Hello everyone, I'm new here, this is my first post, so treat me gentle! Also, excuse my terrible English please!
I'm upgrading my everyday desktop PC occassionally, and now with the latest addition (replacement of 9800GT OC + Arctic Accelero TwinTurbo against a GTX460 OC from Gigabyte), it got to the point where the heat building up inside the case is too much for the basic ventilation. This shows as the GPU is running ~75°C under full load (while others with the same GPU report 60-is temps), the CPU is getting to around 62-63°C too, and the MCP78 to 70°C, even the PSU is running warm (even at idle!), and once I stop the load on everything, it cools down rather slowly.
Unfortunately, most of my components are quite hot running parts, as if I were deliberately picking the hottest possible components... My case also has a lot of drives and wires and cables inside, and sound damping too, and no room for hiding the cables either. So it's far from being optimal.
When it gets hot (mostly when gaming), it is also terribly loud, but that's because of the PSU - it's got to be a hidden Jumbo Jet inside!

All in all, I want to improve cooling, and if possible, reduce the noise, and hiding those cables would be nice, too.

This is what I got: http://img529.imageshack.us/img529/5399/desktoppcn.jpg [Large picture warning]

Case: Cooler Master Sileo 500, rear fan replaced with a Coolink SWIF2 with slightly better airflow @1000RPM
Motherboard: Gigabyte GA-M720-US3 rev 1.0
CPU: Phenom II X4 940 (@stock, as with everything else)
Video: gigabyte GTX460 1GB OC rev 1.0
PSU: Chieftec CFT-650-14CS
CPU Cooler: Arctic Freezer 64 Pro (PWM)
RAM: 2x2GB blank Kingmax DDR2-1066
HDDs: 2x WD3200AAKS in RAID0, 1x WD5001AALS in the top rack, 1x WD2500xxxx (Caviar Blue 250GB, ATA100) <--this one is very rarely used, and is turned off most of the time. (And I'm planning to add another 2 to the raid -> need a bigger case!)

The rear fans (fan+psu fan) suck the air out of the case, the front fan blows inside the case, and there are two small 4cm fans on the hard drive racks (top two units), that are also sucking air from outside, through the rack and into the case (but one of these racks is mostly turned off completely, so no fan.)
I think, the intake is inadequate. There's a "block of HDD's" right in front of the intake fan. There are a few vent holes on the side and on the bottom of the front bezel and some on the back of the case. No vent holes on the side panels or anywhere else.

I thought of getting a new side panel that has vent holes, for more fresh air.
Was looking for a new case, that's bigger and better ventilated too, but due to constraints, I only found the Antec 1200.
If you think, a new case would be the way to go, could you recommend me one?
[*] no front door
[*] 5x or 6x external 5.25" slots minimum!
[*] 1x 3.5" external slot
[*] room for 4-6 HDD's
[*] depth should not exceed 520mm -> this is where the most big tower cases fail... (width and height are not limited)
[*] cable management
[*] if possible, PSU not on the bottom, or at least not sucking air from outside -> there's way too much dust where I live
[*] no lighting, LED's, fireworks, acrylic side panels if possible.

Any help and ideas are much appreciated! Thanks in advance!

I believe an adapter that converts 3 5.25" bays into a 120mm fan mount will be best. That will create a strong and straight path through the case with the CPU and possibly some suspended hard drives in the flow. It would probably push the stagnant air at the bottom of the case around too.

You can move your DVD drive up and remove the CD drive and second hard drive adapter to make room. The drive you don't use can be placed in the hard drive cage or in an external enclosure, there are adapters that will allow that old CD drive to operate externally as well. I imagine you don't use it often anyway.

2.5" laptop drives can be used for your future upgrade as they're generally quieter and there are accessories that will let you mount them over unused PCI sockets.

Note that the stated CPU and GPU temperatures mean lots of headroom still. Yeah, you can bring them down by adding a fan (or speeding up the exhaust) and removing the slot covers below the graphics card (it's cooking in its own heat). However, you shouldn't expect the PSU to become silent without a fan swap.

