[leojharris]

scythe infinity: single fan vs. dual fan
--------------------------------------------------------------------

as discussed in several other threads, i ordered some ducting from thermalright along with a second silverstone FM121 to mount with my already installed Scythe Infinity and originally purchased, identical FM121 fan.

as it turns out and after a bit of installation testing ... it ends up that the ducting won't work because the heat-sink end of the duct was about 1.5" lower than the case exhaust portal and the particular thermalright ducting i had purchased had nearly zero vertical flexibility. with ducting that had equal 360 degree flexibility, it would work ... but not so with what i ordered. this of course is no fault of thermalrights ... this is, afterall a hack job of sorts.

here's an illustration of the ducting problem:

...

the ducting problem:



...

so, though bummed, I still mounted the second 110 CFM silverstone alongside the first, in a push/pull configuration ... so as to determine what sort of performance increase could be gained by using two fans; also, i was/am interested in maximum cooling with the lowest sound for 24/7 usage at a fairly high overclock of 3.6ghz on my 2.1ghz E6400.


the main cooling goal
------------------------------------------------------------------

the main goal is twofold:

1. fairly big 24/7 overclock on a E6400 with the lowest sound possible.
2. when very cold room ambients are available (depending on weather) i want max heat dissipation without any regard for sound, simply for the purposes of benchmarking and maybe melting a component or two.

keep that in mind when reading on ... because for benching sake and when running both silverstone's at max RPM's ... schwew! ... things get pretty loud.


fan's used
------------------------------------------------------------------

2 x 110CFM Silverstone FM121's

These fan's will be mounted on either end of the scythe infinity cooler in a push/pull configuration. i'll be measuring idle and load temps at the fan's max concurrent speed of 2400RPM's as well as readings at a lower setting of 1200RPM's or so.


zip ties (necessary evils)
------------------------------------------------------------------

since scythe only shipped me one set of mounting wires for a single fan, so i had to improvise and use zip ties. furthermore, since my zip ties wouldn't fit all the way around the cooler, i had to hook two together by inserting the male end of one into the female end of a second, thereby forming one *really* long zip tie.

the dual fan mount via this method is quite solid and secure and it's unlikely that the strap's infront of each fan are hindering the air flow in anyway.

also, i mounted the push-fan a bit lower than the pull-fan because it would more directly hit the NB and NB-heat pipe in that configuration; i lost like a degree or so on the CPU temps, but because the NB was heating up quite a bit, i thought it would be a decent compromise:

...




up and running / test data
------------------------------------------------------------------

after finishing mounting the dual fans, i turned the computer on and began the series of tests:

...



...

*** all tests conducted at room ambient of 22C (+/- .05C) ***
*** all tests run on E6400 at 3.6ghz ***


first i tested the dual fans at 2400RPM's, then at 1200RPM's; between each test, the system was powered down for thirty minutes with the case door open.

idle temps ... obtained after 20 minutes system inactivity.
load temps ... obtained after 30 minutes orthos (in-place large FFT's)

and after completing a full suite of tests for both max and low RPM dual and single fan speeds, here's what i found:

...

SINGLE fan at 2400RPM's (max):
-------------------------------------------------------------------
idle: 38C
load: 54C

DUAL fan at 2400RPM's (max):
-------------------------------------------------------------------
idle: 36C
load: 52C

....

SINGLE fan at 1200RPM's:
-------------------------------------------------------------------
idle: 39C
load: 55C

DUAL fan at 1200RPM's:
-------------------------------------------------------------------
idle: 36C
load: 53C


...

Final Thoughts...
--------------------------------------------------------------------

across the board, at whatever speed, the two fan's in a push pull set up reduced cpu temps by around 2 degrees on average.

the main thing to note, however, is that the two fan's running at a low RPM setting, still outperformed the single fan at it's max RPM setting; and the two fan's running at 1200RPM were significantly, even extremely more quiet than the single fan running at max RPM's. both fan's running at 1200RPM's were near imperceptible to the ear ... i had to kill the rest of my case fan's to hear them.

so all and all, i'm happy with the results; now i'm running even cooler and i can't even hear the heat sink fans.

another thing i'm interested in seeing is just how much heat this dual fan set up can dissipate once i can access room ambients appropriate for 4.1ghz+ benchmarking. keep in mind that all temps were obtained with a moderately low room ambient of 22 degrees and since air benchmark's are *totally* dependant on very low room ambients ... it will be interesting to see just how much of a difference 12 degree celsius ambients will make when trying to run benchmarks at 100% overclocks or higher.

hope this info is helpful.

happy new year!

[enz660]

Beautiful review

I was expecting two slow fans to be better than one fast one since there was some review on that same situation done on another site...

