3 ghz AthlonXP/Barton?

[dejacky]

Currently, I'm running my AthlonXP 1700+ (133mhz bus & 1.47ghz) at 2.1Ghz 200mhz FSB using 1.775v. I'm just curious if anyone know of any specific 32bit amd chips that could run at 3.0ghz on my Shuttle XN45G nforce 2 computer. Thanks in advance.

-dejacky

 

[Megatomic]

Even with LN2 you will not be able to get 3GHz out of an AMD chip at this time.

 

[dejacky]

LN2? How close could I overclock to 3ghz? mayb 2.8 on air cooling?

-dejacky

 

[Megatomic]

LN2 is the abbreviation used for liquid nitrogen. I think the best mere mortals can hope to attain on air cooling with an AMD chip is just under 2.5GHz.

 

[DAPUNISHER]

If you checkout majormav's post here he gets a couple of 1700+'s he picked up a hair from 3ghz with a prometia setup. Also, if you checkout prometheus's AKA Wesley Fink *new Anandtech staffer* review of the DFI LANParty he did for bleedin'edge before he took the job here, you'll see he had his 2500+@2.5ghz on air Linkage with 2v vcore and water cooling you may be able to do 2.7-2.8ghz with your cube but obviously the pump and all would be external and the longevity of your CPU short. I'm not certain if the ACHME PSU my SN45G came with could handle the load either.

 

[DAPUNISHER]

I take that back about the water cooling having to be external I forgot about the sweet mod this guy did to his cube link

 

[batmang]

hehe, i thought it was awesome when hardocp got 4.44ghz off a p4-3.06 using vapochill cooler.

P4-3.06 @ 4.44ghz - word.

 

[Megatomic]

Originally posted by: DAPUNISHER
If you checkout majormav's post here he gets a couple of 1700+'s he picked up a hair from 3ghz with a prometia setup. Also, if you checkout prometheus's AKA Wesley Fink *new Anandtech staffer* review of the DFI LANParty he did for bleedin'edge before he took the job here, you'll see he had his 2500+@2.5ghz on air Linkage with 2v vcore and water cooling you may be able to do 2.7-2.8ghz with your cube but obviously the pump and all would be external and the longevity of your CPU short. I'm not certain if the ACHME PSU my SN45G came with could handle the load either.

These are by far exceptions and not the norm. It's all good work though. :beer:

 

[Jeff7181]

No, you won't get 2.7-2.8 with water. You guys don't understand... this isn't an issue with heat... it's an issue with the design of the CPU. Because of the Athlon XP's relatively short pipeline it physically can't run that high without super cooling it like with liquid nitrogen or prometia... the reason those work is not because they get rid of the heat generated, it's because they cool the CPU down so that electrical resistance is a lot lower, and the transistors can switch faster. Kryotech called it thermal acceleration when they were in the biz.

 

[DAPUNISHER]

Hey Jeff, you are right in 99% of the cases but I have seen 2.7-2.8ghz on water with the thoroughbreds and bartons around the web. All water coolers aren't built equal and there's setups that can provide superior cooling to others, some combine the peltiers with them as well. So it can be done without phase change or liquid nitrogen but it certainly is the exception as Megatomic pointed out.

 

[Jeff7181]

I seriously doubt it's stable at that speed on just water. A Peltier is not water cooling. A Peltier cools the the CPU down well below 0 C if I'm not mistaken... which falls into the category of "thermal acceleration" more appropriately than just a really good cooling solution.

*EDIT* Hell, my XP2500 will POST at 2.5 Ghz, and BIOS temps don't go above 60C when I'm just sitting in the BIOS... but the chip physically isn't capable of operating at that speed unless you freeze the damn thing.

 

[DAPUNISHER]

Originally posted by: batmang
hehe, i thought it was awesome when hardocp got 4.44ghz off a p4-3.06 using vapochill cooler.

P4-3.06 @ 4.44ghz - word.

This is specifically a sktA related question so please don't post irrelevant data, thank you

 

[SinfulWeeper]

Originally posted by: batmang
hehe, i thought it was awesome when hardocp got 4.44ghz off a p4-3.06 using vapochill cooler.

P4-3.06 @ 4.44ghz - word.

:Q

 

[DAPUNISHER]

Originally posted by: Jeff7181
I seriously doubt it's stable at that speed on just water. A Peltier is not water cooling. A Peltier cools the the CPU down well below 0 C if I'm not mistaken... which falls into the category of "thermal acceleration" more appropriately than just a really good cooling solution.

Yes Jeff, I know peltiers aren't water cooling hence the term "combine" I can't comment about the stability beyond what the users claim since I haven't done it or used their systems so you are certainly entitled to your skepticism since the internet is full of FUD. CuteHamster posted some nice benchies using superchilled water@almost 2.8ghz but as to wether it was what you or I would qualify as stable is another matter. Never-the-less it's evidence you should qualify your remark better next time IMO

 

[Jeff7181]

Super chilled water would be right around 0 C. That's hardly enough to "thermally accelerate" a CPU.

I can complete benchmarks at 2.4 Ghz on 1.75 volts... does that qualify as stable even though it fails Prime95 as soon as it starts?

 

[DAPUNISHER]

Originally posted by: Jeff7181
Super chilled water would be right around 0 C. That's hardly enough to "thermally accelerate" a CPU.

I can complete benchmarks at 2.4 Ghz on 1.75 volts... does that qualify as stable even though it fails Prime95 as soon as it starts?

