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

AMD FX-8350

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Last random question of the night:

What's the difference between offset voltage mode and enabling CnQ? I've been trying to do some research and can't find anything concrete. They seem to accomplish the same goals - downclocking your CPU at lower loads. Will utilizing both be redundant?

Also, I've read that C1E specifically affects how your computer sleeps. I almost never completely shutdown my computer and use sleep mode quite a bit. Obviously C1E is helpful here, right?
 
What's the difference between offset voltage mode and enabling CnQ?
1. CnQ is an AMD designed method of lowering the multiplier and lowering the cpu voltage when load is not great. Vcore drop is pre-designed into the cpu by AMD.

2. Offset Vcore is a user chosen method of ADDING Vcore to the cpu by an amount the user chooses to use when the cpu goes under load. Take the load away and the Vcore "loses" the offset chosen amount but there is no dropping of the multiplier.

3. Turbocore is monitoring load/temp and can UP the multiplier and move to a chosen AMD programmed voltage when Turbocore senses that 1/2 but not all cores can have their speed raised.

4. Now throw in Windows power manager that can set "no load" cpu to a % of cpu computing power when not loaded and can utilize cpu power by a % amount up to 100% when loaded when in "balanced mode".

5. So the FX processor of today has TurboCore trying to raise the speed of some cores by upping the multiplier and Vcore by a set amount. BUT CnQ is looking to see what is the load and intends to reduce all the multipliers and Vcore by a set amount. Windows is also looking at the load and in balanced mode can reduce the cpu power output by dropping the cpu by a % amount that lightly translates to reduced multiplier and possibly the Vcore but when coupled with use of the Offset Vcore settng mode, the Vcore is dropped when the load is reduced.

6. You get all those things that seem to be arm-wrestling each other for control of what the cpu gets to do? The arm-wrestling power/performance/speed managers is why we don't try to 'teach' offset vcore setting in a forum by typing text into a forum post.

I run C6 and TurboCore Disabled. C6 is required for TurboCore to function.

I run Offest Vore mode with the resulting voltage enough to supply the Vcore needed to run my FX-8350 at 4.8Ghz P95 Stable. This Vcore includes accurate setting of CPU_LLC.

I run C1E enabled and Windows power manager set to balanced mode with the settings for balanced mode set to a low of 15% and the high to 100%.

The result of this balancing act is that when my cpu goes under load the cpu leaves 1400Mhz and moves up to 4800Mhz with enough Vcore adjusted with CPU_LLC to run the 4800Mhz P95 Blend mode for 2 hours.

Trying to teach this to many forum goers is too dang complicated most of the time. On the surface it seems complicated and most don't seem to get it because it takes some trial and error and recording of results at "their" end. Mostly they want gimme A and B so I can get C and go back to gaming. So we suggest turning off all the Green Stuff to allow for overclocking since removing all the Green Stuff from the setup, makes for easier overclocking for the masses of gimme A and B and get me to C.

Using my setup I get ~1400Mhz when not in heavy use and when needed the cpu can ramp up to a stable 4.8Ghz and last I checked my system would sleep and then wake-up with a push of my power button.

I have written about the Offset Vcore mode setting inside a thread in the AMD mobo forum section and once here in the CPU forum section in response to and with "johan45". Not sure which actual threads those were but I did do some how to setup offset Vcore explaining.

RGone...ster.
 
Yay, just passed 20 minutes of Blend. Snips are attached, but relevant BIOS settings were as follows:

CPU Ratio: 23
FSB: 200
Memory Freq: Auto (1600)
CPUNB Freq: Auto (2200)
HT Link Freq: Auto (2600)
CPU LLC: Ultra High
CPUNB LLC: High

Vcore: Manually set to 1.4625 (max load 1.476)
Everything else on Auto

I'm going to P95 Blend these exact settings for a couple hours tomorrow to establish real stability. I think this is the end of the road for me, though. I'm perfectly happy with a 4.6 OC and the H80 performed way above expectations (maybe the non-stock fans helped). I maxed out on temps with that 212 on 1.425 vcore.

After stability is established with a two hour Blend, I'm going to turn back on the green stuff: CnQ, C1E, Balanced Windows power. I would like to look into offset mode, too, though. If my VID is 1.375 and my current OC manual vcore is 1.4625, would my offset vcore input be 0.0875?

Anything else I'm missing to properly set up a 24/7 OC? Do I leave the LLCs on Ultra High/High, respectively?

Appreciate all the help and feedback, as always.

I see you say your VIDD is 1.375 which is exactly the same as mine and for me to be stable at 4.64 I was using 1.476 v coincidence ?? I think not.
What you say about people getting this or that is true. These chips seem to be all over the map when it comes to GHz and required Volts.
 
