How to find OC sweetspot?

[Naweo]

Hello!

I have been prime (100%) stressed my temperatures up to max 97 and 98 degress, however, using my system for a week never doing primetest m temperatures get to max 85, and stays below 80 more than 99% of the timer. This is to my satisfaction.

However, it appears that while my system is stable 95% of the time, sometimes a random program freezes, my mouse freezes (needs to be unplugged) etc. Is there any way I can resolve this without changing GHZ on my CPU?

CPU: i7-3770k
MB, z77ag45
Windows 7

I understand you can increase or decrease voltage but in my case increasing voltage might make temps a bit too high, I canīt know for sure without further experimentation.

CPU overclock: 4.4 ghz
Cpu cooler: hyper evo 212
Cpu volt: 1.200 vcore std, cpu-z shows 1.176

[user 18]

I have never overclocked a modern intel processor, so take everything I say here with a sizable grain of salt.

Have you delidded the CPU? If you do, it tends to drop temps by a significant amount. Intel's stock TIM is low quality and imperfectly applied.

What are you using the CPU for? How long have you run prime for when it topped out at 98C? What test of Prime did you use? Did you change any of the other BIOS settings other than multiplier, voltage, etc.? Is turbo enabled or disabled, what process are you using to overclock?

The short of it is that you're running a CPU that isn't 100% stable, and the easiest way to deal with that is to either increase voltage or decrease speed. Higher voltages will result in higher temperatures. I'd suggest using a better cooler -- 212 evo is a good budget choice to be sure, but it can't hit the performance you would expect from a higher performing air cooler or closed loop water cooler. Depending on the answer to some of the questions above, there might be another way to make it stable, but I wouldn't count on it.

[Naweo]

Thanks for answer!

Non-delid.
I use the CPU for 100 FPS sc2 streaming, which is necesary. I only barely get that with 4.4 ghz.
It takes between 1 and 10 minutes to get to 98 max, probably more during long runs.
Not sure which test of prime, but i think its max cuz it says 100% load in realtemp.
I changed multiplier and voltage as well as ram timing, but I will put ram back to stock right now.
Turbo is enabled and enhanced.
I used bios to overclock.

What air coolers do you recommend?

[tp4tissue]

If it's crashing, it's NOT STABLE......

delid, and pump the volts.. stop making threads... it won't delid for you.. LOLOL

[Naweo]

Its very, very, very easy to delid.

[user 18]

Thanks for answer!

Non-delid.
I use the CPU for 100 FPS sc2 streaming, which is necesary. I only barely get that with 4.4 ghz.
It takes between 1 and 10 minutes to get to 98 max, probably more during long runs.
Not sure which test of prime, but i think its max cuz it says 100% load in realtemp.
I changed multiplier and voltage as well as ram timing, but I will put ram back to stock right now.
Turbo is enabled and enhanced.
I used bios to overclock.

What air coolers do you recommend?

Prime 95 has a number of different tests (small/large fft, blend). They all test different aspects of the CPU/platform, but they will all load the CPU 100%. I personally don't use prime for testing CPUs, because I find it sometimes doesn't truly demonstrate stability under short runs. I've run Prime for 12 hours with no issues and still had an unstable system. Instead, I recommend testing using LinPack (LinX), which tends to give quicker and more accurate results for system stability. Run the test in safe mode, or after a reboot. If you're running after a boot, close any unnecessary processes directly after system startup, then let the machine sit idle for about 10 minutes. Run only one temperature monitoring application. Tell the test how much memory it can use by looking at free physical memory in task manager and select 200-300MB less. Run the test for at least an hour, and for at least 15 passes. (instructions source)

Why is it necessary to stream at 100fps? Most monitors can't display more than 60fps in any case. If you were to drop the streaming to 60 or 80, you could probably save yourself a lot of headache.

The practically undisputed master of air cooling is Noctua's NH-D14. Their NH-U14S is a slightly cheaper proposition as well. Scythe makes some good designs as well. Can't speak about any of these from personal experience though.

The fact that you overclocked the RAM as well brings up an interesting point. Try running HCI Memtest on your OC'd RAM. Again, run it after a reboot or in safe mode. Run one instance of Memtest per thread on your CPU, and split the available physical memory approximately evenly between them. It could be a RAM problem you're dealing with, after all.