Now its time to lap the beast & pressure mount it for another (at least) 5c improvment.

As for the fan clips, I was thinking of making another pair out of a coat hanger, hopefully it will work. Again, good review.

Happy New years eve! Hope it will be as least sober as possible

edit: what vcore did you use for 3.6ghz?

[leojharris]

Originally Posted by enz660

Beautiful review

I was expecting two slow fans to be better than one fast one since there was some review on that same situation done on another site...

Now its time to lap the beast & pressure mount it for another (at least) 5c improvment.

FYI ...

two slow fans were better than one fast fan.

and also, way way more quiet than one fast fan ... that's why i'm leaving everything mounted.

give the temp findings a closer look, the title's may have you thrown off.

also ... i've already done the pressure mod (remember, i bent my motherboard all out of whack by tightening too much?) ... and the cooler has been seriously lapped as well.

thanks for the word, though ...

[leojharris]

Originally Posted by enz660

Beautiful review

edit: what vcore did you use for 3.6ghz?

for my 3.6ghz 24/7 settings ...

my vcore is set at 1.48 in bios but it really only runs at around 1.44 according to cpu-z (P5B-Dlx vdroop and all) ...

...

by the way, again ... the two fan, low speed setup totally outperformed the 1 fan, high speed configuration; so nice ... couldn't even hear the dual fan set up and it was still 2 degrees lower at idle and load than the quite loud 40dba single fan!

can't wait to see how your own set up performs!

...

and regarding 'least sober' new years ...

HELL! i've already started!

[Drinkyoghurt]

temps are a lil bit high aint they?

[trodas]

Thanks for the review. It show pretty nicely that two slow are not only more silent, but are also better that one fast fan in cooling effect

Great!

And a happy new year to you too!

[leojharris]

Originally Posted by Drinkyoghurt

temps are a lil bit high aint they?

as far as i'm concerned, not at all ...

a 2.1ghz E6400 running at 3.6ghz with dual prime load temps of 53C in a 22C ambient room ... is pretty darn good for air cooling.

my old modified zalman would load at around 65 to 66 or so if not higher in the same ambients.

thanks for the word ...

[Drinkyoghurt]

im just amazed because my p4 2.8 northwood with a 20% oc on air was 30c idle and 40c load, same with my amd 4200+, but that was with watercooling

[enz660]

dual core's are A LOT hotter than single core cpu's. I used to run my a64 which was oc'ed to 2.5ghz @1.6vcore on a tiny thermaltake venus 12, and the temps never went over 50c. His temps are very good.

[enz660]

crap...they sent me the AM2 bolt through kit instead of the lga 775 one. hopefully jabtech can send me the correct one soon.

[enz660]

well, I mounted the scythe with the same vantec stealth I had on my ac freezer 7 pro and compared to that, this cooler SUCKS! at 20c ambient (same as with the freezer) at 3.4ghz, 1.43vcore, idle temps were about 38c. With the stock scythe, they are 42! Load temps are now 10c hotter than before... Im a little upset . Hopefully the temps will drop lower than my old cooler once all of my modifications are finished.

edit: wow, they also sent me 24" sata cables instead of 36"..

[leojharris]

Originally Posted by enz660

well, I mounted the scythe with the same vantec stealth I had on my ac freezer 7 pro and compared to that, this cooler SUCKS! at 20c ambient (same as with the freezer) at 3.4ghz, 1.43vcore, idle temps were about 38c. With the stock scythe, they are 42! Load temps are now 10c hotter than before... Im a little upset . Hopefully the temps will drop lower than my old cooler once all of my modifications are finished.

edit: wow, they also sent me 24" sata cables instead of 36"..

sorry to hear about the shipping nightmare.

and dude! mounting an infinity and finding it idles at 5 degees hotter than a freezer 7! aww, hell ... the pain. maybe the pressure mod really does make a huge difference ... i never mounted it with the push pins, so don't know.

it has to be a mounting issue; without looking for the exact review sites i think it's safe to say the infinity has more than once been shown to be more efficient the the AC 7 ...

how does the mount feel with those push pins? does it seem like a crappy mount ... not much pressure? try remounting and having a look at how well the thermal compound distributed across the CPU ...

when i apply AS5, i use the tiny dab in center approach ... when using this method on my lapped heatsink and cpu, the pressure from the mount spreads the dot of AS5 to all edges of the cpu ... fully covered ... reseat it and see what it looks like.

[enz660]

the ppl at jabtech are sending me the correct kit via priority mail. I will lap & relap my cpu (found some odd scratches) and now with the backplate comming, I can mount the thing VERY tight. Im praying that it does at least 5c better than my old cooler. The pin mount SUCKS so bad...I pitty the fool who wants to save 5min of mounting time over 10c of cooling performance. I counted the fins of each cooler...42 fins on the freezer, 58 on the scythe. Yes, once I get this thing mounted correctly, it will be better.

