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

Another FX-8350 on a Saberkitty

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

Tetrall

Registered
Joined
Jun 27, 2014
Location
Alaska
Hey guys!

Seems like as good as time as any to start working on understanding the overclocking business a little better, here's my baseline run: no overclock, power saving options disabled, turbo disabled starting off at 4.0 Mhz for a baseline. Anything raise a red flag?

20 minutes of Prime95 Blend:

first run.jpg

P95blend20.PNG
 
NO red flags that I see on first inspection. Bump cpu multiplier up by 0.5 and run P95 Blend for 20 mins with HWMonitor open logging min/max temps and volts. Always good when reporting how it goes to post CPUz > Cpu TAB, Memory TAB and at least a couple of times the SPD TAB and put the HWMonitor capture there as well. NoT necessary to see P95 Blend Image. You know if it passed or failed.

You seem to have one of those LOW P-State cpus by the Vcore showing as used at 4.0Ghz. That has meant the cpu speed will go to a certain level and then quit NO matter how much voltage you feed the cpu. Time will tell but that is one thing that I do see.

So give it a multiplier bump by the 0.5 and run your P95 and get your captures for attaching. WE do not have to see each 0.5 bump up. Only the last one that does not fail P95 Blend mode, which is the one before the P95 Failure as you move upwards in cpu speed. Other tweaks can be taken care of after you find a pretty fair cpu speed.

Welcome man.:welcome:
RGone...ster
 
Ok... did fine at stock voltage at 4.1Mhz but failed at 4.2Mhz. I upped the voltage by .0625 with the CPU offset and running another round of prime95 blend for 20 minutes.

4.2 +0.625volts stable.jpg

The results of .0625 additional volts definitely increased the temp on my socket, I wonder if I'll be able to push it much more than this due to the fact it's only P95 blend and not Small.

Off to the vets! Pup has an appointment.
 
Last edited:
A bunch of us put fans on the VRM heatsink and also mount a fan on the backside of the motherboard. This will help get the Socket temp closer to the package temp.
 
So everything has been really stable... up to 4.7 MHz I only needed .0625 additional voltage for it to be stable on prime... for both small FFT and blend, at least 20 minutes.

It also was giving noticable returns on my 3DBenchmark 11 scores, I saw the biggest gains from 4.4 to 4.5 as I went from an overall score of 11411 to 11524, however from 4.5 to 4.6 it was only about 11538, after that I expected to see some need for more voltage.

It failed right out of the gate on 4.7MHz and I added some voltage (+.00625X3 coming out to 1.37 something) at load but didn't want to go too high, but I'm stopping here until I get some advice. 4.6 seems rock solid stable... at almost stock voltage too.

Here's a couple questions on voltage:

1. Should I leave most of my other voltage settings on "auto" or switch them to baseline? I've heard of people doing this but I don't know the reason.
2. CPU load line: I've heard that setting this to very high or extreme helps... any danger in that?
3. I've disable all the power saving stuff. AMP, Turbo, Core6, CoolNQuiet.
 
Last edited:
Alright - here's the data from 1.380V, appears to be stable, 20M of Pr, do I still need all the renditions of CPUZ or is that just for the start?

4.7 MHz 1.38V load.jpg

Thanks guys, this is a lot of fun :D, much more heat on the socket as you can see at 63C... gonna need that fan on those VRMs and possibly on the backside if I want to go much higher like you've been saying.

Oh, back when I was first trying to overclock here last month, I saw MUCH higher temps even at stock on my cores... I think the Kraken X61's stock thermal paste is one that needs to actually bake, that and my stuff is starting to get broken in.

Oh, so far, the only thing I've been messing with is the multiplier... I'll take it back down here eventually as I've seen others do on here and mess with the HT and NB for some fun.
 
Tetrall, looks good so far, take into account that 20 minutes prime is only tentative stability. Most of us will only call our system stable is we can run at least 2 hours of Prime Blend. If you're able to pass two hours at that voltage, it would be one heck of a chip. Additionally seeing the memory speed tab would be helpful, so we can see where the Nb frequency is set.
 
yeah, I have seen that on several of these, I don't think that number is right... none of my other monitoring software registers anything even close to that. Anyone else care to weigh in?