Well, first of all I think any wanna-be-quiet system have to run warmer than a more noisy one.
Then, according to me there should be nothing REALLY wrong with your temps (about the noise, I have some mental reservations).
Moreover, the Sileo - being quiet - properly lacks of ventilation.

About the MCP, it's always been a pain IME: according to me the best way to go is for a new heatsink. Something like a Thermalright HR-05 might give you a 15-20°C improvement, without any fan, far more with a slow spinning (1000rpm) 80mm one (a Sythe SFF80A, or a Nexus Real Silent Case Fan maybe undervolted). Even a top down cooler for the CPU may help to lower the temp of about 3-5°C.
Obviously you may go cheap and fix with a zip tie a fan on the original heatsink (I don't recommend to do so).

Your CPU temps would seem fine to me: to lower these temperature the only thing you may do is to swap the heatsink itself, as AC ones are quiet and cheap but also mediocre coolers. I'm afraid that no intake may help you enough.

Your GPU temp might benefit from an intake but I don't see any need to cool it down with just 75°C (how much is the load temp out of the box?). I would know how much headroom you may have: which are the temps under torture (I mean four instances of Prime95, FurMark or OCCT)? If it should be under 90°C, I think you should be fine even now.
Moreover, I don't see any straight path from any possible intake up to your GTX: so even using something like a Silverstone Air Penetrator or a medium speed Scythe Gentle Typhoon might turn out a bit uneffective.
A thing you may do for free is to rip off those two slot blanket you have near it.

About the PSU, it's just as horrible as its cousin Thermaltake ThoughPower QFan (or any CWT made PSU), but here we are. Obviously a modular one will help airflow but I think you wouldn't change it.

Eventually, the case. An Antec 1200 will help you, for sure (but your Chieftec should sound unberable), in particular for the intakes effectiveness (with reference to the Sileo). But I don't think it may give much more than an average of 5°C, in particular on the GPU (maybe as it should have the PSU under itself?). Any smaller one should work equally good (so, according to me, a 900 or a 600/300 should be preferred with your components).

At the end, I don't think you have to change the case: and if you rather to stick with this idea, I'm afraid there aren't better ones with PSU on top, and you should go for an Antec 900/600/300 (or the new Antec DF-10, if you may like it), aiming at something similar to the SPCR Mid-gaming PC.

Obviously. But not above specifications.


That's a higher clocked CPU and it is still running cooler than mine in that test. And the temperature difference (56°C vs my 63°C) may not seem too much, but I'm afraid my CPU is just crossing the upper temperature limit. The AMD Datasheets define the max temp as 62°C for the Tcase, which I guess is measured on the IHS. Obviously, I can ionly measure the core temperature, but that can't be that much different from the IHS temp, since it's a very small package. And I'm getting 63°C sometimes. (But the maximum temperature of the CPU is somwhat unclear to me. I don't know why AMD defined it like that. They could have said, XX°C core temp is the max or something...)
But since the CPU didn't yet shut down or throttle itself, I guess, I'm still okay with tihs temp.


Umm, did you look at the pictore?
I can't replace the heatsink on the MCP. Because it's under the sound card. It is very thin (only about 3-4mm thick) because of this, and it is already thinned out on half of its surface, to allow the card (the X-Fi here) to be installed. The other (upper) half of it is about 10mm thick, and because of this, the other PCI-e slot can't be used: if there was a card in it, it wouldn't fit because of the heatsink. For this reason, I can't put a fan on it either - it wouldn't fit under the sound card, or anywhere else.
Any other heatsink would be too tall ("thick") to fit on the MCP and under the sound card.


See above, with AMD specifications. the 63°C and above worries me a bit.
I'm thinking of replacing the heat sink with a Noctua NH-C12P or something, that blows air onto the board, cooling hte MCP too (maybe).


yeah, the GPU temp is not too bad (but still higher than what everyone seems to be getting (low 60's), and Nvidia clearly defines it's upper Temp limit as 104°C, so I'm not worried. Unfortunatelly, CPU and MCP temps aren't specified as clearly or I haven't got any exact information.


