[mabaile2]

So I've been playing with my FX-9370 for a while trying to get it beyond stock clock and I've had no luck, I've posted here, Reddit pcmasterrace, and Reddit overclocking. So at first I found it weird that changing my voltage didn't actually change anything, which I still find really weird, but my voltage would only change when I changed my LLC. After a few posts someone from the overclocking Reddit added me on Skype and we tried for almost 2 hours to get it to work but we couldn't. This is where it gets weird it would finally boot and run p95 at 4.7ghz but would throttle down to 1.4ghz, which doesn't sound to bad other than the fact that it idles at 4.7ghz and jumps between 4.7ghz and 1.4ghz after a couple minutes in p95. I honestly have no clue what to do now since changing my LLC didn't help, setting it to extreme which gave the most voltage (up to 1.512v) caused an instant crash and the lowest setting which turned out to be low/standard and not the low setting only gave 1.41v which also caused a near instant crash. In my bios no matter what I set my vcore to it never exceeds 1.48v in HWMonitor during p95 testing, unless LLC is set to extreme in which case it crashes. Anyone have any idea whats causing this? In CPUZ my CPU multiplier shows up as 7-23.5 when it is set to 23.5 in the bios.

Bios settings:

http://i.imgur.com/I7BX487.jpg (APM and HPC are now disabled)
http://i.imgur.com/PuRw2BO.jpg (stock voltage settings yet vcore is still 1.48 or lower in p95)
http://i.imgur.com/B1Rs1IZ.jpg (ignore the NB core settings I had just changed and then reverted it to auto before even testing)
http://i.imgur.com/oWGw0S9.jpg (after saving and loading the above settings this is the vcore reading I got)

Setup:

CPU: AMD FX-9370
MOBO: Gigabyte 990FXA-UD3 rev. 4
PSU: (OCZ) FirePower ModXStream Pro 700MXSP 700W 80Plus
GPU: XFX R9 390
RAM: Corsair Vengeance Pro 16gb 1866mhz
CPU Cooler: Cooler Master Seidon 240m

Honestly I think its a mobo problem but I'm not sure. This board also had an issue with the newest F3 bios where it would only boot if my drives were set to IDE, I have since reverted to the F2 bios and no longer have that problem. Would something like this be worth trying to contact Gigabyte? Also if all of this is something I can't really change or would be a long difficult process I may just wait it out and switch to Zen when it comes out but I would still like to get this thing OCed and get a good daily OC setup. Thanks all.

 

[p4inkill3r]

It sounds like you're throttling down due to VRM overheating possibly.
Apply spot cooling via a small fan directly to the VRMs and check temps using hwinfo64.

Download IBT AVX also, run it on very high. http://www.overclock.net/attachments/13202

 

[mabaile2]

Yeah the vrm overheating is the only thing I could think of but it still bothers me that the bios had boot issues as well since I couldn't find others having the same issue. Any idea why my voltage won't change no matter the setting?

 

[The Sandman]

What do you have for a cooling solution? This is always a huge deal with a 9xxx chip!
You haven't listed any temps and this is my 1st suspect.

What you're seeing is most likely the VRM's causing throttling due to high temps, giga's are known for this.
I've read about buggy, hard coded bios causing early throttling but not exactly sure which mobo's are more prone to this.
Perhaps a couple giga owners will chime in on this part.

You'll definitely want to have active cooling over the VRM's and CPU Socket (both sides works best)
I do, and my rig has CPU, VRM/NB, and GPU are under H20.

If you don't have any active cooling best get some quick.
Not dising giga (too bad) but the only two mobo's I'd run a 9xxx series in would be either a Saborkitty or CHV-Z, especially if I wanted to OC it.

You might want to read this too http://www.overclock.net/t/1258253/how-to-put-your-rig-in-your-sig as it will help those trying to help you.

Fill us in with this info for starters.

also disable APM and "maybe" CPU Core Control (set at least to a manual setting?) (1st pic)
Also seeing to many settings on auto. Best to manually enter as much info (even stock) as possible. (2nd pic).