It *sounds* to me like your OC technique is okay -- I haven't read anything that's a red flag, at least -- but I can't say for sure. There are recommendations floating around out there about disabling turbo entirely for OC, but there are recommendations in the other direction as well. I'd say just stick with what you have, or find a specific guide and follow it.

[dorkvader]

Its very, very, very easy to delid.

note: make sure your processor is possible to de-lid and not one of those ones where the lid is soldered on.

In other news: I have a sweet processor keychain on my keyring now.

[user 18]

Its very, very, very easy to delid.

note: make sure your processor is possible to de-lid and not one of those ones where the lid is soldered on.

In other news: I have a sweet processor keychain on my keyring now.

*are* there 3770Ks that are soldered on? I thought the possibility was determined entirely by model number.

[Lanx]

It takes between 1 and 10 minutes to get to 98 max, probably more during long runs.

your cooler isn't up to snuff, what you want is very taxing on the hyper212, which is a very high quality barebones cooler.

you're basically really thermal loading that poor h212

go watercooling if have the budget, the all in one, closed loop ones are great and give serious dedicated 250$ user created water cooling for it's money

corsair is what i use, so thats what i'm familiar with

the cheapest is 60bucks for a single rad, i'd recommend for your setup a double rad, 240mm, thats like the h100 for 99bucks
http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=-1&IsNodeId=1&Description=corsair%20hydro&bop=And&Order=PRICE&PageSize=100

my 8core amd is stock 3500 and my max overclock is 4230, of course it can actually go 4400 and like you i've had the odd blue screen and 4230 is max i'm comfortable with (since it's my work computer i never want it to bluescreen)

i have it run with speedfan, and 99% of my day it's on silent.

most high end air coolers are 60 to 70 bucks anyway and thats usually without fans, and super loud, once these self contained watercoolers became super good i stopped buying high end air coolers.

[Beca]

Hm, do you have a razer mouse by any chance?

[tp4tissue]

It takes between 1 and 10 minutes to get to 98 max, probably more during long runs.

your cooler isn't up to snuff, what you want is very taxing on the hyper212, which is a very high quality barebones cooler.

you're basically really thermal loading that poor h212

go watercooling if have the budget, the all in one, closed loop ones are great and give serious dedicated 250$ user created water cooling for it's money

corsair is what i use, so thats what i'm familiar with

the cheapest is 60bucks for a single rad, i'd recommend for your setup a double rad, 240mm, thats like the h100 for 99bucks
http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=-1&IsNodeId=1&Description=corsair%20hydro&bop=And&Order=PRICE&PageSize=100

my 8core amd is stock 3500 and my max overclock is 4230, of course it can actually go 4400 and like you i've had the odd blue screen and 4230 is max i'm comfortable with (since it's my work computer i never want it to bluescreen)

i have it run with speedfan, and 99% of my day it's on silent.

most high end air coolers are 60 to 70 bucks anyway and thats usually without fans, and super loud, once these self contained watercoolers became super good i stopped buying high end air coolers.

It's not the cooler... it's the damn chip..

ivy and haswell are very small chips, AND they're not soldered..

the cooler has more than enough capacity.. but the contact is bad due to the above 2 reasons.



None of the ivy and haswell series were soldered.

[Skull_Angel]

Thanks for answer!

Non-delid.
I use the CPU for 100 FPS sc2 streaming, which is necesary. I only barely get that with 4.4 ghz.
It takes between 1 and 10 minutes to get to 98 max, probably more during long runs.
Not sure which test of prime, but i think its max cuz it says 100% load in realtemp.
I changed multiplier and voltage as well as ram timing, but I will put ram back to stock right now.
Turbo is enabled and enhanced.
I used bios to overclock.

What air coolers do you recommend?

Prime 95 has a number of different tests (small/large fft, blend). They all test different aspects of the CPU/platform, but they will all load the CPU 100%. I personally don't use prime for testing CPUs, because I find it sometimes doesn't truly demonstrate stability under short runs. I've run Prime for 12 hours with no issues and still had an unstable system. Instead, I recommend testing using LinPack (LinX), which tends to give quicker and more accurate results for system stability. Run the test in safe mode, or after a reboot. If you're running after a boot, close any unnecessary processes directly after system startup, then let the machine sit idle for about 10 minutes. Run only one temperature monitoring application. Tell the test how much memory it can use by looking at free physical memory in task manager and select 200-300MB less. Run the test for at least an hour, and for at least 15 passes. (instructions source)

I find that neither P95 and LinPack tests are great on their own for a rock solid OC, but together they work amazingly.

