Overclocking 8370

[jaxapollo]

Hello! I just installed a water cooler and was wondering what the ideal voltage and OC would be for my rig. Right now I'm at 4.3 but I'm new at this so I don't want to mess it up..

Specs:
AMD FX-8370
GeForce 980
msi 970a-g45
corsair h80i
650W PSU.

I ran 95 and passed .. These are the screens.





Thanks in advance for all of the assistance. I am excited to be a part of this community.

 

[trents]

No one can answer your questions as such because every CPU will be different, even from the same model. Variations in the quality of the silicon used for a particular production run plus variations in the manufacturing process are responsible for that. All CPUs (or at least samples from every production batch) are tested at factory and as long as the variations in them are within a certain envelope they will be sold as a certain model. If they fall outside of that envelope they will be sold as a different model. This is called "binning." So there is going to be variation between individual CPUs even when being sold as a certain model like your FX-8370.

But then as far as overclocking goes there are other variables outside the silcon it self. Like: 1. Motherboard quality (are the power producing components on the board weak or strong?) and cooling capacity. The amount and speed of RAM also influence this equation as does the "cleanness" of the power being supplied to the motherboard by the PSU.

Finding what the ideal voltage is will need to be done by experimentation. Keeping the package and CPU socket temps within safe and stable limits is more important than actual voltage. Don't allow package temps to exceed 60-65 and CPU socket temps to exceed 70c.

The motherboard you have chosen is not a good match for overclocking the 8 core FX CPUs. MSI boards for the FX CPU just don't have strong enough power producing components and we hear of many failures of MSI boards running overclocked FX eight cores.

Your Corsair H80i will also limit the overclocking of that CPU but the board may be the greatest limiting factor.

What is the make and wattage of your PSU?

 

[Bluefalcon13]

Everything Trents said is spot on. Just some food for thought though. My H80 was running outta gas at 4.5ghz. again, it's all CPU and board dependant.

 

[jaxapollo]

Thanks for the thoughtful responses. I really appreciate it. The PSU is an Antec Earthwatts Green 650W.

I did overclock a little bit to get to 4.3.. I used Overdrive to do so. I think I did it right but I'm confused on how to tell which temps are the package temps and which are socket..

See the issue arised when I got the HTC Vive. Everything was working great with my 770 GTX, but when I upgraded to a 980 I started to get some significant performance issues. when I brought up the graphs it showed that when the picture began to stutter, the CPU would turn red, indicating "Slow Start".. So I checked on the temps and they would shoot up when this would happen, so I replaced the stock cooler which was very loud with the h80i.. It worked really well at first, then it started to do the same thing. so I thought, maybe I should overclock a bit, since I've heard this CPU is good for Overclocking

Now I'm unsure of what to do next. Feeling a bit discouraged.

 

[Bluefalcon13]

First AOD (amd overdrive) is about the worst way to overclock. Take a look at the sticky and familiarize yourself with the Bulldozer overclocking guide. One you get that read, ask questions here and goto bios take a look at what your options are in there. Ask more questions about the options in bios. Since most of us do not have your board, it may be helpful to screen shot your bios and post the pics (most bios' support saving screen shots to a fat32 formated flash drive)

 

[trents]

To capture screen shots of the bios to a flash drive as Bluefalcon13 suggested, most bioses use he F12 key for that. It will only capture what you can see on the display so you will have to scroll down to get the whole page of settings for each tab. Use the attachment tool I have circled to add the pics to your posst. You may have to downsize the pics as there is a file size limit on each attachment built into the forum. Irfanview is a great free program for manipulating images in this fashion.

If you cannot afford a beefier motherboard, you really do need to look at putting some small "spot" fans to blow in the VRM section of the motherboard and on the backside of the socket.

The picture stutter you describe and the CPU "turning red" (I assume you mean the temp bar graph of the CPU cores in AMDOverdrive) probably means the CPU is going into thermal throttle. AMDOverdrive can be a good supplemental tool for experimenting with bios changes without having to reboot every time to enter bios. We do not recommend it as the primary method for overclocking, however. You should be using the bios at least for any final values you decide on in your overclocking. AMDOD uses resources and sometime causes instability itself.

AMDODrive reports the temp room left before hitting "TJ max" which is the point at which thermal throttling takes place. In HWMonitor, TMPIN1 is probably your socket temp at a max of 75c. That is definitely borderline. You need to cool that down. Unfortunately, HWMonitor is not correctly reporting your package temp. Those values can't be true. So you might have to use AMDOD to monitor room to TJMax when you are doing something stressful. Or, MSI may offer a custom hardware monitoring utility that will work with your motherboard. Check the motherboard installation disk.

