Intel Core i7 920 Overclocking and Heat @ [H]

n3g471v3 d3c1b3l said:
another most excellent video, Kyle. I think I like them more than the idea of a [H]podcast that has been brought up before. Kudos good sir. Kudos.
Click to expand...

Love the videos when I get the time to watch them. I guess I missed the [H] podcast suggestion because I would like to dig it back up! I listen to many podcasts on my commute and very few people know how to do this stuff much less verbalize the results in a manner that is understandable. I want to cast another ballot for the podcast possibilities but, if that never transpires, I will find time in the wee hours of the morning to watch the videos :)
 
jonneymendoza said:
great video review mate cheers for that. so i assume 80c is an acceptable temp for these cpu's it seems. now to find a good case that provides good airflow
Click to expand...

80c on-die for this CPU seems to be just fine.

darklcd said:
Dethklok FTW
Click to expand...

Knew some of you would get it. If not, check here.

EvilAlchemist said:
Kyle,

Many thanks for the Cooling Following Up. I enjoyed the fact you showed so many cooling options.
It really show how important spending a little extra for cooling can make a lot of difference in temps.
I also was impressed you reminded viewers that test bench temps can be different then temps in a case.

The [H]ardOCP videos adds so much to the site, and I hope they continue for a long time to come.
It presents so much information and I feel it gives so much more information then a text only review.
Click to expand...

Thanks for the support I hope to keep doing more of this when I see it fits the information that I think is important that you guys get quickly.

Sojuuk said:
If you're clever like me you might have ducts that take the heat out of the case, or you could just leave the door off and aim a fan at it.
Click to expand...

The reason I took notice of Steve was him using plastic rain gutters to build ducting for his dually Celeron BP6 so many years ago. :) Check this towards the bottom of the page!

n3g471v3 d3c1b3l said:
another most excellent video, Kyle. I think I like them more than the idea of a [H]podcast that has been brought up before. Kudos good sir. Kudos.

Anyway, have I missed some [H]core benchmarks with that setup and some gaming?

BTW, your choice of music was great. and That beast of a setup is awesome with those 12GB of ram. I guess 12GB is going to be the new standard of overkill that the 8GB has been for a while.
Click to expand...

Figure if we are going to do this we will keep it video format. I might do ipod format video too, been kicking that around.

Check the front page.

I think 6GB is the new 4GB for sure. And yes 12GB is the new 8GB. :) Awesome stuff.
 
These are great.

[H] is the best!

I know many a folder that are drooling over these new systems.

 
Guess I was the only one who thought the music was cheeeeesy. :p

Thanks for the follow-up. Looks like 4.0ghz will probably be a bad idea on a 920, contrary to some reports. 3.8ghz is certainly enough of an overclock regardless. Can you bring that vCore down at all to maybe help the high temps?
 
Nehalem meets Nathan Explosion :)

This is one point where ATX layout falls short - the CPU ends up in the center of the case, making it tough to duct unless have one of those cases with the PSU at the bottom, or nothing to the left of your rig. What's a reasonable estimate at the 3.8GHz * 100% power dissipation from the CPU ...somewhere around 185W I'd guess? It won't heat a room, but it will definitely heat the inside of a confined case in a hurry if you don't get that hot air out!
 
Kyle,
I've been following the website from the first Celeron 400A days, anyway I just wanted to drop a line.

I really enjoy the new video segments you are doing! Keep up the good work.

Matt
 
Very nice review, and it does shed some light on one of my questions regarding heat from the last review.
I thought I remembered INtel was going to box a different cooler withe the i7's than the C2d's (not the TRUE)...
 
I just can't stop myself...
First and foremost, once again, great video review. You work well on camera, Kyle. You also get the information across so that it makes sense. Which for some of us (ME!!) is very important.

The info at the end about H2O vs Air was a great sidebar, and a very important point. Definitely something to keep in mind if you're going to OC one of those bad boys.

Now then, how about some real world test numbers with that beast of a computer you have there...