For the primary stability test [on default settings] I like to run LinPack for an hour (or 13~15 tests) and follow it up with P95 blend test for ~30 hrs (should be enough time for nearly 3 full passes of tests) and MemTest for about the same amount of time, this should tell you if your hardware is alright to begin with. While overclocking I'll only run LinPack (1 hour or 15+ tests) until I find a max stable setting and proceed to the 30 hour P95 blend tests. Should any workers on P95 fail at any time; diagnose, try to correct it (there should be lists of error codes floating around that will give you an idea of what went wrong), and start back over with LinPack > P95.

P95's Small FFT test has been said to be good for quick stability testing, but I find LinPack is better at making unstable settings fail quickly.

[Naweo]

Thanks for answer!

Non-delid.
I use the CPU for 100 FPS sc2 streaming, which is necesary. I only barely get that with 4.4 ghz.
It takes between 1 and 10 minutes to get to 98 max, probably more during long runs.
Not sure which test of prime, but i think its max cuz it says 100% load in realtemp.
I changed multiplier and voltage as well as ram timing, but I will put ram back to stock right now.
Turbo is enabled and enhanced.
I used bios to overclock.

What air coolers do you recommend?

Prime 95 has a number of different tests (small/large fft, blend). They all test different aspects of the CPU/platform, but they will all load the CPU 100%. I personally don't use prime for testing CPUs, because I find it sometimes doesn't truly demonstrate stability under short runs. I've run Prime for 12 hours with no issues and still had an unstable system. Instead, I recommend testing using LinPack (LinX), which tends to give quicker and more accurate results for system stability. Run the test in safe mode, or after a reboot. If you're running after a boot, close any unnecessary processes directly after system startup, then let the machine sit idle for about 10 minutes. Run only one temperature monitoring application. Tell the test how much memory it can use by looking at free physical memory in task manager and select 200-300MB less. Run the test for at least an hour, and for at least 15 passes. (instructions source)

Why is it necessary to stream at 100fps? Most monitors can't display more than 60fps in any case. If you were to drop the streaming to 60 or 80, you could probably save yourself a lot of headache.

The practically undisputed master of air cooling is Noctua's NH-D14. Their NH-U14S is a slightly cheaper proposition as well. Scythe makes some good designs as well. Can't speak about any of these from personal experience though.

The fact that you overclocked the RAM as well brings up an interesting point. Try running HCI Memtest on your OC'd RAM. Again, run it after a reboot or in safe mode. Run one instance of Memtest per thread on your CPU, and split the available physical memory approximately evenly between them. It could be a RAM problem you're dealing with, after all.

It *sounds* to me like your OC technique is okay -- I haven't read anything that's a red flag, at least -- but I can't say for sure. There are recommendations floating around out there about disabling turbo entirely for OC, but there are recommendations in the other direction as well. I'd say just stick with what you have, or find a specific guide and follow it.

Itīs imperative that I stay at least above 90 fps.

Right now I am sitting on stock clocks to ram since I believe I donīt need them overclocked. It sounds like itīs not a ram problem because I have run it for more than a year with standard OC genie overclock, and now itīs back to stock 1333 mhz and 23 dram and standard timings.

Please note that it can be other things that has caused my mouse to freeze or other things to crash as well, this has only happened a couple of times, but better safe than sorry.

I will attempt the linx guide thing now. As far as I understand I have 5.4 gb of ram, but intelburntest (running linpack x64 or x86 applications) allows me to set 5 different values that seems to be consistent values, maximum harvests all memory, very high harvests about 80% (leaving 500 mb ram as rest) I suppose this is what I want to select? Also I will attempt 15 runs as per your and the guideīs request.

http://gyazo.com/5a40b6a19ef81be01c82b0edff44b502

Thanks for the warnings about headache and reading the linpack guide Im starting to understand that symbiotic overclocks of FSB/RAM/CPU gives the most accurate gigaflop results, which Im starting to understand.

Hm, do you have a razer mouse by any chance?

Yes, I have a razer deathadder on the almighty synapse. Why? To be honest I have not had any instability problems at 4.4 ghz and 1.2 vcore, standard ram clock except my mouse freezings.

