• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

***official amd 83x0 overclocking thread***

So appear to have taken a step forward.

Set my memory speed via DOCP at 2400MHz and timings are all correct.

Bumped the multi up and it's sitting at 4.8GHz at the moment. Set voltage to 1.5v for the time being just to make sure everything is stable and sweet. Will try higher core speed later on and then play with volts to see what is the lowest stable.

Ran prime95 for 30 mins and temps never went above 47 degrees.
 
Sweet setup James, very clean. Well done.

I ran the setup with Nvidia cards (GTX780s) so not 100% sure why the catalyst drivers are posing issues. I have to say my best overclocking and stability with this CPU/mobo combo was in win 7 ultimate 64 rather than Win 8.1 (tried both).
 
Damn it, but I like Windows 8.1 too much haha

With your 5GHz profile, was that prime stable or did you have a core drop out?

and what voltage did you use?

Could be a windows/driver error then I guess.

I thought you had a 7990 installed too?
 
Nothing at 5GHz seems to work for me.

Memory at 1600 instead of 2400, even upping the voltages to the limits has no effect, one core shuts down in prime or the system freezes.

I did notice that the total power drain was close to 500W when running prime, maybe a 1000W PSU isn't enough? (I have two, one for the 2nd 7990 and the other running the main 7990 and the system)

Update

FSB x 300 works, currently have 4.8GHz, 2400NB, 2400HT and 2400MHz memory.

Does timings effect performance with AMD cpus much? I've got 11-13-13-35 at the moment so it's not the best memory around.
 
Last edited:
The 5ghz profile in the bios I don't think was totally stable. I did have a totally stable version at 1.51v but it was in a previous bios rom (only used for benching). The 4.8ghz one was totally stable at much lower voltage. The bios version on the board (which was the latest one at the time) I felt wasn't the most stable but I switched to it as I wanted to try win 8.1.

You could check for any bios updates. If there aren't any I'd be tempted to roll back and flash a previous bios version and win7. I used 2133 memory with tighter timings. Also I found the system prefers only 2 memory slots occupied and max 4gb per dimm. Beyond that and it wasn't totally stable when heavily OCed and benching. I never use prime, instead only synthetic benches and gaming.

Good you've found the bus limit. I'd start with that while keeping everything else almost stock. Then take it back a couple of notches and start raising the multi.

I had a 7990 briefly but sold it in favour of 780s.
 
Last edited:
Okies cheers buddy

I updated the bios last night. I've tried 1.51v at 5GHz - stock 200 FSB and just adjusted the multi - and the system kept locking up.

What bench did you use just so I can compare?

Watching afterburner last night, High settings on a 64 player BF4 server the CPU doesn't reach 100% on all cores so i'm not being bottlenecked that way. Just the 7990 stays at around 70-80% usage too. I get over 100fps but was just wondering why this was (I have a bottleneck before with an i7) so i'm used to it.

Praying Mantle can improve things a lot
 
In my experience with this mobo/cpu combination a balance between FSB and multi is required for a stable OC. Try an FSB of around 210-215 and work on a multi to reach 5GHz. This usually also results in lower Vcore requirements. I'm assuming the digi+ settings are aggressive as you will need that.

In terms of BF4, what you are seeing may well just be poor quad-fire scaling. When BF4 first came out, I tested it with this setup and a 7990. Some of my results below. I also found GPU utilisation wasn't 100% but neither was CPU, so I'm inclined to believe there is a driver issue still. Mantle wasn't out at the time so I couldn't test it under that.

3D Mark Firestrike result with 1 x 7990:

http://www.3dmark.com/fs/1064530
Very similar physics score to you and when you get the CPU to 5GHz, should see close to 10k. Your graphics score is nicely above a single 7990 as I would expect. Again, being effectively a quadfire setup, I don't think it scales linearly from a single 7990.

3D Mark 11 result:
http://www.3dmark.com/3dm11/7406477

Bench summary:


BF4 CPU/GPU utilisation - soon after game launch (click twice for larger image):


BF4 CPU clock comparison with 7990 - BAKU level:


 
Also, just to note, in the CPU/GPU utilisation screen above, I ran the BF4 level with the 8320 at 5GHz using an FSB of 233, multi 21.5, Vcore 1.5V. This was totally stable.

Try those settings and see if it works. If I remember correctly, I couldn't replicate this easily on win 8.1 without bumping the voltage up.
 
JamesM said:
So with Windows 7 it does seem more stable but my combined score has dropped from 3500 something to 2400 something?

http://www.3dmark.com/fs/2101517
Click to expand...

Awesome GPU and physics scores though. Better than previously. I think Firestrike far prefers win 8.1 in terms of the combined test. You can see that, although you scored better in all the other tests, in the combined test under win 7, you were at 11 fps vs 16 in win 8.1. I wouldn't worry about this at all. Try 3D Mark 11 and see what you get and more importantly, games.
 
Well I played BF4 (14.4 drivers actually install in Windows 7) with quad fire and after a brief spell it was running smoothly at 1080p.

FPS drops were noticeable but I guess cause it's Mantle that will sort itself out.

System is getting there! Will get 3dmark 2011 installed in a sec

Thanks for your help Alex btw
 
Good job James. Something funny is going on in the benches with quad fire. Both fire strike and 3dmark 11 seem to show poorer combined scores even though your individual test scores are massive.

Could you disable your second 7990 via bios or catalyst and run the 2 benches again? That should give us direct comparisons. I suspect quadfire scaling is the issue.
 
You must log in or register to reply here.
Back
Top Bottom