Gotta say those results are impressive for the cost, but I agree with everyone that the numbers don't make sense. It shouldn't be performing that well..
Shrug -- I agree, impressive, and I agree they seem too good to be true for what it is. I've said a few times, I was and still am shocked. But the numbers don't lie.
Apparently the Corsair H50 performs much better when mated with a pair of 120mm fans in push/pull and makes for about a 5c drop in idle; and load temps and most seem to be going that way and also installing so the heat it still exhuasted out the rear of the case.
Maybe so.. I don't install that way though for these results.
I use a 1500-1600rpm 120mm fan pulling cool air from the outside to the radiator on the inside of the case (that is on the ground). I do not have a second fan pushing against this on the other side of the radiator. If it would improve temps I might consider it, but temps are more than stable for now.
I have the single 120mm pulling air from the outside, as is shown in picture (if you know how the H50 has to install you know how it has to mate against a fan).. and I have the 120mm exhaust above (exiting the top of the case) -- simple principle of hot air rises. Using thermodynamic air flow principles.. Cool intake on lower front, then aforementioned exhaust on top. In the middle is the radiator fan that has little to do with case flow as it is directly blocked by the radiator block itself.
What cpu voltage setting are you using Logikos?
I run 1.36v with my 4.011Ghz setup that is shown above.
Whats your ambient temps? Also, stock non-overclocked voltage I understand low temps (especially when you show us it when its not even running its rated speed and voltage lol).
Huh? I don't have a thermometer in my case,... but ambient temps in the house run between 21 and 22c -- in the case it is lower due to cooling, so it makes sense that idle would be lower, slightly. Which it is. With the AC on it will be 18-19c.
I posted stock temps because that was asked! It was a direct part of the conversation and something someone didn't 'believe'... I posted my overclocked temps as well.. I posted it ALL so there wouldn't be any of this 'doubt' or comments like you just mentioned. I knew it was coming.
The Intel i5 at Idle runs drops the multiplier to 9xFSB -- It is just smart/energy efficient like that. Immediately upon sensing load it goes to proper clock speeds. Which if you look in the images again, you will see idle 9xFSB.. 100%load.. 21xFSB. Make sense now?
And I'm still baffled with the voltage comment... where are you getting this not running at rated voltage?
Everything is running at specification, and OC'ed to specs I say.
Besides that, you are still idling at 20c. Ambient must be minimum 15c, which is quite low and can bring temps down. 15c and I will load at less than 50c.
So? The temps this is produced at, is inside a home. And were actually with the heat on because it is winter time. So it isn't going to get any hotter in regards to ambient temps. This isn't a competition anyways..
-- I am just backing up claims, one time.
lemme put it this way... wats the longest time your computer is every on? My computer failed prime after 3 hours but it doesn't worry me cause im never doing anything at full load for more than 3 hours
I never turn it off. When I change a setting and test for stability my routine goes like this.
Prime95 Blend Torture for 15 minutes initial.. if pass , stop. Boot with Memtest ISO,.. Run for ~hour. If pass,.. Run Prime95 for 24hours straight. If no errors, I consider it completely stable.
If a setting errors out after 8 hours of P95, I won't keep it. But I am just anal like that.
Logikos, if you are replying to several posts in the same thread, you can click the "quote +" button and it will quote each post in the same response. That way, you don't have to fill the thread up with consecutive posts.
Thank you! Most helpful response yet!
Now, I have no idea what the typical temps are like for the 1156 i5 and i7 CPUs are like, but I can confidently say that to achieve temps as low as 20*C, you have to have some pretty low ambient temps. I would love to see how a top-end air cooler would perform under the same conditions.
Well the i5 and i7's are very common around here.. but the H50 isn't so common 'yet'.. I think that is where the perplex is here: Very low temps with a WC setup that isn't a monster Radiator, full-tower, etc etc.. (typical) water setup. Instead it is a 2hour Corsair 120mm block/pump/radiator/fan/input-ouput plumbing and wiring (with MB backplate, rings, and so on) that is producing results closer to what a bigger rig WC setup does. Just unexpected is all.
If my ambient temps are so low then it is only because of the way I built my computer. I spent alot of care on airflow, cable management and so on (and have 4x140mm fans otw to replace the 2 of the 120's and to add 2 side fans for further cooling to drop temps further *and noise*)..
My point is.. the temps in the computer room aren't 'cold' or 'frigid' .. hardly.. they are warmer than most other rooms because of the computers in the room producing the heat. I would say an easy 22c.. perhaps 22-23ish. It is the location in the corner, on the ground where the air is coolest, and on a self built heat dissipating platform (I built due to the bottom mount PSU, not wanting it to pull off of carpet),.. and the airflow inside the case.
IE: I don't have this Rig hanging out the window where it is 0c, to run these tests lol. (Would be fun to do though).
Top end air coolers -- aye,.. I would love to see what the best of the best (loud as hell) air coolers do with a 4Ghz OC i5 on 100% load temps in P95.
That brings me to my next point; the Corsair H50 has been shown to perform on par with the top air coolers like the TRUE1 120 and Prolimatech Megahalems, and costs about the same or more. The Corsair H50 will also suffer from all the same problems that any other water setup will, like the pump going bad, fluid leaking or evaporating, etc. An air cooler, on the other hand, will basically last as long as you could ever need it to. Not that it's a bad cooler, but for those reasons I personally wouldn't recommend the H50.
I'll have to respectfully disagree on some points here.
The H50 will not suffer the exact fate as any other WC setup. It is sealed. ..And due to it being sealed, the Water should not evaporate nearly as easily.. perhaps not at all. .. but it isn't quite the same as self plumbed WC setups. Yes, the pump can go bad,. so can a fan motor. Fluid can leak, yes.. so can bearing grease.. but again, the design of this mass manufactured 'already built and sealed' design prevents alot of those common "over-time" flaws. I would personally guess the pump motor would die before the fluid leaked or evaporated in this setup. And the pump is rated for many , many years.
A fan will not last as long as you ever need it to. It, like any other 'moving part' is subject to wear and tear.. and will and do break down. I can't tell you how many fans I have replaced on just about every conceivable piece of computer equipment you can think of. They go bad over time.. just like anything else. No, usually not for 4-5+ years.. but they go bad.. (bad is really the wrong word.. 'wear out' is more accurate'.) -- I fully expect to replace this in 3-5 yrs... Likewise I usually build another computer before that time frame has passed, and will clock this computer back to stock, and if this fails put a stock HSF on it.
And another thing to note, OP has a 1366 i7, which normally runs hotter than the 1156 i5 and i7. Temps of 83*C are perfectly fine for a 1366 i7 and really, you still have quite a bit of headroom with that. Perhaps the Corsair H50 is beginning to reach its limit with the 1366 i7's TDP but does a better job with a smaller chip like the i5; I don't really know, though. Just putting it up there.
Good point,.. I'll take your word for it -- I haven't researched the temp differences on the 1366 sockets vs 1156.
Back on-topic:
With i7, small FFTs isn't very helpful. Large FFTs stresses the i7 WAY more and is much better for testing for CPU stability. And yes, stability is in the eye of the beholder.
To me, any instability is an instability. If a stability test fails, I don't know why I would ignore it and consider it stable anyhow, at least when below 24 hours of Prime testing (and small FFTs, at that). If you think 15 hours is enough for you, then go ahead and try it. If you encounter any problems, make some adjustments and try again.
Good advice..