Fx 8150 with H100i OC! Help

[tungureanu]

not to be mean. GOOGLE. you have to know your system

 

[trents]

Freezecz, I think you have enough information to make a good decision about which direction you should go. This thread is losing focus now and I think it's time to close it out. Post back when you have made your decisions and purchases and give us an update.

 

[tungureanu]

+1 trents

 

[Freezecz]

Okey, Im going to buy SABERTOOTH 990FX R2.0 Thx all

 

[BeepBeep2]

Too much anything burns your CPU.

Intel typically sets the PLL too high..

AMD has dropped the ball on PLL voltage on the FX processors-- You basically HAVE to increase PLL voltage to get a good OC beyond 4.8GHz, and it makes a smaller OC easier with less core voltage.

It is a lot less stressful on your CPU than continuing to increase voltage Give it a shot when you have time. Best way to increase cpu performance overall is the following:

1. Increase CPU Frequency
2. Increase NB Frequency (Not as important on FX as Thuban / Deneb)
3. Increase FSB (Don't do this until you get familiar with multiplier-only OC)

I do all 3 when I OC. This also requires you to be able to monitor multiple variables at the same time, on the fly (Which is terrible for somebody asking for a cookie cutter OC solution)

Lets get you a stable OC before worrying about diving deeper into the game, alright?

AMD didn't drop the ball on anything, your board simply lacks any type of useful Load-Line Control. Raising PLL voltage can help combat the symptoms of vdroop/low vcore and load stability. I'm not an engineer so I can't really explain how the PLL would affect stability that way but I've confirmed this phenomenon with several people in the past.

On Intel IMHO they dropped the ball on PLL voltage. When up around 54x multi+ it's a must to have PLL Overvoltage enabled and/or add PLL voltage...

With LLC enabled, upping PLL volts isn't going to do a whole lot for AMD Phenom II and FX, whether on air, or LN2. I've tested from 2.5 to 2.8v on air/water/LN2.

Now, I've missed 9 pages of discussion here and it seems the only solution or advice that's been given was for the OP to buy a new motherboard. Unfortunately, I think he will find that a new motherboard does not change things for him very much.

If the OP was only having the issue "ILLEGAL SUMOUT" as described in post #4, then the only thing that will solve the problem is more vcore at load, either by reducing any vdroop that may be occurring or raising vcore significantly all together. If a new motherboard has better LLC options (not sure what LLC does on the M5A97 R2.0, though I've tested the 1st version), then it will be more overclocker friendly (better calibrated Load-Line options), but outside of that there is nothing wrong with the board.

 

[tungureanu]

the issue was the folowing(if i remember correctly):"after he raised vcore, the system became unstable to a point that it not even post , after adding 2 bumps on it vcore. from what i remember the max was 4-4.2 and it whould boot. anything past that post failed. +the board has 4+1(if i remember correctly) power phases. wich in my opinion and experience is not really enough for a OC`d 8150.
Please correct me if i`m wrong about the power pahse part.

 

[BeepBeep2]

the issue was the folowing(if i remember correctly):"after he raised vcore, the system became unstable to a point that it not even post , after adding 2 bumps on it vcore. from what i remember the max was 4-4.2 and it whould boot. anything past that post failed. +the board has 4+1(if i remember correctly) power phases. wich in my opinion and experience is not really enough for a OC`d 8150.
Please correct me if i`m wrong about the power pahse part.

Thanks tungrureanu.

RGone attacked me an another thread saying that my advice/opinion that this board *may* be fine was way off base and also attacked me for not reading the thread (making mention of the fact that I seem to important to read it).

Because of this, I've read every post in this thread to see what was happening. It took almost 15 minutes. This thread, in a sense, started out on the wrong foot, which is why it was 9 pages long in the first place.

I think a lot of you know that I seem rather arrogant and bitter around this thread and that I often argue with others and criticize their process or what they think. While I apologize for being rude and/or snide, I'm going to get this out of the way right now:

This is what I surmise as a plot of the thread:

OP posts issue: Can't overclock past X.X MHz, because it's unstable. They are a bit vague about it and more information is needed to diagnose the problem.

RGone is the first to respond, he starts his post with:

"Not just a single fish for a meal but Fishing Lessons by "trents" for a lifetime of meals. With these CPUs you would need to be concerned with and deal with certain things when overclocking..."

Touting himself with this statement as someone ready to solve the problem, and lists the general methodology of "post X, Y, Z screenshots so we can help you", having CPU-Z, HWMonitor, etc. giving the information instead of the user. Ironically, this teaches the user nothing but to open some programs.