Into my second hour of P95 blend... no failure, and the socket is holding under 65C... my cut off is 67C right now due to the fact they say you shouldn't let the socket get above 70C... just in case the system is a touch off.
 
Well, there it is, 2 hrs 2 minutes of P95 Blend, stable.

Now, I can't go any higher than this because at this point my VRMs are getting quite toasty. Two more fans will be coming shortly... any other known way of getting my temps down guys?

2 hours 4.7 MHz p95 blend stable.jpg
 
Well, there it is, 2 hrs 2 minutes of P95 Blend, stable.

Now, I can't go any higher than this because at this point my VRMs are getting quite toasty. How do you know this or what is saying to you that VRMs are toasty.?

Two more fans will be coming shortly...Fans for mounting where?
any other known way of getting my temps down guys?MORE cooling or lower the speed; which can reduce the power needed to run the higher speed.

RGone...
 
Haven't uninstalled the AI Suite II yet... I know, it causes instability issues but I just haven't gotten around to removing it yet... still like seeing all the temps and having the alerts, it's on my list of things to do. That's the one that gave me the 80C temp warning on the Vcore1.

One fan for the heatsinks on the Vcores to give it a little extra cooling which I've seen on this website on other "help me overclock this CPU" threads. I only figure I'll actually need it for when I'm running prime though as for daily use the Vcores have never gotten very hot.


The second I'm planning on mounting on the backside of the mobo over the socket, again found it on this site and it seemed an interesting suggestion and I wanted to try it out.
 
The only way to lower Vcore1 is to lessen the heat at the VRMs or lessen the draw thru the VRMs by lowering the clock so you can lower the Vcore or use a lesser CPU_LLC. That temp comes from an Asus only sensor not normal for most AM3+ motherboards. That sensor is not read/reported by other software either, so you cannot validate against other software.

Also the overflow heat from the cpu socket can influence heat registered by the Vcore1 sensor. So more cooling or less load is about it.

In addition if Vcore1 is not exceeding 80c there is no real need for concern. 80c is at least 10c below the place where worry should increase. If P95 Blend mode or equivalent is the only thing that has pushed Vcore1 to nearly 80c then most other normal applications will never approach that high range. And since I know this...I never use Ai Suite on my CHV since I do not like the other issues of reporting that I have experienced. YMMV. Your desires may vary.
RGone...
 
Well, there it is, 2 hrs 2 minutes of P95 Blend, stable.

Now, I can't go any higher than this because at this point my VRMs are getting quite toasty. Two more fans will be coming shortly... any other known way of getting my temps down guys?

View attachment 148829
Wowzers Tetrall, 2 hours at 1.38 max Cpu V is a pretty darn good OC. I'd love to see the P-States on that chip.
 
RGone... said:
You seem to have one of those LOW P-State cpus by the Vcore showing as used at 4.0Ghz. That has meant the cpu speed will go to a certain level and then quit NO matter how much voltage you feed the cpu. Time will tell but that is one thing that I do see.

C_D had one of those things with a x20 =1.29xxV since I don't remember the last 2 of that voltage, but it would not budge beyond 4.8Ghz no matter how much Vcore he put on it. And you know that C_D is not voltage shy. Hehehe.
RGone....
 
C_D had one of those things with a x20 =1.29xxV since I don't remember the last 2 of that voltage, but it would not budge beyond 4.8Ghz no matter how much Vcore he put on it. And you know that C_D is not voltage shy. Hehehe.
RGone....
I was thinking the same RGone but I guess well see?!?!
 
yea, please have him post the reported p states, I really want to see one with 20x and lower than 1.3 get into the mid 5 ghz range.
 
ok - so I'm not sure exactly where to find the P-states or even what they are. I'd be happy to post them though if you tell me where to look for them!
 
Back