I was thinking more like buying the side panel of a CM Centurion 5 (RC-5xx) (like this) with air vents on it.
Will take those slot covers off (without high hopes though.)


yeah, it is really that bad. "Funnily", this is a recommended PSU around here, on local forums, in particular because of it's quietness. LOL!


Er, this is already a modular PSU, but I need to use all of it's cables due to many installed components, so effectively, I could just use a PSU with fixed cables too. (As I was doing so for long.)
I was thinking of buying a Seasonic X-650 or X-560, but since it's turning off its fan under my typical idle loads, it would only worsen the ventilation of the case.

Eventually, the case. An Antec 1200 will help you, for sure (but your Chieftec should sound unberable), in particular for the intakes effectiveness (with reference to the Sileo). But I don't think it may give much more than an average of 5°C, in particular on the GPU (maybe as it should have the PSU under itself?). Any smaller one should work equally good (so, according to me, a 900 or a 600/300 should be preferred with your components).[/quote]
I don't see why a 900 or an even smaller case should be preferred with my components? They wouldn't even have enough bays for what i've got! The 300 and the 600 has only 3 bays. Only the 900 would take my 4x 5.25" and 1x3,5", but if I'd put all of these there, plus the other hard disks internally, i'd have to remove one of the front intake fans and the bay with it, plus there wouldn't be any open surface to breathe on the front panel!
I really only found the 1200, that would take everything I've gotten.


I don't want to change the case, but I don't see any other option. That's why I came here to ask, if there's anything else. Maybe I'll try to find a side panel with air vents and see if it would help.
Looking at the dust traces, the case has very low negative pressure, it's sucking in air on all the tiny holes and gaps. This is why I thought, more intake would help fresh air to get in.

As for temps: I was just playing Crysis for an hour or less, and the Bios set teh temp warning sound off again, so I quickly opened Nvidia System Monitor, and it showed that the MCP temp was 79°C (was probably 80 when the alarm was set off), GPU was 75°C and the CPU was 60°C (but it cools down a few degrees quickly, so it probably cooled down a bit while I was quiting the game.)

Well, first of all forgive me to have been a bit shallow answering you.
I don't know if it's above specifications, either as I'm used to higher specs for my AMDs (70-72°C), and as I've not checked before answering the TCase max of your 940BE (just 62°C).
However, as you already know, it's not the absolute maximum but the maximum safe operating temp on the heatspreader: AMD isn't used to publish the equivalent of the TJ-max for Intel.

Currently I still think you aren't out of any specs: apart there are several reports from users on the web on temperature about 60°C for the 940BE, to be sure there are several test - IMO - to run: first of all I would use a trusted pair of apps, Prime95 and FurMark to fully stress the system. Then I would check the fans voltages, as on your first post it isn't clear how you run them (particularly the CPU and the front ones).

Personally I would run 4 Large FFT + FurMark two times, with and without the side panel on, while monitoring the temperatures: if the recorded CPU temperature won't differ so much (I guess at most around 3-5°C, as a first instance) then I would be tempted to think that the Sileo (its airflow) don't impact too much the CPU cooling.
The further test I would done, in order to determine if the Sileo is the culprit, is to run the same test without the case at all: if CPU temp should be substantially similar to the one recorded with the mobo inside the case, I would turn my attention to the interface between the heatspreader and the heatsink, and probably I would reinstall this latter giving a fresh application of a good TIM.
If even after reinstalling the heatsink the CPU temp hadn't to lower enough, I would think that the AC isn't enough to cool down the 940BE.

Without recording a consistent data set about temperatures (and fan speeds) I think it's not possible to properly troubleshoot this situation, nor I see a quicker way to do so.


It's a more efficient and far more powerful heatsink, CPU temp have to be lower.