[nd4spdbh2]

Originally Posted by leojharris

couldn't even hear the dual fan set up and it was still 2 degrees lower at idle and load than the quite loud 40dba single fan!

LOL coulnt hear 2 of thoes things.... even at 1200rpm! NO THANKS i bet it was still way loud... i dont even want to kno how loud the system was with both thoes runnin at full bore

n e ways nice results but, call me ***** i would back off the oc and pop a yate loon on there... i like QUIET! and NICE clocks btw.

[leojharris]

Originally Posted by nd4spdbh2

LOL coulnt hear 2 of thoes things.... even at 1200rpm! NO THANKS i bet it was still way loud... i dont even want to kno how loud the system was with both thoes runnin at full bore

n e ways nice results but, call me ***** i would back off the oc and pop a yate loon on there... i like QUIET! and NICE clocks btw.

for real . not loud at 1200rpm's ... my 24/7 overclock is 3.6 and that's what i run the fan's at.

the noise in my system comes from my antec tri-speed 120's at intake and exhaust ... might replace those with some yate loons (i also really like the noctua fans).

i'm totally into benchmark races so need some high output when i want it ...

[nd4spdbh2]

Originally Posted by leojharris

for real . not loud at 1200rpm's ... my 24/7 overclock is 3.6 and that's what i run the fan's at.

the noise in my system comes from my antec tri-speed 120's at intake and exhaust ... might replace those with some yate loons (i also really like the noctua fans).

i'm totally into benchmark races so need some high output when i want it ...

pff benchmark races... lol been there done that... got my e6600 to 4 ghz with a yate loon on a ultra 120... now my 24/7 oc is 3.0ghz @ 1.25v (8% undervolt)

[leojharris]

Originally Posted by nd4spdbh2

LOL coulnt hear 2 of thoes things.... even at 1200rpm! NO THANKS i bet it was still way loud... i dont even want to kno how loud the system was with both thoes runnin at full bore

oh i just couldn't help myself!

i made a video.

TWO FM121's at 1200rpm vs. 2400RPM ... yes, the 2400RPM setting is ungodly loud ... but the 1200RPM is quite lovely and quiet.

not a yate loon ... but acceptable for government work.



you can download the video at the link below:

...

file download:

Dual Silverstone FM121's on a Scythe Infinity

...

same video, on youtube.com:

View on Youtube.com



always looking for an excuse to edit video.

[leojharris]

Originally Posted by nd4spdbh2

pff benchmark races... lol been there done that... got my
e6600 to 4 ghz with a yate loon on a ultra 120...

max overclock is one thing; *benching* is another ...

consider yourself hereby Officially Invited(tm) to the Sciencemark 2.0 benchmark race over at TechPowerUp forums, here's a link to the current scores: http://forums.techpowerup.com/showpo...postcount=2091

and, here's a link to the general thread (the latest post's anyways, the score listing moves around): http://forums.techpowerup.com/showth...=14736&page=85

i had the number one spot, against a handful of 6600's and 6700's until a fellow by the AKA of "dominick32" waltzed over and handily dethroned me with his sickeningly overclocked 6700 ... (i didn't stand a chance). i think he's from OCForums, actually ...

anyway, we need more competition ... so since you've 'been there and done that' c'mon over with your faster (stock anyway) cpu and knock me off the second place spot!! (that is, if you think your up to it)



(all in good fun, btw, but you simply can't compare "max overclock" with "benchmarking".)

[S3CUR3]

Thanks for a really great write-up, leojharris. Informative, clear, and to the point. Kudos.

[shard]

Do you think that I could do this and gain much on my Scythe Ninja?

[enz660]

you can do this to any heatsink that uses pins to mount it. Heck, the stock pin mounts on this infinity make it perform WORSE than my pressure mounted ac freezer pro (both with the same fan, 7c higher load temps)

[shard]

Originally Posted by enz660

you can do this to any heatsink that uses pins to mount it. Heck, the stock pin mounts on this infinity make it perform WORSE than my pressure mounted ac freezer pro (both with the same fan, 7c higher load temps)

I am just using the clip mounts... Unless that is what you are referring to?

[enz660]

oh whoops....i thought u were talking about the heatsink mount, not the fans :P

edit: still waiting for my bolt through kit...tried two yate loons instead of one vantec stealth, load temps dropped about 3-4c. Still not as good as before..