Questions to ask instead would be:

"What is your Core Voltage, CPU-NB Voltage, DRAM Voltage, DRAM Frequency, CPU and CPU-NB Frequency, HT Link Frequency? These can be found with a program CPU-Z*link* and HWMonitor*link*. Also, can you give a whole list of your computer specifications and components? You can put these in your signature with this format:
*signature template*"

The above quoted accomplishes all of this post without confusing the user and simply having the screenshots report the data to the "experts". The user is now actively involved in solving the problem, knowing exactly what variables we're looking at right off the bat and having already looked at them himself. At this point, if the user has any questions as to what these do, he is able to ask. With the other methodology, the user simply starts some programs, takes a screenshot and posts them here, and other users go "You need to change vcore to 1.xxv", the user goes "Okay, but I still have the problem"...

When the person does it otherwise, he is given the tools to also actively troubleshoot.

Once he reports this information to everyone, then it is a good time to ask also if Turbo, APM, C6, C&Q are disabled and also explain what they do and why they should be disabled...not just telling the user to disable them and leaving the possibility for him to assume why they were supposed to be disabled outside just "You should disable them for overclocking." This also does not overwhelm the OP by throwing 50 things at them at once in one large post.

Anyway, after these two initial posts, the thread continued with the OP posting screenshots showing ILLEGAL SUMOUT on 2 cores (4 and 8).

In this case, the initial advice should have been for him to increase vcore or back down the overclock.

The OP decided on his own to increase vcore and came back telling us that increasing vcore made it freeze instead.

Anonaru responded with advice about CPU PLL.

Why? This does not make sense to me.
The OP was already using quite low vcore, and his board has LLC options, why didn't we check if he was using LLC? CPU PLL like I described helps only when LLC isn't available on some boards due to the VRM setup for some obscure reason.

The OP was sceptical about this, saying he heard overvolting PLL can kill CPUs. This is true, it can happen, but anything can kill CPUs, voltage in general.

RGone backed up Anonaru (no offense to Anonaru) by announcing that "Anonaru gives very good advice" <---what basis outside the fact that he likes the guy, when he may not understand why the guy is advising this?


AFAIK, the next few pages are giving the OP vague advice such as "Increase vcore by two steps", which on Gigabyte boards would be .05v, and on many ASUS would be .0026v...
Here and there, when the problem isn't solved, people start saying "Maybe RAM is the problem, maybe the board just sucks...etc etc etc."

If you'd look at the behavior that repeats itself time and time again, Illegal Sumout in Prime95 points to one thing - vcore. However, when the OP increased vcore more, generally the CPU lasted a bit longer, but then Windows just hung. Why did Windows hang?

At this point, were we asking the OP what temperatures he had when it hung, if once in a while, ILLEGAL SUMOUT would still occur?

I hate to say it, but I think the issue once again in that case is low vcore or CPU temperature. This may sound silly, but when ILLEGAL SUMOUT happens (and usually pretty quickly, in the first tests, that thread does not even start testing. After about 15-20 minutes too, Prime95 users will note that there is a temperature increase regardless of cooling. Since the OP didn't get that far in the beginning (ILLEGAL SUMOUT) and does not know what kind of load temperatures he would have if he were stable, then what are we basing this on?

Later, (not 100% sure if the freezing was ever solved) the OP noted that cores started throttling after pushing frequencies more without the CPU overheating and with all power saving features disabled.

WHY?
Because in this case, the VRM's OCP or OTP is triggering. This is a safety feature, so that your motherboard does not catch fire or kill itself. (Just Google "MSI Motherboard Fire HardOCP" for an example.)

I've already described this behavior in detail. To solve the issue, active cooling needs to be put on the VRM for stress testing.

Once the OP would put a fan over the VRM to stop it from overheating, and the issue would still happen, then it would be time to buy a new motherboard.

 

[tungureanu]

Ok. Did you read about the throttleing (OCP/OHP)? On the other thread you wrote
"In that thread, I did not read the whole thread but I did state as follows:" here "Because of this, I've read every post in this thread to see what was happening. It took almost 15 minutes. " BUT thats not the point. The point is as folows . The mobo has 4+2 power phases. yes it can handle a STOCK fx8150, maybe even a slight OC. but he want to go enthusiast on the 8150 and get as much as he can. IF he watercools the VRM`s maybe he can do it, BUT a single CPU wc loop is minimum 120E(here in the EU). Maybe he gets it stable somehow while a fan is blowing on the heatsink, but what will he do when the wear on the vrm`s is too much to take and he sees magik smoke and fire(perhaps 1 week after warranty expires), ain`t it just better to rma it and sell it (or keep it as backup) , and get a board that is adequate for his needs?