Also...about the pics of Steve's old school computer with the rain gutters...those are amazing. There was a time when my friends and I measured how great out computers where by how MANY fans we could jam into a case...I think I won that one with somewhere around 18 or 19...talk about LOUD.

I really think every once in a while you should dig up a dead old thread of something someone did like 5 years ago...OOOOH THAT GIVES ME AN IDEA!! Just like Car and Driver..or is it Road & Track? What ever... they do a Years Ago thing...you could to like a 8 years ago and 4 years ago today...

Sorry...off topic...
Thanks again for the review! I'm looking forward to the next one
 
I can't see the video from work, but I can see the graph and my jaw is on the floor when I saw the difference in the delta C for the True120 vs water was only 5%. I understand from the comments that it is an open system, but still, that's pretty sick. The water you used was the "Koolance Exos-2 LX" right? The only thing you had in the loops was the CPU right? Were the fans at 100%? I'm just really shocked because koolance claims that unit can do 750 Watts at a 30C delta. Last I checked those CPUs were rated around 140Watts. Even with OC, you shouldn't even be close to 750 and your at almost twice the delta.
 
What surprises me are the stock temps. At 75C full load on stock speeds, this is one hot little chip. Those are GPU territory right there. Did the CPU become unstable at 90C or did you just determine to be unsafe? If it's hitting 75C at stock it may have fairly high temperature tolerances.
 
You are getting better with every new video review. Great job.

One thing though, its pronounced jigahertz not guigahertz, didn't you watch back to the future?
 
Great video and good information. Music was cool, too.

I agree that the real benefit of the water system is (at least in my system) all the heat is taken away and out of the case.

Looking forward to putting together a new i7 once the MBs are more available.
 
vengence said:
I can't see the video from work, but I can see the graph and my jaw is on the floor when I saw the difference in the delta C for the True120 vs water was only 5%. I understand from the comments that it is an open system, but still, that's pretty sick. The water you used was the "Koolance Exos-2 LX" right? The only thing you had in the loops was the CPU right? Were the fans at 100%? I'm just really shocked because koolance claims that unit can do 750 Watts at a 30C delta. Last I checked those CPUs were rated around 140Watts. Even with OC, you shouldn't even be close to 750 and your at almost twice the delta.
Click to expand...


Keep in mind that we are still just basically dealing with ambient temp cooling on both the water and the air systems. The air cooler when in this state of having ambient temperature air continually fed through it is doing as good a job as the water cooler. The water cooler is simply using the same temperature air.

The 120 is VERY efficient though, no denying that, but realistically keeping the interior of a chassis to ambient room temp while dumping a full system's wattage in it is not realistic. In this case shown, around 500 watts in the case is what you would be dealing with. The water cooler has access to ambient temperature air and dumps the heat of whatever it is attached to outside the chasses.

The Exos 2 LX can be rated for 1000000000watts, but that is not going to change the fact that the face of that water block is never going to be cooler than ambient and that is going to impact how fast it can dissapate the amount of heat. You might find that it can keep 4 920s cool at 3.8GHz? See my point? If the 120 can dissapate the CPU heat as quickly in an open test, and obviously it can, that hardly says that it can handle a 750w load. (And a small aside, the Exos system is rated for 750w, I am not sure what the block is rated for.)

Hope that all makes some sense. There are simply limits when you are talking about active cooling using ambient temps for the heat exchangers.
 
Stoly said:
I beg to differ. The X58 works fine with 2 channels, yes you loose bandwidth but real word difference is neglible. Price made 4 and 8gb affordable which is why it quickly became stardard. I don't see DDR3 prices falling at that level anytime soon.

Besides AFAIK windows 7 won't need more memory than Vista. Actually its supposed to work better with less memory.
Click to expand...

You don't need to beg to be wrong. ;) You do that fine all on your own.
 
How do the noise levels between the water cooler and that monster air cooler compare?
 
EvilJohn said:
How do the noise levels between the water cooler and that monster air cooler compare?
Click to expand...

