FX-6200, A little chronicled journey

[UrbanCamper]

If u had read carefully what i said earlier you would have noticed that the vcore increases ONLY in .25v increments. This means that since my stock voltage is 1.385 the best i can do is raise it to 1.485 without going over 1.5v. Which is what I did when also I took your advice and raised the nb to 1.25v The rest of that experment was trying the multiplier at first 22.5 which failed and then at 22 which failed.

Well i kept some of your advice. I raised the the nb voltage to 1.25. I hope this is safe. The temp looked ok as u can see. The vcore i left at stock settings. I did however put the multilier back at 21.5 to see what might happen. This would be the 4328 i ran for 25 day without knowing i had errors. Here are the results after a little less then an hour.

I would have gone longer but the Bulls game is on and my tv is my msi tv tuner in the system. Will run a more complete test when i go to sleep tonight. But for now this looks kind of promising. At least to get me back where i was without the errors i didn't know i was getting.

Oh and ya mine is a budget board, but i live on a disability check so i guess that would mean i am on a budget. Oh and it is a Good budget board. I had my 965be running at [email protected] stable on this board for 4months. On my old ddr2 board i could only get the 965be to 3.8

 

[skpwn]

Well i kept some of your advice. I raised the the nb voltage to 1.25. I hope this is safe. The temp looked ok as u can see. The vcore i left at stock settings. I did however put the multilier back at 21.5 to see what might happen. This would be the 4328 i ran for 25 day without knowing i had errors. Here are the results after a little less then an hour.

You're looking better! Now you're going to have to really chunk that vcore up for that extra 100 mhz. After 4.2, this processor needs massive amounts of VCore to go higher. So start slowly raising it by as little as you can, and keep priming. Post your results as you feel fit. Don't go higher than 1.5 though...you do sound like you know that!

Oh you can also up the bus a tad, something like ~230 might be the highest you can go. Remember that the GHZ is based on multiplier MULTIPLIED BY the bus speed. So if you up the bus, you'll have to change the multiplier accordingly, to hit 4.5ghz. This could make your computer be a bit snappier when it comes to bench marking, so don't be afraid to play with it.

Oh and ya mine is a budget board, but i live on a disability check so i guess that would mean i am on a budget. Oh and it is a Good budget board. I had my 965be running at [email protected] stable on this board for 4months. On my old ddr2 board i could only get the 965be to 3.8

Budget board has been a joke throughout this whole thread. My m59a7 is also a budget board (~$100), so I wasn't poking fun at you .

 

[UrbanCamper]

Did u take a close look at that picture? I raised it to 4328 without raising the vcore. I did however raise the nb voltage to 1.25 as u suggested. So i am thinking that 4500 i will have to raise frequency to 205 and the voltage which i will try to do in .25v increments since i cant do it in .10v. Unfortunatly this will also raise the memory to 1680 if i remember correctly. i forget what the math was but the old 965be(which is gathering dust)was running this memory at 1680. Anyways I know this memory can clock to 1866 so 1680 should be no problem.

 

[skpwn]

My knowledge of boards wanes a bit when it comes to some of the Gigabyte boards, so I'm unsure what is going on. I have a hunch that your motherboard is automatically changing the vcore (you have it auto I'm assuming) and allow it to boot into Windows @ 1.41. So you're not at stock voltage when you're firing up Prime / Booting into windows, but instead at 1.41 for 4.3ghz.

I'm going to defer to one of the other overclockers on this forum for confirmation. I could be off on how that is working, but I clearly see 1.41 max, meaning your chip had/has that power coursing through it.

Now you need to either choose a vcore you're comfortable with and see how how you can kick that multiplier OR you need to stutter your step your way up for another 100 or 200 mhz.

 

[skpwn]

Let's compare you run to my run. Your run:

My run:




If you honestly didn't up the vcore in the bios, then your motherboard is upping it for you to 1.41 with a drop to 1.36. While my 4.2ghz run is 1.39 with a drop to 1.34.

Your board is within range of mine, meaning you're probably going to be following nearly the same vcore steps that I offered earlier. I will do a 1.41 run with 1.25 CPUNB voltage to try for 4.3ghz tonight and compare. I am fairly confident that it'll run without issue.

Was that 4.3ghz run still in the room with no air conditioning? What is the ambient temp in the room? This question is just for curiosities sake .

 

[gnusounduave]

So i am thinking that 4500 i will have to raise frequency to 205 and the voltage which i will try to do in .25v increments since i cant do it in .10v.

