Overclocking an FX6100

genthug

Honorable
FX 6100
Asus M5A99X EVO
12 GB DDR3
Zalman CNPS9500

I'm currently working on OCing my cpu. I've got it up to 4.3GHz right now air cooled, however the temps are nearing what I believe (and will be questioning about) are thermal limits. Currently (as I'm running Prime95) the highest temps that Open Hardware Monitor and Speed Fan have detected on my mobo socket are 66 degrees, the max on my CPU cores is 64.5 degrees. I'm not a huge fan of those temps, as I would enjoy keeping it under or around 60, but if those are still under the thermal limit, I shall not complain. The computer is three years old, after all.
Now for questions: as a bit more relevant info: my CPU multiplier is currently sitting at 17. My FSB is sitting at 253.8 (per CPU-Z), or 253 (per BIOS), 1.3875 CPU voltage
First question: How are those temps for an air cooled system and how much I'm OCing?
Second question: Does changing the multiplier increase thermals? I had it set to 253, 16.5(stock) and was running at around 60. I changed to 17, and now my temps are about 5 degrees warmer. Coincidence or?
Third question: Probably up for debate, but what is the best way to OC? I've read that using the FSB is more reliable, but that the best way to do it is by using both the CPU multiplier and the FSB. True?
 
Solution
1. What is your cooler model?

2. 65 degrees should be the max you want to see in Prime with that chip, and make sure you are running version 26.6 ONLY. No newer version. Newer versions run AVX instructions which create unrealistic thermal conditions.

http://windows-downloads-center.blogspot.com/2011/04/prime95-266.html

3. 4.3Ghz is extremely good for an FX-6100.

4. Increasing the multiplier does not increase thermals. Increasing the multi without increasing voltage creates, or can create, instability. Increasing voltage increases thermals. Since you're already a full 1Ghz over the stock configuration, I'd be happy with that. Be sure to disable turbo core. You don't want turbo behavior on an overclocked CPU. I'd probably also leave...
1. What is your cooler model?

2. 65 degrees should be the max you want to see in Prime with that chip, and make sure you are running version 26.6 ONLY. No newer version. Newer versions run AVX instructions which create unrealistic thermal conditions.

http://windows-downloads-center.blogspot.com/2011/04/prime95-266.html

3. 4.3Ghz is extremely good for an FX-6100.

4. Increasing the multiplier does not increase thermals. Increasing the multi without increasing voltage creates, or can create, instability. Increasing voltage increases thermals. Since you're already a full 1Ghz over the stock configuration, I'd be happy with that. Be sure to disable turbo core. You don't want turbo behavior on an overclocked CPU. I'd probably also leave CnQ enabled and set my min processor power state under processor power management in the advanced power options section of control panel to about 10-20%.

It will help save power but will also help keep the CPU cool when it doesn't need to be full throttle.

If you're not already using version 26.6, do so, and report back with the results. It will likely be many degrees lower than if you're running a newer version. I would not bother overclocking using the FSB on that chip or board. Using the FSB on that architecture will also increase the bus speeds of all other lanes like PCI, USB and SATA and can cause serious instability if not done exactly right.

 
Solution

genthug

Honorable
So, thank you for the quick response, but I did a slight dumb. In bios, the voltage increases are increments of .00625. I... read that as .0625 and forgot that it was .00.. from the last time I was working on OCing the chip. The cooler is the Zalman. The exact model in the system specs, should've specified what all those pointed to, my apologies. So, when I got it to 4.3, it looked to be very stable (Prime 95 ran for an hour (not ideal, but I'm looking for fast system failures in tweaking)), however it was sitting at 1.38 volts. Running 4.0GHz from last time I was sitting at 1.33 volts. Currently working on getting back up to that... It is good to know that my thermal limit occurs around 1.38 volts though.
I'm running v 27.7. What do you mean by unrealistic thermal conditions? I want pretty much the highest thermals I can get just in case I ever for some ungodly reason reach 100% CPU usage.
I made sure to go through the BIOS settings and disable all of that to leave it up to manual input. My CPU is currently idling at around 20 degrees.
I've been using the FSB and resetting all of the other changes in BIOS as needed (CPU/NB link, HT link, RAM speed) I didn't know about the PCI, USB, and Sata connections so I will definitely take a look at those and (if needed), reset my bios to doublecheck.
I will take a looksy about using the multiplier to OC instead of the FSB, possibly knock the FSB down a bit and turn to the multiplier instead.
 