Don`t forget everyone here is trying to help each other. As for RGone he has followed th thread from post no.1 to post nr 160, and he told you what he thinks/knows as a guy who folowed the whole thing.

 

[BeepBeep2]

Don`t forget everyone here is trying to help each other. As for RGone he has followed th thread from post no.1 to post nr 160, and he told you what he thinks/knows as a guy who folowed the whole thing.

I know people here are only trying to help each other. I ask questions too sometimes, because as in any case, if I can't figure something out on my own or I'm stumped, I'll ask someone else. This is how the world works...and this forum is here so we can all help each other.
___________________________________________________________________
However, I don't know why users here must take the longest approach to every issue, and why they do not start recognising obvious patterns after it happens right in front of them time and time again.

I believe this is because they do not experience it themselves (they only have read about it), nor do they fully understand the advice they are handing out, but are handing it out with full confidence and making very broad and general statements/assumptions because of it.

Because of this, you start getting "forum experts" that make statements like "those generation of CPU don't like you do change Xoeujffr voltage" when that "expert" never ever touched one of those CPUs themselves, nor did the guy that told them that. Each time it gets passed along, it gets more vague, and more people become confident in that "forum expert", because it seems to work every time. They do seem to solve all the issues, but they don't necessarily know why they are solving it either.

If you were to ask the guy why it works, he couldn't tell you anything but "Because you upped the voltage, because you lowered the multiplier, etc etc"...which may be true but does not "show any work", they just happened to come up with the correct answer.
_____________________________________________________________________

When someone like me comes along, wondering why vague advice is being given or why experts are going about problem solving in the most roundabout way, those experts are quick to get upset with me when I point out the issues. They tell me things like "What you say does not validate anything to me" and "Stop interjecting yourself half way through solving something" and "You don't need to be such a smart*** and correct every little thing I say wrong"..."You have some kind of superiority complex and you have to realize that there are other ways to solve the problem"..."Don't you realize there is more than one answer to the problem and that doesn't make yours right and mine wrong..." etc, etc, etc.

And furthermore, if they inherrently start ignoring the patterns and equations after they are made known, to keep going about things the way they have been, and because they don't like the person that pointed it all out and crashed the party.
____________________________________________________________

Ok. Did you read about the throttleing (OCP/OHP)? On the other thread you wrote
"In that thread, I did not read the whole thread but I did state as follows:" here "Because of this, I've read every post in this thread to see what was happening. It took almost 15 minutes. " BUT thats not the point. The point is as folows . The mobo has 4+2 power phases. yes it can handle a STOCK fx8150, maybe even a slight OC. but he want to go enthusiast on the 8150 and get as much as he can. IF he watercools the VRM`s maybe he can do it, BUT a single CPU wc loop is minimum 120E(here in the EU). Maybe he gets it stable somehow while a fan is blowing on the heatsink, but what will he do when the wear on the vrm`s is too much to take and he sees magik smoke and fire(perhaps 1 week after warranty expires), ain`t it just better to rma it and sell it (or keep it as backup) , and get a board that is adequate for his needs?

Don`t forget everyone here is trying to help each other. As for RGone he has followed th thread from post no.1 to post nr 160, and he told you what he thinks/knows as a guy who folowed the whole thing.

...Yes, I didn't read about it the first time, however I did think that if that wasn't covered, I'm afraid a new motherboard wouldn't help. Obviously it was covered so I don't understand why people would get butthurt over what I said, but do you do realize, there were about 120 of these 160 posts that were all bullcrap, because people were too quick to take random stabs at the problem?

Lets combine two idioms:
The math does not lie and the truth is in the pudding.(in this case, the math is logic)

1st idiom meaning:
If you do the math correctly, you will always get the same answers. A pattern becomes apparent.

2nd:
If the truth is in the pudding, you may have a hard time finding it right away.

To find the truth however (the answer to the math), you can either guess and check (the idea of pudding) or you can follow an applicable equation. If you follow the equation, the pudding does not matter, You would be able to find it the same with or without the pudding.

________________________________________________________________________

This thread, by the OP's screenshots and feedback, should have gone like this:
Post 1, initial problem

Post 2, gathering the extra data needed to solve the equation

Post 3, explaining/solving the equation and testing an answer

In 3 steps, the initial problem is solved. ---> (ILLEGAL SUMOUT at X.X clock solved by adding vcore because integer processor in CPU is outputting junk to Prime95 and it can not start a test)...