When making adjustments to your vcore are you using the manual mode or offset mode to adjust voltages?

Also your 4.3 run with 1.41vcore looks solid.

 

[UrbanCamper]

There is no offset mode only auto and manual. I am using manual. Also there is no llc control though according to a post i read, earlier versions of the bios on this board had it.

Yes i am stable at [email protected]. In the bios while i am at this vcore it is sitting at normal which reads 1.385. So what reads as 1.385 in the bios is 1.408 which rounds up to the 1.41 that you see in hardware monitor. I would attribute this to vdroop. Of course we also see the 3.6 that it goes to when I run prime.

Heres where it gets interesting. This board has no llc control. Which means if i wanted to control the vdroop, well i cant. There is something wrong with the vcore voltage control. I believe it has something to do with the, first f5b, then f5h, and finally the f5 bios that came out this morning. When i was running my 965be on this boards' f4 bios i never had such wild fluctuations. Here are the voltage increases and all the numbers. Its a wild ride. Ok remember normal according to the bios is 1.3875

Reading#1/ 1.3875v+.075v=1.4625v/cpuid reads 1.41v>1.39v low\vdroop while testing p95 1.36v

Reading#2/ 1.3875v+0.10v=1.4875v/cpuid reads 1.44v>1.42v low\vdroop while testing p95 1.36v

Reading#3/ 1.3875v+.125v=1.5125/cpuid reads 1.46v>1.44v low\vdroop while testing p95 1.36v

Reading 1 shows no difference then stock, even though i have raised the voltage .075

All 3 readings vdroop to 1.36v under prime95. As we can see none of the vcore readings are as they have been set in the bios, and something is throttling every vcore back to 1.36v under prime95.

The f4 bios that was on this board never had problems raising the voltages and the vdroop was rock solid at .015 up and down. I am toying with putting it back on even though the cpu may not work( i guess i would see ). Also the f4 bios had a core unlocker on it and i suppose it might be possible to unlock the 2 locked cores making it, dare i say an 8 core cpu. The worst that would happen i think is that i would have to reinstall the 965be to put the f5 bios back on the board.

Or i could say screw it all, live with 4328mhz, and rip a nasty letter to Gigabye about the crappy f5 bios, hoping they fix the crappy vcore voltage control.

One last possibilty(that i can think of) would be to increase the frequency and decrease the multiplier, hopefully making the vdroop footprint smaller. This may be the only way to go on this mb. To do this i am thinking i would have to raise the nb voltage even higher then the 1.25 it is presently at. I am not sure about this though.


Before i do any of this does anyone have any other ideas? Sigh

 

[skpwn]

I'm almost 100% confident your board is the issue here . The m5a97's are showing similar readings at 4.2 and 4.3, but will allow us to continue adding vcore while yours is acting really strange.

 

[UrbanCamper]

The board performed flawlessly with the 965be on the f4 bios. The vcore voltages stayed where you put them with and exact vdroop of .015+-. It also had llc control so u could eliminated the vdroop altogether. This bios version f5 including the f5b and f5h versions do NOT put the vcore voltage where you set it to, and does not have llc. I thought i made it clear with all those numbers i put down in the last post. I guess in a sense you are correct in saying it's the board. However from what i can tell it is the bios that is flawed. I will be ripping a letter to Gigabyte with my finding. Hopefully i will not be alone in this and they will repair the bios. Sigh

 

[skpwn]

When I say your board is having the issues, that encompasses the bios. Good luck with your Gigabyte customer service adventure sir!

 

[UrbanCamper]

Well a little update. I figured out the vcore increments on this wacky board of mine. DARN BIOS!!! If u recall i can only increase the vcore in .025 increments. The problem is the bios .025 is not the true increase even though it says so. For every .025 I increase it. It is really only increasing .018. Pretty messed up eh? So this is what i did to get it stable at 4410mhz.

I increased the vcore from 1.385 to 1.485. which really was.... Now stay with me, this is the fun part.! On this board past the the stock vcore setting .025=.018 \ .018x4+.072\ 1.385+.072=1.457 Which as you can see is pretty darn close to the core voltage, as it sits now.

Well it worked. I pushed the multilpier to 21 and the frequency to 210. Raising the frequency had the added benefit of halvin the vdroop to 1.41 as opposed to the 1.36 I was getting under prime before. It also raised the memory speed to 1680mhz.