 

[p4inkill3r]

Quote:

Originally Posted by mabaile2

Yeah the vrm overheating is the only thing I could think of but it still bothers me that the bios had boot issues as well since I couldn't find others having the same issue. Any idea why my voltage won't change no matter the setting?

The motherboard you have can handle the 9370, and you need to download hwinfo64 or CPU-Z and monitor the voltages during load.

enter BIOS, set to optimized defaults
go to advanced frequency settings
RAM to XMP
Go to advanced cpu core features
CPU clock ratio to 22
cool & quiet, c1e support, c6 state, APM disabled
HPC mode enabled
go to advanced voltage settings
CPU vcore to 1.50v
vcore loadline calibration to medium

You are going to need to actively cool the VRMs and the CPU socket on the back of the motherboard for optimal performance as well.

 

[mabaile2]

How would you recommend cooling the socket and vrm without it just looking flat out terrible? As long as I don't have to spend to much I can put pretty much put whatever I need to together so that's not an issue.

 

[p4inkill3r]

The fan that comes on the AMD stock cooler works great for me; small, black, and PWM controllable.

 

[mabaile2]

I'll have to go dig out my old cooler from my 960T because I got the 9370 that came with no fans. What's a good way to mount the fan over the vrm that will be fairly stable?

 

[p4inkill3r]

Good old fashioned tie straps work, although it is possible to lodge the fan directly between your cooler and the heatsink and have it be fairly secure.

 

[mabaile2]

Alright I'll work on that either tonight if I can find some zip ties or tomorrow if I can't I'll also stress test and upload a picture of the temps before and after adding the fan.

 

[mabaile2]

Looks like I may be delayed in adding a fan and posting the temps of the vrm with a fan because I cannot find my old stock heatsink anywhere so I'll have to get a fan sometime to mount there because all I have is a couple 230mm fans laying around.

 

[cainy1991]

Cooling the VRM is a MUST with the UD3... honestly I'm kinda surprised it hasn't blown out yet..
I found two 80mm high rpm fans to do an okay job that being said, my extreme 4 with no additional fans doesn't heat up as much
as my UD3s did with the additional fans.

 

[mabaile2]

I just tested making only 2 passes because I saw just how hot it was getting but I still wanted to let it go long enough to get fully warmed up.

First image is at full load at 4.4ghz 1.5v

http://i.imgur.com/ul2sxXG.png

Second image is all the same settings but the CPU started to throttle

http://i.imgur.com/914iaKp.png

I don't know much about VRM temps but I'm assuming in the images I linked the VRM temp is the VR T1 and T2 which puts it at ~103c which I don't need to know much about VRM temps to know that it is way to high and needs to be cooled somehow.

 

[AcEsSalvation]

IIRC, 105°C is when silicon starts to get damaged. Your system is overheating so bad right now, I would be surprised if there isn't already permanent damage to your system.

 

[mabaile2]

Yeah that's what I figured that's why I kept it up long enough to get images then closed it asap. Its rarely under that kind of load on daily use, most games I play are much more GPU intensive vs CPU intensive so my CPU very rarely crosses 50% unless I'm stress testing, even recording (or streaming) with OBS while playing my usage doesn't cross 50%. If there is damage as long as it can hold out til Zen or until I can get something else later this year that will be alright. I haven't seen any signs of damage physical, or performance, and like I said my temps getting that high is a fairly rare occurrence. At idle the temps are around 66c so much more reasonable but still way to hot for idle.

 

[AcEsSalvation]

66°C is close the the absolute maximum your chip should run at. Also, damage may not be directly noticeable. Over time you could find your system starting to have random issues because your CPU is failing to perform properly from degradation (which is a reality at this voltage as well). Also, your VRMs could fail at any moment, which could kill your CPU too.

I would highly advise on downclocking and under volting for the time being.

 

[mabaile2]

Honestly it should have failed by now if its that bad because its been running like this for a little over 2 years.

 

[Dhoulmagus]

You have to remember that the FX-9370 is a 220W chip factory default. Pushing 1.4+ V through an fx 83xx / 93xx is a massive load for the VRMs. I use that same board with an overclocked 4350 and I have a side panel 200mm fan mounted straight over it blowing out 150+ CFM, this is in a CM haf set with a single GPU and some serious fan power for positive airflow (another high power 200mm in the front, still running scythe ultra kaze 3krpm fans on the bottom, top rear, and two up top to make the thing a wind tunnel.

