Overclocking E7200 Wolfdale

[karkas]

You can try a lower multiplier, though the gains you realize may not translate into much which is noticable. If you want a stable system tomorrow I wouldn't start messing with settings today.

 

[Adaman]

Guess I couple days to fool around with it actually. It's stable at 7x410FSB (2870 mhz) right now though. Should just leave ram at 1:1 until I figure out the best stable CPU speed and all that?

 

[Craxbax]

The e7200 I just OC'd did better with a 9 multi then a 9.5. I think the mobo/chip combo has a lot to do with it. I was using ASUS P5k Pro.

 

[jason4207]

Yep, leave it 1:1 for now. You should be able to get it going a good bit faster...not sure what's holding you back at this point.

Does 7x400 work? If so leave the FSB at 400, and try using the multi to OC. See if you can get 7.5 or 8 to work. Ultimately, I'd think even 9 would work, but work up 0.5 at a time and see what happens.

You can probably set the BIOS vcore to 1.4 or even a bit higher (but only go up if you need it). You'll see much less than that in CPU-Z, and you should go by the CPU-Z value, not the BIOS value.

 

[Adaman]

Well it seems kinda weird. I'm currently running stable at stock vcore with 6x480 (2880) but even with 1.3vcore I couldn't get it to boot at 6x500. Maybe that's not too weird though, I dunno. Thanks again for your replies and patience.

edit: ahhh nevermind all that. For some reason setting the vcore differently in the bios doesn't actually change anything, so I've never actually tried anything but stock voltage even though I thought I was. Not sure why it won't change, but they have an application you can run on your desktop that can change it I think so I'll give that a try.

 

[jason4207]

If the vcore won't change despite your BIOS changes I would call GB tech support. If they can't get it working for you I'd get an advanced RMA. That's me, though...I won't settle for a board that doesn't work like it should.

 

[Adaman]

If the vcore won't change despite your BIOS changes I would call GB tech support. If they can't get it working for you I'd get an advanced RMA. That's me, though...I won't settle for a board that doesn't work like it should.

It has got me a bit worried. After reading the manual over a few times now and trying many different bios settings, the Vcore doesn't want to change. Off to the motherboard forum i guess - I better check to see if I have the latest bios and all that first. I'll give Gigabyte a call if all else fails.

 

[Evilsizer]

have you tried changing the bios in easytune with cpuz open to see if it does/doesnt chagne?

 

[raIDERgeek]

A lot of people are having hard time OCing the newer e7200 most will max out at 3.2-3.6 to go higher you need 1.4v+ ouch!

 

[Adaman]

Updated the bios and I can change Vcore easily now...kind of odd but I'm happy now. But another obstacle has arisen with the new bios...although it shows a 7.5x400 multiplier, on the bootup and in windows sytem properties...CPUZ has it as a 6x multiplier and at 2400mhz. Not sure why that is. Back to work I go

A lot of people are having hard time OCing the newer e7200 most will max out at 3.2-3.6 to go higher you need 1.4v+ ouch!

I've had this CPU for a while, just ran it at stock for a while. So it's one of the older ones, same batch as all the OC champs.

 

[jayfella]

It shows a 6.5 multiplier because of speedstep - turn it off in your BIOS. What it does is lower the multi to as low as it can go (6.5) when your computer is idle to save energy, but it can cause instability when you OC. You can try turning it back on after you hit your limit to see if it remaing stable, if not, turn it back off.

 

[Adaman]

Thanks jayfella.

I'm stable now at 3.2ghz (7x460mhz). Using 1.3 vcore though since I'm going to eventually use it anyways. So nothing too exciting yet. Might play with it some more now - What to try next I wonder...

So using an fsb over 500 isn't a good idea because it'd be "hammering my NB"? What's that mean exactly? Was thinking of doing 6xfivehundred something. I guess people say a higher FSB doesn't make much of a difference anyhow.

 

[jayfella]

synthetic benchmarks report an increase in speed with a higher FSB, but in everyday use, you wont notice it at all.

I would try for 3.4GHz - 3.6GHz - i dont think you will get much more past that. Maybe 3.8GHz. If you have an early chip like you stated, and it was early enough you could probably get 4.0 or higher, but for now, just keep doing what your doing until you hit a wall.

For what its worth, i have to throw over 1.4v at it to get 3.75GHz - it wont go past that unless i put 1.6v+ through it. I would say 3.4-3.6 is a decent OC because you wont have to put so much voltage through it.

When they say you will be hammering your FSB its because your gonna have to put so much voltage through it to achieve that speed, its not worth the degradation when you can get the same CPU speed with a higher multiplier/lower FSB.

 

[Adaman]

Currently Stable at 3400mhz (8.5x400) using 1.15 vcore. Yay! I'm happy with this. I doubt it has hit the wall (haven't tried anything higher, I assume it's pretty close to the wall) but I'm done fooling around with the CPU for now as I'm satisfied with voltage and temps.

But now I'm wondering what I should do with my memory. It's still at a 1:1 ratio. Am I supposed to overclock that now too see how high it goes? I guess the default settings would be 1066 5-5-5-18.

 

[jayfella]

A 1:1 ratio is always the fastest. For example: If i run my ram at 1:1 with my settings (in sig) the spped of my RAM will be 790MHz - underclocked by 10MHz - but if i change the divider to 5:6 - it will be 940MHZ - but in actual performance tests, the 1:1 always wins by a few marks. So, long and short, always try to keep it 1:1 - but if you can, lower the multiplier and increase the FSB, so it is still 1:1 but at a higher rating;

395 x 9.5 = 790MHz (mem) and 3752MHz CPU

OR

417 x 9 = 834MHz (mem) and 3753 CPU

Same CPU speed (there abouts) but higher memory speed.

 

[jason4207]

A 1:1 ratio is always the fastest. For example: If i run my ram at 1:1 with my settings (in sig) the spped of my RAM will be 790MHz - underclocked by 10MHz - but if i change the divider to 5:6 - it will be 940MHZ - but in actual performance tests, the 1:1 always wins by a few marks. So, long and short, always try to keep it 1:1 - but if you can, lower the multiplier and increase the FSB, so it is still 1:1 but at a higher rating;

395 x 9.5 = 790MHz (mem) and 3752MHz CPU

OR

417 x 9 = 834MHz (mem) and 3753 CPU

Same CPU speed (there abouts) but higher memory speed.

1:1 is not always the fastest. All of my testing has shown the complete opposite. You are correct that a higher FSB speed w/ a lower multi can give a bandwidth boost, though, but it sounds like Adaman is happy w/ his FSB setting.

IMO the best thing to do (once you're satisfied w/ your CPU/FSB OC) is to try each memory divider in turn (don't skip 1:1). At each memory divider you need to take the time to tighten up the timings (otherwise this level of tweaking is kind of pointless)...use memset to do this in real-time, and use Everest to test your bandwidth along the way. Run P95-large FFT for a minute or so to ensure you somewhat stable at each change.

Once you find the best timings for each memory divider you can compare the Everest bandwidth scores and choose the best. Also run memtest and OCCT2.0 custom 2hr RAM test to ensure your RAM is 100% stable.

Remember that all of this tweaking will give you a small boost, but it is nothing compared to more CPU speed. It is a lot of work for only a small gain, but it is nice knowing your RAM is doing all it can do.