Yes, from the photo the standard heatsink didn't seem cutted down, but however I've looked at it: as said, I think an HR-05 might be mounted, maybe the SLI one.
I have larger Scythe Kabuto on a mATX board with that infamous Nvidia chip, and I was able to swap the heatsink (but the mounting seem different from your pushpins), placing the TR slightly slanted (with reference to the motherboard's sides).
Probably using the SLI one you should move the RAM in the currently empty slots: however, the copper base of the HR05 is usually smaller as footprint than regular mobo's heatsink, about 3x3cm with an about 5mm height (thickness), so it might be possible to fix it under your card.

About the proposed fan, you have just to secure it in order to let it blow against the heatsink, it shouldn't be horizontally out of necessity.

However, there's no other way to substantially lower the MCP78 temp, even without the case it would run very hot.
Maybe your system is just too much crowded, and maybe you should rationalize it.


It may help, but IME usually not dramatically, probably a single digit improvement. The C14, if it fits, it's better suited at cooling the mobo.


I can tell you that above 86°C your MCP should reboot the system: so it did mine (this is why now that mine sports an HR-05 SLI).


Do the tests with and without the side panel, to best determine your choices.


Shit, this is one of the shallowness I wouldn't have made.


The PSU can't have an exhaust role: a fan have to spin to freshen the PSU, not the case.
An help may come from PSUs such the CoolerMaster Silent Pro with their flat cables, but the more I look your picture, the more I think you should rationalize it.


Because they aren't the hottest parts in town (despite of your high load temp on CPU) and in that way you might have spared some money.


AFAIK the 902, the 600 and the 300 have the about the same bays (9 to 10), even if I obviously forgot that you have those ugly swap caddy for two 3.5" disks.


You may give a look to the quick research made by MikeC for the SPCR mid-tower file server:

• AeroCool VX/VS-9 Pro
• Antec 1200
• Antec 902
• Chieftec Smart WH-01
• Coolermaster Centurion 590
• Lianli PC-P80
• Lianli PC-P50
• NZXT Evo
• Sharkoon Rebel 12 Eco / Value
• Silverstone KL01 & KL02
• Silverstone TJ-07
• Sunbeam AC9B-T Acrylic
• Thermaltake V5
• Xigmatek Utgard CPC-T90DB
• Zalman MS1000
• Zalman Z7 Plus

I think you need to do some tests as the ones I pointed you out above.
I'm also thinking that if the rear exhaust fan is costantly run at 1000rpm, it might just be turn out inadequate.


Usually a front intake add noticeably more noise but cannot lower so much the temps. Usually, not always.

But otherwise you should have the first four 5.25 slots still filled by the optical and swappable drives, so at first sight I don't think you could substantially lower those CPU and MCP temps with an intake fan under them. Maybe the GPU ones.

I strongly advise you to perform some various stress tests (with the side panel on, without it, on open-air), as you might find out that just convection won't help you enough.

Gonna run the tests tonight when I get home. (I already did earlier, btw.) What I noticed is, the tests didn't raise the MCP temp as much as a bit of gaming, but we'll see. I'll post the results later.


Yeah, it's not really good for these energy wasters, but many people use this cooler here, and I'm using it since years, and nearly 2 years on this CPU, it survived two hot summers already without shutting down, so it is on its limits but still copes.


Will do some torturing and record as much as possible. What app can do that? (logging and graphing everything?) I can only think of speedfan now, but it's graphing is a bit lame.


That is exactly my problem. But I'm using everything in it. they aren't there just because I didn't have an empty drawer to sink all those components. In fact, i'd go for a 4-disk raid0 array, if I'd have enough room. The one drive in the upper rack would have to be built in properly too (but the rack would stay there if ever a "guest" hard drive have to be installed quickly.) I'd also use that free pci-e slot too, if the heatsink on the MCP would allow, etc. (The other empty pci slot is only empty, because before the GTX i had a 9800GT with an AC Accelero on it that occupied that slot too, so I had to remove my other sound card from there and go USB, and I didn't yet put back that soundcard - probably won't even.)
Anyway, yes, I have a lot of stuff, and this is a problem. But can't live without them. The only way to rationalize is to get a bigger box for them, I think.
(I used to laugh at case reviews where they praise the nice roomy interior, when they only put in a single disk and a single DVD.)


will be done.

edit: found a pic of the MCP heatsink and the area around it. Not sure if that Thermalright thingie would fit, first, it looks more than 5mm thick (unless it's using 3mm heatpipes) plus the bridge on top of it, and that capacitor seems to be in the way too. Arrrggh, this mobo is a terrible design.