It takes between 1 and 10 minutes to get to 98 max, probably more during long runs.

your cooler isn't up to snuff, what you want is very taxing on the hyper212, which is a very high quality barebones cooler.

you're basically really thermal loading that poor h212

go watercooling if have the budget, the all in one, closed loop ones are great and give serious dedicated 250$ user created water cooling for it's money

corsair is what i use, so thats what i'm familiar with

the cheapest is 60bucks for a single rad, i'd recommend for your setup a double rad, 240mm, thats like the h100 for 99bucks
http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=-1&IsNodeId=1&Description=corsair%20hydro&bop=And&Order=PRICE&PageSize=100

my 8core amd is stock 3500 and my max overclock is 4230, of course it can actually go 4400 and like you i've had the odd blue screen and 4230 is max i'm comfortable with (since it's my work computer i never want it to bluescreen)

i have it run with speedfan, and 99% of my day it's on silent.

most high end air coolers are 60 to 70 bucks anyway and thats usually without fans, and super loud, once these self contained watercoolers became super good i stopped buying high end air coolers.

I can budget a 250 dollar cooling system, hydro or air, easily. The only problem is that I am not sure my medium tower case supports it. Is there any chance you can determine that through screenshots of my case? I will try to borrow a high-quality camera if that is possible.

This is the cabinet I am using:

http://gyazo.com/224acdc8f21a0a06921640ea686d71f2 - note that 1 fan was removed due to heatsink height.

Thanks for answer!

Non-delid.
I use the CPU for 100 FPS sc2 streaming, which is necesary. I only barely get that with 4.4 ghz.
It takes between 1 and 10 minutes to get to 98 max, probably more during long runs.
Not sure which test of prime, but i think its max cuz it says 100% load in realtemp.
I changed multiplier and voltage as well as ram timing, but I will put ram back to stock right now.
Turbo is enabled and enhanced.
I used bios to overclock.

What air coolers do you recommend?

Prime 95 has a number of different tests (small/large fft, blend). They all test different aspects of the CPU/platform, but they will all load the CPU 100%. I personally don't use prime for testing CPUs, because I find it sometimes doesn't truly demonstrate stability under short runs. I've run Prime for 12 hours with no issues and still had an unstable system. Instead, I recommend testing using LinPack (LinX), which tends to give quicker and more accurate results for system stability. Run the test in safe mode, or after a reboot. If you're running after a boot, close any unnecessary processes directly after system startup, then let the machine sit idle for about 10 minutes. Run only one temperature monitoring application. Tell the test how much memory it can use by looking at free physical memory in task manager and select 200-300MB less. Run the test for at least an hour, and for at least 15 passes. (instructions source)

I find that neither P95 and LinPack tests are great on their own for a rock solid OC, but together they work amazingly.

For the primary stability test [on default settings] I like to run LinPack for an hour (or 13~15 tests) and follow it up with P95 blend test for ~30 hrs (should be enough time for nearly 3 full passes of tests) and MemTest for about the same amount of time, this should tell you if your hardware is alright to begin with. While overclocking I'll only run LinPack (1 hour or 15+ tests) until I find a max stable setting and proceed to the 30 hour P95 blend tests. Should any workers on P95 fail at any time; diagnose, try to correct it (there should be lists of error codes floating around that will give you an idea of what went wrong), and start back over with LinPack > P95.

P95's Small FFT test has been said to be good for quick stability testing, but I find LinPack is better at making unstable settings fail quickly.

Do you think itīs reasonable that I make the necessary cooling/reapplying TIM/revoltage etc to see if I can hold good temperatures for at least 10 minutes before attempting these tests?

[User Was Banned]

Itīs imperative that I stay at least above 90 fps.

Why?

[Naweo]

Itīs imperative that I stay at least above 90 fps.

Why?

Because I am not going to stream Skyrim or Watch dogs, I am going to stream competetive level RTS. With 4.2 ghz I wasnīt satisfied, on 4.3 ghz it was better, but it seems like 4.4 ghz or above gives me around the stuff I need.

Note that the key areas of which I need higher fps are the most crucial points in my game. The huge fights with lots of phsysical and processing activity. Streaming these loads your CPU a lot as well. I did want to go above 4.4 all the way to maybe 4.7 or 4.8, but I do not have the tools atm or the environment to do delidding so I wanted to see what I could achieve without first.

[User Was Banned]

I'm just curious where you're streaming this competitive level RTS that allows you to stream in 90+ fps. Twitch?