I've seen a lot of people around the Internet (disclaimer) claim to push a 4.6 multiplier OC on stock voltage.

lostcondottiere, we have seen and heard so many claims of X Voltage for X Ghz. I would be hard pressed to believe most of those claims. The group of people here, running these Fx chips have not only overclocked them, we test them through so many scenario's and post our results. I can tell you the differences between most of them and what voltages they need to run X Ghz. For example I know Rgons 8350 usually will require slightly less V core to run the same Ghz as I do as well as Johans will usually require more then mine. While I know I can boot into windows and do things such as navigate the internet at 4.7 at a lot less voltage then the 1.46 that I use for 24/7. When it comes to brass tacks, it isn't stable when I'm really doing work. What I'm trying to say here is, a lot of people overclock their chips just to be able to say it, sort of like a badge of honor, "I run X Ghz". When it comes down to it I would bet we have done more work in this forum testing these processors then the majority of people and can tell you that the claims you read about running 4.6 are most likely false or not "stable".
 
mR. Clean you are on it today. I cannot see a single bit of your post above that I would not agree with just about 99.9999999999999%. Hehehe.

mR. Clean said:
While I know I can boot into windows and do things such as navigate the internet at 4.7 at a lot less voltage then the 1.46 that I use for 24/7
Why don't U just mince your words? Sounds like what I have experienced myself. I can get into Winders and do a lot of stuffz at 4.5Ghz with 1.375 Vcore but if I wish to be fully stable, it takes a good bit more as you have already set forth.

So when I went to look for some procedures to use in testing with P95, I got the same ole BS that has been around forever. At least since I came over to computering instead of motorcycle racing and tournament softball years ago it has been the same way when stability testing is discussed.

I saw a poll last night in another forum and as regards P95 Blend mode testing, there were 71% that believed in P95 Blend mode and well you know how many that left somewhere else.

As cpu construction was downsized and the heat went up, there are more and more people that cannot get P95 Blend mode stable. Not enough cooling and so they begin to say that stability testing is not so necessary. Or perhaps they move to some form or front-end of IBT which for me and many takes less Vcore to show stability. So they like IBT variants better.

Now as you well know there is a heads-up for our BS. Goldfish in "johan's" geo thermal pool. Hehehe. Otherwise it is as straight and true and honest as we can possibly make it! We do that for every last one of these people that come here asking for help. No BS just as plain with no sugar or spices as we can make it. Those conditions make some of us feel like the "mobo-police" at times.

How we treat computing amongst the group of us that hang and help most of the time is what draws us together. Even over the many miles and varying occupations previous or persent. Testing stability to us is not just P95 Blend mode even if we suggest it most of the time. What we all really mean is that P95 Blend is about the least we would test with. Length of testing time varies.

I have two computers that must pass various stability tests for at least 12 hours. They are hold-overs from my days of employment in the computer industry. The other three rigs need only pass P95 Blend for 2 to 4 hours. So there are variances, but no move to be released from stress testing.
RGone...ster.
 
Not just here in the forums but in real life as well. I had popped into the Local Canada Computers and was talking with one of the sales .... specialist .... ehm. We got talking about our OC's on my FX 8350 He was just floored that I required 1.525v to get stable @ 5GHZ and was running temps pushing 60*C at that. When I asked him why he told me he was running STABLE an FX 8150 fully loaded @ 5.2GHZ and no higher then 35*C temp that with an H100 for cooling. When asked about what stress test he was using ..... cant remember now ...... but was something that I never head of before. His comments about Prime was , well its a different BEAST.

We have seen it day in and day out here in the forums where someone comes in expecting high OC's with their gear. They have read on other sites they can get X GHZ with this stuff but for whatever reason they can't. You really have to be carful on what you read and listen to. Here at OCF whether it is in the AMD CPU section or over in the water cooling section, go check it out conumdrum is even more pointed in his advice then we are ..... you will get good advice on what questions you ask and any help we can give you.
 
It certainly is funny how this "stability" thing comes back around every few months. Prime 95 is a staple around the AMD sections here. But even it isn't perfect. We've had some who were stable with P95 but after running 3 days of Fold@Home they'd get an error. So like RGone said if you're going to use your system for surfing/gaming. A couple hours of prime is good enough. But if you're doing something intensive and important and can't afford to have any glitches at all much more testing is needed. Or you just leave it all stock.
Yeah I said stock, once you're in the realm of OC"ing there is no guarentee.
 
over in the water cooling section, go check it out conumdrum is even more pointed in his advice then we are
Ha Ha Bassnut, when I was building my H2O setup I made sure I read and re-read that section numerous times. I was afraid of the "Conumdrum Wrath" if I posed a question that was already in the stickies.
 
