Begining To Oc

[gotdamojo06]

psycho, were you using stock cpu cooling or no?

[Verran]

ok, ill have to try to set the memory divider at 133, after i set the voltage down to 1.6, and see how that does.

So like I asked before, how did you get to 1.7v? Did you just kinda set it there? Or did you keep upping it to improve your OC slowly? It seems unlikely that you'd continue to see improvement past 1.65v or so, especially with the normal temps on air.

has anyone else heard anything about the "hidden" settings with my bios (shitf+f2 and alt+f3)?

Yeah, I've heard of them. It adds (or reveals, rather) a few settings in the OC settings area. I don't remember using any of them on a regular basis, but it has been a long time since I did any tweaking to that system.

Not saying that you are wrong Verran but when I had my A64 3000+ (Newcastle) that thing LOVED volts. I had that thing pumping out 1.75 with an OC of 2680mhz ON AIR

Yes, but that's a Newcastle chip. It's 130nm. It's two full generations away from the Venice that he has. The farther back you go, the more voltage a chip will take.

 

1.6v has kind of always been the unwritten limit on an air cooled Venice or Winchester. (Or at least, it was back when those chips just hit the market.) Honestly, I wouldn't want to run 1.7v 24/7 even on water. That just seems like too much voltage on that chip to me. I haven't seen many people run that kind of voltage on those chips that weren't using extreme cooling.

 

Like I said, if the temps really are OK, and stability is fine, then it's up to you. It's not like it's going to explode in a week at 1.7v, but to me I just wouldn't want to run it that high.

[gotdamojo06]

yeah i sotra just set it there, thinking that it would help the stability, which from what i could tell it didnt change much, so i didnt bother lowering the voltage down any.

 

so here when CS:S is done being reinstalled, will be setting the Voltage down to 1.6v and the memory divider at 133 with the ht multipiler at 4x, and the bus speed at 245 to see if it will work

 

hopefully ill get better than a 2684MHz oc

[Verran]

hopefully ill get better than a 2684MHz oc

Honestly, 2.7ish is not bad at all for that chip. Especially on stock cooling. But the thing is, if it's unstable, then you can't really go by that speed. Hell my Opteron will do 3.5ghz... just not for very long

 

If you can get a 3500+ Venice to 2.7ghz on a neo2 completely stable with stock cooling, I'd say that's quite an achievement. I would shoot for that first, then think about going higher.

 

What you really need to do is start over. Just picking speeds and voltages randomly doesn't really help anything. Start at stock and raise the speed a bit. Then TEST IT. When you get unstable, up the voltage. Rinse and repeat until temps or voltage get past what you consider acceptable. This process is all over the boards, and for good reason. It's the only way to really do it right. If you just pick settings randomly, you have no idea how to fix them when they fail.

[gotdamojo06]

they arnt very random, i started by upping the bus speed after i set the ht multipier at 4, when it wouldnt boot and would just go black i found out that i wasnt "locking" the vga so i had to set that at 67. when i would get stuck booting windows i would then up the voltage and i got to 1.7 then i couldnt go past 244 with the bus speed so i asked questions. i was going up by one mhz with the bus speed at a time, not just random ones. so im now deciding to lower the voltage, and change the memory divider and try at where i left off with the bus speed + 1 mhz

 

if you think that i should do something else, or what setting you think i should start over with, suggestions are very well welcomed

[Encrypted_God]

psycho, were you using stock cpu cooling or no?

No...I was/am using the XP-90. Good luck to you...and have fun!

 

I understand verran... :thumbs-up:

[gotdamojo06]

dont get me wrong, i understand him as well, but i was just trying to say that i was not just picking random settings and changing random settings to achieve a higher oc, i actually sorta had a method.

[BTSchnarfy]

I am running stock cooling on the cpu, sucking air off the heatsink, with a duct taking it out the back of the case with a fan adding more suction

 

Even though you have a duct for the air to go straight out the back of the case, it would probably still help to change that setup a little bit. First, flip the CPU fan over so it will blow air ONTO the heatsink, not pull it away from it. Second, lose the duct and just go with the regular exhaust fan on the back. Generally, having it set up the way you do hinders cooling performance rather than helping it. Also, as long as you're using stock cooling I wouldn't run it at much more than 1.5V 24/7 unless you are pretty content with the temps (IMO).

[gotdamojo06]

even if i have a side exaust, i should loose the duct?

[Verran]

I definatley have suggestions. Like I said, start over. Right now you have absolutely no idea why your tests fail. That's because you've gone about your process in a very haphazard sort of way. If done properly, you should always know where your problem comes from (or be able to find it quickly).

 

And for the record... booting into windows is NOT a stability test.

 

--Set everything back to stock. Use the "reset factory defaults" if you have to.

--Set your memory divider. If your memory is crappy, set it to 100 (1:2 ratio). If it's decent, then 133 or 166. This is done so that your memory speeds always stay below DDR400. That way, you know your ram is never the cause of the failure. If you want to see your ram speeds, use CPU-Z, which you should DEFINATELY be using anyways.

--From stock, move your FSB (HTT) up 5mhz. Boot into windows and run a stability test. Honestly, I don't run full stability tests when I'm doing a "test climb" like this. I'll usually run a few SuperPI 1M's, and then a few 4M's. I've found that documenting these results can REALLY help later. Write down all your settings, what tests you did, and then what the results were (pass/fail, superPI times, fail times in Prime or OCCT, etc).

--If you pass the tests at 205, then redo them at 210. Then 215. 220. Etc.

--Eventually, you will fail them. When you do, check your ram speeds in CPU-Z. If it's above 400, then set a bigger divider and retry. If the memory's OK, it's time to step the voltage up. Do this in SMALL increments. 0.025v or so at a time. Jump the voltage up and keep going.

--Just keep doing this until you either get temps that are too high, hit a voltage that you're not comfortable with, or your speeds won't climb anymore.

 

This is how you find the ceiling for your CPU alone. Once you've found that, you can start dialing your ram speeds back in. That's a whole different thread though.

 

If you want more, I wrote a whole guide on this. It's pinned in the Overclocking section.

 

EDIT###

SuperPI is not the end of the testing phase, for the record. When you find the ceiling, then you can start testing a lot harder with OCCT, Prime95, etc.

 

Also, he's not saying to get rid of the duct. He's saying that your heatsink fan should blow air down onto the chip, not suck air off of it. This is how they are set up from the factory. So unless you changed it, you're fine. A lot of people assume it sucks because heat rises, but that actually doesn't work as well. I think you just made that assumption, but I could be wrong.

[BTSchnarfy]

If you have a side panel fan, it should be an intake, not exhaust. You'd want the fans to pull the cooler outside air into the case (expecially right over the CPU) from the front and side, then exhaust out the top and rear.

[gotdamojo06]

thank you, i will have to do that and i do know that booting into windows is not a stability test, i was just saying that it wouldnt boot into windows, so i knew that something was very wrong.