Take your time to do so. It's the most important part of your homework, I mean.


It seems quite unusual. You might also play Crysis in an open-air scenario to see if there's actually a heat build out (while it's inside the case).


What about OCCT? I think it can read data from SpeedFan (at least the temperature ones: it should also be able to read data from Everest, Motherboard Monitor and SysTools).


As you aren't using all the PCI slots, won't be more effective to change your motherboard with one with more PCI Express slots? Maybe something aimed at triple or dual SLI sets (even a second hand one)? I don't know about Vista/W7, but at least with WXP should be doable with relative ease to swap the mainboard without reinstalling all the software.
However, swapping the AC with something like a Mugen 2 HAVE to lower the CPU temp.

I can only say, that I'm getting the bios temperature warning beeps when playing Crysis or BFBC2, and not, when doing prime95 and furmark together. (At least, didn't beep on me yet. But will see tonight.)
I tried OCCT yesterday, and it displayed a few temps and fan speeds and made some very nice graphs indeed, but it didn't display the MCP temp, which is the most critical here.
The problem with the mobo change is: it took me some time for me to find this one, as there aren't many with nvidia chipsets anymore, there's not many to choose from. Others were either mATX, or had even worse layouts, or lacked IDE connectors etc. The other problem is, I have Win7 OEM registered to this board. If I'd change it, and if I cared about staying perfectly legal, I'd have to buy another copy of Win7 for the new board (or even get a boxed version of win7 for even more $$$), which is something I won't do. I rather let this cook itself.

Btw, I think I have one of these laying around in a drawer. I coud just lay this on top of the sound card (hopefully it's shielded), blowing on top of the MCP heatsink - would be probably the best (and loudest too), right? I could even screw it to the soundcard with a long screw.

IIRC if SpeedFan can see MCP temp, you should be able to tell OCCT to graph it in its Options tab, just reading from the SpeedFan logged sensor.


AFAIK you may just say that's a replacement (and sensu lato it would be quite so, as the older one won't work properly), and reactivate it for free.


Don't ask me for those things: if in case, use something like a Scythe SFF80B, if it does fit.

OK, so this is what I got:
OCCT couldn't find the MCP temp, and I didn't want to install any external app, so I used AIDA64 (successor of Everest), which not only identified everything correctly, but also has graphing, logging and stability testing (but only for the CPU). So I used this for the test.
First, after powering on, I checked my e-mails etc, warmed up my dinner and left the PC to warm up to it's idle temp, with the side panel on, also in its everyday state. Room temp was 23°C.
After that, I begin the stability test (1), which hammered the CPU. Let it run until it reached a stable temp. After this, I started Crysis (2) and played a bit (was fun with the CPU 100% loaded...). Soon, I heard the motherboard beeping (overheating warning) intermittently so I quit Crysis (3), but kept the CPU stress running til the GPU cooled a bit. Then I stopped the cpu stressing too (4), and waited until the temps went back to the roughly their original levels.
Following this, I removed the side panel (5), and repeated the above process. First I waited a bit to see what happens with the temps - nothing really interesting, they didn't change much. Then started CPU stressing (6) until stable CPU temps, then ran Crysis (7). This time there wasn't any motherboard warning beeps, so I just quit Crysis (8), but the CPU stressing was still on. Let it run a bit more, until the GPU cooled down. Then stopped stressing (9) again and waited until the temps normalize again.
Then I put back the side panel (10).
And here are the graphs: I put them into one image, aligned the markers to each other. Graphs are: Temps, Fan speeds, Voltages.
>>>CLICK<<< (Warning! Image is big, but only 90kB)
(Attached is a short summary of the above, with min, max and avg values.)

