Sign in with
Sign up | Sign in
Your question

OC newb trouble

Last response: in Overclocking
Share
July 14, 2005 6:49:03 PM

Got my new system all setup flawlessly - booted up great on the first try, so thanks everyone for all the help on that. Now, with all the reading I've been doing, I wanted to go mess around and try to OC it just a little bit. My system is

Athlon 64 3200+
2x512 OCZ ValueVX
Epox 9npa Nforce 4 Ultra mobo
FSP 500w bluestorm

and by increasing the FSB by 5 at a time, I got up to a stable 210 before it started crapping out on me. I rebooted and raised the Vcore temp, but I couldn't get it to run at 215 FSB even at +.025, .05 and even .075. Even after the highest raise, the Vcore was only at like 1.55v and temps were still sub40 for both CPU and chipset.

I went in and upped HT (wth is this, by the way?) to 4 like someone said to do, then increased power again it still crapped out.

It would either give me the blue screen of death, or, once, boot into windows and not run correctly.

I'm wondering why I cap out so ridiculously low when the temps are fine and I'm increasing voltage as per instructions. At 210 FSB without increasing voltages I ran the prime95 torture test perfectly for 6 hours, so I'm failing to understand what's kicking my butt about another 5mhz in such an otherwise solid system (good OCing mobo and CPU, good temps etc.).

I still have the original BIOS.. would upgrading to the newest ones make a difference?

I'm still a total newb to this, so if anyone could throw me a few things to check out or steps that I might be doing wrong/missing, I'd really appreciate it.

More about : newb trouble

July 14, 2005 9:17:12 PM

what are your ram timings and speed? Also voltage.

Ah, I think I see it: VX = Voltage Extreem. They like the volts - 3.0 or so - I think 3.6 for the best timings and performance (but that wasn't for the value series - they should take 3.2 or so though - and your epox tops out at 3.2 I think.)

Try raising your RAM voltage, and slack the timings to the maximum. If it still doesn't work, drop the RAM back to 166 and see if it helps.

Mike.

<font color=blue>Outside of a dog, a book is man's best friend. Inside the dog its too dark to read.
-- Groucho Marx</font color=blue>
July 14, 2005 9:31:23 PM

Awesome.. I think the RAM defaulted to like 2.7v, and the max setting I had was 3.1. I'll crank it up to 3.0 and test it out when I get home. The timings are, if I remember correctly, 2.5,4,4,7/8.

I have no real setting to adjust RAM speed, all I can do is set it to like 100, 133, 166 and 200. So in order to figure out what it's at, and then change it, do I have to use the equation of (CPU-FSB) x (RAMfsb:CPUfsb ratio) to calculate what it is? That might make no sense, but another guy who posted below mentioned it, and I believe I remember reading about it.

In the end, am I aiming to have an equal FSB with my CPU and RAM? Say I could max out my RAM safely at 230mhz, and CPU at 240, would it be best to drop the CPU down to equal the RAM? These are things I remember reading about, but they weren't explained very well, thus the confusion.

Thanks again Fishmahn.
Related resources
July 14, 2005 10:09:31 PM

That's the setting I'm talking about - if voltage doesn't work, set your ram to 166 and as you OC the FSB, the ram will follow along at 5/6th of the speed. (so 240fsb w/ram at 166 = 200 ram) You can also try setting the CAS to 3.0 - the others are already pretty relaxed. (3-4-4-8 is usually as far as timing adjustments go for DDR)

That looks like the right thing - basically 200 = 1:1 (RAM:FSB), 166 = 5:6, 133 = 2:3, 100 = 1:2

In the end, the best performance is RAM = FSB. But, if your ram doesn't do it, A64s don't lose much (less than 1% I think) by running the RAM on the 5/6 divider).

Mike.

<font color=blue>Outside of a dog, a book is man's best friend. Inside the dog its too dark to read.
-- Groucho Marx</font color=blue>
July 14, 2005 11:34:22 PM

Ok.. got back home and did the following:
1. HTT up to 4
2. CAL to 3
3. DIMMv to 3.0
4. CPUfsb from 210 to 215
5. Vcore + .025

So upped the power, loosened the timings and went for a minimal overclock. Comp fails to boot into windows and beeps at me long and hard a few times. No registry damage, as all I had to do was reset the CMOS jumper, but I went back and lowered the FSB to 211 and dropped the divider to 166 and it did the same thing.

I'm stumped. With more DIMMv and a lower divider, it should be able to get over 210, but that's all I can get to and remain stable. It's running at 210 now with a DIMMv of 3.0, but I'm getting frustrated as nothing's working. Temps are still fine...

Am I forgetting something key?
July 18, 2005 6:24:47 PM

Ok, no one else responded, so I'll throw out a few wild pitches...

If I get one that's totally off base, I'm sure I'll hear about it. lol

I forget what mobo you have, but how's the temp of the northbridge? Is the chipset fan cooled or passive?

Try HTT at 3 - it should be fine at 4, but now you're in the fun part where you get to determine what's holding you back. With the RAM set at 166, it shouldn't be limiting you, so look elsewhere.

Also try lowering the CPU multiplier to 8 or 9 - it'll slow your chip below stock speeds, and take that possible limitation out of the picture.

