Fanles VIA C3 Nehemiah 1200Mhz machine done!

[trodas]

Hello, dudes.
This is reaction on the orginal tread:
http://forums.silentpcreview.com/viewtopic.php?t=8851
...about how to get VIA C3 CPU

With the big help of TheEagleCD (i realy appreciate your help, dear friend - YOU make it possible!), who send it me to Czech Republic, i was able to finish my dream and make the server machine noiseless. There are lots of pictures and hopefully funny and enough descriptive comments bellow them, so feel free to take a look what the hell i did:
http://ax2.old-cans.com/galerie.php?p=s ... 8&d=1&v=v2

The machine have got some pretty interesting behaviours, witch i will post there later. At this moment, at the machine running Memtest-86 v3 ( http://www.memtest.org/ ), witch i find more overheating the CPU/machine that just a CPU burn (P6 or MMX tests at high priority still not match this one), and it with slight pause running from yesterday. From todays noon it running non-stop and as i stated into the orginal tread, i don't stop torturing the machine for exactly 3 days

If she can manage it, it's entierly stable. Up to now not a single memory error or a glitch happen, however the machine is overheating pretty hard

When i first yesterday put it all togeter, i hit 75 degrees with the CPU burn MMX/hi priority test (!) and 77 degrees with the Memtest (i run both for 30 min, Memtest later, and with CPU burn i use MBM5 (SpeedFan not working yet, waiting for beta to test) for just-in-time temp checking and the Memtest values are from bios-reboot

Well, 77 degrees is too hot even for me, a CPU-torturing geek, so i removed all the bay casings on the case to open its back. That cooled machine i leaved for night running and at the morning (about 9:30) are there 70 degrees.
Better, eh
But still not good.

So, i made a hole to the case, to allow the cold air to get in trough bays and get out just there - see pictures. Pre-tests showed up to 68 degrees temperatures, so we see, what temp are there after 3 days of burning

Now some more-interesting values. When i have there Tulation 1300 stock cooler, slowed to 1670RPM by Zalman FanMate, i got on CeleronII CPU (900Mhz, Vcore 1.75) 41 degrees in iddle and 54 on load. Average, for the slow RPM, suxxking AL cooler and small case with CPU just bellow the PSU, what are on side into the case (see pics)...

With VIA C3 (1200Mhz, Vcore 1.4) i got starting temperatures as low, as 24 degrees on CPU (bios) and 32 on chipset. When Win2k server booted, these are 32 degrees for CPU and 29 for chipset (chipset get lazy, seems )
But the most weird thing is, how the CPU reacting to the CPU burn. All system, including friend and my watercooling machines, inlcuding my past machines, including brother AthlonXP aircooler and little sis Duron aircooled - every these machines react very quickly on the CPU burn. All raise their temps and in few minutes reaching the almost top temperature.

Not VIA. Either it's thanks to the ultra-low 18W of heat (don't P4 got almost 5 x more? http://www.tomshardware.com/cpu/20031230/index.html - 84W for the 3.2Ghz version for those, who don't want to be bothered to read up to page 3 ) or it's the huge passive cooling?
But i used exactly same cooler on the CeleronII once and on my Tualatin 1300 as well, and everytime it reacted the same way... (well, now W/O fan...)
Interesting.

And this bring me to to last thing - the cooler. For VIA passive cooling, Zalman made its CNPS6100Cu-Al low-cost version of CNPS6000.
However because even the CNPS6000 it's still not delivered yet, not to mention that the CNPS6100 is "yet to come", i used my older CNPS3100-Plus one.
As you can compare:
CNPS3100-Plus
http://www.zalman.co.kr/english/product/cnps3100P.htm

CNPS6100Cu-Al
http://www.zalman.co.kr/english/product ... 00alcu.htm

CNPS6000
http://www.zalman.co.kr/english/product/cnps6000Cu.htm

...the 3100-Plus and the CNPS6100 is not such much diferent. Two facts. Heat dissipation area (the only one important thing in passive cooling) of 3100 is rated as 2000 square cm, where 6100 is rated as 1800-2200 square cm. Interesingly - 6000 and 6100 looking same for me - but the CNPS6000 is rated as 2600-2900 square cm
Anyway, my CNPS3100-Plus is all-cooper, onlike the CNPS6100Cu-Al, so i dediced to not waiting anymore for CNPS6100 and giving CNPS6000 it a try.
The relatively high temperatures are a bit weird for me. Either i have such badly overheating case (check the case on pictures), or i have no clue, how the cooler can passively cool the CPU into more extreme cases, that's the mine...

