SO CLOSE! Stability issues on FX8320 at 4.6Ghz

[ShrimpBrime]

Ah, well...now after about 40 minutes of prime95 I have CPU throttling. But my temps are at about 48. This just keeps getting better and better.

VRMs running a little warm perhaps? Actively cool them!

 

[fueloficarus]

VRMs running a little warm perhaps? Actively cool them!

I have a fan on them...and they're not very warm to the touch. Never was a problem before...

 

[fueloficarus]

I've adjusted multiple settings, still happening (max temp under load 55.5-56C) about two minutes into Prime95 (small FFTs). I've definitely disabled APM master mode, before anyone asks ;-)

Could it be my onboard chipset? The 970?

 

[fueloficarus]

(Posted to other thread I created when I thought this one had been deleted)

Guys! I think I may have got it. I'm running AIDA64 right now on a 4.4 OC and it's doing great. No throttling, temps have yet to exceed 43C (and comp room is hot as hell right now). I think there were two problems: I was not in performance mode (I effing forgot I upgraded to Win10Pro) in my Control Panel settings, ALSO I've done about 12 hours worth of reading.

It appears that this particular combination of board and processor HATES prime95 between 4.2 and 4.5. What I managed to figure out is that this problem isn't a result of prime95 or my board sucking, it had to do with the VCORE offset I had. I think I can boost to 4.6 just to see if I can do it (shouldn't be a problem), but I'll probably run 24/7 on this 4.4 I got. I did it with one RAM stick, I'm about to restart and try it with the other (different brands of RAM sticks, same CAS latency and timing strings though). Also, I ran out and got a 140mm fan to blow at my VRM and the 970 heatsink (the 970 heat sink was getting hot as balls, but in my experience they do regardless of OC or stock).

The point is, the offset and (in many cases) the CPU and CPU/NB Frequency can be modified to correct what was happening to me. It took a lot of tweaking (filled up all the profile slots in BIOS) to get me here, but it seems locked down.

 

[RJARRRPCGP]

More vcore. Your chips a ****ty clocker, and what are your load temps?

If folks were to think like the old school AMD days, then 8320s usually would be sucky OC'ers... (And 8350s better and 9590s the best) But we know that the 9590s are ridiculously ironic for having a tendency of clocking poorly...

And it's probably not the chipset causing problems. Because I saw a good OC with the same motherboard, IIRC! Looks more like a poor batch, more likely if an early one. (anything pre-2014) (and 2012s are possibly the hardest to OC)

It they're anything like the old school AMD days, then the first Visheras (2012) are likely real finicky...

I had a 2014 8350 with a low VID...

I would also try again with air from outside into the case...

 

[fueloficarus]

Nah, manufactured in 2015 in Germany. BUT! I got it 4.4 Stable, no temps exceeding 49C, using all 16 ram, and Prime95 and Aida64 didn't cause one lick of throttling! I've never used "lick" as a noun before...it didn't feel right. It didn't cause on MF'ING BIT of throttling! I'm on about 40 minutes of continuous testing now, but I'm about to boot down, restart, reactivate C'n'C and Core C6 (want to be able to put my comp to sleep) and play some TF2 and BF1 to push the RAM, then Dragon Age Inquisition and a few others to push it, but I think I got this.

IDLE:
Socket - 36C
MB - 34C
Core - 18-22C

LOAD:
Socket - 71C
MB - 34C
Core - 47C (peaked to 49 a couple times)

Below are my BIOS settings for stress testing. There are a few things that people don't often talk about changing in the "Beginner's OC guides" that I had to mess around with (NB voltage, for example).

 

[RJARRRPCGP]

I would try for more once you confirmed 4.4 being stable... Make sure you have good air flow on that rad!

Then you can fall back to 4.4 if you can't get it to budge...

I would use air from outside, because it looks like Asus AM3+ motherboards may be more finicky with temps than others and shut down at only about 65 C to about 70 C! Just a heads up!

Despite it shouldn't be triggered at that low of a temp... The shut down temp should be more like 80 C...

 

[fueloficarus]

I would try for more once you confirmed 4.4 being stable... Make sure you have good air flow on that rad!

Then you can fall back to 4.4 if you can't get it to budge...

I would use air from outside, because it looks like Asus AM3+ motherboards may be more finicky with temps than others and shut down at only about 65 C to about 70 C! Just a heads up!