What I don't understand is: I have 2 overheating warnings set in the BIOS: 60°C for the CPU and 70°C for the "system". The CPU can be either the cores temp, or the other CPU temp (whatever it is). The "system" can be what's called "motherboard" on the graphs, or can be the MCP, dunno, really.
At point 3:
- the "Core" reaches 62°C, which is Tcase temp for this processor,
- the "CPU temp" (whatever it is) reaches 60°C, which is the overheating warning point set in the bios,
- the "MCP Temp" reaches 70°C, which is the overheating warning point set in the bios.
I'll have to find out which one sets the alarm beeps on.

What I observe:
- with the side panel removed, the CPU and core temps are 3-5°C lower, and they do not go above 60°C
- the GPU heats the system up
- the MCP heats up quickly (probably because the GPU heats the air up?) but cools down very slowly.
- the hard disk monitored, slowly warms up from 33°C to 37°C, interestingly even with the side panel removed, it gets warmer! but it starts to cool back with the side panel put back on. hmmm. There must be a wind channel or something.
- when the side panel is removed, the rear exhaust fan spins faster!? does it have to work against the underpressure when the case is closed?

Even if we haven't clearly identified the temperature sensors, I think we may do some guessworks.


As I previously told you, you have to say to OCCT in the options to collect data from SpeedFan (or some other program), pointing OCCT out which sensors it have to read: so if SpeedFan see the MCP, even OCCT will see it.


And it is what (IMO/IME) you should had to observe: if you haven't reached the system thermal limit, this difference have to be small, as there would not be an actual heat build up inside the case.
On a more conservative system than your one (E5200 mildly overclocked to 3,33GHz and a GeForce 9300 IGP but in a more cramped space, an Antec NSK3480), I've observed a limited 2-4°C improvement in similar conditions.


As expected: high performance GPUs are usually more decisive than modern CPUs heating up the whole system: if they haven't a blower/external exhaust they will heat it up even more.
I've not understood if the test has been conducted with the slots blanket or not.
But with reference to the hot MCP (and the so-called motherboard sensor) however it seems to me that the CPU stressing is more related to its heat up than playing Crysis, so maybe a top down CPU cooler might actually improve your situations.


As expected: if the recovery is slow it's usually a clear index of an inadequate heatsink. On my MCP78 based system, with a fanless HR-05 SLI on chipset, this recovery is indeed quick, much faster than your one (but to be fair, even without it the operating temp went down quicker than your).
On another - cooler - system, an Athlon II 605e mildly overclocked to 3GHz and an AMD 785 IGP / SB 750 with the stock heatsinks, I have observed a similar slow recovery from heating up, even if not worrisome at all, due to its far higher efficiency with reference to your nForce 720D (so the 785G/750 run much cooler than your one under every conditions).

However, that MCP would seem to run a bit too much hot even without heating it up, IMO (you have just an about 10°C rise, haven't you?): so it should be another point that its cooling is barely enough, or almost inadequate.


I don't understand very well: is that case fan temperature controlled? However, it seems to me it spins faster by a very small amount, so at first sight I wouldn't be worried about.

Now you should have to do the same tests with your system outside the case and repeat the same with a fresh TIM on CPU (I know it's pity), in order to see if CPU temp may improve furtherly: up to now I'm quite convinced that on the contrary the MCP temp won't be substantially lower than those ones, while for the GPU your mileage may vary. If its temp should improve noticeably, than it's a clear indication it may benefit from a (more) focused intake fan (or from removing/moving the bottom cage) and that currently the one you have is blocked and therefore uneffective (but not for the lower HDD). With reference to the GPU, a more spacious case, without a cage in front of the GPU itself, might actually work (but I'm not fond about it).

Another test you may do for free is to apply a fresh rice grain of Arctic Silver Ceramique (or other any not conductive high performance TIM) on the MCP, in place of the current thermal interface (pad or paste): if it should be very effective, IMO it's the third clear indication you need more some better heatsinks than a new case.
On the contrary, you may try to swap the current upper drives cage with the lower intake fan (I'm referring to the photo you posted) to see if actually an intake fan may improve the MCP temp (and so if a new higher case along with a different placement for your drives may help you enough).