[Naweo]

I'm just curious where you're streaming this competitive level RTS that allows you to stream in 90+ fps. Twitch?

Yeah, twitch. You misunderstand the conception of fps here. I am streaming "60" fps, but my computer fps drops when i stream, from which i need more than 100 fps

[User Was Banned]

I'm just curious where you're streaming this competitive level RTS that allows you to stream in 90+ fps. Twitch?

Yeah, twitch. You misunderstand the conception of fps here. I am streaming "60" fps, but my computer fps drops when i stream, from which i need more than 100 fps

Ahhh, I see. Welp, sorry for derailing your thread then.

You should look into watercooling though. I had a blast making the switch from air to water. It's like the thrill of building a PC for the first time all over again, but better.

[Naweo]

Well water cooling is the same installation as air, right? no need for maintenance or pumping water into the system?

[Lanx]

Well water cooling is the same installation as air, right? no need for maintenance or pumping water into the system?

no, unless it's on a closed loop system, like the ones i recommended, otherwise it's all build a water

is this you?

http://www.hardwareonline.dk/traad.aspx?fid=11&tid=631417

well it has your case and no a dual rad (240mm) will not fit in it, you have a top mounted power supply. usually most cases have gone to bottom mounted power supplies and many have a 240mm double fan slot on top for a dual rad (240mm)

but you can fit a single rad (all the 60$ kinds) that accepts a 120mm and it'll fit at the back of the window

of course you can always just replace your case with a more modern cooling arrangement,

newegg happens to have corsair cases with coupon code and this cheap one would probably cool better than what you got

http://www.newegg.com/Product/Product.aspx?Item=N82E16811139038&ignorebbr=1

look at the top of it, it has room for a 240mm dual radiator water cooling setup.

of course you can always spend more.

[Naweo]

No itīs not lanx but excellent job finding those pictures, probably useful for me (I assume the legalty rights here are in place.)

I understand that water cooling 240 mm rads fans canīt fit my case, and i will definitely keep this in mind.

Thank you.

[TacticalCoder]

Why is it necessary to stream at 100fps? Most monitors can't display more than 60fps in any case.

Sometimes I feel like technology went backwards... Nearly 15 years ago I used to play Counter-Strike on an old Pentium II 300 Mhz (if I recall correctly) and I was using a "low polygon mod" to be able to get the framerate up to 99 fps (my system wasn't powerful enough, but by removing polygons I could get more fps). It was night and day compared to playing @ 60 fps or 72 fps (now I'm more than forty years old and still have great eye vision on both eyes and still no glasses: I know some people couldn't tell the difference between 60 and 99 but I wasn't one of them. To me 60 was unbearable while 72 was ok and 99 was smooth as silk).

Of course that was on a good old CRT monitor.

Now I don't game at all anymore since basically ten years or so so I haven't followed what's up.  But what's with all these flat panels monitors claiming 8ms and 5ms etc.?  I know they work in a different way than good old CRTs but still, I'm surprised by your claim that most monitor can't do better than 60 fps.

[user 18]

Monitors can run at different refresh rates up to 144 hz. So yes, it is possible to run at higher rates. However, most monitors today still run at 60hz. 60hz = 60 updates per second, so anything more than 60fps will have frames rendered by the GPU faster than the monitor can display them. Nvidia's gsync is one technology that actually changes the refresh rate of the monitor to work with different framerates. 60fps on an LCD is a different experience than 60fps on a CRT because of the underlying technologies behind the different display types. I can't manage with a 60hz CRT either, but 60hz LCD is a different story.

[tp4tissue]

Monitors can run at different refresh rates up to 144 hz. So yes, it is possible to run at higher rates. However, most monitors today still run at 60hz. 60hz = 60 updates per second, so anything more than 60fps will have frames rendered by the GPU faster than the monitor can display them. Nvidia's gsync is one technology that actually changes the refresh rate of the monitor to work with different framerates. 60fps on an LCD is a different experience than 60fps on a CRT because of the underlying technologies behind the different display types. I can't manage with a 60hz CRT either, but 60hz LCD is a different story.

gsync is stupid because it doesn't work with Light boost smooth motion...

and Lightboost is the FUTURE...

[tp4tissue]

I'm just curious where you're streaming this competitive level RTS that allows you to stream in 90+ fps. Twitch?