Ha Ha Bassnut, when I was building my H2O setup I made sure I read and re-read that section numerous times. I was afraid of the "Conumdrum Wrath" if I posed a question that was already in the stickies.

Ya there's no sugar coating on that one is there?
 
Stability is a relative term. Just because your 'puter is stable in IBT doesn't necc mean it's going to be stable in P95 either and I would assume that the inverse would be true as well. I was in a discussion with another member (can't remember who right now) about which was better IBT or P95. The fact is it doesn't matter. as long as your system never fails when it is doing it's job, be it gaming, encoding or mining.

:salute:
 
That's exactly why I've tried to listen to you guys and take your direction to heart. I'm a stickler to doing things the right way. I've put a lot of money, effort, and time into my computer, it's something I love to tinker with and spend time on, and the last thing I want to do is take the wrong shortcut and screw something up.

I was talking to a guy on another forum who suggested not even running stress testing on an overclock. I told him that heat was becoming an issue when I was running Blend. He said my temps would be way lower when I game (duh) and I shouldn't worry about. I asked him how would I establish a stable OC if I don't run a stress test to determine said stability. His answer, verbatim, was "just wing it."

Lot of bad advice around the Internet. It started to make me feel like I was doing something wrong seeing all these claims that people had their 8350s up to 5ghz on 1.4 vcore. But then I remember that Johan's chip is probably broken off the same sheet as mine and we both have to pump a lot of voltage to get to stable OCs.
 
As suggested earlier "stability" is what works for your pc. I myself use Prime Blend 2 hours to establish stability. I know if I can pass 2 hours, my pc will take anything that "I" throw at it. Some people this isn't enough, there are people here that do "real" work on their pc's and if it crashed while doing it they would lose many hours of work. The person that said that your temps would be a lot lower while gaming is correct, in most instances your pc will never be stressed as hard a prime will stress it. Though, also game on my Pc and if I cannot pass 2 hours of prime I find that I will also crash during a game from time to time. Furthermore, the amount of FPS that you get from really pushing the pee out of these processors while gaming sometimes isn't worth the Cpu V you need to add to get it there. The two reasons why I run 4.7 24/7 is because I have ample cooling that allows me to do so with reasonably low temps. The other reason is that the increase in Cpu V that I need to run stable at 4.8 isn't worth it to me.
 
Gimme a break or I got the wrong board and cooling...

.
I just went to do a little g00gling to see what speed some were getting on FX-8350 so I am not too too far behind.

Right off I found a thread and began to read since the OP was having trouble getting past 4.0Ghz which is really stock speed of an FX-8350. A few told him he needed to up Vcore and the general responses. About 1/2 way thru the thread this was the only thing in one post. Awesome.
RGone...ster.

5.1 FX-8350 EXT9.jpg
 
Woo hoo, just passed an hour and five minutes of Blend. Temps totally stable throughout, with a brief spike to 65c on the CPU (which last for maybe 3-4 seconds).

Had everything green turned on during the Blend test (CnQ, C1E, Windows balanced power). I was also using offset mode. My 0.0875 offset was right on the money, as it jacked my voltage to the exact number I had it in manual. :D

See attached snips if interested.

I know I've asked this question before, but with all the green stuff turned on, even in offset, my clock/vcore will jump all over the place. I remember when I was using stock settings it'd only jump between max clock (4.0 at the time) and 1.4. Is there anyway to set this? If you guys are telling me the random jumps (1.4, 2.8, 3.3, and 4.6 are the most popular) are completely harmless, I'll just leave it alone. But maybe there's a way to set a more consistent fluctuation?

This H80 has been a beast for me, though. At idle with all the green stuff turned on, my CPU is sitting at 8c. :D

Thanks again for all the help.
 

Attachments

  • cpu22.JPG
    cpu22.JPG
    54.1 KB · Views: 27
  • hw22.JPG
    hw22.JPG
    98.2 KB · Views: 27
  • p22.JPG
    p22.JPG
    304.3 KB · Views: 27
I know I've asked this question before, but with all the green stuff turned on, even in offset, my clock/vcore will jump all over the place. I remember when I was using stock settings it'd only jump between max clock (4.0 at the time) and 1.4. Is there anyway to set this? = That sounds like you still have TurboCore enabled and C6 even now. Mine stays as you first describe in the underlined area the few times I have looked at it.
 
My system isn't 24\7 stable at 3.9 but I am 'game' stable. The reason I no longer game at that speed is the voltage/heat increase are not worth the minimal FPS. I cannot see or feel the .2Ghz during game play so why not let my system run cooler and most likely longer. That's my 2 cents.
 
Back