Yes, this is another thing I've been thinking about...
Any idea on size I'd need (no gfx cooling needed) for a reasonable O/C on a 95W TDP processor and what would be a recommended passive rad?
Overclocking silently... anyone done it?
- Thread starter arad85
- Start date
Any idea on size I'd need (no gfx cooling needed) for a reasonable O/C on a 95W TDP processor and what would be a recommended passive rad?
Soldato
- Joined
- 13 Oct 2011
- Posts
- 11,881
- Location
- Melbourne, Australia
so i removed 2 top fans and 1 front fan
set to cpu fans to 400rpm via motherboard
now pc is slient
That's the way to do it. massive heatsink and slow silent fans. Also case insulation if you havnt already
a 1080 or 1260 should be good enough to do that
Thanks.... A 1080/1260 what? (sorry, not up on w/cooling).
Soldato
- Joined
- 13 Oct 2011
- Posts
- 11,881
- Location
- Melbourne, Australia
That's enough to cool a forest fire. Extremely big, extremely expensive for no reason..
A big Air cooling will be a lot better. The £200+ you will spend on watercooling against a £50-£60 air cooler...
And we're back to square #1... I'm looking at spending £90 on a passive air cooler (Nofan 95). The easy solution is to live with what I have now (Scythe Ninja) or moving to another heatsink (pick one from: http://www.silentpcreview.com/Recommended_Heatsinks or your fave review site - although the Ninja doesn't do too bad) and just controlling the fans intelligently. But what I'd like to do is for the system to remain silent when running flat out...
I'm not after ultimate O/C nor am I after lowest temps, I'm after some overclock (4.4-4.5 on the 2600K), reasonable temps (~80 deg C max most days to allow some headroom for warm nights) and as quiet as possible all the time.
I guess no one has tried entirely passive cooling for this.... I may just wait for Haswell (will suit my setup better), spend the £90 and see how far I can get passively
But you cant have everything i want a 100mpg, 1000 pound purchase price DB9 in black and orange but aston just wont give me what i need.
Overclocking creates heat to remove that and prevent cooking you need either a big heatsink with fans (loud) or water and a large external radiator without fans (expensive).
Take your pick, oh and waiting for haswell wont help yes each new cpu gets a little more energy efficient but more work done is more heat produced and dont forget haswell is set to be a 84 tdp part compared to the 77 of ivy and 95 of sandy youl still need sufficient cooling to achieve temperatures good for overclocking
Overclocking creates heat to remove that and prevent cooking you need either a big heatsink with fans (loud) or water and a large external radiator without fans (expensive).
Take your pick, oh and waiting for haswell wont help yes each new cpu gets a little more energy efficient but more work done is more heat produced and dont forget haswell is set to be a 84 tdp part compared to the 77 of ivy and 95 of sandy youl still need sufficient cooling to achieve temperatures good for overclocking
Do you have any candidates for a quiet 200/230mm fan?
Well I hit on a decent one in the sprectre pro 230mm. But manufucaturing variance comes into play I guess so there are always going to be better or worse ones within even the same models. Another 230mm, an nzxt, was horribly balanced (really feel it when holding it) and on a sidepanel those vibrations where exacerbated, but it was a cheap case supplied one.
Only trouble is the R2 doesnt have any mounts so it would be a mod job. And the design of the grill/airholes has a huge effect on the level and type of sound. Honeycomb while not the worst is also far from ideal.
The increased tdp takes into account the igpu being active and the higher stock clocks. It's reasonable for sure to assume that it's no worse than ivy. Whether thay have made it easier to get that heat out by fixing the IHS mount (apparently it's not the paste that is the major casue) is another matter.
Last edited:
I think you'll have to be the guinea pig! What sort of usage did you have in mind if you'll never be heavily loading it it might suffice, it will just always be on the toastier side of warm ofc.
I'm willing to give up MPG here....
You need to have a method of removing the heat. A big enough heatsink will remove the heat without the fans, just very few people make those - plus it seems most of the overclockers are gamers (or want to use the compute engine of the graphics card) so are worried about cooling that too. The NoFan works well with an unclocked i7-2600K, I was just asking whether anyone has tried it overclocked...
Yes it will - for two reasons. Firstly, I'm expecting mobos with 3 PCI-e 16 slots and I'll be able to move my graphics card away from the CPU as it has a cooler that laps over the back of the board. Moving it further away will allow me to fit a NoFan on my board without having to spring for an E-ATX card (only ones I can find with PCI-e 16 not in slot 1 or 2)...
Secondly, 11W is still 11W less to dissipate at the same clock... I'm not sure how accurate throttlestop is, but at 4.4GHz, it is saying my TDP is 95W which is just on the edge of what a NoFan can support.... Doing a similar overclock on an Ivy/Haswell device might be enough to be within margin...
Fundamentally, I understand overclocking, what it does, how it affects temps etc... If I had ALL the information (actual power output, case temps, heatsink efficiency) I could probably work it all out myself. The problem is, I don't have all the information.... I guess I'm trying to to understand whether big heatsink+loud fan or water is religious dogma or absolute truth for an averagely overclocked part...
Now... where did I put my thermocouple probe.....
Looks that way
Main heavy usage is for video recoding.... Everything else is small peanuts to that.
How about a 230mm hole?