Also I'm up to discussion, whatever the "BeatTheHeat.wmv" animation showing that P3 crash 5 sec after removing heatsink (no dubt about it ) is real or is not it. Because given the temperatures with my Zalman into now relatively well-ventilated case ( rate my case mod, please ) and the fact this is the new core, witch should generating less heat (perhaps not, its 1.4V and the previous Ezra-T cores are 1.35, IIRC) I'm slightly start dubting about it.
Don't get me wrong, i still LOVE my VIA cpu, trully and deeply. I just don't think that it can make it W/O heatsing at all and the words from the anim "...your is cold as ice..." facing the 77 degrees (thats record, i never measured such high temperature on anything before) sound me now a bit weird.

So, either i did a pisspoor job, or have overheated room (about 17-18 celsius at nights and 20-22 on days - that's at least what i think it is ) or the anim claiming 24h+ running VIA C3 800Mhz Enzo-T core under full load (Q3 demo don't allow the CPU to breath, I'm sure about it) is fake.

Let me know, what do you think

Galery link in case anyone missed it
http://ax2.old-cans.com/galerie.php?p=s ... 8&d=1&v=v2

[silvervarg]

Great to hear about the success!
I don't think the temps sound too bad. The VIA C3 is know to take very high temps without any problems, like 95C or so.
Do you intend to run 100% load all the time?
Have you done any measuring on temps at normal load for that server?

[trodas]

No "sucess" yet It made (checking if the Memtester did not hang... it still working ) barely 12 hours from planed 72 hours burn-in benchmark
And ofcourse nope, it will run most of the time iddle, about 1-5% load or so... downlading, marcos running and so on, so i expect real temps will be totaly different, like aroun 40-50 degrees....

Glad to hear that the VIA C3 tan take that temps! Do you know it for sure? I already mailed VIA about what temps one can expect as safe... The 95 is valid for what core? Ezra-T or Nehemiah?
I would like to see more detailed specs on the VIA page, eh...

And no, as i say, i just completed it and under normal load are the temps around 32 degrees after windows boot and few programs launched... I supposed that the temps will go to about 40 or so in normal usage, but we have first wait - i said 3 days and i mean it. It just passed first 12 hours. 60 hours left

[pangit]

Great project, well done!

I think you should open up the case a bit more near the bottom, to allow more cool air in. For instance near the HDD or at the back (perhaps just remove some of the empty PCI slot covers).

The hole in the top is great for letting the heat out, but unless there is somewhere for cool air to come in, it will not be used to maximum effect.

Silvervarg is right about the temps though, VIA chips are renowned for being able to run at much higher temps than normal.

[trodas]

pangit - thanks for the kind words With some help from friends - everything is possible
And yes, i already removed all the PCI slot covers (i tought i write about removing the bay casings into the galery, however "bay casing" is perhaps not right name for them?) so there the cold air come in, and exhust is right on top, over the CPU
W/O dubt it works - when i place hand there, i feel the hoit air flowing out of the case! Hoooray

About the temperatures, i just checked the VIA specs PDF and there is mentioned 0-85 degrees. I hit 77 at most, 68 at most now - so i think its okay

However, i come to realize that the "too high temps" are caused by also the different (passive) cooling behaviour. No no, i know that there is no forced airflow with fan, and therefore temps get higher. I was talking about wrong measuring of the temperatures with fans/watercooling
Let me explain - some boards measuring about 10-15 degrees higher temps that others. Its because they using the internal CPU termal diode, instead of the usual sensor bellow the CPU.
So now, let me explain, where the difference come. On hot CPU you put cold air or watercooled heatsink. So the temp you got is fake, because first - you measuring it bellow the CPU, where is lower, and second - the cold heatsink cooling the CPU down from the top side, creating big termal diference between cooled top (heat gets right away to the heatsink) and the measured bottom.
Both the fact lead to the known and proved about 10-15 degrees higher actuall CPU temperature that we proudly measuring, boosting how low temp we god with that or what cooling