Yeah, twitch. You misunderstand the conception of fps here. I am streaming "60" fps, but my computer fps drops when i stream, from which i need more than 100 fps

Ahhh, I see. Welp, sorry for derailing your thread then.

You should look into watercooling though. I had a blast making the switch from air to water. It's like the thrill of building a PC for the first time all over again, but better.

Water is not worth it on any current -INTEL- CPU.. it doesn't even make much a difference if you don't delid..

and AFTER you delid.. it'd be a total waste of money, because any $20 air-cooler can give you the same temperatures as the $100 AIO water-coolers..


on GPU then, water is completely worth it if you're using AMD cards..

on Nvidia, they don't run hot, and they are voltage locked tightly such that even if you got water, you can't push them much further anyway,  so water is a waste on Nvidia Cards.

So on AMD cards...  Water cooling is especially good, and I'd even say NECESSARY on the new 290 and 290x cards...  The performance difference is around 10-15% with and without water because of heavy throttling..

[User Was Banned]

Water is not worth it on any current -INTEL- CPU.. it doesn't even make much a difference if you don't delid..

and AFTER you delid.. it'd be a total waste of money, because any $20 air-cooler can give you the same temperatures as the $100 AIO water-coolers..


on GPU then, water is completely worth it if you're using AMD cards..

on Nvidia, they don't run hot, and they are voltage locked tightly such that even if you got water, you can't push them much further anyway,  so water is a waste on Nvidia Cards.

So on AMD cards...  Water cooling is especially good, and I'd even say NECESSARY on the new 290 and 290x cards...  The performance difference is around 10-15% with and without water because of heavy throttling..

I was referring to a GPU+CPU loop, not just CPU; I would never recommend that. Water is not a waste on Nvidia cards, as you can simply flash a custom BIOS and unlock the voltage to push them much further. I run both of my 780s under water at 1.325V. However, I think the 780 Ti is locked indefinitely, and cannot be unlocked with a custom BIOS; at least that was the case last I checked...I haven't payed much attention to it lately.

AMD cards, yes water is extremely beneficial. Their newer R9 cards are hot as lava. But afaik they do not throttle much, if at all. They were designed to run hot.

Also, with custom water my 3570k runs much cooler than with a close loop cooler, or under air. I was also able to push more voltage.

[tp4tissue]

Water is not worth it on any current -INTEL- CPU.. it doesn't even make much a difference if you don't delid..

and AFTER you delid.. it'd be a total waste of money, because any $20 air-cooler can give you the same temperatures as the $100 AIO water-coolers..


on GPU then, water is completely worth it if you're using AMD cards..

on Nvidia, they don't run hot, and they are voltage locked tightly such that even if you got water, you can't push them much further anyway,  so water is a waste on Nvidia Cards.

So on AMD cards...  Water cooling is especially good, and I'd even say NECESSARY on the new 290 and 290x cards...  The performance difference is around 10-15% with and without water because of heavy throttling..

I was referring to a GPU+CPU loop, not just CPU; I would never recommend that. Water is not a waste on Nvidia cards, as you can simply flash a custom BIOS and unlock the voltage to push them much further. I run both of my 780s under water at 1.325V. However, I think the 780 Ti is locked indefinitely, and cannot be unlocked with a custom BIOS; at least that was the case last I checked...I haven't payed much attention to it lately.

AMD cards, yes water is extremely beneficial. Their newer R9 cards are hot as lava. But afaik they do not throttle much, if at all. They were designed to run hot.

Also, with custom water my 3570k runs much cooler than with a close loop cooler, or under air. I was also able to push more voltage.

While it may be true that a GPU + CPU loop would run cooler..  the question is WHY.....

cpu blocks cost money, tubing costs money, setup time, time is money..

No matter how good it is, if you get nothing out of it.. no increased performance from the CPU.. then that's just wasted capacity..

It makes more sense to solely cool the GPU, and avoid the risk of hosing your Cpu with water all together.

ontop of that, with only the GPU to cool you can run the pump and fan at lower rpms, extending their lifetime..

The CPU and GPU will be obsolete long before they die, even under Heavy OC.. so.. there's really no point in wasting your cooling capcity on the CPU.

[JaccoW]

Simple;
Space constraints that require the heat to be carried away. Some cases are exceptionally narrow or low and might make it difficult to use taller well-performing air coolers.
HTPC comes to mind as well as some enthusiast designs such as the NCASE M1.
A rad will carry the heat away and spread it over a much larger area, making it possible to use lower spinning fans. = less noise.