Despite it shouldn't be triggered at that low of a temp... The shut down temp should be more like 80 C...

IRONY! I finally crashed when typing a response to you! I re-activated Cool n Quiet and Core c6 though, so maybe that did it. Also I was running 10 tabs on Chrome, Dragon Age Inquisition and about five programs (I was pushing it as hard as I could without rendering a video as well). Point is! Check this out (sorry for the messy setup, just got this working and now I must tuck the cables). That room is an outdoor storage area letting in cool air, and as you can see I have a 120mm fan pointed at my VRMs and my 970 heat sink. The 970 gets HOT, which isn't unusual I hear.

Tomorrow I'm going to clean it up and make a mount for that fan. I think I can fabricate a clip for the PCI port from some old bits and pieces that the fan can snap on and off of.

 

[Mr.Scott]

If folks were to think like the old school AMD days, then 8320s usually would be sucky OC'ers... (And 8350s better and 9590s the best) But we know that the 9590s are ridiculously ironic for having a tendency of clocking poorly...

Looks more like a poor batch, more likely if an early one. (anything pre-2014) (and 2012s are possibly the hardest to OC)

It they're anything like the old school AMD days, then the first Visheras (2012) are likely real finicky...

I had a 2014 8350 with a low VID...

The early chips were high leakage and usually OC'ed better.

My 8320
http://valid.x86.fr/j4r93y

 

[Mr.Scott]

I would use air from outside, because it looks like Asus AM3+ motherboards may be more finicky with temps than others and shut down at only about 65 C to about 70 C! Just a heads up!

Despite it shouldn't be triggered at that low of a temp... The shut down temp should be more like 80 C...

Max temp directly from AMD whitepapers says 62c. Anything above that does damage. That's why they shutdown where they do.

 

[RJARRRPCGP]

Max temp directly from AMD whitepapers says 62c. Anything above that does damage. That's why they shutdown where they do.

I wasn't talking about the inner core temp sensor. There's 2 sensors and the other one is good to about 80 C with a little OC...

And that sensor for me stayed cool. That wasn't the big problem... I usually had low readings, compared to the socket sensor, which is vice-versa from socket 462!
(The socket 462 sensors were a slow responding sensors that have trouble detecting sudden temp jumps) I did stop the shutdowns at a measly 4.4 by just sucking air from outside...

The Sabertooth 990 FX R 2.0 I used to have, hated room temp air!

 

[ShrimpBrime]

I wasn't talking about the inner core temp sensor. There's 2 sensors and the other one is good to about 80 C with a little OC...

From what I had experienced with AMD and Asus boards was that the socket temp while board was set auto all would throttle processor at only 65c the core temps never seem to matter unless you reach that thermtrip temp of 85-90c. I found this out after a de-lid or two where I could get the cores to usually run hotter than the socket temp.

The point is running these FX chips past 65c (socket/board) is likely to create instability while overclocking.

My de-lidded FX-9590 does not like temperatures past 50c while running beyond 5ghz.

However I have seen higher socket temps on some people's set ups, but if you look closely there was either no or very little overclocking taking place.

FX needs the cold and this ultimately keeps your v-core lower also.

 

[RJARRRPCGP]

From what I had experienced with AMD and Asus boards was that the socket temp while board was set auto all would throttle processor at only 65c the core temps never seem to matter unless you reach that thermtrip temp of 85-90c. I found this out after a de-lid or two where I could get the cores to usually run hotter than the socket temp.

The point is running these FX chips past 65c (socket/board) is likely to create instability while overclocking.

My de-lidded FX-9590 does not like temperatures past 50c while running beyond 5ghz.

However I have seen higher socket temps on some people's set ups, but if you look closely there was either no or very little overclocking taking place.

FX needs the cold and this ultimately keeps your v-core lower also.

I was frustrated, because of the shutdowns I had at only 4.4!

 

[Mr.Scott]

I wasn't talking about the inner core temp sensor. There's 2 sensors and the other one is good to about 80 C with a little OC...

And that sensor for me stayed cool. That wasn't the big problem... I usually had low readings, compared to the socket sensor, which is vice-versa from socket 462!
(The socket 462 sensors were a slow responding sensors that have trouble detecting sudden temp jumps) I did stop the shutdowns at a measly 4.4 by just sucking air from outside...

The Sabertooth 990 FX R 2.0 I used to have, hated room temp air!

There is only one sensor on die. The rest are on the board.