The temps got a bit high at the end. But i think it is ok. I was not using arctic silver when i put the h/s and fans on, and also I have them pushing out the back. Got more arctic silver and tomorrow i am going to use it and also face the h/s and fans upward to the 200mm fan on top. I am fairly confident that it will shave a few degrees off the top. 4410mhz is probably the highest I will be able to go air. Though if the temps drop enough i may try for 4500mhz now that i have a handle on this board. Woohoo

 

[UrbanCamper]

I have yet to see anyones elses results at 2hrs prime95. Is it possible that no one else in this thead has found a way to get this cpu stable over 4200mhz? Would like to see the proof. Also the numbers that got u to your overclock. I will try to hit 4500mhz sometime later today.

 

[gnusounduave]

UrbanCamper, perhaps you might want to begin your own thread topic discussing the exploits of your 6200. I mean I can't speak for SKPwn but at the present we seem to be getting derailed a bit from his main topic which is the documentation of his OC experience.

I haven't minded jumping in and helping you a little here and there as I'm sure SKPwn hasn't minded helping either. However now might be a good time to start a new thread with something like FX-6200 @ 4200Mhz what's your max OC?

 

[UrbanCamper]

While it is true that this is SKPwns' thread. He has submited no proof of any stable overclock to his fx6200. I on the other hand have done exactly what was asked as far as documentation goes. I have shown solid evidence of a stabe overclock at 4410mhz. All im asking for is proof all i have heard is wind. Several people have contributed to this post and so far i am the only one to show 2+hours stable on an overclock. First at 4228mhz, then at 4410mhz. I did learn one thing here. Raising the nb to 1.25v Helped to stablize the memory. Thanks for that. Sorry SkPwn if i did steal your thread it was not intentional. I hope you can find a stable overclock for yours.

 

[skpwn]

Are you serious? Have you not read a single post in this entire thread? I've got a good 30+ minute blend at 4.4 ghz, 52c, on page one, which is what you just came in and did.

Some of us don't have the time like you have to do constant 2 hours stable OCs more than one day in a row. I wish I didn't have to work during the day, but I can't afford that luxury, and am stuck doing more than just ****ing with my PC at night. I'm lucky if I have time to stop and overclock my computer for a 20 minute test funny little overclocks.

Please jump off your high horse and respect the fact that we helped you run your budget board to 4.4ghz. If you want to have a **** measuring contest, you can create a thread asking what the best OC people have gotten on a 6200 is.

edit;

I'd like to point out that these are the pics 4.4ghz the first week I got this chip. Subsequent posts had me running to the conclusion that really trying to crank out an extra 200mhz isn't going to do jack ****. So while you're over here coarsing 1.46 vcore through you're chip, I'm happily doing the same OC at 1.45 vcore (oh, and look! It blends at 1.4 like yours).

 

[UrbanCamper]

I was going to make the this private just to you Skpwn. But since you have decided to curse in a post and since this is the last post i am going to make here i think i will just let it rip.

Not one piece of advice u personally tried to give me helped. I had to figure out the vcore differential on my computer on my own while u sat back and first declare that 4200mhz was the highest i could go because of the heat. Then u decided to tell me my little budget board was crap.then u made a little remark about my contacting the manufacturer.

If,ya if u ever had a stable oc at 4400mhz on your 6200after running prime on it. It would not have been hard at all to push that little button called printscreen and documented it on the spot at the 2 hour mark. Very simple. 2 steps Stop prime95, hit printscreen. 20 minutes is not proof. As I said U HAVE NOT SHOWN ANY PROOF.

As far as peoples max oc on the fx6200 goes, No one was asking to measure, well you know what u said. I am sure that many people have overclocked their fx6200 STABLE way above what i have done so far.

As for your little curse fit. You just keep on cursing i am sure that will get you all sorts of places real fast. Cursing is not Proof. 20 Minutes is not proof You Sir have no PROPER DOCUMENTATION and therefore no PROOF.

Now maybe you can quit raging and get some......

I'm done here. Thank you, thank you very much! Elvis has left the building

 

[Automata]

I think it is time for both of you to cool down.

20 Minutes is not proof You Sir have no PROPER DOCUMENTATION and therefore no PROOF.

Now maybe you can quit raging and get some......

Drop it. I'd advise against posting in here if you are going to keep going in this direction.

 

[skpwn]

Prelimanary tests look great, David.


30 minutes: