Gaming 7 6600k OC Voltage

[GreenSmoke]

I am using the Gigabyte G1 Gaming 7 motherboard to overclock a i5 6600k. Right now I have the i5 running at 4.6Ghz with vcore set to 1.35v in BIOS. Lowering the voltage to even 1.33v will cause my cpu to not be stable as only 3/4cores run to their max potential during prime95 testing. My question is, since I have my vcore set at 1.35, during idle, CPU-Z is saying it's only pulling 1.32v, which is fine, and i've seen it top out at (I think) 1.332v or something very close to that, but what I really don't understand is this:

When I run a prime95 stress test, play any game, or just in general do anything somewhat CPU intensive, my voltage runs below 1.3 and my system remains stable. Right now i am running a prime95 stress test as I type this and my vcore according to CPU-Z is sitting right on 1.236 and sometimes boosting to 1.248v, even though my BIOS is set at 1.35v. The literal second i press stop on prime95 my vcore boosts up to 1.320v again during idle according to cpu-z.

Basically what I want to know is, since my cpu actually runs below 1.3v during 100% usage, why do I NEED to have it set at 1.350v in the BIOS to get my CPU to use 4/4 cores, since running 1.330V will eventually cause 1 core to almost quit working during prime95, although it still does something sitting at like 20% usage instead of 100%. I have never even seen my vcore utilize the entire 1.35v I have it set to in the BIOS either. I've searched my BIOS for energy modes such as adaptave, high performance, etc, but I can't find anything on it. I'd like to lower my voltage or even up the frequency (at this 1.35v setting) even more if possible since I don't see why it idles at 1.32v but goes under 1.3v during full load. I can't run 4.7Ghz stable at this setting so 4.6 is as high as I can go, and I'm not willing to push my vcore any higher.

I'm a beginner overclocker, just built this system about 2 weeks ago, but I have done a ton of research over the internet about overclocking on the z170 system so I have a pretty good idea I know what I'm doing. I'm happy with my 4.6Ghz with 1.35vcore setting, tempertures are great, but i'm either missing a piece of information I should have known before i overclocked, or something isn't quite right since it idles with more volts than it uses during a full load.

Please, somebody out there has the same motherboard and CPU, Does this happen to you? And is there a better solution for my overclocking voltage settings?

 

[John_485]

Hi John,

The MSI is truly a great graphics card and would be my choice too !!! As I wrote above, I don't believe that your CPU is sub-par but if you want to take a chance in replacing it, go for it (I often take risks like that too). However, I don't want to push it without knowing what you are going to get, because the replacement chip could very well be much worse (although unlikely IMO). If you do choose to replace it, let me know and wait until I give you a word of advice on the way you best 'paste' your EVO Cooler on it. This is important for heat dissipation and the overall efficiency of the CPU Cooler.

This is Friday, I have to sign-off now. Please let me know what you decided to do.

Hey Phil, sorry for the belated reply brother.

Well, I didn't take back the processor or the cpu. Technically, due to bad motherboards, switching up processors, adding new fans, etc. I've built about 5 systems in the past two weeks and I'm kind of sick of pulling my rig apart and then putting it all back together again, simply to exchange it for yet another board. I decided to keep what I have and try and figure out a way to get to the best overclocking configuration possible. I do plan on buying the Micro Center extended Warranty, which gives me the option to swap out any part I like. So if the processor fries or I just feel like I lost the silicon lottery I can go exchange it, no questions asked.

I'm glad you brought up my cooler. I was thinking on trying some new thermal paste. I've used the old standby (Arctic Silver 5) for over a decade (maybe even longer?) and its always treated me good. I've read a few articles and looked at some reviews and I hear something called Thermal Grizzly Kryonaut is the best out there right now, and Gelid is right behind it, then Arctic MX-4. For whatever reason, Gelid is what I plan on buying.

I really didn't do anything special when I installed my 212 Hyper, I just put a little rice-shaped blob of thermal paste (approximately the same mass as a small pea) onto the processor heat-sink and then torque down the screws using a diagonal torque pattern (top left, bottom right, etc.) until the 212 was torqued down tightly, yet not so tight to damage anything. My temps have been totally average (idle at 21c-27c and under load at 40c-60c). I'm not counting stress test temps though, but they've been pretty good for the most part also, except doing a torture test at 4.4Ghz in Prime95, when I hit 85c for a second and stabilized at around 82c during the hottest part of the test.

(Note: The last time I installed a cooler I used the "spread" method, meaning I applied the thermal paste across the entire surface of the processor heat sink. I guess I got lazy, hence the "pea" method, then tighten).

I'm interested in hearing what you used and your method of properly affixing the cooler to the processor? Oh yeah! I finally installed that Cougar 140mm fan and so far I'm not really impressed, nor am I sure if I should be? It seems like it pushes less air then the little 120mm fan did, at least that's how it feels when I put my hand over it and compare it to what I remember the 120mm feeling like.

And what is really strange is the smell. Yes, the smell!

After installing that fan I began smelling a very distinct smell, similar to ozone or some type of strange oil. I checked to make sure nothing was burning, seized, etc. and found no problems. I even nosed my power supply to make sure the smell wasn't coming from there. Finally my nose led me right to the Cougar fan. It is a hydraulic bearning design so maybe that is playing a role in the smell? I'm not sure if the smell has lessened overnight or if I've gotten used to it, to be honest. I'm just hoping it's part of the break-in period but I have my suspicions (My wife says the smell has lessened). I used the left over 120mm fan as a "pull" fan on the other end of my 212 Hyper. I didn't notice a difference it temps either way, to be honest. Maybe my case isn't designed for optimal airflow (it does have some strange slot covers, lots of gaps and drilled "vent" holes everywhere at the back of the case)?

So far, the best I can do is an overclock of 4.2GHz that is fairly stable (meaning it passes 15 minutes of Prime95 without any errors) Prime95 was the only stress test I couldn't pass before, so any amount of success there is a good sign. I've been fooling around a little bit today with 4.4GHz and it ended badly. My mouse froze a few times in the Bios, something I've never experienced before so I figured it had something to do with an odd setting and the semi-high voltage I set (1.40v with an Override offset+ of 0.030mv to start with). I just decided I am not ready for 4.4GHz and am trying to get a 4.3GHZ overclock stable.

The best overclock (in terms of speed and performance) I had was a 4.3GHz OC but I forgot to save the configuration! Man am I a moroon (as Bugs Bunny would say)! It wasn't stable, according to Prime95 and the high setting on Intel Burn test but it worked nicely on my desktop. I never did get around to testing it with a game though, something I plan on adding to my testing regiment. I'm basing performance off of Cinebench runs, just like you told me to do. It's a good indicator of how well the overclock is in the beginning stages. If an overclock can't pass the Cinebench test then it isn't very stable at all.

Anyway, according to my results my chip is slightly sub-par (lottery loser) or this MSI M5 mobo is the culprit, which some people say is the most likely cause, aside from a bad chip. I can't believe that though, I mean this motherboard has a 12-2 power phase so it should be golden, unless MSI needs to bump out a Bios update to fix problems they are aware of but their customers are not? Most motherboard manufacturers don't like to talk about problems with their products until they have already fixed them. That means I'll be in the dark for awhile, wondering where the problem lies.

I'd like to hear what you think Phil,

-laz.

 

[grimsin]

When I has a 212 cooler I tightened it till it wouldn't tighten any further, I believe the screws don't stop turning but just keep turning and you feel a "clicking"... If that makes any sense lol

 

[philipew]

Hi John,

The MSI is truly a great graphics card and would be my choice too !!! As I wrote above, I don't believe that your CPU is sub-par but if you want to take a chance in replacing it, go for it (I often take risks like that too). However, I don't want to push it without knowing what you are going to get, because the replacement chip could very well be much worse (although unlikely IMO). If you do choose to replace it, let me know and wait until I give you a word of advice on the way you best 'paste' your EVO Cooler on it. This is important for heat dissipation and the overall efficiency of the CPU Cooler.

This is Friday, I have to sign-off now. Please let me know what you decided to do.

Hey Phil, sorry for the belated reply brother.

Well, I didn't take back the processor or the cpu. Technically, due to bad motherboards, switching up processors, adding new fans, etc. I've built about 5 systems in the past two weeks and I'm kind of sick of pulling my rig apart and then putting it all back together again, simply to exchange it for yet another board. I decided to keep what I have and try and figure out a way to get to the best overclocking configuration possible. I do plan on buying the Micro Center extended Warranty, which gives me the option to swap out any part I like. So if the processor fries or I just feel like I lost the silicon lottery I can go exchange it, no questions asked.

I'm glad you brought up my cooler. I was thinking on trying some new thermal paste. I've used the old standby (Arctic Silver 5) for over a decade (maybe even longer?) and its always treated me good. I've read a few articles and looked at some reviews and I hear something called Thermal Grizzly Kryonaut is the best out there right now, and Gelid is right behind it, then Arctic MX-4. For whatever reason, Gelid is what I plan on buying.

I really didn't do anything special when I installed my 212 Hyper, I just put a little rice-shaped blob of thermal paste (approximately the same mass as a small pea) onto the processor heat-sink and then torque down the screws using a diagonal torque pattern (top left, bottom right, etc.) until the 212 was torqued down tightly, yet not so tight to damage anything. My temps have been totally average (idle at 21c-27c and under load at 40c-60c). I'm not counting stress test temps though, but they've been pretty good for the most part also, except doing a torture test at 4.4Ghz in Prime95, when I hit 85c for a second and stabilized at around 82c during the hottest part of the test.

(Note: The last time I installed a cooler I used the "spread" method, meaning I applied the thermal paste across the entire surface of the processor heat sink. I guess I got lazy, hence the "pea" method, then tighten).

I'm interested in hearing what you used and your method of properly affixing the cooler to the processor? Oh yeah! I finally installed that Cougar 140mm fan and so far I'm not really impressed, nor am I sure if I should be? It seems like it pushes less air then the little 120mm fan did, at least that's how it feels when I put my hand over it and compare it to what I remember the 120mm feeling like.

And what is really strange is the smell. Yes, the smell!

After installing that fan I began smelling a very distinct smell, similar to ozone or some type of strange oil. I checked to make sure nothing was burning, seized, etc. and found no problems. I even nosed my power supply to make sure the smell wasn't coming from there. Finally my nose led me right to the Cougar fan. It is a hydraulic bearning design so maybe that is playing a role in the smell? I'm not sure if the smell has lessened overnight or if I've gotten used to it, to be honest. I'm just hoping it's part of the break-in period but I have my suspicions (My wife says the smell has lessened). I used the left over 120mm fan as a "pull" fan on the other end of my 212 Hyper. I didn't notice a difference it temps either way, to be honest. Maybe my case isn't designed for optimal airflow (it does have some strange slot covers, lots of gaps and drilled "vent" holes everywhere at the back of the case)?

So far, the best I can do is an overclock of 4.2GHz that is fairly stable (meaning it passes 15 minutes of Prime95 without any errors) Prime95 was the only stress test I couldn't pass before, so any amount of success there is a good sign. I've been fooling around a little bit today with 4.4GHz and it ended badly. My mouse froze a few times in the Bios, something I've never experienced before so I figured it had something to do with an odd setting and the semi-high voltage I set (1.40v with an Override offset+ of 0.030mv to start with). I just decided I am not ready for 4.4GHz and am trying to get a 4.3GHZ overclock stable.

The best overclock (in terms of speed and performance) I had was a 4.3GHz OC but I forgot to save the configuration! Man am I a moroon (as Bugs Bunny would say)! It wasn't stable, according to Prime95 and the high setting on Intel Burn test but it worked nicely on my desktop. I never did get around to testing it with a game though, something I plan on adding to my testing regiment. I'm basing performance off of Cinebench runs, just like you told me to do. It's a good indicator of how well the overclock is in the beginning stages. If an overclock can't pass the Cinebench test then it isn't very stable at all.

Anyway, according to my results my chip is slightly sub-par (lottery loser) or this MSI M5 mobo is the culprit, which some people say is the most likely cause, aside from a bad chip. I can't believe that though, I mean this motherboard has a 12-2 power phase so it should be golden, unless MSI needs to bump out a Bios update to fix problems they are aware of but their customers are not? Most motherboard manufacturers don't like to talk about problems with their products until they have already fixed them. That means I'll be in the dark for awhile, wondering where the problem lies.

I'd like to hear what you think Phil,

-laz.

Hi John,

You won't hear much I am afraid because this is in writing ;-) but I hope is is still useful to you and whoever reads it.

It's good to know that you still hang in there... and without continuously swapping parts with the store. Concerning you cooler, I prefer the spread method to make sure that the entire surface (including the very edge of each corner) of the chip is in good contact with the (preferably copper) pipes of the cooler. Just dropping a pea-size blob in the centre does not even guarantee that every edge of the four corners of the CPU plate does actually get any paste at all, which is not likely to contribute anything towards cooling. The cooler should also ideally have a spring arrangement so that the pressure on the CPU plate is not excessive and remains constant over time (like my Cooler Master Hyper 212 X).

As for your fan, the smell usually comes from lubricant running down on the carbon brushes of the motor, or lubricant on the rotor and is quite common with "hydraulic" (lots of oil) fans in particular. It should go away after a week or so. Women (my wife is the same) have a more developed sense of smell (yes, that too) than men in general. Dogs are still better but can't tell you if it's from the fan ;-)

Manufacturers of fans tend to exaggerate the CFM (air transfer) performance (like they do with cars) to sell more. As it is not regulated (unlike cars), they can claim pretty much what they want on paper. That's why I only buy fans from reputed (well known) brands who do not have to use those tricks to sell, like Corsair, Thermaltake or of course the "BMW" of them: Noctua (quite expensive though IMO), even for a few dollars more. However, (like for cars) price 'alone' is not a reliable indication of performance / longevity, beware. But there is no need to go out and splash a handful of cash on fans with fancy names, physics-defying performance and Xmas-tree lighting (been there).

I still believe that in your particular situation, because this exhaust fan is very quiet, located at the back of the case (can hardly hear it) and you need absolutely all the cooling you can get most of the time (who doesn't), you should preferably connect it to a 12 V Molex plug coming directly from the PSU, let it run at full speed continuously (you can never have 'too much' cooling...) and put it vertical, just parallel to the CPU Cooler fan which it will assist then.

For your PC case, leave the intake dust filters at the bottom and front vents (no need at the top), remove (cut off with pliers if necessary) anything (perforated plates for example) in the way of exhaust fans, and close any opening (like unused fan locations, unused PCI slots at the back, holes between the surrounding edges of a fan and the case, etc.) which is not there to be used by either an intake or an exhaust fan with (black) duct tape (great invention), so that YOU control the air flow as being: entry exclusively via the designated front intake fan and exit exclusively at the back top corner via exhaust fan, nothing else. Of course there will always be a few fine cracks and crevices (the case is not water proof) but that's OK. By the way, the (apparent) weakness of your fan could also come from a lower air pressure, as with a more concentrated air flow, the pressure of 120mm fans tends to be a little higher, perhaps to compensate for their reduced size (just saying). I wouldn't worry too much for now.

For my Fractal Design case, all intake fans are forced to take exclusively external air (not uselessly re-stirring hot internal air) with self-adhesive (dark grey) foam (common house window-seal bought from a local hardware store) fitted tightly all around their outside edges. A plastic separator is used to separate the cool air intake (at the front) from the hot air exhaust (at the back) of the two SP fans at the top of the CPU (both at the top panel of the case) as can be seen on the picture of my case in a previous post - updated picture: http://imageshack.com/a/img923/3196/PjaqTS.png

To give you a pointer to performance, although not all processor chips are identical as you well know, I initially got 4.2 GHz by not changing anything in the BIOS either than the CPU multiplier from 35 to 42, then 4.3 GHz by also setting VCore at 1.29 GHz, nothing else, and 4.5 GHz with VCore at 1.35 V still with no other change (except FCLK and XMP). It's only when I started playing with the LLC (offset) that I could get 4.6 GHz at VCore 1.355 V but then also changing other parameters (8 in total) as I described (at length) in an earlier post of this thread.

If I were you, I would perform a reset of all the BIOS settings (take the battery of the MoBo out for 30 seconds), and restart fresh (following what I explained previously) because I think that by now you would have changed (and re-changed) so many parameters that you may have left something behind, like "an odd setting" as you wrote, which prevents you from effectively reaching 4.4 GHz with VCore set around 1.35 V. I am convinced you can do it... "seek and you shall find". Bugs Bunny is wrong.

If, as I think, your MoBo is OK (just make sure that you updated it to the latest BIOS - search on the internet), your chip is OK (does not have to win the million dollar silicon lottery for this), your cooler is OK, your control of the cooling (airflow in the case) is OK, the only thing to consider now is your Power Supply (PSU) and it's stability in particular (more than its max power delivery). This is why I bought a Corsair (yes again, I know) RM650x 650W 80PLUS Gold Modular. Because the stability of the voltage is related to the max. frequency (GHz) you can get on the CPU. It all becomes quite jittery when high (an oscilloscope gives you a good idea of this).

Cinebench is used for benchmarking (i.e. officially determining) the performance you have reached with your CPU (don't use the OpenGL setting yet), not for testing stability. By the way, what score did you get? You should be somewhere around 735. At the moment I consistently reach 767 at 4.6 GHz which is good for my particular setup but not 'truly crazy mind-blowing' either. There are much higher results out there, mostly because of better cooling.

The "extended Warranty" for the hardware is a good idea, I wish we had that here too. The last piece of advice I can give you at this time is to proceed methodically, one thing at a time, take notes, save your profiles, stay in control and measure your progress as you go. You'll get there.

 

[John_485]

You won't hear much I am afraid because this is in writing ;-) but I hope is is still useful to you and whoever reads it.

It's good to know that you still hang in there... and without continuously swapping parts with the store. Concerning you cooler, I prefer the spread method to make sure that the entire surface (including the very edge of each corner) of the chip is in good contact with the (preferably copper) pipes of the cooler. Just dropping a pea-size blob in the centre does not even guarantee that every edge of the four corners of the CPU plate does actually get any paste at all, which is not likely to contribute anything towards cooling. The cooler should also ideally have a spring arrangement so that the pressure on the CPU plate is not excessive and remains constant over time (like my Cooler Master Hyper 212 X).

As for your fan, the smell usually comes from lubricant running down on the carbon brushes of the motor, or lubricant on the rotor and is quite common with "hydraulic" (lots of oil) fans in particular. It should go away after a week or so. Women (my wife is the same) have a more developed sense of smell (yes, that too) than men in general. Dogs are still better but can't tell you if it's from the fan ;-)

Manufacturers of fans tend to exaggerate the CFM (air transfer) performance (like they do with cars) to sell more. As it is not regulated (unlike cars), they can claim pretty much what they want on paper. That's why I only buy fans from reputed (well known) brands who do not have to use those tricks to sell, like Corsair, Thermaltake or of course the "BMW" of them: Noctua (quite expensive though IMO), even for a few dollars more. However, (like for cars) price 'alone' is not a reliable indication of performance / longevity, beware. But there is no need to go out and splash a handful of cash on fans with fancy names, physics-defying performance and Xmas-tree lighting (been there).

I still believe that in your particular situation, because this exhaust fan is very quiet, located at the back of the case (can hardly hear it) and you need absolutely all the cooling you can get most of the time (who doesn't), you should preferably connect it to a 12 V Molex plug coming directly from the PSU, let it run at full speed continuously (you can never have 'too much' cooling...) and put it vertical, just parallel to the CPU Cooler fan which it will assist then.

For your PC case, leave the intake dust filters at the bottom and front vents (no need at the top), remove (cut off with pliers if necessary) anything (perforated plates for example) in the way of exhaust fans, and close any opening (like unused fan locations, unused PCI slots at the back, holes between the surrounding edges of a fan and the case, etc.) which is not there to be used by either an intake or an exhaust fan with (black) duct tape (great invention), so that YOU control the air flow as being: entry exclusively via the designated front intake fan and exit exclusively at the back top corner via exhaust fan, nothing else. Of course there will always be a few fine cracks and crevices (the case is not water proof) but that's OK. By the way, the (apparent) weakness of your fan could also come from a lower air pressure, as with a more concentrated air flow, the pressure of 120mm fans tends to be a little higher, perhaps to compensate for their reduced size (just saying). I wouldn't worry too much for now.

For my Fractal Design case, all intake fans are forced to take exclusively external air (not uselessly re-stirring hot internal air) with self-adhesive (dark grey) foam (common house window-seal bought from a local hardware store) fitted tightly all around their outside edges. A plastic separator is used to separate the cool air intake (at the front) from the hot air exhaust (at the back) of the two SP fans at the top of the CPU (both at the top panel of the case) as can be seen on the picture of my case in a previous post - updated picture: http://imageshack.com/a/img923/3196/PjaqTS.png

To give you a pointer to performance, although not all processor chips are identical as you well know, I initially got 4.2 GHz by not changing anything in the BIOS either than the CPU multiplier from 35 to 42, then 4.3 GHz by also setting VCore at 1.29 GHz, nothing else, and 4.5 GHz with VCore at 1.35 V still with no other change (except FCLK and XMP). It's only when I started playing with the LLC (offset) that I could get 4.6 GHz at VCore 1.355 V but then also changing other parameters (8 in total) as I described (at length) in an earlier post of this thread.

If I were you, I would perform a reset of all the BIOS settings (take the battery of the MoBo out for 30 seconds), and restart fresh (following what I explained previously) because I think that by now you would have changed (and re-changed) so many parameters that you may have left something behind, like "an odd setting" as you wrote, which prevents you from effectively reaching 4.4 GHz with VCore set around 1.35 V. I am convinced you can do it... "seek and you shall find". Bugs Bunny is wrong.

If, as I think, your MoBo is OK (just make sure that you updated it to the latest BIOS - search on the internet), your chip is OK (does not have to win the million dollar silicon lottery for this), your cooler is OK, your control of the cooling (airflow in the case) is OK, the only thing to consider now is your Power Supply (PSU) and it's stability in particular (more than its max power delivery). This is why I bought a Corsair (yes again, I know) RM650x 650W 80PLUS Gold Modular. Because the stability of the voltage is related to the max. frequency (GHz) you can get on the CPU. It all becomes quite jittery when high (an oscilloscope gives you a good idea of this).

Cinebench is used for benchmarking (i.e. officially determining) the performance you have reached with your CPU (don't use the OpenGL setting yet), not for testing stability. By the way, what score did you get? You should be somewhere around 735. At the moment I consistently reach 767 at 4.6 GHz which is good for my particular setup but not 'truly crazy mind-blowing' either. There are much higher results out there, mostly because of better cooling.

The "extended Warranty" for the hardware is a good idea, I wish we had that here too. The last piece of advice I can give you at this time is to proceed methodically, one thing at a time, take notes, save your profiles, stay in control and measure your progress as you go. You'll get there.

Hey brother Phil,

I noticed you have one Noctua fan (92mm I believe?). Did it impress you because I was thinking of replacing this 140mm with a high powered Noctua 120mm for top exhaust or back exhaust. Remember, I used the fan that was doing top exhaust work and moved to become a pull through for my CPU cooler (one push and one pull on my Cooler Master 212 Hyper LED). The Noctua I'm thinking about getting is expensive but if the specifications don't lie then it's one helluva fan!

It's funny that you mentioned cutting things away on the exhaust ports of my case as I was thinking the same thing about two days ago. I was looking at the top exhaust design which has honeycombing holes at approximately 3.00mm in size to match a fan anywhere between 120mm and 140 mm. How much air is being pushed back into the case because of this design? Same setup for the back exhaust. If I removed the metal honeycombing on the back exhaust I wonder if the cooling efficiency of my rig would see a significant gain in performance?

About resetting the Bios. I can hit F6 and it will load what the system calls "Optimized Defaults". Wouldn't it be easier to hit F6 rather then taking out the CMOS battery or shorting the CMOS jumper? Wouldn't doing that give me the same optimized defaults?

My best Cinebench score was well below what you expected, it was 685 and that isn't even what I'm hitting right now. I'd have to run the bench test and tell you what my score is according to the overclocking profile I have loaded. Right now I'm running at a steady 4.0GHz. I could load another OC profile, say 4.3GHz and then run Cinebench and tell you the score (but I doubt I'd get much higher 680 or so since that was the last score I remember from Cinebench at 4.3GHZ). Now I'm a bit concerned, having not passed 700 on this rig makes me feel like things are running very efficiently.

I'm confident that if I removed my CPU cooler the thermal past would most likely not have reached each corner. It would look more like a circle, laving at least 40% of the processor without coverage. Now you've got me thinking I should reapply my paste (and I think it would be a good idea, I just doubt Micro Center carries anything better then Arctic Silver 5. I'd rather buy the Arctic MX4 if it is available).

Heat is the enemy of overclocking, no doubt about that, though my heats doesn't seem excessive, it isn't great either. I did a game test the other night and my processor was about 55c-60c at load. Not bad but definitely a lot of room for improvement.

-laz.

 

[philipew]

You won't hear much I am afraid because this is in writing ;-) but I hope is is still useful to you and whoever reads it.

It's good to know that you still hang in there... and without continuously swapping parts with the store. Concerning you cooler, I prefer the spread method to make sure that the entire surface (including the very edge of each corner) of the chip is in good contact with the (preferably copper) pipes of the cooler. Just dropping a pea-size blob in the centre does not even guarantee that every edge of the four corners of the CPU plate does actually get any paste at all, which is not likely to contribute anything towards cooling. The cooler should also ideally have a spring arrangement so that the pressure on the CPU plate is not excessive and remains constant over time (like my Cooler Master Hyper 212 X).

As for your fan, the smell usually comes from lubricant running down on the carbon brushes of the motor, or lubricant on the rotor and is quite common with "hydraulic" (lots of oil) fans in particular. It should go away after a week or so. Women (my wife is the same) have a more developed sense of smell (yes, that too) than men in general. Dogs are still better but can't tell you if it's from the fan ;-)

Manufacturers of fans tend to exaggerate the CFM (air transfer) performance (like they do with cars) to sell more. As it is not regulated (unlike cars), they can claim pretty much what they want on paper. That's why I only buy fans from reputed (well known) brands who do not have to use those tricks to sell, like Corsair, Thermaltake or of course the "BMW" of them: Noctua (quite expensive though IMO), even for a few dollars more. However, (like for cars) price 'alone' is not a reliable indication of performance / longevity, beware. But there is no need to go out and splash a handful of cash on fans with fancy names, physics-defying performance and Xmas-tree lighting (been there).

I still believe that in your particular situation, because this exhaust fan is very quiet, located at the back of the case (can hardly hear it) and you need absolutely all the cooling you can get most of the time (who doesn't), you should preferably connect it to a 12 V Molex plug coming directly from the PSU, let it run at full speed continuously (you can never have 'too much' cooling...) and put it vertical, just parallel to the CPU Cooler fan which it will assist then.

For your PC case, leave the intake dust filters at the bottom and front vents (no need at the top), remove (cut off with pliers if necessary) anything (perforated plates for example) in the way of exhaust fans, and close any opening (like unused fan locations, unused PCI slots at the back, holes between the surrounding edges of a fan and the case, etc.) which is not there to be used by either an intake or an exhaust fan with (black) duct tape (great invention), so that YOU control the air flow as being: entry exclusively via the designated front intake fan and exit exclusively at the back top corner via exhaust fan, nothing else. Of course there will always be a few fine cracks and crevices (the case is not water proof) but that's OK. By the way, the (apparent) weakness of your fan could also come from a lower air pressure, as with a more concentrated air flow, the pressure of 120mm fans tends to be a little higher, perhaps to compensate for their reduced size (just saying). I wouldn't worry too much for now.

For my Fractal Design case, all intake fans are forced to take exclusively external air (not uselessly re-stirring hot internal air) with self-adhesive (dark grey) foam (common house window-seal bought from a local hardware store) fitted tightly all around their outside edges. A plastic separator is used to separate the cool air intake (at the front) from the hot air exhaust (at the back) of the two SP fans at the top of the CPU (both at the top panel of the case) as can be seen on the picture of my case in a previous post - updated picture: http://imageshack.com/a/img923/3196/PjaqTS.png

To give you a pointer to performance, although not all processor chips are identical as you well know, I initially got 4.2 GHz by not changing anything in the BIOS either than the CPU multiplier from 35 to 42, then 4.3 GHz by also setting VCore at 1.29 GHz, nothing else, and 4.5 GHz with VCore at 1.35 V still with no other change (except FCLK and XMP). It's only when I started playing with the LLC (offset) that I could get 4.6 GHz at VCore 1.355 V but then also changing other parameters (8 in total) as I described (at length) in an earlier post of this thread.

If I were you, I would perform a reset of all the BIOS settings (take the battery of the MoBo out for 30 seconds), and restart fresh (following what I explained previously) because I think that by now you would have changed (and re-changed) so many parameters that you may have left something behind, like "an odd setting" as you wrote, which prevents you from effectively reaching 4.4 GHz with VCore set around 1.35 V. I am convinced you can do it... "seek and you shall find". Bugs Bunny is wrong.

If, as I think, your MoBo is OK (just make sure that you updated it to the latest BIOS - search on the internet), your chip is OK (does not have to win the million dollar silicon lottery for this), your cooler is OK, your control of the cooling (airflow in the case) is OK, the only thing to consider now is your Power Supply (PSU) and it's stability in particular (more than its max power delivery). This is why I bought a Corsair (yes again, I know) RM650x 650W 80PLUS Gold Modular. Because the stability of the voltage is related to the max. frequency (GHz) you can get on the CPU. It all becomes quite jittery when high (an oscilloscope gives you a good idea of this).

Cinebench is used for benchmarking (i.e. officially determining) the performance you have reached with your CPU (don't use the OpenGL setting yet), not for testing stability. By the way, what score did you get? You should be somewhere around 735. At the moment I consistently reach 767 at 4.6 GHz which is good for my particular setup but not 'truly crazy mind-blowing' either. There are much higher results out there, mostly because of better cooling.

The "extended Warranty" for the hardware is a good idea, I wish we had that here too. The last piece of advice I can give you at this time is to proceed methodically, one thing at a time, take notes, save your profiles, stay in control and measure your progress as you go. You'll get there.

Hey brother Phil,

I noticed you have one Noctua fan (92mm I believe?). Did it impress you because I was thinking of replacing this 140mm with a high powered Noctua 120mm for top exhaust or back exhaust. Remember, I used the fan that was doing top exhaust work and moved to become a pull through for my CPU cooler (one push and one pull on my Cooler Master 212 Hyper LED). The Noctua I'm thinking about getting is expensive but if the specifications don't lie then it's one helluva fan!

It's funny that you mentioned cutting things away on the exhaust ports of my case as I was thinking the same thing about two days ago. I was looking at the top exhaust design which has honeycombing holes at approximately 3.00mm in size to match a fan anywhere between 120mm and 140 mm. How much air is being pushed back into the case because of this design? Same setup for the back exhaust. If I removed the metal honeycombing on the back exhaust I wonder if the cooling efficiency of my rig would see a significant gain in performance?

About resetting the Bios. I can hit F6 and it will load what the system calls "Optimized Defaults". Wouldn't it be easier to hit F6 rather then taking out the CMOS battery or shorting the CMOS jumper? Wouldn't doing that give me the same optimized defaults?

My best Cinebench score was well below what you expected, it was 685 and that isn't even what I'm hitting right now. I'd have to run the bench test and tell you what my score is according to the overclocking profile I have loaded. Right now I'm running at a steady 4.0GHz. I could load another OC profile, say 4.3GHz and then run Cinebench and tell you the score (but I doubt I'd get much higher 680 or so since that was the last score I remember from Cinebench at 4.3GHZ). Now I'm a bit concerned, having not passed 700 on this rig makes me feel like things are running very efficiently.

I'm confident that if I removed my CPU cooler the thermal past would most likely not have reached each corner. It would look more like a circle, laving at least 40% of the processor without coverage. Now you've got me thinking I should reapply my paste (and I think it would be a good idea, I just doubt Micro Center carries anything better then Arctic Silver 5. I'd rather buy the Arctic MX4 if it is available).

Heat is the enemy of overclocking, no doubt about that, though my heats doesn't seem excessive, it isn't great either. I did a game test the other night and my processor was about 55c-60c at load. Not bad but definitely a lot of room for improvement.

-laz.

Hi John,

"brother Phil" makes me feel like a priest or something... ;-)
The little Noctua (from Austria) is fine and very quiet (they perform exactly as written on their box, no selling tricks, no lying). I took this one mostly because of its size though, just what I was looking for. When you write "used the fan that was doing top exhaust work and moved to become a pull through for my CPU cooler", this fan must be a SP (static pressure) fan which concentrates the airflow towards the aluminium blades of your heatsink. Otherwise it may be close to useless (also the one pulling the air out, although to a lesser extent).

The 120mm fan from Micro Center is very good (just look at the reviews) although it is a case fan, not a SP fan for a heatsink. It's quite expensive because we get it here for $30 too but in Australian dollars (!). And I don't know where they get their claimed "93.4 CFM" from because here it is quoted at only "55 cfm" [ https://www.ple.com.au/Products/611566/Noctua-NF-F12-PWM-120mm-PWM-Cooling-Fan ]. You can buy it here for as low as $28 AUD. Oh... I got it... it's only when you check on the Noctua site [ http://noctua.at/en/products/fan ] that you realise how Micro Center are quoting "93.4 m³/h" as being "93.4 CFM" which it is definitely not. It is "55 cfm" (do the conversion) as stipulated on the Australian site PLE which I trust. It's just another selling trick (I have seen it before, the error is always that way, never the other way around). But then I wonder what else they are up to... beware.

As for "cutting things away on the exhaust ports", it is particularly important for the exhaust fans, not so much for intake fans (air gets in easily without the fan having to suck harder). From my experience, fans can lose as much as 50% of their claimed CFM air transfer (what a waste) with certain grids, even having relatively large holes. Remember that just as for the claimed economy of a car, the claimed CFM of fan is measured in a lab under "optimum" (often unrealistic) conditions, including no grid whatsoever of course (otherwise they would have to give us the full specifications of that grid too). Exhaust air bounces on the perforated grid and creates a buffer zone where pressure increases and prevents air from escaping properly through the holes. I saw a significant improvement via cutting off, and so continued cutting off everything I could. It is also very good for the noise generated by the air being pushed through the holes (this may be true for intake fans too - that's why I cut them off too, leaving only the dust filters - largely sufficient - who needs more in front of fans?).

Resetting with F6 is fine IMO. Just go ahead, you don't have to remove the battery (an extreme solution), it will do the same.

The more you touch your CPU heatsink (like to replace a fan), the more likely you are to lift it ever so slightly from the plate of the CPU and create an air bubble. So after you have fitted your CPU fans and you are happy, it may be a good idea to refresh your paste (clean and re-apply - everywhere - in a very fine layer). The type of paste is not really that important, they are all more or less the same. The type of application and the quality of the application are much more important. After that, don't touch your CPU heatsink/fan combo anymore.

55 to 60 C is not bad, that's what I usually get (mind you, at 4.6 GHz after one hour of gaming at 90 fps). But anything up to about 75 C is OK. You are still a long way from that critical point. So you still have plenty of headroom to OC your rig. Cinebench performance will come after you reset your parameters, and go back to the BIOS to start methodically re-applying them one by one by following the settings that I have detailed previously. Even with a 'below average' CPU, you should get 4.4 GHz at around 1.31 V and 4.5 GHz at around 1.36 V. The LLC is what brought my voltage from 1.41 V to 1.355 V at 4.6 GHz. With that, of course the temp went right down, the fans slowed, the noise eased and the power consumption dropped.

Good luck with your project, it is a great challenge. Once you get a breakthrough you feel good.

 

[John_485]

Hi John,

"brother Phil" makes me feel like a priest or something... ;-)
The little Noctua (from Austria) is fine and very quiet (they perform exactly as written on their box, no selling tricks, no lying). I took this one mostly because of its size though, just what I was looking for. When you write "used the fan that was doing top exhaust work and moved to become a pull through for my CPU cooler", this fan must be a SP (static pressure) fan which concentrates the airflow towards the aluminium blades of your heatsink. Otherwise it may be close to useless (also the one pulling the air out, although to a lesser extent).

The 120mm fan from Micro Center is very good (just look at the reviews) although it is a case fan, not a SP fan for a heatsink. It's quite expensive because we get it here for $30 too but in Australian dollars (!). And I don't know where they get their claimed "93.4 CFM" from because here it is quoted at only "55 cfm" [ https://www.ple.com.au/Products/611566/Noctua-NF-F12-PWM-120mm-PWM-Cooling-Fan ]. You can buy it here for as low as $28 AUD. Oh... I got it... it's only when you check on the Noctua site [ http://noctua.at/en/products/fan ] that you realise how Micro Center are quoting "93.4 m³/h" as being "93.4 CFM" which it is definitely not. It is "55 cfm" (do the conversion) as stipulated on the Australian site PLE which I trust. It's just another selling trick (I have seen it before, the error is always that way, never the other way around). But then I wonder what else they are up to... beware.

As for "cutting things away on the exhaust ports", it is particularly important for the exhaust fans, not so much for intake fans (air gets in easily without the fan having to suck harder). From my experience, fans can lose as much as 50% of their claimed CFM air transfer (what a waste) with certain grids, even having relatively large holes. Remember that just as for the claimed economy of a car, the claimed CFM of fan is measured in a lab under "optimum" (often unrealistic) conditions, including no grid whatsoever of course (otherwise they would have to give us the full specifications of that grid too). Exhaust air bounces on the perforated grid and creates a buffer zone where pressure increases and prevents air from escaping properly through the holes. I saw a significant improvement via cutting off, and so continued cutting off everything I could. It is also very good for the noise generated by the air being pushed through the holes (this may be true for intake fans too - that's why I cut them off too, leaving only the dust filters - largely sufficient - who needs more in front of fans?).

Resetting with F6 is fine IMO. Just go ahead, you don't have to remove the battery (an extreme solution), it will do the same.

The more you touch your CPU heatsink (like to replace a fan), the more likely you are to lift it ever so slightly from the plate of the CPU and create an air bubble. So after you have fitted your CPU fans and you are happy, it may be a good idea to refresh your paste (clean and re-apply - everywhere - in a very fine layer). The type of paste is not really that important, they are all more or less the same. The type of application and the quality of the application are much more important. After that, don't touch your CPU heatsink/fan combo anymore.

55 to 60 C is not bad, that's what I usually get (mind you, at 4.6 GHz after one hour of gaming at 90 fps). But anything up to about 75 C is OK. You are still a long way from that critical point. So you still have plenty of headroom to OC your rig. Cinebench performance will come after you reset your parameters, and go back to the BIOS to start methodically re-applying them one by one by following the settings that I have detailed previously. Even with a 'below average' CPU, you should get 4.4 GHz at around 1.31 V and 4.5 GHz at around 1.36 V. The LLC is what brought my voltage from 1.41 V to 1.355 V at 4.6 GHz. With that, of course the temp went right down, the fans slowed, the noise eased and the power consumption dropped.

Good luck with your project, it is a great challenge. Once you get a breakthrough you feel good.

I'm definitely not giving up, not yet. But since I have no LLC (maybe MSI will add it in a BIOS revision in the near future?) I'm left with Override Offset and other tricks (that you've enlightened me to). I just can't imagine how a motherboard would limit the quality of an overclock? My board had twelve power phases plus two for memory and was built for overclocking/gaming so you'd think I wouldn't be having any problems, yet I am.

What about setting things to Adaptive to reign in the voltages during idle brother Phil (a bright light engulfs the Ozzie and peels of thunder are heard for miles around)?

I'd really love to see the voltage go down at idle and peak during load. A guy from another forum said a combination of Dynamic and Adaptive might do that but I'm a little skeptical. I guess I'll just have to try your basic recipe and then try other things if I'm still getting wonky results.

I'm at 4.0GHz right now with a Vcore of 1.280v in the Bios. My VID always reads 1.2197-8v. It never budges. I'm doubtful I'll reach 4.4GHZ at the voltages you listed but I will surely give it the college drop-out try. I really think I must be missing something because the voltage I'm at just to get to 4.0GHz is over the norm for this chip. There must be a problem I just have to identify where it lies. It could be the board or the Bios, a bad chip (doubtful) or something I've done without realizing it.

Amazing at how Micro Center took the liberty to lie like a rug in that product specification sheet. By the way, isn't that particular fan a made for cooling a CPU (Static Pressure Fan)?

I think you misunderstood my fan setup. The fan that used to be my top exhaust fan (120mm NZXT case fan) is now used as a pull fan on my cpu cooler. Remember though, my cpu cooler already has a fan at the front pushing air through. So the case fan is now doing duty helping to draw air away from the cpu heat-sink fins and pushing air directly towards the back exhaust . I really haven't seen any decrease in cpu temps since doing this so I doubt having two fans on a cpu cooler is really as effective as so many people believe. On the other hand, maybe I'm not seeing results because it isn't a SP fan?

Well, looking for that break through my friend,

-laz.

 

[philipew]

Hi John,

"brother Phil" makes me feel like a priest or something... ;-)
The little Noctua (from Austria) is fine and very quiet (they perform exactly as written on their box, no selling tricks, no lying). I took this one mostly because of its size though, just what I was looking for. When you write "used the fan that was doing top exhaust work and moved to become a pull through for my CPU cooler", this fan must be a SP (static pressure) fan which concentrates the airflow towards the aluminium blades of your heatsink. Otherwise it may be close to useless (also the one pulling the air out, although to a lesser extent).

The 120mm fan from Micro Center is very good (just look at the reviews) although it is a case fan, not a SP fan for a heatsink. It's quite expensive because we get it here for $30 too but in Australian dollars (!). And I don't know where they get their claimed "93.4 CFM" from because here it is quoted at only "55 cfm" [ https://www.ple.com.au/Products/611566/Noctua-NF-F12-PWM-120mm-PWM-Cooling-Fan ]. You can buy it here for as low as $28 AUD. Oh... I got it... it's only when you check on the Noctua site [ http://noctua.at/en/products/fan ] that you realise how Micro Center are quoting "93.4 m³/h" as being "93.4 CFM" which it is definitely not. It is "55 cfm" (do the conversion) as stipulated on the Australian site PLE which I trust. It's just another selling trick (I have seen it before, the error is always that way, never the other way around). But then I wonder what else they are up to... beware.

As for "cutting things away on the exhaust ports", it is particularly important for the exhaust fans, not so much for intake fans (air gets in easily without the fan having to suck harder). From my experience, fans can lose as much as 50% of their claimed CFM air transfer (what a waste) with certain grids, even having relatively large holes. Remember that just as for the claimed economy of a car, the claimed CFM of fan is measured in a lab under "optimum" (often unrealistic) conditions, including no grid whatsoever of course (otherwise they would have to give us the full specifications of that grid too). Exhaust air bounces on the perforated grid and creates a buffer zone where pressure increases and prevents air from escaping properly through the holes. I saw a significant improvement via cutting off, and so continued cutting off everything I could. It is also very good for the noise generated by the air being pushed through the holes (this may be true for intake fans too - that's why I cut them off too, leaving only the dust filters - largely sufficient - who needs more in front of fans?).

Resetting with F6 is fine IMO. Just go ahead, you don't have to remove the battery (an extreme solution), it will do the same.

The more you touch your CPU heatsink (like to replace a fan), the more likely you are to lift it ever so slightly from the plate of the CPU and create an air bubble. So after you have fitted your CPU fans and you are happy, it may be a good idea to refresh your paste (clean and re-apply - everywhere - in a very fine layer). The type of paste is not really that important, they are all more or less the same. The type of application and the quality of the application are much more important. After that, don't touch your CPU heatsink/fan combo anymore.

55 to 60 C is not bad, that's what I usually get (mind you, at 4.6 GHz after one hour of gaming at 90 fps). But anything up to about 75 C is OK. You are still a long way from that critical point. So you still have plenty of headroom to OC your rig. Cinebench performance will come after you reset your parameters, and go back to the BIOS to start methodically re-applying them one by one by following the settings that I have detailed previously. Even with a 'below average' CPU, you should get 4.4 GHz at around 1.31 V and 4.5 GHz at around 1.36 V. The LLC is what brought my voltage from 1.41 V to 1.355 V at 4.6 GHz. With that, of course the temp went right down, the fans slowed, the noise eased and the power consumption dropped.

Good luck with your project, it is a great challenge. Once you get a breakthrough you feel good.

I'm definitely not giving up, not yet. But since I have no LLC (maybe MSI will add it in a BIOS revision in the near future?) I'm left with Override Offset and other tricks (that you've enlightened me to). I just can't imagine how a motherboard would limit the quality of an overclock? My board had twelve power phases plus two for memory and was built for overclocking/gaming so you'd think I wouldn't be having any problems, yet I am.

What about setting things to Adaptive to reign in the voltages during idle brother Phil (a bright light engulfs the Ozzie and peels of thunder are heard for miles around)?

I'd really love to see the voltage go down at idle and peak during load. A guy from another forum said a combination of Dynamic and Adaptive might do that but I'm a little skeptical. I guess I'll just have to try your basic recipe and then try other things if I'm still getting wonky results.

I'm at 4.0GHz right now with a Vcore of 1.280v in the Bios. My VID always reads 1.2197-8v. It never budges. I'm doubtful I'll reach 4.4GHZ at the voltages you listed but I will surely give it the college drop-out try. I really think I must be missing something because the voltage I'm at just to get to 4.0GHz is over the norm for this chip. There must be a problem I just have to identify where it lies. It could be the board or the Bios, a bad chip (doubtful) or something I've done without realizing it.

Amazing at how Micro Center took the liberty to lie like a rug in that product specification sheet. By the way, isn't that particular fan a made for cooling a CPU (Static Pressure Fan)?

I think you misunderstood my fan setup. The fan that used to be my top exhaust fan (120mm NZXT case fan) is now used as a pull fan on my cpu cooler. Remember though, my cpu cooler already has a fan at the front pushing air through. So the case fan is now doing duty helping to draw air away from the cpu heat-sink fins and pushing air directly towards the back exhaust . I really haven't seen any decrease in cpu temps since doing this so I doubt having two fans on a cpu cooler is really as effective as so many people believe. On the other hand, maybe I'm not seeing results because it isn't a SP fan?

Well, looking for that break through my friend,

-laz.

Hi John,

Yes, I believe that this fan, although not qualified as an "SP" fan is good for pressure as it has the following characteristics:
- Focused Flow frame
Designed for pressure demanding applications such as heatsinks and radiators, the Focused Flow frame features eleven stator guide vanes that straighten, channel and focus the airflow, which allows the NF-F12 to rival the performance of conventional fans running at much faster speeds.
- Inner surface microstructures
With the tips of the fan blades plowing through the boundary layer created by the inner surface microstructures, flow separation from the suction side of the blades is significantly suppressed, which results in reduced blade passing noise and improved airflow and pressure efficiency.

Given the exchange rate and the fact that everything tends to be more expensive for Australia, it should be sold for around US$18 without any discount. This is for the Noctua-NF-F12-PWM-120mm-PWM. It's a good fan by any measure.

As for your CPU cooler (heatsink), I am not convinced that a push-pull system is all that effective unless it has been specifically designed for it like: https://www.ple.com.au/Products/623709/Cooler-Master-MasterAir-Maker-8-CPU-Cooler-
And even so, I am not convinced that paying almost three times more will even give you twice as much cooling. Maybe 50% more at a guess. But I must say, some of these coolers are impressive and "look" the part.

Take care

 

[John_485]

Hi John,

Yes, I believe that this fan, although not qualified as an "SP" fan is good for pressure as it has the following characteristics:
- Focused Flow frame
Designed for pressure demanding applications such as heatsinks and radiators, the Focused Flow frame features eleven stator guide vanes that straighten, channel and focus the airflow, which allows the NF-F12 to rival the performance of conventional fans running at much faster speeds.
- Inner surface microstructures
With the tips of the fan blades plowing through the boundary layer created by the inner surface microstructures, flow separation from the suction side of the blades is significantly suppressed, which results in reduced blade passing noise and improved airflow and pressure efficiency.

Given the exchange rate and the fact that everything tends to be more expensive for Australia, it should be sold for around US$18 without any discount. This is for the Noctua-NF-F12-PWM-120mm-PWM. It's a good fan by any measure.

As for your CPU cooler (heatsink), I am not convinced that a push-pull system is all that effective unless it has been specifically designed for it like: https://www.ple.com.au/Products/623709/Cooler-Master-MasterAir-Maker-8-CPU-Cooler-
And even so, I am not convinced that paying almost three times more will even give you twice as much cooling. Maybe 50% more at a guess. But I must say, some of these coolers are impressive and "look" the part.

Take care

What's up Phil long time no see no hear (never really!).

I made a mistake and wasn't logged into Tom's so I apologize for my late reply, that and I had to take my poor mother to the emergency room two days ago, thank God I got there in time.

Anyway, I agree. I'm not very convinced of the "extra" performance gains by adding a second fan to a cpu cooler. As I said I haven't seen any value after adding a second fan (pull) to my 212 Hyper LED. I like the Noctua NH-15S though I'm interested in the CM vapor chamber technology.

I bought myself the MSI GTX 1060 Gaming X 6GB GDDR5 yesterday. It's a beast of a card and looks sweet too!

I have a question for you Phil: If cpu voltage is set slightly lower then what would make an overclock stable, yet high enough that the computer won't crash, can it cause odd artifacts in the system, like audio crackling or other strange behaviors? I'm fiddling around with Adaptive Offset+ with an offset of 0.050 (used to be at 0.060). I did this in the last ten minutes and noticed some strange behavior, like the audio crackling a little bit, though it did go away very quickly?

Thanks,

-laz.

 

[philipew]

Hi John,

Yes, I believe that this fan, although not qualified as an "SP" fan is good for pressure as it has the following characteristics:
- Focused Flow frame
Designed for pressure demanding applications such as heatsinks and radiators, the Focused Flow frame features eleven stator guide vanes that straighten, channel and focus the airflow, which allows the NF-F12 to rival the performance of conventional fans running at much faster speeds.
- Inner surface microstructures
With the tips of the fan blades plowing through the boundary layer created by the inner surface microstructures, flow separation from the suction side of the blades is significantly suppressed, which results in reduced blade passing noise and improved airflow and pressure efficiency.

Given the exchange rate and the fact that everything tends to be more expensive for Australia, it should be sold for around US$18 without any discount. This is for the Noctua-NF-F12-PWM-120mm-PWM. It's a good fan by any measure.

As for your CPU cooler (heatsink), I am not convinced that a push-pull system is all that effective unless it has been specifically designed for it like: https://www.ple.com.au/Products/623709/Cooler-Master-MasterAir-Maker-8-CPU-Cooler-
And even so, I am not convinced that paying almost three times more will even give you twice as much cooling. Maybe 50% more at a guess. But I must say, some of these coolers are impressive and "look" the part.

Take care

What's up Phil long time no see no hear (never really!).

I made a mistake and wasn't logged into Tom's so I apologize for my late reply, that and I had to take my poor mother to the emergency room two days ago, thank God I got there in time.

Anyway, I agree. I'm not very convinced of the "extra" performance gains by adding a second fan to a cpu cooler. As I said I haven't seen any value after adding a second fan (pull) to my 212 Hyper LED. I like the Noctua NH-15S though I'm interested in the CM vapor chamber technology.

I bought myself the MSI GTX 1060 Gaming X 6GB GDDR5 yesterday. It's a beast of a card and looks sweet too!

I have a question for you Phil: If cpu voltage is set slightly lower then what would make an overclock stable, yet high enough that the computer won't crash, can it cause odd artifacts in the system, like audio crackling or other strange behaviors? I'm fiddling around with Adaptive Offset+ with an offset of 0.050 (used to be at 0.060). I did this in the last ten minutes and noticed some strange behavior, like the audio crackling a little bit, though it did go away very quickly?

Thanks,

-laz.

Hi John,

I really hope that your mother is OK - that's more important than any OC. The MSI GTX 1060 card is fantastic. I would give you my 2 GTX 970 for yours ;-).

To answer your question, I don't think that the audio crackling is coming from OC (at any level). There can be display artifacts when you start overclocking graphic cards. I had some when pushing the GDDR5 memory too far (I saw that on Valley and Heaven benchmarks). That's when you know you pushed too far. But I don't think you'd have to overclock your 1060. It will be more than enough to play any game on Ultra (max settings) at least for the next 5 years (unless of course 3 D takes off a lot with VR and AR... but even so). I like to use this Nvidia "face" benchmarking app. to test my GFX card: https://www.youtube.com/watch?v=kCPMaROnrTc. The level of detail when zooming close is amazing. Well... with the latest version of course. Sometime it shows a little "jolt" but that's part of it, not to worry (a bit as if it were on a repeating loop). There should be no artifacts.

By the way, in order to further reduce the power (which is dissipated as heat... therefore causing temp to increase), I set the following:
PSH = 1.000 V -> VCCIO = 1.000 V -> VCCSA = 1.150 V (...close to default values but I was surprised)

After testing for over two hours on Prime95 Blend test without any issue, I decided to leave it at that. It reduces my power and temps by approx. 10 %. When progressively going down in small steps from VCCIO/VCCSA = 1.250/1.250 V to a (very) low 1.000/1.150 V, the power continued to go down but then when I tried VCCSA at 1.100 V (by curiosity), the effect on power was quite reverse (going up a lot). I guess I had reached the limits and came back to VCCSA = 1.150 V. Cinebench R15 performance is unchanged at 767. All good.

However, after much more use I think that leaving PCH / VCCIO / VCCSA on "Auto" is best because although their voltage will go quite high under load (possibly a little too high), it will also get very low when not under load and at idle (which is most of the time). For example, VCCIO / VCCSA will go to 1.250 V on Auto under load but drop down significantly when not under load (check their value in the BIOS). This contributes to lowering the overall temperature. It's a bit like Speedstep (under Windows' "Balanced" power management setting). It gives you the frequency you need (up to the maximum you have set) under load but drops down significantly (to 800 MHz) when not under load and this greatly reduces the overall temp. of your rig.

Keep well...

 

[John_485]

Hi John,

Yes, I believe that this fan, although not qualified as an "SP" fan is good for pressure as it has the following characteristics:
- Focused Flow frame
Designed for pressure demanding applications such as heatsinks and radiators, the Focused Flow frame features eleven stator guide vanes that straighten, channel and focus the airflow, which allows the NF-F12 to rival the performance of conventional fans running at much faster speeds.
- Inner surface microstructures
With the tips of the fan blades plowing through the boundary layer created by the inner surface microstructures, flow separation from the suction side of the blades is significantly suppressed, which results in reduced blade passing noise and improved airflow and pressure efficiency.

Given the exchange rate and the fact that everything tends to be more expensive for Australia, it should be sold for around US$18 without any discount. This is for the Noctua-NF-F12-PWM-120mm-PWM. It's a good fan by any measure.

As for your CPU cooler (heatsink), I am not convinced that a push-pull system is all that effective unless it has been specifically designed for it like: https://www.ple.com.au/Products/623709/Cooler-Master-MasterAir-Maker-8-CPU-Cooler-
And even so, I am not convinced that paying almost three times more will even give you twice as much cooling. Maybe 50% more at a guess. But I must say, some of these coolers are impressive and "look" the part.

Take care

What's up Phil long time no see no hear (never really!).

I made a mistake and wasn't logged into Tom's so I apologize for my late reply, that and I had to take my poor mother to the emergency room two days ago, thank God I got there in time.

Anyway, I agree. I'm not very convinced of the "extra" performance gains by adding a second fan to a cpu cooler. As I said I haven't seen any value after adding a second fan (pull) to my 212 Hyper LED. I like the Noctua NH-15S though I'm interested in the CM vapor chamber technology.

I bought myself the MSI GTX 1060 Gaming X 6GB GDDR5 yesterday. It's a beast of a card and looks sweet too!

I have a question for you Phil: If cpu voltage is set slightly lower then what would make an overclock stable, yet high enough that the computer won't crash, can it cause odd artifacts in the system, like audio crackling or other strange behaviors? I'm fiddling around with Adaptive Offset+ with an offset of 0.050 (used to be at 0.060). I did this in the last ten minutes and noticed some strange behavior, like the audio crackling a little bit, though it did go away very quickly?

Thanks,

-laz.

Hi John,

I really hope that your mother is OK - that's more important than any OC. The MSI GTX 1060 card is fantastic. I would give you my 2 GTX 970 for yours ;-).

To answer your question, I don't think that the audio crackling is coming from OC (at any level). There can be display artifacts when you start overclocking graphic cards. I had some when pushing the GDDR5 memory too far (I saw that on Valley and Heaven benchmarks). That's when you know you pushed too far. But I don't think you'd have to overclock your 1060. It will be more than enough to play any game on Ultra (max settings) at least for the next 5 years (unless of course 3 D takes off a lot with VR and AR... but even so). I like to use this Nvidia "face" benchmarking app. to test my GFX card: https://www.youtube.com/watch?v=kCPMaROnrTc. The level of detail when zooming close is amazing. Well... with the latest version of course. Sometime it shows a little "jolt" but that's part of it, not to worry (a bit as if it were on a repeating loop). There should be no artifacts.

By the way, in order to further reduce the power (which is dissipated as heat... therefore causing temp to increase), I set the following:
PSH = 1.000 V -> VCCIO = 1.000 V -> VCCSA = 1.150 V (...close to default values but I was surprised)

After testing for over two hours on Prime95 Blend test without any issue, I decided to leave it at that. It reduces my power and temps by approx. 10 %. When progressively going down in small steps from VCCIO/VCCSA = 1.250/1.250 V to a (very) low 1.000/1.150 V, the power continued to go down but then when I tried VCCSA at 1.100 V (by curiosity), the effect on power was quite reverse (going up a lot). I guess I had reached the limits and came back to VCCSA = 1.150 V. Cinebench R15 performance is unchanged at 767. All good.


Keep well...

Hey Phil,

Sorry for the belated reply. My mother is doing alright, though not as well as I would want. It looks like my mother may have hit her limit but I hope and pray that isn't the case. She is in a sub-acute care facility and is receiving rehabilitation therapy. I have to take her to a doctor's appointment tomorrow so I'm hoping everything is going to go well. Thank you for your concern.

As for the 1060 I have no plans to overclock her just yet. As you mentioned there really isn't a reason to do it. The card eats up everything I throw at it with ease (this is the best card I've ever bought, and the most expensive to boot). I've never paid more then $150.00 for a GPU so I feel like a king owning a GTX 1060.

I haven't had much time (or enthusiasm) to overclock my 6600K past 4.2GHz but the bug is returning, meaning I want to try and take the chip past 4.2GHz this week. I still haven't used those memory timing you provided for me yet since I haven't settled on 4.2GHz being my chip's limit. If I can get to 4.4GHz and under 1.35v I'll be a very happy camper, but I have my doubts.

As for Prime95 I have been told to use two specific older versions: v27.9 & v26.6. I keep getting told that the newer versions are not good for the Skylake chips. What is your opinion on this? I'll think about adjusting those additional voltage levels after I come to a stable overclock that I will use on a daily basis.

Thanks,

-laz.

 

[philipew]

Hi John,

Yes, I believe that this fan, although not qualified as an "SP" fan is good for pressure as it has the following characteristics:
- Focused Flow frame
Designed for pressure demanding applications such as heatsinks and radiators, the Focused Flow frame features eleven stator guide vanes that straighten, channel and focus the airflow, which allows the NF-F12 to rival the performance of conventional fans running at much faster speeds.
- Inner surface microstructures
With the tips of the fan blades plowing through the boundary layer created by the inner surface microstructures, flow separation from the suction side of the blades is significantly suppressed, which results in reduced blade passing noise and improved airflow and pressure efficiency.

Given the exchange rate and the fact that everything tends to be more expensive for Australia, it should be sold for around US$18 without any discount. This is for the Noctua-NF-F12-PWM-120mm-PWM. It's a good fan by any measure.

As for your CPU cooler (heatsink), I am not convinced that a push-pull system is all that effective unless it has been specifically designed for it like: https://www.ple.com.au/Products/623709/Cooler-Master-MasterAir-Maker-8-CPU-Cooler-
And even so, I am not convinced that paying almost three times more will even give you twice as much cooling. Maybe 50% more at a guess. But I must say, some of these coolers are impressive and "look" the part.

Take care

What's up Phil long time no see no hear (never really!).

I made a mistake and wasn't logged into Tom's so I apologize for my late reply, that and I had to take my poor mother to the emergency room two days ago, thank God I got there in time.

Anyway, I agree. I'm not very convinced of the "extra" performance gains by adding a second fan to a cpu cooler. As I said I haven't seen any value after adding a second fan (pull) to my 212 Hyper LED. I like the Noctua NH-15S though I'm interested in the CM vapor chamber technology.

I bought myself the MSI GTX 1060 Gaming X 6GB GDDR5 yesterday. It's a beast of a card and looks sweet too!

I have a question for you Phil: If cpu voltage is set slightly lower then what would make an overclock stable, yet high enough that the computer won't crash, can it cause odd artifacts in the system, like audio crackling or other strange behaviors? I'm fiddling around with Adaptive Offset+ with an offset of 0.050 (used to be at 0.060). I did this in the last ten minutes and noticed some strange behavior, like the audio crackling a little bit, though it did go away very quickly?

Thanks,

-laz.

Hi John,

I really hope that your mother is OK - that's more important than any OC. The MSI GTX 1060 card is fantastic. I would give you my 2 GTX 970 for yours ;-).

To answer your question, I don't think that the audio crackling is coming from OC (at any level). There can be display artifacts when you start overclocking graphic cards. I had some when pushing the GDDR5 memory too far (I saw that on Valley and Heaven benchmarks). That's when you know you pushed too far. But I don't think you'd have to overclock your 1060. It will be more than enough to play any game on Ultra (max settings) at least for the next 5 years (unless of course 3 D takes off a lot with VR and AR... but even so). I like to use this Nvidia "face" benchmarking app. to test my GFX card: https://www.youtube.com/watch?v=kCPMaROnrTc. The level of detail when zooming close is amazing. Well... with the latest version of course. Sometime it shows a little "jolt" but that's part of it, not to worry (a bit as if it were on a repeating loop). There should be no artifacts.

By the way, in order to further reduce the power (which is dissipated as heat... therefore causing temp to increase), I set the following:
PSH = 1.000 V -> VCCIO = 1.000 V -> VCCSA = 1.150 V (...close to default values but I was surprised)

After testing for over two hours on Prime95 Blend test without any issue, I decided to leave it at that. It reduces my power and temps by approx. 10 %. When progressively going down in small steps from VCCIO/VCCSA = 1.250/1.250 V to a (very) low 1.000/1.150 V, the power continued to go down but then when I tried VCCSA at 1.100 V (by curiosity), the effect on power was quite reverse (going up a lot). I guess I had reached the limits and came back to VCCSA = 1.150 V. Cinebench R15 performance is unchanged at 767. All good.


Keep well...

Hey Phil,

Sorry for the belated reply. My mother is doing alright, though not as well as I would want. It looks like my mother may have hit her limit but I hope and pray that isn't the case. She is in a sub-acute care facility and is receiving rehabilitation therapy. I have to take her to a doctor's appointment tomorrow so I'm hoping everything is going to go well. Thank you for your concern.

As for the 1060 I have no plans to overclock her just yet. As you mentioned there really isn't a reason to do it. The card eats up everything I throw at it with ease (this is the best card I've ever bought, and the most expensive to boot). I've never paid more then $150.00 for a GPU so I feel like a king owning a GTX 1060.

I haven't had much time (or enthusiasm) to overclock my 6600K past 4.2GHz but the bug is returning, meaning I want to try and take the chip past 4.2GHz this week. I still haven't used those memory timing you provided for me yet since I haven't settled on 4.2GHz being my chip's limit. If I can get to 4.4GHz and under 1.35v I'll be a very happy camper, but I have my doubts.

As for Prime95 I have been told to use two specific older versions: v27.9 & v26.6. I keep getting told that the newer versions are not good for the Skylake chips. What is your opinion on this? I'll think about adjusting those additional voltage levels after I come to a stable overclock that I will use on a daily basis.

Thanks,

-laz.

Hi John,

Happy for your mother, no drama. As for the memory settings, don't forget to first "ease" your motherboard settings with Extreme Memory Profile (XMP) on Profile1 and FLK on 1GHz. Preferably don't touch PCH, VCCIO and VCCSA. I don't know where those comments come regarding Prime95 but whatever version you use, it should be good enough to tell you if your rig is stable, otherwise you will find out when playing games. For the voltage, remember that it's LLC (Load Line Calibration) which will seriously reduce it.

To keep the temperature down, make sure that you set your Power Plan in Windows to "Balanced" (this is important). It is the default I believe, but make sure that it is that way because it triggers "Speedstep" which allows the frequency of the CPU to drop (down to the minimum 800 MHz per core) when not needed (you should see this happening in real time with CPUID HWMonitor). It's a great saving and has absolutely no impact on performance because when you need the higher frequencies, it gives them to you immediately, up to the maximum. But if you set your Power Plan to "Performance" in Windows, it kills the Speedstep and keeps your CPU continuously at the highest frequency (like 4.2 GHz in your case) even when not needed. This also ages the CPU uselessly.

Your graphics card is absolutely fabulous (all my friends agree) and whatever the frequency you end up getting on your CPU, it is the GPU that does all the work for games anyway. So don't worry too much about getting any higher (I did because in my case with 2 GPUs in parallel, the SLI is heavy on the CPU). From 3.5 GHz to 4.2 GHz it's already a good 20% increase by overclocking. Good luck with your tuning.

 

[John_485]

Hi John,

Happy for your mother, no drama. As for the memory settings, don't forget to first "ease" your motherboard settings with Extreme Memory Profile (XMP) on Profile1 and FLK on 1GHz. Preferably don't touch PCH, VCCIO and VCCSA. I don't know where those comments come regarding Prime95 but whatever version you use, it should be good enough to tell you if your rig is stable, otherwise you will find out when playing games. For the voltage, remember that it's LLC (Load Line Calibration) which will seriously reduce it.

To keep the temperature down, make sure that you set your Power Plan in Windows to "Balanced" (this is important). It is the default I believe, but make sure that it is that way because it triggers "Speedstep" which allows the frequency of the CPU to drop (down to the minimum 800 MHz per core) when not needed (you should see this happening in real time with CPUID HWMonitor). It's a great saving and has absolutely no impact on performance because when you need the higher frequencies, it gives them to you immediately, up to the maximum. But if you set your Power Plan to "Performance" in Windows, it kills the Speedstep and keeps your CPU continuously at the highest frequency (like 4.2 GHz in your case) even when not needed. This also ages the CPU uselessly.

Your graphics card is absolutely fabulous (all my friends agree) and whatever the frequency you end up getting on your CPU, it is the GPU that does all the work for games anyway. So don't worry too much about getting any higher (I did because in my case with 2 GPUs in parallel, the SLI is heavy on the CPU). From 3.5 GHz to 4.2 GHz it's already a good 20% increase by overclocking. Good luck with your tuning.

Hello Phil,

If the overclock is unstable how will games reveal it (I mean the specific behavior like the game crashing or other similar behaviors)?

Thanks for the power plan advice. I was scratching my head and wondering why I wasn't seeing my cpu speed stepping down even though I have been using Adaptive+ Offset mode in the Bios! Since changing my plan to balanced I am seeing the cpu speed fluctuate like I originally expected.

Once again, thank you for the help Phil,

-laz.

 

[philipew]

Hi John,

Happy for your mother, no drama. As for the memory settings, don't forget to first "ease" your motherboard settings with Extreme Memory Profile (XMP) on Profile1 and FLK on 1GHz. Preferably don't touch PCH, VCCIO and VCCSA. I don't know where those comments come regarding Prime95 but whatever version you use, it should be good enough to tell you if your rig is stable, otherwise you will find out when playing games. For the voltage, remember that it's LLC (Load Line Calibration) which will seriously reduce it.

To keep the temperature down, make sure that you set your Power Plan in Windows to "Balanced" (this is important). It is the default I believe, but make sure that it is that way because it triggers "Speedstep" which allows the frequency of the CPU to drop (down to the minimum 800 MHz per core) when not needed (you should see this happening in real time with CPUID HWMonitor). It's a great saving and has absolutely no impact on performance because when you need the higher frequencies, it gives them to you immediately, up to the maximum. But if you set your Power Plan to "Performance" in Windows, it kills the Speedstep and keeps your CPU continuously at the highest frequency (like 4.2 GHz in your case) even when not needed. This also ages the CPU uselessly.

Your graphics card is absolutely fabulous (all my friends agree) and whatever the frequency you end up getting on your CPU, it is the GPU that does all the work for games anyway. So don't worry too much about getting any higher (I did because in my case with 2 GPUs in parallel, the SLI is heavy on the CPU). From 3.5 GHz to 4.2 GHz it's already a good 20% increase by overclocking. Good luck with your tuning.

Hello Phil,

If the overclock is unstable how will games reveal it (I mean the specific behavior like the game crashing or other similar behaviors)?

Thanks for the power plan advice. I was scratching my head and wondering why I wasn't seeing my cpu speed stepping down even though I have been using Adaptive+ Offset mode in the Bios! Since changing my plan to balanced I am seeing the cpu speed fluctuate like I originally expected.

Once again, thank you for the help Phil,

-laz.

Hi John,

You are welcome. Anomalies coming from instability manifest themselves usually as crashes like a game that freezes or is not very fluid at times (particularly in demanding games), or of cause as a CPU crash with a blue screen and some error message and number (both mostly useless). Then you can always reboot (sometimes after a minute or so of wait with a black screen) but it should restart (I never experienced a complete inability to reboot). This is not really a cause for concern.

To tell you, after running over an hour fully stable on Prime95 with my super-reduced values for PCH/VCCIO/VCCSA, I noticed that one game (Assassin's Creed) kept on freezing occasionally (particularly after panning the camera a lot). So I came back to more "reasonable" values (PCH 1.1100V - VCCIO 1.150V - VCCSA 1.200V) and it never occurred again. Artifacts can also appear like spikes (flashes) in the texture of the displayed picture but as I wrote previously, those appear mostly when overclocking the GPU which you certainly don't need to even consider, spoiled as you are with the amount of power your 1060 gives you.

By the way, any time you change a memory timing, with your "Memory Boot Mode" on Auto (as it should ideally be), the reboot will take longer because the MoBo will (Auto)matically re-train the memory and check if the new settings can stick (i.e. are viable). This doesn't mean that the settings are solid under load, just that they are "compatible" with everything else. The delay can be suspenseful. Then the computer usually stops by itself and after a short while, restarts and the settings are all taken in. If they are not "viable", you will be sent back to the BIOS. It's good also to save your last BIOS settings as you progress (and make notes of what exactly you change without changing too much at a time).

 

[John_485]

Hi John,

You are welcome. Anomalies coming from instability manifest themselves usually as crashes like a game that freezes or is not very fluid at times (particularly in demanding games), or of cause as a CPU crash with a blue screen and some error message and number (both mostly useless). Then you can always reboot (sometimes after a minute or so of wait with a black screen) but it should restart (I never experienced a complete inability to reboot). This is not really a cause for concern.

To tell you, after running over an hour fully stable on Prime95 with my super-reduced values for PCH/VCCIO/VCCSA, I noticed that one game (Assassin's Creed) kept on freezing occasionally (particularly after panning the camera a lot). So I came back to more "reasonable" values (PCH 1.1100V - VCCIO 1.150V - VCCSA 1.200V) and it never occurred again. Artifacts can also appear like spikes (flashes) in the texture of the displayed picture but as I wrote previously, those appear mostly when overclocking the GPU which you certainly don't need to even consider, spoiled as you are with the amount of power your 1060 gives you.

By the way, any time you change a memory timing, with your "Memory Boot Mode" on Auto (as it should ideally be), the reboot will take longer because the MoBo will (Auto)matically re-train the memory and check if the new settings can stick (i.e. are viable). This doesn't mean that the settings are solid under load, just that they are "compatible" with everything else. The delay can be suspenseful. Then the computer usually stops by itself and after a short while, restarts and the settings are all taken in. If they are not "viable", you will be sent back to the BIOS. It's good also to save your last BIOS settings as you progress (and make notes of what exactly you change without changing too much at a time).

Hey Phil,

Thanks for the info on the GPU and CPU overclocking/negative manifestations, etc. Is my 1060 really that good, meaning in regards to the other two cards? I mean it is the low end of the GTX 1000 series so the 1070 & 1080 are better (more Cuda cores, more memory, though the clock speeds seem to be about even with my 1060)

I have a voltage/Bios question for you:

On my board in the MSI Click 5 Bios there is a summary shown above all settings. It shows the speed of the processor, voltage, RAM, etc. What is kind of confusing me is the voltage readout in the summary.

I set my voltage to 1.27 (Adaptive Offset+) yet in the summary it reads 1.32v? Now my offset is 0.040mv so added to 1.27 I get a voltage of 1.31v overall volts. There is still a difference of 0.010mv so what's going on? Why does the summary say 1.32v?

One more small problem. I have a cordless keyboard (Logitech K350). Lately (past day or so) I'm getting some delay. It is intermittent, sometimes it is there and sometimes it isn't. I put in some fresh batteries two weeks ago so it isn't a battery issue. What could be causing the delay? On my old computer I moved the unified receiver to a front USB port and delay (the time it takes for my keystroke to register and show up on screen) was virtually eliminated. Maybe I'm experiencing some interference from another electronic device? Maybe signals from an outside source (radar, radio, ???).

Thanks,

-laz.