Thanks for the info. I don't - yet - own a Dremel... perhaps after some experimenting I may
Ok put it this way before I ran water I ran and overclocked ivy at 4.3ghz off an nzxt havik 140 at 4.3 I was well under the 77w tdp of ivy a good 15-20w in fact however if i primed or ran folding for any decent length of time (under half hour) with no fans turned on in winter with a low ambient i was breaking 80c and my heatsink was quite warm to the touch. The problem is air just cannot shift 50+ tdp without the help of fans, also remember watts tdp doesnt = heat output nor does it always equal watts drawn.
for example
http://media.bestofmicro.com/U/O/338064/original/load power.png
http://media.bestofmicro.com/U/P/338065/original/load temp.png
In that instance yest the sandy draws about the same as the difference between the 77 vs 94 watts however look at the temp difference and that is all at stock once you start adding performance and voltage temps sky rocket
Dont get me wrong it can be done but the reason intel offer to low watt non clockable cpus is for your wants and needs, and if your not fully loading then you dont need all the performance so leave it at stock
for example
http://media.bestofmicro.com/U/O/338064/original/load power.png
http://media.bestofmicro.com/U/P/338065/original/load temp.png
In that instance yest the sandy draws about the same as the difference between the 77 vs 94 watts however look at the temp difference and that is all at stock once you start adding performance and voltage temps sky rocket
Dont get me wrong it can be done but the reason intel offer to low watt non clockable cpus is for your wants and needs, and if your not fully loading then you dont need all the performance so leave it at stock
Last edited:
OK. Good info. Thanks.
And this is where we potentially disagree. The tower heatsink with vanes that you tried and that relies on airflow to remove the heat can't dissipate it. That doesn't mean no practical heatsink can dissipate it.... The NoFan is a radically different design - using widely spaced thin tubes (maximising surface area) to dissipate the heat rather than narrow vanes.
Nope, it's the thermal envelope. Basically what you are doing with heatsinking is wicking away the heat from the transistor junctions. To do that, you transfer the heat out of the chip to the heatsink. The thermal resistance (how much it heats up per watt input) of that heatsink is well defined and the tower heatsinks will have a value that lowers a lot (lower = better efficiency) when you blow air over the heatsink. Can I find these figures for PC heatsinks - not easily.
The NoFan site states the thermal resistance as 0.51 deg C per W, which means that put 100W into it, it will heat up 51 deg C from ambient. Ambient in the room is clearly different to ambient in a closed case. It should just about work trying to dissipate 95W continuous... but you'll be in the 80's at the junction temps assuming perfect heat transmission to the cooler...
I know it can be done - I've done it, both with a 65W TDP Intel and 95W AMD processor - just not overclocked. I don't need the performance, but I want it so video recodes happen quicker - so it firmly falls into the line of "just because"
Well... it depends.
If the passive heatsink is good enough and the airflow is good enough, then convecting the heat away will do it. As far as I can see all "traditional" heatsinks and most water cooled rads rely on forced airflow to keep things cool. With passive rads/heatsinks (like the NoFan) you are relying on a radically different heatsink design and convection to do the work for you. If you do need to move air to help cool it, then you should be able to do it with a significantly lower flow rate and hence much, much quieter case.
As I said above, I have built two boxes already (that are on 24/7) using a NoFan cooler. Both are essentially convection cooled and have no fans in them to keep the box cool - but they are not overclocked....
Soldato
- Joined
- 30 Nov 2011
- Posts
- 11,521
even tiny amounts of forced air flow can massively improve the capability of a heatsink or radiator to remove heat from a device... if you wanted to go with the nofan but also overclock, you could use something like a 180mm (the same size as the heatsink) and mod a hole in to the case directly above it to vent the heat out
a 180mm fan on minimum will be basically inaudible, but will hugely increase the capacity of that heatsink to move heat
if you are going to overclock then the motherboard will need some kind of forced cooling for the VRMs anyway, so you aren't going to get away with having absolutely no fans in the system
the trick will be to have a fan controller, put any fans on minimum and then turn them up until the fan actually spins, and then spins to a level where you can actually hear it over any background noise
a 180mm fan on minimum will be basically inaudible, but will hugely increase the capacity of that heatsink to move heat
if you are going to overclock then the motherboard will need some kind of forced cooling for the VRMs anyway, so you aren't going to get away with having absolutely no fans in the system
the trick will be to have a fan controller, put any fans on minimum and then turn them up until the fan actually spins, and then spins to a level where you can actually hear it over any background noise
I'm not convinced slow fans will have any impact on a VRM buried deep beneath a heatsink though... Still I have a Aquaero 5 and 4 temperature probes, so I will be able to instrument things and experiment when the new case comes (this is likely a 2 stage process - stage 1: move to new case with Sandy and monitor temps, stage 2: move to Haswell and see what I can get away with passively).
I have to say I think people are way too conservative when it comes to cooling - especially if you have a board with decent heatsinking on the VRMs. Anyone got any links to "decent" mobos expiring early due to temperatures being too high?
Why not wait for some reviews of this and just run with the 2 front intakes:
http://forums.overclockers.co.uk/showthread.php?t=18510891
I'm running a TJ08-E with the 180mm on low with its speed further reduced by a fanmate2 (~300rpm) - everything else is passive, inc. the cooler on the x4 905e. The only thing I can hear is the PSU squeal up to around ~0.5m with c1e enabled. I actually tried 3 different PSUs to make the machine dead silent with no luck because of this.
I've got a Scythe Kama Master fan controller in another PC which works wonders. Its able to start up fans with silly high starting voltages and drop them to inaudible levels.