Read more.Quote:
Quietly capable.
Printable View
Read more.Quote:
Quietly capable.
I've stated this before. This should be compared to other air coolers in a similar price bracket to get a proper comparison of value and performance, not AIO's or others at twice the price.
I agree to a large part, but also remember that when a dirt cheap thing perform really well, nobody complain about seeing it VS more expensive parts it then beat or match.
But yeah, the chart should have a bias towards similar priced coolers of the same kind, and beside that maybe old favorites one could assume this might replace.
I think part of the problem is all the coolers in the chart are the ones that have been tested on the new Hexus testing rig, so as to keep things fair between them.Quote:
Originally Posted by mers
I agree that all the coolers need to be tested on the same rig.Quote:
Originally Posted by Terbinator
Whit I would really like to see tested though are the AMD and Intel stock coolers. I know the 3950X doesn't come with a stock cooler but most people will be buying fewer-cored CPUs that do have a boxed cooler and for me at least, the difference between stock cooler & aftermarket cooler is more important. I'm saving up for a new system right now and I don't know if I can get away with stock or if setting aside £75-£100+ for an aftermarket cooler is worthwhile.
I bought a ryzen 3600, and the box cooler is eminently usable. Hottest i've gotten is 65c whilst intense gaming. (according to HWinfo} It's pretty quiet too. I hear the latest box coolers from Intel are just as good, according to reviews.Quote:
Originally Posted by EuphoniumI agree that all the coolers need to be tested on the same rig.Quote:
Originally Posted by Terbinator
Whit I would really like to see tested though are the AMD and Intel stock coolers. I know the 3950X doesn't come with a stock cooler but most people will be buying fewer-cored CPUs that do have a boxed cooler and for me at least, the difference between stock cooler & aftermarket cooler is more important. I'm saving up for a new system right now and I don't know if I can get away with stock or if setting aside £75-£100+ for an aftermarket cooler is worthwhile.
Mind you...I've got 5 case fans too. Two 140mm Arctic for intake and 3 120mm Lian Li for exhaust.
Depending on what software you plan to run - Then you're mileage may vary.
Don't worry...The stock coolers are decent.
Looks to be a decent cooler.
CPU coolers can be a challenge to test objectively as their performance will depend on the motherboard's fan curves which in my experience can be terrible. I've mentioned this before, but using my own setup as an example, I have a 7700 CPU and a Hyper 212 cooler. With the motherboard I have, the default fan curve is annoyingly loud under load as it ramps fan speed right up even though the heatsink itself is just lukewarm - the extra fan speed doesn't really achieve anything other than making more noise. I had to manually drag down the fan curve to make it reasonable.
In that scenario, the cooler could be blamed for being loud when in reality it is the fault of the fan curve for being unoptimised for the CPU - it is unrealistic to expect that CPU to maintain a very low temperature under heavy load regardless of cooler because of the poor thermal conductivity between die and heatspreader.
An easy win for a cooler manufacturer would be to equip the cooler with a slower fan to compensate for the poor fan curves out there, however that could impact its performance on higher power CPUs with a soldered IHS.
To that end, I wonder if we might see some solutions (perhaps there are some and I've just not noticed?) where fan speed is set relative to heatsink temperature. I know this is the case with some AIO coolers (e.g. basing fan speed on coolant temperature as increasing fan speed with already cold liquid is pointless), but seems to be absent as an option for air coolers. Even if motherboards gave us something like a temperature probe header to connect up to the heatsink, it would be much better!
Edit: I think I've heard of some motherboards having a sort of tuning feature for coolers, but I've not tried any out myself.
Mine has that. It helps a lot, but is still quite conservative.Quote:
Originally Posted by watercooled
My asus board has "fan tuning", but that's just it running all the fans 0-100% to see how fast they go and what the start voltage/duty cycle is - nothing cleverer than that. Motherboard fan control is just bad in general - they're wedded to the idea of linear interpolation based on CPU temp so the users can move the line around on the graph, whereas a nice PID controller where you set the desired maximum temp would be a whole lot betterQuote:
Originally Posted by watercooled
As I suggested earlier it would be interesting if we saw some air cooler set fan speed based on heatsink temperature - the manufacturer could spend some time properly tuning the curve then. But in reality, I suspect that would start pushing the price into AIO territory anyway.