[User Was Banned]

Water is not worth it on any current -INTEL- CPU.. it doesn't even make much a difference if you don't delid..

and AFTER you delid.. it'd be a total waste of money, because any $20 air-cooler can give you the same temperatures as the $100 AIO water-coolers..


on GPU then, water is completely worth it if you're using AMD cards..

on Nvidia, they don't run hot, and they are voltage locked tightly such that even if you got water, you can't push them much further anyway,  so water is a waste on Nvidia Cards.

So on AMD cards...  Water cooling is especially good, and I'd even say NECESSARY on the new 290 and 290x cards...  The performance difference is around 10-15% with and without water because of heavy throttling..

I was referring to a GPU+CPU loop, not just CPU; I would never recommend that. Water is not a waste on Nvidia cards, as you can simply flash a custom BIOS and unlock the voltage to push them much further. I run both of my 780s under water at 1.325V. However, I think the 780 Ti is locked indefinitely, and cannot be unlocked with a custom BIOS; at least that was the case last I checked...I haven't payed much attention to it lately.

AMD cards, yes water is extremely beneficial. Their newer R9 cards are hot as lava. But afaik they do not throttle much, if at all. They were designed to run hot.

Also, with custom water my 3570k runs much cooler than with a close loop cooler, or under air. I was also able to push more voltage.

While it may be true that a GPU + CPU loop would run cooler..  the question is WHY.....

cpu blocks cost money, tubing costs money, setup time, time is money..

No matter how good it is, if you get nothing out of it.. no increased performance from the CPU.. then that's just wasted capacity..

It makes more sense to solely cool the GPU, and avoid the risk of hosing your Cpu with water all together.

Show Image

ontop of that, with only the GPU to cool you can run the pump and fan at lower rpms, extending their lifetime..

The CPU and GPU will be obsolete long before they die, even under Heavy OC.. so.. there's really no point in wasting your cooling capcity on the CPU.

Why? Because it's fun, it's quiet, it looks cool, and it allows me to push more voltage through my CPU+GPU so I can achieve higher overclocks with lower temps. It's more practical in a small form factor case with poor airflow, or with 2+ GPUs.

You don't really run the pump and fans at lower speeds with a GPU only set-up. Especially if you have plenty rad space. And with today's components, 'hosing your CPU with water' is highly unlikely unless you're completely incompetent.

Anyway, it was just a suggestion because I enjoyed the process myself, and I think anyone that enjoys building stuff should try out watercooling.

[ApocalypseMaow]

If you're freezing and hanging up at 4.4 then you either have to downclock to a stable clock or use a different cpu.

Some Ivy just won't handle those speeds and it seems like you got a "lemon"...

off topic: Ohhh the AquaComp pump/res combo... Nice caselabs too!!!

[tp4tissue]

If you're freezing and hanging up at 4.4 then you either have to downclock to a stable clock or use a different cpu.

Some Ivy just won't handle those speeds and it seems like you got a "lemon"...

off topic: Ohhh the AquaComp pump/res combo... Nice caselabs too!!!

he's probably got all the vital settings wrong.. but since he's too lazy to post his bios shots, we will never know..

I've Never seen any K series that couldn't do at least 4.7...

All the low reports are by n00bs who don't know what they're doing..


A few of my friend's friends got ivys a while back,, they couldn't get past 4.0...

I show up.. boom 4.7, 4.8 all around..

[ApocalypseMaow]

If you're freezing and hanging up at 4.4 then you either have to downclock to a stable clock or use a different cpu.

Some Ivy just won't handle those speeds and it seems like you got a "lemon"...

off topic: Ohhh the AquaComp pump/res combo... Nice caselabs too!!!

he's probably got all the vital settings wrong.. but since he's too lazy to post his bios shots, we will never know..

I've Never seen any K series that couldn't do at least 4.7...

All the low reports are by n00bs who don't know what they're doing..


A few of my friend's friends got ivys a while back,, they couldn't get past 4.0...

I show up.. boom 4.7, 4.8 all around..

I wouldn't be as harsh, but I can't deny you're right...  )

Now, If you move up to haswell... That was the easiest OC of my life.

[tp4tissue]

If you're freezing and hanging up at 4.4 then you either have to downclock to a stable clock or use a different cpu.

