E7200 Overclock Problem...333 FSB Wall

aj6065

Distinguished
May 31, 2006
142
0
18,680
Ok, I have a new E7200 with a Gigabyte GA-G31M-S2L. My memory is 2x2gb of G.Skill DDR2 1000. I'm fairly new to overclocking but have done a ton of reading and know a lot about it.

I have hit a wall at about 333mhz FSB. The highest overclock I have been able to reach is 3.23ghz (9.5 x 340FSB). Any higher than that, and the motherboard will reboot and reset to default settings. It doesn't matter how much voltage I pump into the CPU, FSB, or RAM; I still can't get any higher than 340mhz. Right now I am running at 9.5 x 333 = 3.16ghz. My CPU voltage is 1.06v, my FSB is normal/auto, and my RAM is 2.1v. I am completely stable with Prime95 and Orthos running simultaneously for 24 hours.

One thing I have noticed while doing some research is FSB straps and how they can limit the FSB to 333mhz. I know this is a problem generally associated with the e2xx0 line, though.

The E7200 is such a good overclocker and this is driving me crazy because I can't get anything out of it. What is even stranger is that I have read about other people with the same motherbaord and CPU who have gone well beyond 340mhz FSB. Any help would be appreciated, thanks!
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790
you need to increase vcore and other voltage as you approach higher FSB. if you have read the OC guide you should know that you cant just keep upping the FSB without adjusting other settings.
 

aj6065

Distinguished
May 31, 2006
142
0
18,680

As I stated in my OP, I tried increasing all voltages available, with no success. I went as high as 1.45v for my CPU, +.3v for the FSB, and my RAM is set at 2.1v (rated by manufacturer).

I am currently using a RAM multiplier of 3, which gives me DDR2 1000 at 333mhz FSB. However, when trying to go above 340mhz FSB, I used a multiplier of 2.4.
 

omicron_15

Distinguished
May 1, 2008
56
0
18,630
I have the same ram, but have a p45 gigibyte mobo with a E8400, I also am stuck around only a 345mhz FSB before I eather dont post or get prime95 errors. I also tried increasing all the voltages to the maximum and they also didnt seem to have any effect on increasing the OC. Our common factor seems to be our ram.
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790
and just to let those of you know dont know. a high ram multiplier then 1 or 1:1 you will NOT gain any performance at ram. so no point putting extra stress on the ram.
 

aj6065

Distinguished
May 31, 2006
142
0
18,680

really? Why is that? Is using a 1:1 ratio going to allow me to go beyond 340mhz FSB? Thanks for your help
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790
because the CPU and Ram are cummincate THROUGH the FSB. so even the speed on either or both side is faster, the speed will be restricted to the max FSB speed. make sense now?

and by select lower multiplier it will let you OC higher as most of the time on a "default" setting OC the ram maxed out before all the others do.

but of course if you are so unlocky and got a bad yield chip then nothing will help.
 

aj6065

Distinguished
May 31, 2006
142
0
18,680

But how does that make sense. That would mean RAM multipliers are useless and shouldn't exist. If the RAM is restricted to the FSB, then there would be no point in making RAM that could run at 800mhz if your FSB runs at 266 or 333mhz. You are saying in that case the RAM can only communicate with the FSB at 533 or 667mhz, which is below the stated 800mhz of the RAM. The reason RAM multipliers exist is so you can purposely run RAM higher than the FSB speed.
 

SirCrono

Distinguished
Sep 9, 2006
463
3
18,785
You shoud check you haven't run into a FSB hole, I recently OCed and old Pentium D 805 (133 Mhz stock FSB) and the thing woudn't post at 2.8 or 3 GHz, I cranked it up to 3.2 GHz and it worked (and it's still working) fine.

