Voltage hog of a barton 2500+

[SavageIX]

I'm running on an Barton 2500+ (AXDA2500DKV4D Y907574L3670 PQZFA 0347 MPMW) and this baby is a real voltage hog. I would slowly up the fsb until unstable, then i'd up the vcore a little too when necessary. To get anywhere near a FSB of 200MHz (note: NEAR, not at), I have to set the vcore to 2.0 volts - and as I've read, that's not something that is generally a smart idea. Currently, I'm running at 174x11 @1.65v. I just ran a 30 cycle burn-in test (CPU Arithmetic, Multi-media benchmarks and Memory Bandwidth and Cache & Memory benchmarks) for Sandra and that ran fine.

What I'm looking for is some advice as to getting the FSB up to at least 200 MHz.

Also, what is better to use? SiSoft Sandra or Prime95? I've been having some bad expreinces with prime, and I have both of them, and i'm wondering what gives a more accurate test of system stability.

[obsolete]

Hey Savage,

Not sure why that is happening to you. I have my 2500 (newer Barton) at 223x11 with only a 1.8v applied. I have a different board though. Your board is suppose to be a decent ocer though. Maybe your powersupply? Your memory looks adequate.

I think it's very arguable as to what defys a "stable" overclock. Personally I wouldn't & don't base my conclusion on one program to determine whether or not my OC is stable. Most do with Prime 95, but I think it's all a matter of personal preference. I would at least run some CPU intensive games for hours. Some benchmarks for hours. As your mentioned Sisoftware!! For at least a few hours. Switch it up & run everything continuosly.

Some people have had Prime 95 run fine for 20 hours & then crash after 10 minutes of playing a game......So what does that say, I dunno, up to you to decide.

-ºb§

[obsolete]

Icksnay the power supply comment, didn't see it in your stats before I responded, nm.

[makaka]

borth prime and sandra are good but sandra is great tool for benchmarking and prime95 for stablity test
i have asus a7n8x i can run at 200 fsb with the default voltage
a lot of people can run at 2200 mhz with 1.65v

[SavageIX]

Mayhaps I will quit using P95 and start running UT2003 (and when it comes out, Half-Life 2 :). Come to think of it, P95 was all I was using to benchmakr when I was running. Perhaps THAT is the problem. In any case, I'm going to try a variety of benchmarking programs, as you suggested. Anyone else who has some advice or who favors a particular benchmarking program, even prime, please post! I welcome the opinions of everyone, since I'm pretty new to these forums :).

[obsolete]

Many will say if not P95 stable then not stable at all. So like I said, personal preference Savage. All I'm suggesting is try a variety of tests/benchmarking/CPU stressing. Try on your own & see what you come up with =]

[badge56]

Here is my experience with 2500+ using AN7 and OCZ 3200

0333 2266mhz 11X206 unlocked (stable 1.812) any higher no diff.
0335 2310mhz 11X210 locked (using 1.9 Vcore)

my nephew just got one (week unknown) and got it stable at 3200+ using 1.75VC

They look alike but are not all the same

[Axis]

Originally posted by makaka
borth prime and sandra are good but sandra is great tool for benchmarking and prime95 for stablity test
i have asus a7n8x i can run at 200 fsb with the default voltage
a lot of people can run at 2200 mhz with 1.65v

I can with my 2500 barton and my t-breb B 2600. Both are @ 2.2Ghz 1.65 volts with no issues.

[SavageIX]

Also, i'm wondering at what incriment should I increase the FSB? 1 MHz? 2 MHz? 5 MHz? How much does it depend on how the last change held up?

[Strages]

You have a system that has O/C potential definitely. The NF7-S is the top oc'ing board out there right now, and your memory seems to be decent enough.

Make sure you have the lastest BIOS or at least a newer one (the d21 has given me the best results right now, newest is d22) and that you have the BIOS configured for an o/c. Best results I could get are with settings as folows:
CPU Interface: Enabled
CPU Disconnect: Disabled
FSB Spectrum: Disabled
AGP Spectrum: Disabled
Thermal Throttling: Disabled

If you'd like, you can disable CPU Interface for a more stable o/c, however unless you have a really good overclock (like 2.5) stable with it, then I'd leave it enabled - the added performance is just too great to sacrifice. Also, other people have found that the "Enhance PCI performance: disabled" has helped, but mine shows no difference, so I left it enabled.

