• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

OC on MSI 990FXA-GD80

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

bulldog1963

Member
Joined
Aug 3, 2012
Okay, here's the deal, in the BIOS, the FSB and PCI-E bus speeds are fixed at 200 and 100, respectively. I can adjust Vcore, and multiplier for CPU but nothing for GPU at all. It seems this BIOS restriction happened when they introduced Bulldozers into the family and changed to a mouse-driven GUI Bios (v11.12 for me, AMI)

now, i can set fsb using MSI Control Center. This seems to always casue the bsod or just a plain reboot when i use this online tool.

Also, evga has a program to adjust GPU clock, memory clock and voltage. it is EVGA Precision X. This does nothing. The GPU clock runs from 50 Mhz to 700 Mhz depending on the load regardless whether I override it or not. I see no place to fix a GPU OC in place on the program nor is there anything I can find in th BIOS to shut of this clock throttle.

Anybody have experience with is problem and how did you get by it. I know it can be done in AMIBIO8, but the new bios (uefi) that had to be upgraded to used bulldozer's was changed from an all-cpable text based BIOS to what seems to be a more user friendly and "pretty" hobbled BIOS. I keep reading always do your OC via BIOS. I agree.
 
Oh yeah, 1 thing to add. I can get my CPU (FX-8120) up from 3.1 base to 4.2. It runs EVERY application and onlime game I throw at is without issue. I can run Passmark and a few other bench's, but 3d mark 11 locks up on me about 90% the way through. As for stress tests, the Passmark runs fine. Every other one shuts down core by core - starting with 8, then 7, then whatever after a few minutes.

Now, to me the capability to run every program that I ever do (and some are quite demanding of CPU and GPU) without fail or "hiccup" is the MOST important thing to me. Getting a stress test to run is, truly secondary. It would be nice to see it through the AMDO or Prime95, or OCCT process, but I cant seem to even get them to run beyon stock speed. Even .1 oc locks this stuff up, except Passmark BurnIn Test.
 
but nothing for GPU at all. It seems this BIOS restriction happened when they introduced Bulldozers into the family and changed to a mouse-driven GUI Bios (v11.12 for me, AMI) =There is absolutely no reason for any GPU adjustments in the bios of a 990FX chipset motherboard, because it does n0t have onboard video. Therefore no GPU adjustments in bios.
 
So I guess the PCI-e speed at 100 Mhz is probably I/O port bus, then. I wasn't considering that. Thanks for the nudge.
 
Yeah, you really can't overlock the PCI-e bus. It just don't like it. Guaranteed to produce instability. Sounds like you are using software or bios overclocking genies. We would really recommend using the manual bios controls to get a truly significant and stable overclock, one that will pass Prime95. Wish you had chosen another board manufacturer. We see more problems on the forum from MSI owners than any other major brand.
 
Well, I can and do use manual. I tried the genies and all they did ws give me a bump of about .23. I use the multiplier (successful at getting me to run everyday tasks and some quite stressful) at 4.2, but when I try to stabilize it for stress testing at 4.2, i just can't find a vcore that can help me out. Sometimes i will look at HWinfo64 or CoreTemp and the vcore will have dropped back, on its own, to about 1.28.

I have my ram boosted to 1600, but it naturally runs 1333. should i drop it back to 1333 and is raising vcore the only way to stabilize or can i drop it from the 1.28? Although I have built computers since, like, '92, I never had the urge to OC. Or water cool, but I just tried my first kit and I have my Cores down to 11 C/26 c with offset at 4.2. I am pretty excited about that.
 
I did try a Gigabyte 970a-something, but circumstances forced me to get another board.
 
FX CPUs will easily run 1866 mhz ram so if your ram is capable of 1600 mhz I would run it that way. Sounds like you may have temp-related downthrottling going on. The FX CPUs and the 9xx chipset boards have a very complex power-regualtion system that can be triggered by high socket/core temps or excessive power draw. It can be disabled, however. Have you disabled Cool N Quiet, C1E, C6, Turbo Boost (or whatever they may call it) and Advanced Power Management in bios? You should if your are overclocking because those things interfere with overclocking. And go into Windows Control Panel Power Options and configure it to High Performance. That should take care of all the "green" donwnthrottling stuff.

But something is conspicuously absent from the info you have presented so far. You have said nothing about temps under load. High temps are a major cause of instability. You should donwnload and install HWMonitor. Have it open on the desktop while you run some of your tougher apps or games. It will record min, current and max core and CPU socket temps. The socket temp may be obscured by a label like TMPINx of which there are seveal lines. Do this and get back to us with a report about those two temp readings.
 
Last edited:
Okay, this is what I have. I was using as much of the system resources that I typically use. The Graphics often hover at 77-88% and cpu runs from 22-45-ish% There are graphics levels to 98% in some instances for about 15-30 seconds (depending). I have seen the CPU go to about 50% for short times (Using Core temp and HWinfo64, both) and the use of all 8 cores while I do this is pretty much the norm.

Hope this works. I had a helluva time capping this. I have dual screens and it kept capping the main screen even though i had the secondary screen and the HWMonitor window active.

30l17hk.jpg
 
On the temps, they are a bit higher than I display. I look at the onboard temp sensors.

Fintek F71889a for the CPU temp and ADT7473 for the GPU temp. it looks like HWMonitor uses the same one for the CPU and motherboard and NVAPI for the GPU. The NVAPI displays temps about 11 degrees hotter than the onboard chip reports.
 
You can add about 15c to the core temps with the FX CPUs to get a more realistic core temp reading as they are typically micalculated on the cool side.
 
At 1.28 your vcore is too low. No wonder your overclock is not stable.
 
Back