And now, a PASSIVE cooling! How it works? It just heat the WHOLE CPU (top and bottom as well!) and heating it to the point, where the difference between heat dissipilation to enviroment air and the heat produced by the CPU are zero - or in balance.

Witch means, that with passive cooling, we get even on the bottom of the CPU the actuall CPU temperature, witch is higher by about 10 to 15 degrees, that the temps we are used to them!
(thats because there is no forced airflow or waterflow, taking the heat right from its source (the top core of the CPU) and therefore producing the temperature diference between top and bottom)

Yes, there is the heatsing on the top, but it with proper ArcticSilver3 grease get almost a part of the CPU, so everything is same hot and thereore the temps tend to look high, however they are pretty low. When you substitue 10 - 15 degrees from 68 degrees on full load, you get pretty low temperature for passive coolin, yea?

So, thats what recently come into my mind, hope i made it clear to everyone understand the temperature catch and difference between passive and active cooling

And as for info, my silent machine just pass first 24h W/O a single error or glitch of its planed 72 hour burn-in

[TheEagleCD]

Thanks for the kind words Trodas! I'm as glad as you are that things worked out fine in the end! Even though I'm currently thinking about writing a story called "VIA's Odysee from Germany to the Czech Republic" - it could be a real hit!

Seriously though, your results look very promising at the moment. I'd like to know more about the PSU cooling! A couple of shots from the inside (if possible) would be great. And some temperature readings from inside the PSU would be nice too.
And I agree with the others, you need to get more fresh air into the system. Using a duct could prove very useful in your case (pun intended).

Well, gotta go...

Good luck with the remaining 48 hours of burn-in!

Regards,
TheEagleCD

[silvervarg]

...the 3100-Plus and the CNPS6100 is not such diferent. Two facts. Heat dissipation area (the only one important thing in passive cooling) of 3100 is rated as 2000 square cm, where 6100 is rated as 1800-2200 square cm. Interesingly - 6000 and 6100 looking same for me - the CNPS6000 is rated as 2600-2900 square cm
Anyway, my CNPS3100-Plus is all-cooper, onlike the CNPS6100Cu-Al, so i dediced to not waiting anymore and giving it a try.

Carefully checking the Zalman site and reading the numbers I noticed that the 3100-Plus (all copper) and the 6000 (all copper) differ a lot in weight.
3100-plus: 296g
6000: 462g

That extra weight probably comes from a different design of the finns. So each fin contains a larger area. Perhaps the thickness of the finns could differ some too.
The 6100 (AlCu) should presumably be the same as the 6000 except that aluminium is used instead of some of the copper, giving the weight of 296g. This will also make it pass AMD max 300g recommendation. I guess this in not a coincidence.

If the 6100 had the same shape of fins etc as the 3100 then the 3100 should be a lot heavier than the 6100 (AlCu) version.

So at least one value must be wrong on Zalmans site for all this to make sense. My guess is that the area of the 6100 is wrong and should be the same as the 6000.

I am considering to get one of these heatsinks to put on my northbridge. The question is if I can make it fit. Zalmans measurements is not really good enough here since I don't know exactly what they have measured.
So could someone with one of these flower heatsinks help me out and take some measurements?
What I would like to know:
Measurements of the base: (WxL).
Preferably length with and without bolts.
What is the actual width of the heatsink on top? (Zalman 6000 spec: 95-110mm)
At what height does it pass 13mm height (standard cap on motherboards)?