Actually the Thermalright 120 is a lot quieter with a single low 120mm RPM fan. The Exos has two higher RPM fans and two pumps...all of which can be turned down, but keep in mind the Exos should also handle about 6x the wattage load of the air cooler so it is not a fair comparison.
 
Nice video, you had me worried for the first 1/2 as I looked at and analyzed the chart ahead of your commentary and noticed the relative lack of delta between the high end air and the water but the astute observations at the end really made it an excellent presentation.
 
I actually have my computer in another room from my monitor and keyboard using a small hole in the wall for cables. The case is open on both sides with a 20 inch box fan (that's 508mm lol) delivering somewhere around 1500-2000cfm. The fan is positioned 2 inches from the side of the case. Since I did this I noticed motherboard temps went WAY down by 8C compared to running an open case. The CPU temps went down 5C and under load there is much less delta then there used to be.

This is what all overclockers are going to start doing if they dont want the room to heat up! Great for winter but not so good for those of us who don't use AC in the summer.
 
Nice test Kyle. So far it looks like an overclocked Core i7 can run about 10C to 15C hotter than well overclocked Core 2 Duo based processors can run reliably at.

I'm just working on adding a feature to RealTemp that will let users run a consistent load like Prime95 small FFTs and then have it automatically vary the load percentage so you can do your own load testing like Kyle has done.

Here's a preview:

newsensortestao5.png
 
Temps like that make me want to stay with my current setup for awhile... kind of hard to beat 3.6ghz at roughly 65c max (ambient between 70 and 90F depending on the time of year) plus the fact that we have to wait for new coolers and/or mounting brackets, wait for new boards to come out/BIOS updates, tweaks, etc all for 5-10% performance increases.
 
On the test bench, there isn't much difference between air and water at full load. For a future review why not use a closed case similar to what a typical user has with a couple of case fans and compare the inside of the case temperature between air and water? That would be interesting.
 
Thermalright Ultras dont come with a fan right? What fan was attached to the cooler in the vid? It is it an optional fan that thermalright sells for teh ultra?
 
temps are way too high for my liking! i actually think i7s are hotter than prescotts...i don't remember them getting up to 90 degrees celsius. :rolleyes:

it looks like this is what happens when amd can't compete....intel has no reason to keep thermals in check. sad, sad, sad...

makes me want to pass on i7 altogether and hold onto my current rig. i like being able to passively cool my overclocked c2d right now.
 
unclewebb said:
Nice test Kyle. So far it looks like an overclocked Core i7 can run about 10C to 15C hotter than well overclocked Core 2 Duo based processors can run reliably at.

I'm just working on adding a feature to RealTemp that will let users run a consistent load like Prime95 small FFTs and then have it automatically vary the load percentage so you can do your own load testing like Kyle has done.

Here's a preview:

newsensortestao5.png
Click to expand...

Awesome stuff there, please keep us updated!!!
 
unclewebb said:
On the test bench, there isn't much difference between air and water at full load. For a future review why not use a closed case similar to what a typical user has with a couple of case fans and compare the inside of the case temperature between air and water? That would be interesting.
Click to expand...

Well as we demonstrated with the failing air cooler, it depends on the cooler.

I am not going to start whipping that dog...again That is up to your guys. I am not going to spend my time building up a system to exhaust it properly and show you the same results. Air cooling is all about how well you exhaust the case.

Maybe this article from 1998 will give you some ideas.

http://www.hardocp.com/article.html?art=MjA5LDEsLGhlbnRodXNpYXN0
 
Juggafat said:
Thermalright Ultras dont come with a fan right? What fan was attached to the cooler in the vid? It is it an optional fan that thermalright sells for teh ultra?
Click to expand...

Thermalright part number TR-FDB-12-1600

0.2 amp.
 
I want that fan bracket so I can stop messing around with the stupid clips on my TRUEs. Any word on when/if those are going to be available?
 
Yeah, that's good info, thanks for helping me out.

Noise is really starting to matter to me now, in my computer needs. Maybe I'm just getting old now, and more sensitive to... hold on...