Summarizing: by my own interpretation of your preliminary tests, currently I still think you don't need a more spacious enclosure, but if in case two better heatsinks, one for the CPU (even if it would seem to play safe despite its TCase Max, IMVHO) and above all for the MCP (but just if you want to have it run cooler, 70°C isn't something to be definitely worried about, again IMVHO). I also think that a new case would be somewhat effective, in particular with reference to the GPU, but probably not definitely.

thank you!
- the two case fans are connected to the motherboard, but in the BIOS I turned off their speed ajustment, so they should spin on max (800RPM for the intake and 1200RPM for the exhaust - but this is also seen on the graphs/table.) and yes, the exhaust spins only a little bit faster with the case open, but it is clearly visible on the graphs, and it surprised me - I guess it has a very very weak motor. the intake fan isn't affected by the side panel.
- the CPU fan speed is controlled by the BIOS (also seen on the graphs)
- the rear PCI slot covers are removed but they didn't improve anything (as I expected)
- thought of a bigger case not only to (maybe) improve cooling, but also to have more room to install drives and tidy the cables etc. improving CPU cooling with another cooler is another thing I'm thinking of. I'm also planning to replace that noisy PSU. I was just wondering, if by replacing hte case with say, an Antec 1200 (which i'd do anyway sooner or later), would perhaps improve internal temps that much, that I could get away without a new costy CPU heatsink.
I don't really want to remove the motherboard from the case to test it open air. Only time i'd do so if I'd be replacing the case or mounting a CPU cooler that needs a backplate (yuck!). Am I too lazy?
Btw, after these tests, I played a bit more Crysis, and the warning beep was on again. I quit quickly and checked - the MCP was at 79°C. I think because of the CPU stressing, the GPU could not run at 100%, and it wasn't heating up much, so the case temps was a bit cooler. But when just playing Crysis w/o stressing the CPU, the GPU ran at full speed (and the CPU was loaded quite a bit too), so case temps went higher and the MCP got hotter too. I'll do another graph with this today.

I think you can't do any reasonable guesswork without some open air tests.


Yes you are.
You want to know but you don't want to learn: unreasonable and unfeasible.


I can only advise you to do yourself a favour: to put the mobo outside the case, and apply a fresh TIM on the MCP.
Even if I know that troubleshooting, as the moon, is an harsh master.

Back again.
I did some further testing.
This is what I get with just playing Crysis (without stress testing in the background): (Warning! Big images!)
temperatures graph (fan speeds graph)

Then I did this: (hehe, sorry... )
pic 1, pic 2, pic 3

Which resulted in this: temperatures graph, and I felt like this.
I must somehow apply a 40mm fan on that °&@#!! heatsink.

If you replace the CPU cooler, get a top-down style cooler, it will help your temps.
If it were me, I would put something like an HR-05 IFX on that chip and forget about it. This crappy aluminum one is not worth the trouble IMO...

Update:
ordered the Noctua NH-C14. Measured the available space, and I see it's gonna be a tight fit, if at all.
I'm thinking of an installing orientation, where the heatpipe bends are on the right, above the RAMs, and with a top mounted fan.
Also ordering the Seasonic X660. Edit: it's arrived

Update:
I've got the cooler and the new PSU and put it together. Wasn't that easy, the cooler is a monster! The PSU's cables are very rigid, hard to work with. Also, the all-modular design is... ummm... The motherboard connectors were especially hard to put on the PSU. The pci-e power connector just came loose and felt down... So it wasn't simple.
Here's what it looks like: (Warning! Huge images!)
picture 1, picture 2

And this is a quick test with Crysys:
temperatures, fan speeds, statistics.

I think it'll be fine now! Idle noise didn't really get lower, but the PSU doesn't become a jet engine when under load either. I like this very much. Hopefully it'll last for long!

What do you think?