AVX instructions are not a steady state workload. Steady state workloads are desirable, and if fact should be the requirement, for thermal compliance testing. Use of version 26.6 on Small FFT is the most accurate thermal depiction you're ever going to see. In real world use, at 100% usage on all cores, thermal conditions will be identical or less than when running Small FFT on Prime version 26.6.

I can think of several reasons why x264 encoding or AVX / AVX2 / FMA3 apps won't work as a unilateral metric for thermal testing.

(1) A steady-state workload gives steady-state temperatures; encoding does not.

(2) Simplicity in methodology; most users would find encoding apps unfamiliar and cumbersome to accomplish a simple task.

(3) Most users such as gamers, graphic artists and most general users never run any apps which use AVX / FMA, so adaptive or manual voltage aside, it makes no sense to downgrade your overclock to accommodate those loads and temps.

(4) Standardization; Prime95 has been around since 1996; many users are familiar with it.

For the minority of users who routinely run AVX / FMA apps, then P95 v28.5 can be useful tweaking BIOS for thermal and stability testing.

regardless of architecture. P95 v26.6 works equally well across all platforms. Steady-state is the key. How can anyone extrapolate accurate Core temperatures from workloads that fluctuate like a bad day on the Stock Market?

I'm aware of 5 utilities with steady-state workloads. In order of load level they are:

(1) P95 v26.6 - Small FFT's
(2) HeavyLoad - Stress CPU
(3) FurMark - CPU Burner
(4) Intel Processor Diagnostic Tool - CPU Load
(5) AIDA64 - Tools - System Stability Test - Stress CPU

AIDA64's Stress CPU fails to load any overcloked / ovevolted CPU to get anywhere TDP, and is therefore useless, except for giving naive users a sense of false security because their temps are so low.

HeavyLoad is the closest alternative. Temps and watts are within 3% of Small FFT's.

-Computronix


The whole reason that most AMD boards, and the Intel Z series boards incorporate independent clocks is so the multiplier can be adjusted, and that is so that the FSB, which is a less stable method on systems that don't have separate clock generators for use with BCLK overclocking, can achieve much higher overclocks without upsetting the stability of the rest of the architecture.
 

genthug

Honorable
Okay, that makes sense. Did not take me very long to get back up towards 1.3875V. Sitting at 4275MHz right now with 225 FSB and 19 mult. On CPU-Z, it says my multiplier can range anywhere from 7 to 19.5. Does that mean that my max no matter what I set it to in BIOS is 19.5? Or is that just for AMD's turbo technology?
At time of posting: 26.6 has been running for all of 5 minutes, but max thermals are now sitting around 60 degrees. Occasionally upping towards 61-61.5 but that seems to be the max for right now. I'm sure that will change very soon.
 
Actually, it shouldn't. After the first five minutes, you really shouldn't see significant changes through the 15 minutes that are recommended to establish thermal compliance.

After 15 minutes, you're unlikely to see anything further change thermally regardless of how long you run Prime, so long as a non-AVX version is being used.

That is probably just for the turbo, but some boards have funky, clunky configuration metrics, so who knows. As I said before, I'd turn the automatic turbo features off anyhow. Turbo isn't necessary on a manual overclock and can upset an otherwise stable overclock as the system will take control of the voltage and multi if turbo is enabled.

I'm pretty sure you can set the CPU multi to anything you want in the bios, within reason of course. The only limitation should be what you can reliably cool it with.
 

genthug

Honorable
Well, I'm currently sitting at... around 50 degrees? and dropping? Possibly because it's 47 degrees out (Fahrenheit) and I have a window open right next to my machine. Had to bump the voltage to 1.39375 because I got an illegal sumout... Yeah I made sure to uncheck all of the AMD nonsense and auto OCing stuff in the bios before I started my quest to get this to last another two years.
Now the real question is how high can I overclock this on liquid cooling...
Thank you very much for your help tonight! :D
edit: Yeah, closed my window and my CPU temps shot up to around 56 degrees. Still very good, so I could probably fit in a 19.5 mult if I'm ambitious.
 
Keep in mind as well, that when you move to water cooling, you also remove the residual airflow over the motherboard VRMs that would happen with an air cooler, so when overclocking it isn't uncommon, especially on AMD FX systems and the 9xxx series chips in particular, to experience VRM throttling due to overheating of the VRMs, even if core temps are well within limits.
 

genthug

Honorable
With my case (Antec 900) just by thought, I don't know that I can have more than two radiators on it as it is. I don't actually know if my 200MM top fan pulls air in or if it pushes air out. I know I have one exhaust fan on the back, and I can flip the side fan at will, so those aren't an issue. I could probably get the stock case fans flipped with a DIY solution to get airflow still through the case.
 

genthug

Honorable
I currently do not have liquid cooling. In the process of coercing my mom into going 50/50 with me on the system because she enjoys building computers as much as I do, and neither of us have ever done it :D we would need to buy one third party. The 200MM fan is an exhaust fan though, so that's good.
I'm not very well versed in liquid cooling as of yet.
 