Some Ivy just won't handle those speeds and it seems like you got a "lemon"...

off topic: Ohhh the AquaComp pump/res combo... Nice caselabs too!!!

he's probably got all the vital settings wrong.. but since he's too lazy to post his bios shots, we will never know..

I've Never seen any K series that couldn't do at least 4.7...

All the low reports are by n00bs who don't know what they're doing..


A few of my friend's friends got ivys a while back,, they couldn't get past 4.0...

I show up.. boom 4.7, 4.8 all around..

I wouldn't be as harsh, but I can't deny you're right...  )

Now, If you move up to haswell... That was the easiest OC of my life.

um... depends..

I got a 4770 now, had a 4670, both were pretty easy to oc due to lack of vrm stabilization problems like on 3xxx and 2xxx..

but the Easiest to OC has to be my current G3258, it just works.. plugged it in tweaked 1 voltage, boom 4.8ghz

but I think I might keep it @ 4.6 for summer time operation... since I might throw it some light server duty in an un-ac-ed room

[ApocalypseMaow]

If you're freezing and hanging up at 4.4 then you either have to downclock to a stable clock or use a different cpu.

Some Ivy just won't handle those speeds and it seems like you got a "lemon"...

off topic: Ohhh the AquaComp pump/res combo... Nice caselabs too!!!

he's probably got all the vital settings wrong.. but since he's too lazy to post his bios shots, we will never know..

I've Never seen any K series that couldn't do at least 4.7...

All the low reports are by n00bs who don't know what they're doing..


A few of my friend's friends got ivys a while back,, they couldn't get past 4.0...

I show up.. boom 4.7, 4.8 all around..

I wouldn't be as harsh, but I can't deny you're right...  )

Now, If you move up to haswell... That was the easiest OC of my life.

um... depends..

I got a 4770 now, had a 4670, both were pretty easy to oc due to lack of vrm stabilization problems like on 3xxx and 2xxx..

but the Easiest to OC has to be my current G3258, it just works.. plugged it in tweaked 1 voltage, boom 4.8ghz

Show Image

but I think I might keep it @ 4.6 for summer time operation... since I might throw it some light server duty in an un-ac-ed room

I read your exploits with the anniversary... You can't beat that at $105!!!

[tp4tissue]

If you're freezing and hanging up at 4.4 then you either have to downclock to a stable clock or use a different cpu.

Some Ivy just won't handle those speeds and it seems like you got a "lemon"...

off topic: Ohhh the AquaComp pump/res combo... Nice caselabs too!!!

he's probably got all the vital settings wrong.. but since he's too lazy to post his bios shots, we will never know..

I've Never seen any K series that couldn't do at least 4.7...

All the low reports are by n00bs who don't know what they're doing..


A few of my friend's friends got ivys a while back,, they couldn't get past 4.0...

I show up.. boom 4.7, 4.8 all around..

I wouldn't be as harsh, but I can't deny you're right...  )

Now, If you move up to haswell... That was the easiest OC of my life.

um... depends..

I got a 4770 now, had a 4670, both were pretty easy to oc due to lack of vrm stabilization problems like on 3xxx and 2xxx..

but the Easiest to OC has to be my current G3258, it just works.. plugged it in tweaked 1 voltage, boom 4.8ghz

Show Image

but I think I might keep it @ 4.6 for summer time operation... since I might throw it some light server duty in an un-ac-ed room

I read your exploits with the anniversary... You can't beat that at $105!!!

actually.. there was another deal on Newegg for like 3 hrs or so..  with the same deal $99.. but a slightly better motherboard..

Haha..  I missed it, because I was busy playing with my current one..

[bear95]

Have you tried turning hyperthreading off? I'm not sure if you're going to get a performance hit because your streaming while running starcraft but I do know that starcraft 2 doesn't get any increase in fps above 4 cores. Turning off hyperthreading could give you about 5C headroom to up your voltage.
http://www.bit-tech.net/hardware/cpus/2010/08/18/how-many-cpu-cores-does-starcraft-2-use/2

Monitors can run at different refresh rates up to 144 hz. So yes, it is possible to run at higher rates. However, most monitors today still run at 60hz. 60hz = 60 updates per second, so anything more than 60fps will have frames rendered by the GPU faster than the monitor can display them.

You'll still see the frames rendered by the gpu even if you're going over 60fps. Each frame is a slice of the screen and not the whole screen so above 60fps you're still gonna see more frames updated each screen refresh.