throw that thing above the record.

However, do not expect 3.6ghz, or even 3ghz using a phase change, unless your doing atleast 2.5 already with a fairly moderate voltage as it is.

Cold helps electricity flow better :)

That "3ghz ceiling" on 3dmark is just a default thing said by all programs.. its not any indication of any sort. Theres no way you can actually limit a clockspeed when it is independently controlled, aside from its physical capabilitys. :) This is why overclocking is so widespread.

Yes, possibly -40C water, properly tuned, etc, but a water chiller is just a phase change system dunked into a resevoir of water instead of directly onto the cpu.

Either way, its possible to cool everything, just with direct die phase change, a single head performs best, and dual is possible. More is probably possible, but im betting it would end up very ineffective.

If you want information on buying a phase change, or waterchiller, look for Chilly1 over at xtremesystems.org. He is a god. ;) $500+$70 shipping+ controller price for a regular ohase change. I'm not a master of phase building yet, just mastering the concept still, but I dunno what the controller costs.. ;)

AMDCam
10-10-2005, 07:11 AM
I am doing around 2.5ghz (2420mhz stable, about 36 mhz higher stable but I don't usually do that, and about 2.5ghz max), so I think it might not be a problem. But I would think from ambient temperatures of about 65F in an average room, to about 40F A/C and an overclock that's pretty nice with just air, to a full-powered, constant -40C temperature powered from a machine, you would think that about 100F change would do plenty to overclock, but really if I can't hit 3ghz with a 3200+ with a phase change, and I can hit 2.5ghz with just air, how does that work? I would think I could get like 4ghz out of phase change, wierd. ViperJohn is awesome though, but he isn't doing unbelievable overclocks, I mean yeah a couple 100mhz higher than an average overclock is great, but people push processors to 4ghz from 2ghz, or 3.2ghz to 7.4ghz, so why isn't anyone doing that with a graphics card? I mean think, a 9800 Pro that can outperform an X800 series would be AWESOME. or a 7800gtx that doubles performance with it's clockrates (and has plenty of potential with the 24 pipes and large bus bandwidths). That's what I wanna see, or preferably do.

Dippyskoodlez
10-10-2005, 07:23 PM
I am doing around 2.5ghz (2420mhz stable, about 36 mhz higher stable but I don't usually do that, and about 2.5ghz max), so I think it might not be a problem. But I would think from ambient temperatures of about 65F in an average room, to about 40F A/C and an overclock that's pretty nice with just air, to a full-powered, constant -40C temperature powered from a machine, you would think that about 100F change would do plenty to overclock, but really if I can't hit 3ghz with a 3200+ with a phase change, and I can hit 2.5ghz with just air, how does that work? I would think I could get like 4ghz out of phase change, wierd. ViperJohn is awesome though, but he isn't doing unbelievable overclocks, I mean yeah a couple 100mhz higher than an average overclock is great, but people push processors to 4ghz from 2ghz, or 3.2ghz to 7.4ghz, so why isn't anyone doing that with a graphics card? I mean think, a 9800 Pro that can outperform an X800 series would be AWESOME. or a 7800gtx that doubles performance with it's clockrates (and has plenty of potential with the 24 pipes and large bus bandwidths). That's what I wanna see, or preferably do.

It doesnt work, because the CPU's are at the edge of its limitations, and it does not scale linearly.

You may be doing an easy 2.5ghz, but at what voltage? Temperature greatly helps the overclocking, but it will not scale consistantly with voltage. .15 of a voltage bump from a stock mobile may gain 500mhz when going from 2000 to 2500, but from 1.65 to 1.8 wont net you 500mhz again
;)

if your doing 2.5ghz at ~1.7v, Id say ~3ghz tops using a good phase change system. :)

In doing the math for power usage when it begins to hit those speeds, it tends to make much larger jumps as you go faster. it may take .1v for 250mhz at some point, then .1more for 150mhz, etc, until you hit ~2.3ghz where the CPU just cannot handle anymore and the voltages causes it to fry.

turbopsi
10-10-2005, 09:34 PM
It doesnt work, because the CPU's are at the edge of its limitations, and it does not scale linearly.

You may be doing an easy 2.5ghz, but at what voltage? Temperature greatly helps the overclocking, but it will not scale consistantly with voltage. .15 of a voltage bump from a stock mobile may gain 500mhz when going from 2000 to 2500, but from 1.65 to 1.8 wont net you 500mhz again
;)

if your doing 2.5ghz at ~1.7v, Id say ~3ghz tops using a good phase change system. :)

In doing the math for power usage when it begins to hit those speeds, it tends to make much larger jumps as you go faster. it may take .1v for 250mhz at some point, then .1more for 150mhz, etc, until you hit ~2.3ghz where the CPU just cannot handle anymore and the voltages causes it to fry.