In this thread, that took something like 40 replies.
________________________________________________________________________

After that was solved, the OP said "Okay, but I want to go further now." (What should have been post 4.)

Post 5, advice is given on what to try to help further the OC

Post 6, OP comes back with issues freezing up after 15-20 minutes

Post 7, Problem solved. Advice is given to resolve the problem (not sure 100% what we said solved it, was it more vcore?) ...hard to juggle 160 posts and make sense of every single step taken whether I read it or not which is why I didn't even bother at first, it is still confusing. Either way, problem solved.

Post 8, OP confirms issue is resolved, but now cores are throttling under load.
________________________________________________________________________

Post 9, advice is given to add a high-speed fan directly on the VRM heatsink for stress testing, as the issue is caused by a protection mechanism.

Post 10, OP says they have done so, but the issue still occurs.

Post 11, advice is given for the OP to buy a new motherboard, because the current motherboard is not capable of sustaining the current draw without disabling OCP. (Which is unsafe, and is what kills motherboards.)

__________________________________________________________________________

Real world workloads are not Prime95, it is generally safe to put a fan over the VRM for Prime95 to solve the issue temporarily (so that the user can ensure CPU stability at a held clock) unless the user intends on running World Community Grid or Folding @ Home on their CPU 24 hours a day, 7 days a week.

Current draw during gaming, benchmarking, etc usually will never reach in peak what the current draw for Prime95 is sustained, therefore the VRM should never have an issue. Furthermore, the throttling will occur if temperatures rise too much again without the fan, or it will throttle regardless if it hits OCP, which ASUS sets as the peak current they feel is safe going through the VRM. Maybe it would die in 3 years in that case, but so could any other motherboard. ASUS sets their OTP at 90c which is rather conservative too, considering the MOSFETS should be rated at 125c in the very cheapest, or 175/180c for higher quality parts.

I am here to help people.

However, when I'm trying to help people, just like everyone else, and the threads for which the people I want to help out too, just like everyone else, become 160 posts long and it seems like the "forum experts" are just ****ing around, I start wondering what is going on.

Three months ago, I tried going directly to those "experts", or, those around here that must know what they are doing, and tried to explain either in the threads or in PMs that they could go about it much quicker and that they should start noticing patterns and specific, repeatable behaviors if they pay attention to the details. However, to ANYONE, a single answer was thrown back at me. "Just because you state your opinion as fact doesn't mean it's true, and those here have experienced otherwise."

Except, they haven't. They just repeat what the next guy says and think they experienced it too.

 

[oldbrave]

Now this man^^ speaks words if IRON !!!

 

[DevinR]

i heard overclocking with that motherboard is a nightmare

 

[tungureanu]

I ain`t no EXPERT, but i have experience with the 8150 and A64 x2 6000+ windsor.+asus mobos. If you look around, 80% of my posts , are in 8150 related threads and 20% in more general issue threads. In a last attept to help the OP , i gave him screenshots of mt bios, in the hope he can adapt my settings to his CPU+mobo. What i wrote him was a step by step( because i don`t mind). i do agree there are a lot of BS post but personally I don`t mind.When I "took over"the help desk, i started with let`s go to default and we go from there. Every step i told him to do was what i did (yes i didn`t explain why to do it), i the hope i could get him to 4.4. When nothing i tried worked i gave him the advice i whould give myself and anyone else who has the same issues on that mobo (new mobo). But if you look around the forums,those boards need to be avoided when running an 8150, because i`ve never seen on a forum so many people with the same components having the same issues. I`m sure the OP will get to 4.4-4.5 with his sabertooth , and in the end he will be happy. You shall seeeeeeeeeeeee.

Ps: why are we making this thread longer? BeepBeep if we are to continue this, lets start a new tread in general.

 

[Anonaru]

AMD didn't drop the ball on anything, your board simply lacks any type of useful Load-Line Control. Raising PLL voltage can help combat the symptoms of vdroop/low vcore and load stability. I'm not an engineer so I can't really explain how the PLL would affect stability that way but I've confirmed this phenomenon with several people in the past.

On Intel IMHO they dropped the ball on PLL voltage. When up around 54x multi+ it's a must to have PLL Overvoltage enabled and/or add PLL voltage...