I mention my set up because I want to see what you have for case and airflow, there needs to be a large volume of air cooling that board or you're going to lose the VRM. I guarantee you it's throttling due to either the VRM or socket temperature, 99% likely it's the VRM.
Quote:

Originally Posted by mabaile2

Honestly it should have failed by now if its that bad because its been running like this for a little over 2 years.

Not necessarily, think of how capacitors are rated, a range of hours relative to operating temperature, the higher the operating temperature the shorter the rated life. Just because the VRM hasn't died a firey death yet doesn't mean it won't in the near future. and now you're trying to overclock which is going to put significantly more load through the VRM especially when you're running prime95. Take our word for it, I have the exact board, those VRMs cook even with a 4350 without nice airflow, it's just the nature of the power load going through them, these aren't ~88W TDP i7s we're dealing with. The 9370 is already a heavily overclocked 83xx piledriver CPU.

Also, it would hurt to run an h264 stress test instead of prime95 while VRM heat is a concern, p95 is going to bring your CPU significantly beyond any real world usage, I don't bother with it on my ud3.

 

[mabaile2]

I had actually just changed the side of my Thor v2 to have a side window instead of the mesh with the fan and I checked my temps, except VRM because I didn't even think of it being an issue, and had almost no change at all. Guess its time to take the window off and put the mesh back which sucks because I just ordered about 16.5' of white led strip lighting. Guess I'll just use the leds somewhere else until I change out my CPU and motherboard sometime this year and hopefully have a sizable drop in temp. Well my little lexan window lasted about half a week, you will be missed little guy. On a more serious not though what push pull configuration would you recommend for fans? I have a 140mm in the upper rear, a 230mm red led fan in the front, a 240mm radiator with two fans blowing out the top, and when I put it back on I will have another 230mm fan on the side directly over the motherboard area.

 

[Dhoulmagus]

Quote:

Originally Posted by mabaile2

I had actually just changed the side of my Thor v2 to have a side window instead of the mesh with the fan and I checked my temps, except VRM because I didn't even think of it being an issue, and had almost no change at all. Guess its time to take the window off and put the mesh back which sucks because I just ordered about 16.5' of white led strip lighting. Guess I'll just use the leds somewhere else until I change out my CPU and motherboard sometime this year and hopefully have a sizable drop in temp. Well my little lexan window lasted about half a week, you will be missed little guy. On a more serious not though what push pull configuration would you recommend for fans? I have a 140mm in the upper rear, a 230mm red led fan in the front, a 240mm radiator with two fans blowing out the top, and when I put it back on I will have another 230mm fan on the side directly over the motherboard area.

You could still get away with your side panel window on, even 40-60-80mm fans zip tied to the VRM will be adequate as long as you beef up your airflow and they don't have to look bad, flush against the VRM they actually can look pretty nice. In fact from my own testing I find that having the side panel fan creates a turbulence inside the case and I would probably actually see a benefit from just tying 40mm fans straight to the vrm heatsink because the airflow would stabilize.

Looked up the case you mentioned really quick.

Advanced Cooling System:
-Front 1 x 230mm Red LED Fan (pre-installed)
-Rear 1 x 140mm Fan (pre-installed)
-Top 1 x 230mm Fan (pre-installed) or 2 x 120 / 140mm Fan (option)
-Side 1 x 230mm Fan (pre-installed) or 4 x 120mm Fan (option)
-Bottom 1 x 120 / 140mm Fan (option)

Personally I like to replace 200+mm fans with 120s because typically 120s will move the most air. If this were my case I would see about getting 2x120mm in the front (if possible, it doesn't say), a 120 on the bottom (with the case not on the floor), 2x120mm on the top, 1x120 on the rear. Higher CFM the better for this kind of setup, probably want something pretty beefy with a fan controller so you can crank them when you're loading the system with prime95. Do this with a smaller fan tied flush to the vrm heat sink and you'll be set, even with the window panel still on and you can LED all you want