I dunno bro, it depends on the cpu. I've seen 3.1ghz on an XP cpu on just chilled air(inside of a freezer).

3ghz is very possible, even 3.5 or so on phase, I doubt it would go any further though.

AMDCam
10-10-2005, 11:57 PM
Turbo, in a freezer?!?! I was just asking about that! Awesome, freakin awesome. What kind of freezer though? I'd say it's definitely possible because with about every degree change in my room I get about 1mhz out of the fsb (with is 2 with DDR and 11 CPU with the 11x multiplier), so a freezer dropping it about 20F lower I would think might get me at least 2.6ghz, maybe even 2.7ghz. But what about the graphics card thing? Could I hook up some caps to the gpu core and ram? Dang, I'm thinking that either some freezer idea or chilled water I'm gonna go with (if it's cheaper than phase-change but still gives a great overclock), I mean according to UAMD the processor would be about a 650 or more mhz fsb and if it was a model it would be about a 7500+ with 3.6ghz lol. I mean obviously, it's far from practical but it's awesome to think about. And so far, I've had great luck with my 3200+, I mean according to a lot of people that have the 3200+, I've got one from a good batch, so above 3ghz might be possible with phase-change, especially if someone hit 3.1ghz on air (I know it's chilled, but still, the air isn't making direct contact on the core, and a normal freezer doesn't hit -40C, unless it's one of those subzero freezers.

Dippyskoodlez
10-11-2005, 12:28 AM
I dunno bro, it depends on the cpu. I've seen 3.1ghz on an XP cpu on just chilled air(inside of a freezer).

3ghz is very possible, even 3.5 or so on phase, I doubt it would go any further though.

your assuming what voltage? -40C wont magically make your cpu a god.

Id say 2.5ghz at around 1.65Vcore would net a 3ghz phase changed chip...

ofcourse, im talking a 2.0vcore stable 3ghz, not a useless suicide run to get 3ghz...

Ive seen how the amperage goes from 30 to 60+ on an axp and 3ghz is one of the major walls, and 2800mhz is too. the scaling of the archirtecture just falls apart at 2.5ghz and beyond, thats why it was so hard to breach it.

turbopsi
10-11-2005, 03:56 AM
Turbo, in a freezer?!?!

Yeah, it was on another forum, but it's not a practical thing to do. You will burn up your freezer doing it for prolonged amounts of time. Also, condensation will probably occur.

Dippyskoodlez
10-11-2005, 03:20 PM
Yeah, it was on another forum, but it's not a practical thing to do. You will burn up your freezer doing it for prolonged amounts of time. Also, condensation will probably occur.

yeah, exactly what I explained earlier too ;)

AMDCam
10-11-2005, 03:47 PM
I Know dippy, I know what you explained but I just never heard of anyone trying it in a freezer. I've got 2.5ghz with a 1.7v vcore, and what do you mean scaling of the architecture falls apart? Well I think I'm gonna go try some water, it's cheaper and easily upgradable and repairable compared to Phase-change. I mean even at 3ghz, an 800mhz increase is nothing to be ashamed about, I mean that's about a 30% speed increase, and at the possible 3.6ghz that's about a 60% increase, which is awesome because the 3200+ doesn't perform bad in the first place, plus with how much AMD's perform better with lower clockrates it should increase performance even more. I mean 300mhz, about 10% increase, is cool (what I got), but it'd be sweet to run an old XP quicker than any new processor around.

turbopsi
10-11-2005, 06:04 PM
I Know dippy, I know what you explained but I just never heard of anyone trying it in a freezer. I've got 2.5ghz with a 1.7v vcore, and what do you mean scaling of the architecture falls apart? Well I think I'm gonna go try some water, it's cheaper and easily upgradable and repairable compared to Phase-change. I mean even at 3ghz, an 800mhz increase is nothing to be ashamed about, I mean that's about a 30% speed increase, and at the possible 3.6ghz that's about a 60% increase, which is awesome because the 3200+ doesn't perform bad in the first place, plus with how much AMD's perform better with lower clockrates it should increase performance even more. I mean 300mhz, about 10% increase, is cool (what I got), but it'd be sweet to run an old XP quicker than any new processor around.

You aren't going to get 3ghz on water, unless maybe you try direct-die waterblocks. I'd say 2.7-2.8 might be possible though.