With LLC enabled, upping PLL volts isn't going to do a whole lot for AMD Phenom II and FX, whether on air, or LN2. I've tested from 2.5 to 2.8v on air/water/LN2.

Now, I've missed 9 pages of discussion here and it seems the only solution or advice that's been given was for the OP to buy a new motherboard. Unfortunately, I think he will find that a new motherboard does not change things for him very much.

If the OP was only having the issue "ILLEGAL SUMOUT" as described in post #4, then the only thing that will solve the problem is more vcore at load, either by reducing any vdroop that may be occurring or raising vcore significantly all together. If a new motherboard has better LLC options (not sure what LLC does on the M5A97 R2.0, though I've tested the 1st version), then it will be more overclocker friendly (better calibrated Load-Line options), but outside of that there is nothing wrong with the board.

Well, LLC or not, the voltage preset for PLL seems to be off for the new FX processors, and the chipset doesn't do a good job of correcting this at higher OC. Intel requires the same sort of pampering on some of their chips (Though I had to LOWER my PLL to attain stability with my current i7's OC) You will find that (Not so much on this community, but try the .net varient) that, especially with the 8350s, adjusting PLL based on your OCing needs has generally been extremely helpful for stability-- An easy fix, even on boards with robust LLC (And a good PSU, for that matter.) There's no need to stick up for AMD, I'm not slandering them. I think both companies have some strange PLL issues with their newer chips, something I really couldn't hope to understand :S

There's a lot of information out there stating the board he had from the get-go doesn't allow for a decent OC with the 8150 or 8350, and the main reason was the board's inability to supply a steady / sufficient current at higher OCs. His sumout problem could be fixed by adding v-core, but then he would freeze due to lord knows what (Whatever many number of things that could fail due to vdroop)-- a sabertooth is definitly not going to have this issue, so if he has a decent chip and a sabertooth, it will indeed make a substantial difference between that setup and one that cannot supply a clean enough current for the chip at hand.

As for your other post, I won't go into depth-- I'm not a fan of stirring up unnecessary strife. I would just like to hint at your obvious omissions, especially from that post to this one. Increasing vCore alone wasn't doing the job, this was long ago hit at and addressed. I DO have experience with the 8150, and I even have some experience with the very board he's having issues with. I am not sure if you have objections with the way I helped him, but you can't go hard on and direct with somebody who initially didn't know where the CPU-NB voltage was at on his motherboard, surely you understand that.

 

[miey98se]

hi there, i need some advice on oc 8150, i'm not an expert in the field of oc. anybody using almost same as mine. right now planning to buy h100 to replace my cm v8, with only using multiple to clock up 8150 to 4.2 gzh. Any suggestion. i truly appreciate if somebody can tell me from zero basic.

thanks

http://imageshack.us/photo/my-images/84/49709923.png/

MOBO : ASUS CROSSHAIR V FORMULA
CPU : AMD 8150
CPU COOLER : CM V8
POWER SUPPLY : CM 750
GPU : ASUS DIRECT CUII 6970
MEMORY : CORSAIR VENGENCE 12 GB (3 STICK, BURN 1 SLOT)
HDDs/Optical drives: 4.5 TERA, SATA DRIVE, NO RAID
SOUND DEVICE : Onboard
USB > RAZER TRON KEYBOARD, RAZER NOSTROMO, ALIEN G9, LOGITECH G700, AND ROCAT
O/S : WINDOW 8 64 BIT
Case > Brand/Model/Fans Installed?

 

[tungureanu]

hi there, i need some advice on oc 8150, i'm not an expert in the field of oc. anybody using almost same as mine. right now planning to buy h100 to replace my cm v8, with only using multiple to clock up 8150 to 4.2 gzh. Any suggestion. i truly appreciate if somebody can tell me from zero basic.

thanks

http://imageshack.us/photo/my-images/84/49709923.png/

MOBO : ASUS CROSSHAIR V FORMULA
CPU : AMD 8150
CPU COOLER : CM V8
POWER SUPPLY : CM 750
GPU : ASUS DIRECT CUII 6970
MEMORY : CORSAIR VENGENCE 12 GB (3 STICK, BURN 1 SLOT)
HDDs/Optical drives: 4.5 TERA, SATA DRIVE, NO RAID
SOUND DEVICE : Onboard
USB > RAZER TRON KEYBOARD, RAZER NOSTROMO, ALIEN G9, LOGITECH G700, AND ROCAT
O/S : WINDOW 8 64 BIT
Case > Brand/Model/Fans Installed?

start a new thread. and we will go from there

by the way

to OCF