HEY YOU DAMN KIDS GET OFF MY LAWN

... the volume of things. You're going to need at least a pair of 120s going very fast to push air through the case with an air cooled I7
 
chrisf6969 said:
Is there a big difference in temps between when you have 4 cores at 100% load & 8 cores. do you think it maxes out the 4 true cores first, then utilizes HT?

Just b/c I'm thinking most people will never max out all 8 cores in any real world situation besides possibly a few professional encoding apps that are multithreaded properly.
Click to expand...

I have not specifically tested this, but let me bounce this off you. There are only 4 physical cores paired with 4 virtual (HyperThreading) cores. I have NOT turned HT off to see how it impacts this but I think the impact to power consumption would be small to around 15% (just guessing). So maxing out 4 cores is tough and carrying a full load across them is even harder. Trust me, we are showing a worst case scenario at full load that no all will pull off in real life, UNLESS you are doing encoding of some sort. Premiere Pro CS4 which I used to encode this video pulled 100% of my 3.2GHz OC Q6600 for a good while and it will do the same with i7. So yes, you are right. Guess that is all I should have said....

caw2007 said:
I want that fan bracket so I can stop messing around with the stupid clips on my TRUEs. Any word on when/if those are going to be available?
Click to expand...

No clue, go it from Intel.
 
Update to the article:

The cooler that we showed being used in the above video is the cooler that is shipping with boxed Intel Core i7 processor models 920 and 940. As we joked about in the video, we did not think that cooler was up to the 965. Here is what Intel has to say.

The Core i7-965 processor includes a more powerful air cooled thermal solution called DBX-A that has some additional overhead baked into it. It also has two switch selectable cooling modes we call quiet and performance to allow a little more flexibility for our customers, in addition to the standard 4-wire PWM fan speed control.
Click to expand...
Click to expand...
 
Why is everyone so scared to see higher temps? You guys act like it is the HIV virus or something lol In all seriousness those temps are not that bad. I've been running a prescott for a few years with temps close to that anyways, time to relax guys jeez....
 
Great, video thanks for the update.

Didn't think Intel was going to ship a 120 with all their 965's.... :p
 
maleficarus said:
Why is everyone so scared to see higher temps? You guys act like it is the HIV virus or something lol In all seriousness those temps are not that bad. I've been running a prescott for a few years with temps close to that anyways, time to relax guys jeez....
Click to expand...

Basicly for the same reason people are scarred of programs and games using ram, they want 100000TB of ram yet only wanna have 0.1% of it ever used.
 
Hi Kyle,

I've been a long time reader and lurker. It was this video that honestly caused me to register. As others have said previously, this was really well done and informative.

The discussion at the end about water cooling vs. air cooling is really what grabbed me, tho. It's because of that I'm really going to consider water cooling in my next build. So, thanks for the "food for thought"! :)
 
unclewebb said:
I used to always post a link to the latest beta for my long time friends at [H]ard but somewhere along the way I must have done something bad and had my ability to post links to my download site suspended on this forum.

If you can turn that feature back on for me I promise not to misuse it and will only post links directly to my fileden site when new software is available.

I'm thinking that the new version of RealTemp is going to do a much better job at sniffing out sticking sensors. I think the Core i7 sensors are similar to Intel's previous 65nm sensors which had a lot less issues than the Core 2 Duo 45nm sensors. Thanks Kyle.
Click to expand...

Email me so I know what exactly the domain is and I will look into it.
 
Ive had 2 of those TRUE coolers and nothing but grief with them. In both cases the base was convex and contact with the ihs was shitty. The laughable thing was thermalright actually tried to say theat theyre made that way on purpose because most ihs are concave.

Was a discussion about this on several forums and people had concave, convex and flat surfaces to their TRUE. Its laughable that its luck of the draw as to what base surface you get, and if you get one thats bowed outwards or inwards you have to lap it which takes more time and adds additional cost to the thing.


As to those cpu's though, christ theyre hot little buggers. :eek:
 
Id be interesting in seeing the results on a lesser H2O setup, something like the swiftech compact system. Cheap for water and easy to install.
 
You must log in or register to reply here.
Back
Top