Also, check the northbridge temps, maybe it's too hot and crashing due to that.
 

homerdog

Distinguished
Apr 16, 2007
1,700
0
19,780

Although this is not entirely correct, I agree 100% the you should use a 1:1 ratio with Intel processors. The performance gained by using higher ratios can just as easily be had by lowering the timings :)

Edit: Well, almost (http://www.anandtech.com/memory/showdoc.aspx?i=2989&p=7)
 

aj6065

Distinguished
May 31, 2006
142
0
18,680

A FSB hole could be possible, but I doubt it since I have tried upping the FSB from 341 to 400 in increments of 10. But I will look into that a little more.

I will also check the northbridge temps, which I haven't done. I hope it's something simple like that :) Thanks for the reply.
 

50bmg

Distinguished
Nov 16, 2007
903
2
18,985
I have an E7200 on a gigabyte board OC'd to 3.8G, 400fsb, 1.33v. It has been stable for 6 weeks. The first week i ran prime95 a couple of times and CDmark06 a few times since. No overheats or crashes. I game hard for hours at a time with no issues.

Make sure you set your RAM voltage manually to 2.0v and if you have pc6400 (800Mhz) is set to 1:1 ratio.

Might as well set all other setting to manual. ex: PCIe to 100Mhz.

As your FSB goes up all those default auto settings will go up. That might stress out something.

Edit: i do have an aftermarket coolermaster 40mm NB fan.

I am at work so i cant post upload my cpuz or hwmonitor info to photobucket, but my temps are as follows: CPU idle 33c, load 53c. System idle 38c, 48c load.
 

aj6065

Distinguished
May 31, 2006
142
0
18,680

Thanks for the link. I will definitely change to a 1:1 ratio and see how it goes. I'm at work now but will try when I get home.
 

aj6065

Distinguished
May 31, 2006
142
0
18,680
Well for an update...I messed around with it for a while and got it to boot a FEW times at higher FSB's. With a 9.5 multiplier the highest I got was 355mhz. When I dropped the multiplier to 7 I could get it as high as 370 - 375mhz. These results were all completely random, however, and I didn't even try to stress or test the system. There was no combination of settings that could get consistent results. I changed to a 1:1 RAM ratio but it had no effect. My Northbridge was kinda hot so I threw some AS5 and a fan on it. I don't think that helped too much. Well then again I wouldn't expect it to considering I don't think that is the problem to begin with ;) I might just have got an unstable board or something. But even 3.16ghz is a 25% increase from stock, and it's still fast as hell for normal computing, so I can't really complain. Thanks again for all the help
 

homerdog

Distinguished
Apr 16, 2007
1,700
0
19,780
Make sure the tRD (I think GIGABYTE calles it "static tREAD value") isn't set below 8. Try something like 10 first and a FSB of 400MHz. If you can get the tRD down to 8 at 400MHz you're doing really good, but 10 isn't bad.

And if your G31 is like my P35-DS3L you have to press Ctrl + F1 in the main BIOS screen to access the advanced timings.
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790


of course you can run it at 800mhz with a 200fsb or 266 or even 333. the rated speed of the memory doesnt mean you have to run it at that speed. and even if you DO run it at those speed then the performance gain is minimal because its restricted by the FSB.

with a higher ram speed along with a low FSB you will only gain lower latency but actual read and write performance will ONLY gain about <100MB/s. i think you should read into the architecture of Intel's FSB based processor.

for arguement sake if you cant get any actual performance gain why put extra stress on the ram when its unneccessary? what the ram is stated to run you dont have to run it at that speed. running it at higher speed will be more stressful then run it at lower speed.thats common sense. of course IF you do gain alot of performance then sacraficing its life span its reasonable.BUT NOT WHEN YOU CANT GAIN ANY SPEED.
 

gmanvbva

Distinguished
Jun 7, 2008
75
0
18,630
I have a GA-G33M-S2L with an E7200 OC'd to 3.6 (400x9) on 1.25.

I set +1 to the FSB, if I remember correctly.

Stable for 12 hours on OCCT (Prime).