Also make sure that you have very good cooling. You didn't mention what your cooling setup is, so it'd really help us to know what you've got cooling that thing. If you're running stock fan with only a few (2 or 3) simple case fans, you need to spend some doe on good stuff. Hope this helps!

[Tebore]

There's nothing wrong. It's the chip that stepping sucks. I have 2 XP2500s one the exact same as yours week and all, it wouldn't do what the barton in my sig will do, at least not at sane air cooled voltages. It will do 2330 or so with about 1.9, I primed it for an hour but the temps were insane. The PQZFA chips are terrible. They run hotter and overclock terriblly.

[itshondo]

my 2500 needs 1.825 vcore and mem 2.8 volts to run stable at 3200. Unlocked AQXEA 0330 VPWM. Soon to be for sale- getting a mobile 2500 I think...

[Tebore]

I was wondering is anyone else NF7 under volting? At 1.825v I get 1.8v only. Reading from MBM5.

[DaddyB]

Yeah the NF7 tends to undervolt a little bit, I have my xp2500 set at 2.06V in the BIOS but it reads 2.0V-2.02V in MBM, winbond hardware doctor and even in the BIOS...

My chip is the same stepping (PQZFA) and needs a lot of voltage to OC, at 2.06V my temps go just a bit over 50c with a 50CFM fan on my SLK 800. Anything less then 2.06V will cause reboots at 210FSB/2315MHz(reboots about once a week at 2.03V, once a day at 2.0V). I think I needed about 1.85V just for 200FSB/2200MHz, can't really remember but it was more then I expected.

I use Prime95 for stability testing, if it can run for 3 or 4 days I call it stable. Games are also a good stability tester but IMO nothing is better then Prime, you can play games on a system that is less then 100% stable but if you let Prime run long enough it will get an error. Of course letting your system sit there running Prime for the better part of a week isn't much fun. I only do that when I am near the max OC the chip will do and I want to know if it's 100% stable. While testing the chip I'll run prime for a half hour or so, then up the FSB a few MHz and test again, if it fails add more V and test again, up the FSB...etc.

[Chixofnix]

Originally posted by SavageIX
Mayhaps I will quit using P95 and start running UT2003 (and when it comes out, Half-Life 2 :). Come to think of it, P95 was all I was using to benchmakr when I was running. Perhaps THAT is the problem. In any case, I'm going to try a variety of benchmarking programs, as you suggested. Anyone else who has some advice or who favors a particular benchmarking program, even prime, please post! I welcome the opinions of everyone, since I'm pretty new to these forums :).

I use Command and Conquer Generals at a fast speed with 7 computers on the highest graphics levels to test my entire system as a whole... Gives everything a workout, and I've had OCs that were stable in P95 and 3Dmark, but crashed in CnC...

comes down to what is the most system intensive program that you use? If IT will run without errors, then in truth that means your system is as stable as it needs to be...

I ignore P95 myself, as I didn't build my computer to figure out prime numbers, it's built as a serious tool for gaming entertainment :D

[theedge]

I think its all in the stepping. I have an AQZEA0341 that does not want to do 200x11. I've raised the vcore as high as 1.9v and it still wasn't Prime stable. I recently got an unlocked 2500 AQXEA0331 and it will do 220x10 at 1.725v(BIOS reading) prime stable.

[SavageIX]

My cooling is an AX-7 heatsink, thermaltake's ducting mod, and some fan 80mm fan (I googled it and found out its actually used in some Enermax PSU, and it runs about 40 cfm I believe), and a nice thin layer of AS5 (I bought the whole set as-is for 25 bucks over ebay... not a bad deal). I also have 5 case fans (2 front, 2 back, one side) and the two fans in my power supply. It seems to be quite adequate.

Does the amount of voltage pumping into the chip really matter as long as the chip is relatively cool? What is the absolute maximum I should ever pump into a chip? Please explain.

[badge56]

The Vcore matters. I have tried a few 2500+ and not all are the same. Sometimes you can give it a lot of Vcore and it still will not be stable even if its verry cool.
Using liquid cooling ....

[SavageIX]

How much is "a lot" ?

[malon]

I have the same chip PQZFA 0347, and its a terrible overclocker too, all I can do is 182x11 with 1.65 vcore, more than that and it begins to have artifacts or lock in games :(

[badge56]

I have read that many go as high as 2.+ v
I have gone 1.925 ..... no higher.
You need good cooling, dont forget.
None of this standard HS fan crap. Thermalright, Swiftec etc good stuff :-) or liquid !!!!

[SavageIX]

that's what I said... Thermalright AX-7