Mike.

<font color=blue>Outside of a dog, a book is man's best friend. Inside the dog its too dark to read.
-- Groucho Marx</font color=blue>
July 18, 2005 11:27:51 PM

I've got an Epox 9NPA nforce4 ultra mobo, which is by all accounts one of the top few OCing mobo's..

Will HTT have that big of an effect? I don't even know what it is heh, but I'll try it since I'm still so stuck to 210 that I crash at 212.

Of course, annoyingly enough, my mobo doesn't have a direct option for adjusting the multiplier. There's one that's automatically set at 5, where I can put in anything between 0-10, but if that's it, why would it auto to 5? What's the CPU spread spectrum, by the way?

Sorry, but this is just bugging the hell out of me, as I've tried everything I can think of. I'll go give your tips a shot and see if it makes a difference..

Cpu temp is still under 40, and system temp runs like 34 C.. I've got two 80mm's up front for intake, an 80mm in the back and a 120 in the PSU, plus a side intake cone, so I think everything's ventilated pretty well.

I managed to get the ram do 2.5/3/3/8 at 420 pretty easily, so at least there's room to expand there if I get stuck on the CPU..
July 19, 2005 12:13:32 AM

I'm baffled now..

Lowered HTT to 3, lowered multiplier to 9 and upped FSB to 235 for a 2115mhz OC, and it still crashed on me. Went back and upped the juice to the CPU by 2 notches and tried the same setting.. crashed again so I had to reset the CMOS.

It appears I'm just completely stuck at 2100... because I'm pretty sure I've done everything correctly so far, though it doesn't make sense as CPU temp has yet to break like 37 C at 2100, so you'd think it would have at least a LITTLE more juice in it.

Could I need more than a .05 bump in Vcore? Would chipset power matter? That's about the only thing I haven't raised. So confused :( 

I haven't upgraded my BIOS.. but would that matter in the least?
July 19, 2005 3:19:30 PM

HTT should have no performance effect at 3 4 or 5.

Turn off CPU spread spectrum - what it does when on is slightly vary the speed of the clock signal to reduce EMI emissions, which can effect performance.

Mike.

<font color=blue>Outside of a dog, a book is man's best friend. Inside the dog its too dark to read.
-- Groucho Marx</font color=blue>
July 19, 2005 3:21:39 PM

Was it stable at a 9x multiplier with fsb at 230 (at least, over 210)? If so, I think we have a culprit - the CPU. To confirm, drop the multiplier to 8 and see how far you can push the fsb.



Mike.

<font color=blue>Outside of a dog, a book is man's best friend. Inside the dog its too dark to read.
-- Groucho Marx</font color=blue>
July 19, 2005 3:27:32 PM

I didn't try 230, since 235 was so barely over 2100.. What I'm wondering about is if I should lower my divider to like 133 just to test the RAM, and if I should really juice up my Vcore, as I think it's sitting at under 1.5v, so even if I throw on .05 it's probably only getting up to like 1.52v.. I'm going to try to juice it up to 1.55v at least and see if that does anything.

I DID get it to run at at least 215 (maybe 220)x9.. so it CAN go over 210, but I just can't break a total speed of 2100 mhz.

With my limited knowledge I have no idea what would cap me out so hardcore at such a set speed.. I'm hoping there isn't some BIOS option I don't know about that's holding me back.
July 19, 2005 9:19:55 PM

get a better power supply

<font color=red>gforce mx100/200 @ 230/440 =]</font color=red>
July 19, 2005 9:29:10 PM

What's wrong with that power supply? It's very good, in terms of efficiency it's almost as high as Seasonic power supplies.
July 20, 2005 12:33:39 AM

Ok.. figured it out, and I feel dumb :) 

Bumped Vcore up .125v to 1.59v, and I'm currently running 230x10 perfectly. I'm stopping there to run Prime95 for a while and test it out, but my CPU temp in post was 41 max (sitting at 39-40 mostly) with system still at 36, so it looks like I have some room to keep going!

Thanks a ton for taking the time to answer my questions guys, I've really appreciated it. I'll post my max once I figure it out!
July 20, 2005 7:14:49 PM

cuz i say so =]

<font color=red>gforce mx100/200 @ 230/440 =]</font color=red>
July 20, 2005 7:19:18 PM

Yeah... okay...

NO!
July 20, 2005 9:43:37 PM

you know just about how much better is the pc power&cooling 510w delux better than the fortron?

<font color=red>gforce mx100/200 @ 230/440 =]</font color=red>
July 20, 2005 10:37:09 PM

I know it's better, considering most Fortron made power supplies has a 75%-80% efficiency, PC power&cooling and Seasonic can reach as high as 85% efficiency.
July 21, 2005 12:46:59 AM

not to say the blistering stable voltages when OCing =]

imma get one!

<font color=red>gforce mx100/200 @ 230/440 =]</font color=red>
July 21, 2005 1:11:59 AM

I'm using a trusty Seasonic 600watt, i love the fact the labeled the continious output, not peak. Total stable voltage rails, and it barely gets warm. Thanks to its 85% efficiency, only 15% gets dumped as heat.
!