Looking good !
A couple of things I would try :
- close the top PCI slot (the air gets sucked in by the exhaust fan without cooling anything)
- remove the bottom PCI slot cover (more cool air)

I just noticed that in your first post you mentionned temps around 70°C for your GPU. My GTX460 overclocked runs at 70°C at full speed 24/7 with the fan at 34%. If I leave the card control it's fan automatically, temps will stay around 60°C but noise will be way over my noise tolerance ! 70°C for a GPU is fine at full load, and it's great when it's done quietly.
The 60°C temps that people report are for cards that control their own fan speeds automatically to stay cool, not to stay quiet

One more thing, that bothers me:
now with the Seasonic on the top, running fanless most of the time, and because of the slightly "negative" pressure inside the case, I think, the air is going through the PSU the opposite way. I.e. it's entering the case on back of the PSU and then into the inside of the case, then the exhaust fan blows it out.
And when the fan starts to spin slowly, there must be a point where the PSU fan counteracts this inward airflow and the air stops inside the PSU. Am I wrong? Should I have to worry?
(I hope the fan in the Seasonic is Temperature controlled, so if the air stops inside and things get hot, it ramps up the RPM to blow out the hot air. I'll get a bottom mount case I guess...)

Probably.


Definitely no, IMO.


Usually a bottom mounted PSU enclosure has a somewhat superior cooling power: but usually such style of enclosure is also less quiet than something similar to a Sileo, or to a Solo.

As more probably that not your Sileo has not reached its thermal limits, forget those thoughts and live long & prosper.

I've mounted my Seasonic-X both ways in my test bench case. (No air pressure differences mind you) When mounted upside down, it seems the heat rises out on its own and the fan isn't needed much.

The fan spins up more often when the grill faces downward like yours is in the picture. I assume that means heat builds up in it and a quick (audible) spin flushes it out. I'm planning to mount a fan on the back of the unit to see if it can create enough airflow to combat the spin-ups during games, I'll let you know how it works out if you're interested.

Sure, I'm interested!

Btw, I just checked my PSU and noticed the following: the PSU fan keeps spinning even in idle. Powered down, the fan spun down slowly (surprise! - oh, my chieftec kept the fan running for a while). Then powered on again, the fan spun up quickly, then stopped quickly (does it use reverse polarity for braking??? ), then after the POST screen it started spinning as before the shutdown. So it does spin in idle, when my PC takes like 100W from the wall (that's 88-90W from the PSU).
That paper stripe around the PSU stated that the fan does not spin under 20% (+/-5%) load. I am clearly around/under 20% in idle.

Adding an exhaust fan DOES help. I played a few games today and the PSU didn't try to evacuate the accumulated heat.

The tricky part was suspending a 92mm fan back there. It will probably perform even better if I get a silicone vibration dampener and secure it against the edge of the case.

Just a little update:
I am currently playing with undervolting, and discovered strange things... I guess this is Gigabyte's stupid BIOS maybe...
So, I always had the "global" system voltage setting set to "auto" in the BIOS. With that, the CPU got 1.425V under load (which is I think some 0.05V more than the specified), and 1.025V when idling, which is again about 0.05V more than the normal. The DDR2 was 2.048V too, very high.
Now, I disabled this "global" voltage setting and set it to "manual", so now I can change the different voltages one by one, but these setting also have an "auto" option.
I left everything on auto, except for the CPU voltag, which I set to -0.05V. Under this setting, there's a grayed out informational text, saying that the normal cpu voltage is 1.350V, wehich I think is not correct, but anyway: if the bios thinks that's the normal, why does it set it to 1.425 in "auto" mode?
So, now, with these settings, I see the following: CPU voltage is 0.944V in idle and 1.396V under load. DDR2 voltage is now 1.840V, wihch is strange, as I didn't change it at all...
With this, the CPU temps dropped a good 6-8 degrees!

It seems to me as if the BIOS were stupid, setting stupidly high voltages in auto mode... and maybe that's why I was getting those unusually high temperatures and power consumption...? Hmmm...