Dippyskoodlez
10-11-2005, 07:32 PM
I Know dippy, I know what you explained but I just never heard of anyone trying it in a freezer. I've got 2.5ghz with a 1.7v vcore, and what do you mean scaling of the architecture falls apart? Well I think I'm gonna go try some water, it's cheaper and easily upgradable and repairable compared to Phase-change. I mean even at 3ghz, an 800mhz increase is nothing to be ashamed about, I mean that's about a 30% speed increase, and at the possible 3.6ghz that's about a 60% increase, which is awesome because the 3200+ doesn't perform bad in the first place, plus with how much AMD's perform better with lower clockrates it should increase performance even more. I mean 300mhz, about 10% increase, is cool (what I got), but it'd be sweet to run an old XP quicker than any new processor around.

even at 3ghz, it would only take a 2.5ghz A64 to easily keep up with it.

Not only does the physical scaling of the CPU start to just drop off the face of the earth, and quickly, but so does performance.
http://img.techpowerup.org/051011/sca912.ing

It creates more watts than it uses as it starts t get faster and faster.

(the graph is messed up.. stuck with appleworks at the moment and its worse than ms paint.)

GoLLuM4444
10-11-2005, 07:41 PM
It creates more watts than it uses as it starts t get faster and faster.
You're saying it creates energy? :eek:

wazzledoozle
10-11-2005, 10:57 PM
your assuming what voltage? -40C wont magically make your cpu a god.

Id say 2.5ghz at around 1.65Vcore would net a 3ghz phase changed chip...

ofcourse, im talking a 2.0vcore stable 3ghz, not a useless suicide run to get 3ghz...

Ive seen how the amperage goes from 30 to 60+ on an axp and 3ghz is one of the major walls, and 2800mhz is too. the scaling of the archirtecture just falls apart at 2.5ghz and beyond, thats why it was so hard to breach it.
Exactly.

You will NOT get to 3.6 GHz
Oh cool, you can press ctrl+b for bold :pimp:

Dippyskoodlez
10-11-2005, 11:53 PM
You're saying it creates energy? :eek:

err it releases more than it uses compared to what it was at a lower setting... if the clockspeed increases 15%, it may start releasing 17% or more, more heat, rather than a straight 15/15% scaling like a lot of people assume.
I dont think the link works... my bad... try to look at the link if you can.. i cant fix it with the pc im on atm..

(it was supposed to say heat after watts)

AMDCam
10-12-2005, 01:13 AM
Okay I'm not even planning to get to 3.6ghz and probably not even a cooling upgrade, but uhhh how did the guy that got 3.6ghz do it then? Very annoying that you say I won't if I'm in the same scenario as him, but whatever. I've decided that 2.5ghz is fine, especially if performance actually starts decreasing at higher clocks.

Dippyskoodlez
10-12-2005, 01:51 AM
Okay I'm not even planning to get to 3.6ghz and probably not even a cooling upgrade, but uhhh how did the guy that got 3.6ghz do it then? Very annoying that you say I won't if I'm in the same scenario as him, but whatever. I've decided that 2.5ghz is fine, especially if performance actually starts decreasing at higher clocks.

you think your in the same scenario, but in reality, you dont understand the scenario.

Memesama is a japanese (im pretty sure) overclocker that works for one of the many websites, or maybe just "team japan" which is infamous for setting and holding numerous overclocking record, ranging from 3dmark, super pi to suicide runs.

That 3.6ghz was using a hand picked, im sure, CPU, LIQUID NITROGEN which is ~-150C or colder at times...

Along with the fact that the amount of experience they have torturing numerous types of hardware probably even dwarfs the amount of experience anything you have ever done :eek: Pretty much everyone else tho too :p

They are extremely skilled and talented overclockers, with a fat wallet to go with ;)

www.xtremesystems.org

In the forums, look into LN2 overclock journals and cascade adventures, especially stuff with athlon XPs if you want some realistic results.

LN2 is extremely potential, especially in the hands of a skilled overclocker. :pimp:

I'm workong on an article about covering a lot of this stuff, so heres a copy and paste quote of a preview of the cooling section. Hope this helps others reading aswell.
Cooling
Cooling is definatly one of the most vital components of a good overclock. As power draw increases, to put it bluntly, stuff gets HOT! Fail to keep the components within the operating ranges, will result in either damage, or death of that component very quickly, especially at high voltages. As a general rule, I take the manufacturers specified thermal ceiling, and for every increase in voltage, lower the ceiling. I've concluded this by simply observing a few basic facts:

Mobile Athlon XP's are rated 100C maximum
Desktop Athlon XP's are rated 80C maximum
Mobile and desktop Athlon XP's are as far as we know, Identical, just speed binned for better speeds.

Increasing the speed increases heat output, but when you increase the voltage, it amplifys the heat output by much more.
http://img101.imageshack.us/img101/3930/vcorescaling3qz.jpg
A rough example of how increasing the voltage will effect the heat output/power usage on an Athlon XP. As you can see, mhz itself doesn't have as much of an impact on power consumption, although, large jumps will increase the angle of the line by a significant amount.