[leojharris]

Originally Posted by shard

I am just using the clip mounts... Unless that is what you are referring to?

sure, if the ninja will accomodate two fans in the same sort of arrangement ... you would see similar benefits.

dont' know anything about the ninja, but i think it could use two fan's.

also, the same pressure modification for the mounting process will also improve your ninja's peformance.

[enz660]

Ok well I remounted the scythe and the temps did improve. The performance was about 1c better from my old ac freezer 7 pro with the same fan. I got the new backplate and tried to pressure mount it. I probably messd up becuase the temps are now hotter than before. The metal clips like to bend inwards so I will try to use these thermaltake brackets that came with a big typhoon for someone else's pc. If those dont work, I will have to fabricate something else. Im Praying that I can fix this problem and actually UPGRADE cooling performance by at least 5c. Ill give it another whirl after work.

[leojharris]

Originally Posted by enz660

Ok well I remounted the scythe and the temps did improve. The performance was about 1c better from my old ac freezer 7 pro with the same fan. I got the new backplate and tried to pressure mount it. I probably messd up becuase the temps are now hotter than before. The metal clips like to bend inwards so I will try to use these thermaltake brackets that came with a big typhoon for someone else's pc. If those dont work, I will have to fabricate something else. Im Praying that I can fix this problem and actually UPGRADE cooling performance by at least 5c. Ill give it another whirl after work.

so the time you saw the 1C drop, were you using the stock push/pins? i also noticed that the mounting clips were a bit flimsy; if you figure out a way to do a clean fabrication, let me know.

have you lapped the infinity?

...

keep at it; at least you broke even on that one install ... and even with a 1 degree drop.

just curious, how do you apply AS5? spread it or tiny dab in center? i've done both and found that the tiny dab in the center works best ... i tend to put too much AS5 on when i spread it.

i've got all my new pressure mount hardware in hand as i showed you but i'm afraid to dismantle everything since it's performing so well ... think i'm going to keep my hacked pressure-mod install until i have to take my board out again for something else. i can tell you one thing, though ... the thing is on *tight* ... i ended up having to use regular nuts as my wing nuts hit the base of the cooler; had to tighten the thing down with a wrench! that's why i bent my motherboard so bad. anyway, when i backed them off, i backed them off until the MB was just ever so slightly bent and then assembled the machine.

i'll be glad to put a backplate on ...

[enz660]

I did lap the heatsink, even though it passed the ketchup test with flying colors. I used the zalman stg-1 thermal paste. I believe the reason the pressure is bad is becuase there is nothing under the brackets, which is why they bend. All the pressure is going into the brackets instead of the cpu.


update: OK! so I got a ton of parts, backplate, metal washers, rubber washers, bolts, nuts, brackets..EVERYTHING. Was a major pain to mod the backplate & mount everything. Temps have improved. they are about 4-5c better. Another thing I saw when I was screwing the heatsink tighter & tighter was the BACKPLATE WAS BENDING!! This steel back plate was not even enough for the pressure I have mounted this thing. I unscrewed it a bit so its not bending so much, but I will see what more pressure would do tomorrow. Right now, its almost 3am, im hungry, tired, and just happy I FINALLY have some improvment. More updates to come...

[enz660]

ok time for a new post. Ive done some mods to the heatsink and I will try every possible way to apply thermal paste. I will use as5 and zalman stg-1. Hopefully one them will come out on top. Will make a massive thread on my modding & testing soon.

[leojharris]

Originally Posted by enz660

even enough for the pressure I have mounted this thing. I unscrewed it a bit so its not bending so much, but I will see what more pressure would do tomorrow. Right now, its almost 3am, im hungry, tired, and just happy I FINALLY have some improvment. More updates to come...

where there's a will ... there's a way.

i too had to bend things silly in order to get good temps ...

as i told another person earlier and on a different thread ... i can't believe scythe would make such a good cooler but sabotage it with the mounting system ...

(now you see why i'm *not* messing with my ghetto pressure mod: ... it's working *great* ... and i'm not going to take a chance on f'ing it up)

good to see you're getting somewhere, though ...

once i get the nerve to redo my pressure mount, i'll take pictures of my current set up ... with the sawed-off segments of ball point pen cases in place of the springs ... and you'll laugh your ass off!

[leojharris]

Originally Posted by enz660

I will try every possible way to apply thermal paste

that's good ...

i tried several different approaches and ended up using the rice sized (no more, though) dab approach.

i found that as long as i had lapped both HS and cpu that the tiny dab would spread to the extents of the cpu once applying mounting pressure.

through trial and error, i have found that applying too much thermal compound will seriously degrade performance ... so now, i apply as little as possible while ensuring full coverage of the CPU heatspreader surface; usually, a tiny dab in the center will suffice.

looking forward to your next test results ...