It's just sad that so many coolers receive bad reviews when they're not bad at all, and the generic fan curves on the motherboard have a screaming fan blowing air over a stone cold heatsink, achieving nothing.
For example if my Hyper 212 was reviewed in combination with my motherboard, it would likely be criticised for being noisy. Although it's noisy with the default curve, it's nonsense because with my adjusted curve the fan barely even exceeds 1000rpm under full load in a warm case.
I just think this is something people should consider when choosing and installing a cooler. Does anyone know if some fan controllers do this? Although my solution does work, it's something I'd consider out of interest.
Heatsink temperature isn't a useful figure, when it is CPU temperature that is going to affect performance.Quote:
Originally Posted by watercooled
And there is not going to be any "right" default curve.
It is going to need configured to suit the user's preferences, as some may well tolerate a certain amount of noise more than other people if it means better cooling.
At which point, you may as well use existing options.
And that is before you consider different heatsinks, different CPUs, different cooling from case, ambient temperatures.
No automated "improved" curve is going to be always right as long as it is fixed on assumptions about the conditions and the tolerances of the user.
This is untrue and the reason I made the post in the first place. CPU temperature is almost irrelevant when it comes to setting fan speed - as I've explained already, if the heatsink fins are already stone cold then increasing fan speed will achieve nothing apart from making the system more noisy. If the CPU is so hot it's affecting performance by throttling with a cold heatsink, then something is broken - either it's mounted incorrectly, the heatsink is faulty e.g. leaked heatpipes, or the CPU's heatspreader isn't attached to the die properly. In any case, moar fan speed will achieve nothing.Quote:
Originally Posted by ByteMyAscii
Fan temperature can directly influence heatsink fin temperature, not CPU temperature, so that's the sensible thing to base it on.
There can be far more sensible default curves if motherboard manufacturers would design them around the correct generation of CPUs rather than having a curve where fans are set to 100% at 60C (for example), when that temperature is unrealistically low for a given CPU.Quote:
Originally Posted by ByteMyAscii
It is of course going to depend on user preferences, but it is incorrect to believe higher fan speed will mean higher performance because of what I've already explained.Quote:
Originally Posted by ByteMyAscii
Which is why heatsink fin temperature is sensible - the same reason why higher-end AIO coolers set radiator fan speed based on coolant temperature - just like the heatsink temperature, there is no point having radiator fans screaming away if the coolant temperature is already cold.Quote:
Originally Posted by ByteMyAscii
Again, that's not really true as can be quite easily demonstrated by systems that do it. They might not be totally optimal but they're a heck of a lot better than the dumb default curves set in some motherboards. Quite a straightforward if slightly time-consuming way to do this is to put a heavy load on the CPU with the CPU fan set to 100% and allow it to reach steady-state, then slowly decrease fan speed until you see temperature start to rise appreciably - this is the area at which more fan speed stops being useful, but the exact point would depend on user preference e.g. a noticeable noise difference may only make a tiny and irrelevant difference in temperature.Quote:
Originally Posted by ByteMyAscii
That method doesn't directly factor in higher intake temperatures e.g. with GPU load. That's not impossible to simulate but likely harder to do in BIOS (nothing stopping motherboard manufacturers doing it in Windows software though), but it could be extrapolated if not based on a table of sensible temperatures for supported CPUs.
That's more or less what I did manually, and is an indirect way of basing fan speed on heatsink temperature, which takes us right back to the original point - you could cut out all of this faffing by just measuring it directly. And again, if you think that's not feasible, check out the AIO coolers using coolant temperature for the exact same reason.
Hope that makes sense.