My bad, I forgot you had the Zalman. Appologies. Working several threads related to cooling tonight. So anyhow, when that time comes, a bit of work on the case would be necessary as the 900 does not natively support water cooling. Modification would be necessary, or a different case. Anyhow, with your current configuration you should have two front intakes, a rear exhaust and a top 200mm exhaust, and that should be more than plenty for what you've got going on right now.
 
Grommets like those are intended for externally mounted radiators on custom loops. Can't really run an All in one closed loop cooler line through cutouts since the cooling lines don't detach and an internal custom loop has no need to run lines outside the case unless it's going to an external reservoir. So it probably means those are intended for use with a custom loop but that is a way more extreme, way more involved process than slapping an closed loop cooler in there.

Internally I don't believe the case configuration supports the mounting of radiators, unless it's a rather small 120 or 140mm radiator on the rear exhaust. Those are next to worthless anyhow, as any good mid to high end air cooler with outperform a 120mm AIO cooler, and do it more quietly. If you want to switch to liquid cooling at some point, I'd plan to replace the case as well or be prepared to do some heavy work on case modification.

I could be entirely off base here, but I think it's pretty accurate.
 

genthug

Honorable
Well... shit. Alright, I'll have to factor that into cost then as well. Any recommendations on a good LC case?
Edit: under 100 will be nice, but if at most it is 150 that is doable. If it's gonna be LC I would rather it be a nicer more expensive case than something I skimped on.
Edit edit: and if a nice LC case is more than 150, so be it.
 
I have the Fractal Design Define S, and I like it very much. I modified the heck out of mine though. Even stock it's design is absolutely build for watercooling. There are several others as well. If I had to pick three cases for liquid cooling, under a hundred bucks each, it would be the these.

Fractal Design Define S

Phanteks Eclipse P400

Corsair 400C
 

genthug

Honorable
So, on air cooled systems, you can make the cases look cool. It'd be more of a bonus at this point, but I assume you can't do that due to the fact that you'd sacrifice front/top radiator space?
What would your recommendations be over 100?
Edit: what's your take on a Corsair Obsidian 750D?
 


Not sure I completely understand this question. To me, both air cooled and water cooled systems can look cool, and neither air nor water really has a great advantage over the other in performance unless you get into custom open loop water cooled systems, which are clearly superior performers. All in one closed loop liquid coolers are louder than air coolers for the most part, especially under load, and don't really perform any better than a high end air cooler. Plus, depending on the case, they take up case fan locations and complicate cooling schemes.

But with the right case they look good and can perform very well.


Here is my case, with an air cooler, the Noctua NH-U14S with a Noctua NF-P14 PWM fan on it, but it also supports front and top mounted radiators up to 360mm. Of course, it's far from stock as I modified the front panel with a custom grill and made a custom emblem for it, customized the feet using two sets of feet with one set inverted, colored the feet, colored the PCI slot guards, colored the sides and top of the front panel and case.

But I think it looks "cool". It really all in what you are willing to do with it. Still, even stock it's a really nice case and supports a variety of configurations, plus it's extremely easy to work in. Those other cases I linked are as well.

I don't particularly care for full size cases, as they are often TOO big and are hard to find places to set within reach that will accomodate the size.


10egh35.jpg


 

genthug

Honorable
Would it be worth it to get a watercooled system and a second r7 370? Because essentially, that's what I'm looking at. The CPU loop would end up being a bonus, but I'm looking to keep my GPU temps down so I can crossfire them together.
 

genthug

Honorable
And as for size, I would rather have a larger case now given that high end gaming components are only getting larger. Built my roommate a machine with a 980 in it with my same case and he had about half an inch left in between the GPU and HDD internal bays. That and I just don't mind the size
 
I can't make a recommendation as to the CF 370's. I'm not extremely experienced with water cooling graphics cards, and unless you plan to also grossly overclock those CF'd cards, I wouldn't think it's probably essential to water cool them unless there is a lack of case cooling. For overclocking or aesthetics, it's certainly an option, if waterblocks are available for the particular card model you have.

To ME, and this is just me, it seems like a waste to buy another 370, plus invest a lot of money into a custom loop for those cards, when you could spend less on a bigger card that will outperform them, not have any of the crossfire profile issues that are common and probably not require a new PSU with a much higher capacity.