Types of cooling

None - Not a good option for almost anything out- especially overclocking!

Passive heatsink - More common with older systems, such as pentium's, pentium II's and pentium III's, as they generally can handle higher temperatures and still run fine, along with the fact that their thermal envelope is very very low (in range with most low power laptops today), so a large modern heatsink, such as an XP-120 would keep it cool enough to not even need a fan.

Active air cooling - Probably the most common form of cooling today, relys on fans to amplify the amount of airflow over the heatsink and cools much more effeciently, but can be extremely loud, and has its physical limits, such as size of the heatsink, maximum airflow through the fins on the heatsink, and the ambient air temperatures.

Water cooling - A basic loop of water, running over heatsinks designed for this purpose, known as waterblocks. Much like active air cooling, but instead of air, water is used. Temperatures using this method are generally lower than most air cooling, and a good water setup should be able to outperform the highest end air cooling setups, and quieter, to boot! The main components are the waterblocks(heatsinks), the tubing, pump, and radiator to dissipate the excess heat into the air.

Thermoelectric cooling (TEC, aka peltiers) - refers to a "heat pump" type layer that is sandwiched between the cpu and another cooler (air, water or another TEC). generally needs double the wattage of the heating element - in this case, the cpu. Has a hot side and cold side depending on polarity (cold side on the CPU!) and end up increasing the internal temperature of the case substantially, unless the tec is cooled externally. The advantage is it's noiseless. The cold side is placed against the cpu, and warm side is cooled by another means - water or better for a good setup - since it puts out it's own heat, plus heat from the cpu. (~Beomagi) While generally safer and easier to utilize rather than building your own phase change, it itn't without its dangers! There are many horror stories of people leaving peltiers on, or their water pump dieing, and the peltier then burning up not only the waterblock, but the CPU and motherboard aswell!

Water chillers - Using the evaporator of a phase change system to cool a resevoir of water. This setup is simpler and safer to make for beginners, as it can be done as simply as taking apart a window air conditioner, slightly bending the evaporator, and buillding a resevoir around it. turn it on, (make sure you take the fan out!!) and within inutes, you have chilled water! General expected water temperatures: -15C

Phase change - Phase change is similar to water, as it is a chemical in a closed loop, flowing through a block, but a little more extreme. Instead of water flowing over the block, it uses a gas at a high pressure to create a liquid. As the liquid flows into the CPU or GPU evaporator, it evaporates and in doing so, basicly "sucks" the heat out of whatever it can, to gain the needed energy to turn back into a gas at the pressure. The evaporator line is insulated as much as possible to keep it the coldest where it will meet the CPU. as the gas returns, it created a high pressure, but this time as a gas form, and needs to be cooled. The condensor cools this part of the cycle. The compressor controls the pressures. General expected evaporator temperatures: -40C

Cascades - A cascade, is basicly a dual phase change system. It uses a what you could call "normal" phase change system to super cool the condensor of the second stage. This allows the use of much colder gasses in the second stage, that are unusable in a normal single stage system. For a 3 stage system, it is just repeated once more. This type of cooling isn't suited for 24/7 usage, as running 2 or 3 compressors uses an extremely large amount of electricity and is primarily for long benching runs. Temperatures of a 2/3 stage system can approach and exceed -110C.

Dry ice/liquid nitrogen - Unlike the other methods of cooling this is the least permanent. Using dry ice or liquid Nitrogen requires a "container". It is an open ended tube with copper at the end, which is on the CPU or GPU. Using dry ice, acetone is added to the container to increase surface area and lower overall temps and increase capacity. Liquid nitrogen is a liquid, so no additives are necessary. Both methods sublimate/evaporate very quickly, so refilling often is necessary. General temperatures are -60/-160C respectively.

Ultimately, your cooling choice is up to you. Look at what your goal is with overclocking, and pick whichever best suits you(Along with your budget!). For a light overclock, and learning, its best to start with an upgraded heatsink. For the more experienced users, water cooling is a very good option, and for those looking for excellent operating temperatures, phase change is a good way to go, BUT, those wonderful temperatures come at a price!

http://img328.imageshack.us/img328/5283/prosandconssmall9rv.jpg (http://img328.imageshack.us/img328/7290/prosandcons1ui.jpg)

AMDCam
10-12-2005, 03:02 AM
WELL THANK YOU! You could have explained to me the reason you said "YOU WON'T HIT 3.6ghz" before I got pissed because I thought you were insulting my skill, plus the insult wasn't backed up by anything (which really pisses me off), but obviously you did back it up and thank you very very much. With this chart I can pick what kind I'm looking for. I'd probably run water because it's upgradable to "chilled water" and other liquids, plus it's quiet (my 92mm tornado is killing me when I run benchmarks).