Are you arguing just to past the time? I thought I agreed with you about their interpretation of stability and the tendency of people to post FUD making believability or a lack thereof=skepticism already

 

[dejacky]

Cranking up the FSB speed seems to make the most noteable impact on performance, but I'm worried that higher FSB speeds negatively affect my IDE drives on my Shuttfle SN45G computer (nforce2). I'm thinking of somehow going from 2.1ghz to 2.3 with FSB increases over 200mhz. Any suggestions? Thanks, this thread is starting off great! .

-dejacky

 

[Excelsior]

Originally posted by: Jeff7181
No, you won't get 2.7-2.8 with water. You guys don't understand... this isn't an issue with heat... it's an issue with the design of the CPU. Because of the Athlon XP's relatively short pipeline it physically can't run that high without super cooling it like with liquid nitrogen or prometia... the reason those work is not because they get rid of the heat generated, it's because they cool the CPU down so that electrical resistance is a lot lower, and the transistors can switch faster. Kryotech called it thermal acceleration when they were in the biz.

You are wrong. Actually it is an issue with heat if you are running 2.0V or higher, even 1.9V. True, the design of the CPU limits the overall speed you could ever obtain, but as long as you can keep feeding it voltage w/o killing it, and keeeping it cool, then you will find the true limit of the chip. I know that if I tried to run 1.9V on my 1700+ itd get pretty damn hot, however with chilled water or a peltier/water setup I could do it just fine.

 

[Excelsior]

Originally posted by: dejacky
Cranking up the FSB speed seems to make the most noteable impact on performance, but I'm worried that higher FSB speeds negatively affect my IDE drives on my Shuttfle SN45G computer (nforce2). I'm thinking of somehow going from 2.1ghz to 2.3 with FSB increases over 200mhz. Any suggestions? Thanks, this thread is starting off great! .

-dejacky

It won't affect your IDE drives at all. The PCI/AGP bus on Nforce2 is locked. Go as high as you can .

 

[bjc112]

Originally posted by: Megatomic
LN2 is the abbreviation used for liquid nitrogen. I think the best mere mortals can hope to attain on air cooling with an AMD chip is just under 2.5GHz.

Yep, 2.5 is the highest "reasonable" overclock.

Too bad, if there was an 3ghz AMD chip on air cooling, well, Intel would have something to worry about.

 

[leeperpsu]

bjc are you saying 2.5 is the highest on air or watercooling.

cause i'd think 2.5 is about the limit even with good watercooling setup.

 

[Megatomic]

Originally posted by: leeperpsu
bjc are you saying 2.5 is the highest on air or watercooling.

cause i'd think 2.5 is about the limit even with good watercooling setup.

I had my 2100+ up to 2438MHz (195MHz x 12.5) with a Vcore of 1.75V on air cooling. It wasn't Prime95 stable (many reasons) but it could complete a run of benchmarks without crashing. I think 2.5GHz is reasonable.

Oh, my "air cooling" was a Coolermaster HHC-001 with a quiet fan on it. I can only imagine what it could've done with my current cooler in place.

 

[Jeff7181]

Originally posted by: Excelsior

Originally posted by: Jeff7181
No, you won't get 2.7-2.8 with water. You guys don't understand... this isn't an issue with heat... it's an issue with the design of the CPU. Because of the Athlon XP's relatively short pipeline it physically can't run that high without super cooling it like with liquid nitrogen or prometia... the reason those work is not because they get rid of the heat generated, it's because they cool the CPU down so that electrical resistance is a lot lower, and the transistors can switch faster. Kryotech called it thermal acceleration when they were in the biz.

You are wrong. Actually it is an issue with heat if you are running 2.0V or higher, even 1.9V. True, the design of the CPU limits the overall speed you could ever obtain, but as long as you can keep feeding it voltage w/o killing it, and keeeping it cool, then you will find the true limit of the chip. I know that if I tried to run 1.9V on my 1700+ itd get pretty damn hot, however with chilled water or a peltier/water setup I could do it just fine.

You misunderstood what I'm saying... of course if you use a retail heatsink heat will be an issue. If you can get rid of the heat generated at 2.8 Ghz at whatever voltage it requires, it's not an issue, and the physical properties of the CPU become the issue. And unless you "thermally accelerate" the CPU, you're not gonna hit 2.8 Ghz... even if you're running at room temp.

 

[stevejst]

No, you won't get 2.7-2.8 with water.

Correct. Talking about 2.0V.
Frequency is linearly dependent on voltage, that is physics. Watercooling does not change the physical characteristic of a chip, as said already. So 2.0-1.65 = 0.35 is only 21% increase meaning the possible chip frequency is theoretically at 2.66 GHz, and we are talking about chip that was able to do 2.2 GHz at a default 1.65V, as well as ideal cooling setup.

Now if the chip is able to do 2.3 GHz at 1.65V, you might be able to get to 2.7-2.8 GHz in the ideal setup. Is there such a Barton 2500? I doubt.

However, all of this is very unlikely. Beside how long wll a chip last at 2.0V on a 24/7 basis? I say not very long.

 

[Jeff7181]

Yup... I don't know of a Barton that will do 2.3 on 1.65 volts... mine won't even do 2.3 on 1.750 volts... it takes 1.8 to get it to run Prime 95 for longer than an hour at 2.3. But it runs 2.2 on 1.65 volts for at least 12 hours.