As to the problem going away when you installed the H80i and then returning, have their been changes in the weather that would cause the ambient (room) temperature to increase?

 

[Bluefalcon13]

As a suggestion:. I have always had weird instabilities using AOD for adjusting voltages. I do use AOD for dialing in though. Basic algorithm:
1. Hard set everything to stock, disable turbo.
2. Boot into Windows
3. Launch AOD
4. run Prime 95 blend for 15-20 mins
5. If temps are good and Prime does not fail, bump multiplier 1 notch in AOD.
6. Repeat 4 and 5 until Prime 95 fails (bsod, thread fails, etc)
7. Boot to BIOS, hard set multiplier, bump vcore a notch.
8. Repeat 4-7 until desired clockspeed OR temp limit reached (60C is a nice safe max)
9. Run Prime 95 blend test for 2 hours. If fails you need more vcore.

This is a VERY basic algorithm. It assumes LLC is coving you, and all temps are nominal. The key is starting with a baseline at stock. If you aren't stable at stock, or if your temps are too high at stock, that needs to be fixed first.

 

[jaxapollo]

Actually, the red area is in the SteamVR benchmark tools. I'll try to take a screen capture.

Thanks for the assistance, everyone. I'll attempt to take a crack at it shortly and post screenshots.

 

[trents]

Pretty much the same approach as I use, BF13. Good advice, I think.

- - - Updated - - -

Steam VR? Video Rendering?

 

[jaxapollo]

Virtual Reality. Most games work without a hitch, but a select few work perfectly until they don't, when that happens the chart indicates a red "Slow Start" on the CPU and temps shoot up.

I'll try to take a screenshot.

 

[trents]

Thanks for the clarification. That could also be a software issue on Steam's side. But you might look at disabling Cool N Quiet and other C states in bios that cause the CPU to throttle down when not under load. IF the problem is with Steam I can't help you there but you would do well to focus on your own hardware and overlocking.

 

[jaxapollo]

Sorry for the late reply, been really busy with work. But here are the screenshots from the BIOS. If I could get it to even 4.2 I'd be really happy .. I think a small boost is all I need at this point.



OC genie


Thanks again for all the help everyone!!!

 

[Bluefalcon13]

I don't think you are gonna see the results you want. I just took a look at your board specs. You have a 4+1 VRM section. First you need to get a fan on your VRM sink. That may not be enough to keep from thermal throttling. I would be looking into a new board.

 

[trents]

Ditto

 

[turbobooster196]

No one can answer your questions as such because every CPU will be different, even from the same model. Variations in the quality of the silicon used for a particular production run plus variations in the manufacturing process are responsible for that. All CPUs (or at least samples from every production batch) are tested at factory and as long as the variations in them are within a certain envelope they will be sold as a certain model. If they fall outside of that envelope they will be sold as a different model. This is called "binning." So there is going to be variation between individual CPUs even when being sold as a certain model like your FX-8370.

But then as far as overclocking goes there are other variables outside the silcon it self. Like: 1. Motherboard quality (are the power producing components on the board weak or strong?) and cooling capacity. The amount and speed of RAM also influence this equation as does the "cleanness" of the power being supplied to the motherboard by the PSU.

Finding what the ideal voltage is will need to be done by experimentation. Keeping the package and CPU socket temps within safe and stable limits is more important than actual voltage. Don't allow package temps to exceed 60-65 and CPU socket temps to exceed 70c.

The motherboard you have chosen is not a good match for overclocking the 8 core FX CPUs. MSI boards for the FX CPU just don't have strong enough power producing components and we hear of many failures of MSI boards running overclocked FX eight cores.

Your Corsair H80i will also limit the overclocking of that CPU but the board may be the greatest limiting factor.

What is the make and wattage of your PSU?

oke but looking to the program uesed above for looking at the temps, what is the temp to watch then.

the pacakage temp is the same as the core temp, but what is the socket temp

 

[trents]

Probably the closest one is the socket temp, which appears to be TMPIN1 on your motherboard. It's really the VRM temp that is the concern but there usually isn't a separate temp probe or readout for that one on most boards.

 

[turbobooster196]

Probably the closest one is the socket temp, which appears to be TMPIN1 on your motherboard. It's really the VRM temp that is the concern but there usually isn't a separate temp probe or readout for that one on most boards.

i know the vrms heat is mostley the problem, but thats i thought tmpin1 and i thought that would be no problem going over 70.