[trodas]

TheEagleCD - you are wellcome and as i said - you help it my dream to become reality Thanks!
However you are a bit exaggerating with the story After all, the post did wonderfull job. The worst and only one bad part of the problem are the german "online" shops what don't supporting credit cards payments, and even don't reacting on CANCEL email... Not to mention i never pass the checkout - i just stoped when there is only one option of payment "Nachname" - but yet they are so "kind" to ship the product w/o completing the order and even with my cancelation...
THIS could be a pretty interesting story how shops can work!

Thx for the comment about the result - im pretty sure it make W/O any problem the rest to 72 hours burn test Remember, its noon 20. 1. 04 when i started the test, so right now its 2 days and 3 hours Only 21 hour of torture for the CPU left!

About the PSU - well, i should take some shots on opened one... However i will need help for the temperature measuring. I must admint i wanted for long time to have something for measuring temperatures (maybe i was prepared to spend money even on non-contact laser temperature measuring device), but up to now, i don't have anything.
So, anyone, let's suggest what i should get and WHERE and WHAT - i mean online shop only, and preferable workign and non-germany one... In case its shop that don't send to Czech, then i will need someone to re-send the package to me... Any tips and volunteers for this project?

The fresh air idea helping well, i bet the temperature will hold now bellow 70 degrees on full load, wich i would like call victory
The hole on top and opened all back PCI slots covers are definitively great way

Let's get pass over the rest 21 hours

silvervarg - nice digging on the Zalmans. Frankly spoken, you missed the difference into measurments between CNPS6000 and CNPS6100. It really differ. So, the coolers aren't be same. I would like to buy the CNPS6000... But i don't have where...
The area must be certaingly wrong. Because the measurments of the CNPS6100 are BIGGER that the of CNPS6000, where the area is SMALLER That did not make much sense...
About the measuring - well, when i get my hands on the CNPS6000, i can easily measure it as well, as i can measure the CNPS3100, when i perhaps (never touch things that actually WORKING well!) open my server for exchange
But it curently seems that there will be a problem to get this one for me... Hell, i wish i live into different country

Anyway, i did understand your purpose of northbridge cooling, but not get your last clause. What 13mm height?!?! And what standard cap on mobos?!?!

[silvervarg]

Trodas:
Anyway, i did understand your purpose of northbridge cooling, but not get your last clause. What 13mm height?!?! And what standard cap on mobos?!?!

This is what my motherboard looks like:
http://www.abit-usa.com/products/mb/pro ... =1&model=7

I need the NB cooler to clear the cap that you see between the NB and the fan header for the NB. The cap i 13mm high. There is also a few caps near the AGP port that it has to clear, but I think they are a smaller problem since they are further away.

I could pick up a CNPS6000 Copper cooler when I am in the capital in a few weeks time. So if noone else has helped you out by then just drop me a mail. Shipping from Sweden to Czech should not be too expensive.

[trodas]

First at all, todays at noon i stoped the memtesting. It went W/O single error or glitch. Temperature right after 3 days of maximal CPU/chipset/memory torture was 65 degrees, what is IMHO pretty good result for passive-only cooling

So, i worked a little on the MBM5 and SpeedFan and i make both running, tought SpeedFan reporting wrong -12V Also i added some screenshots, witch show the VIA C3 into pretty bad shape, since the system feels pretty snappy and fast, notably faster that with the CeleronII at 900Mhz More intereting will probably be the screenshot showing typical working temperature 51 degrees
Later it drop bellow to 46, 34 and 42 degrees for CPU, chipset and HDD respectively Its because the machine was overheated from the 3 days, more precisely 4, because i run it one half-day (night) before the final 3 days stability test W/O the hole on top
Updates are of course there:
http://ax2.old-cans.com/galerie.php?p=s ... 8&d=1&v=v2

Anyway, i cannot measure the temperature of the PSU, but when i touch the heatpipes, there are warm, about 40-46 degrees? Nothing hot, just warm. When someone come with any affordable measuring tool, i probably consider it Measuring temperatures "by hand" tend to be VERY HIGHLY imprecise
Anyone, any volunteer?

Any more questions about this project?

Author Message

silvervarg - ah, that cap! I would desolder it from the mainboard Anyway, even close is capacitor, what high it is? And how do you planing to hold the heatsing there? I see there only two holes, so perhaps driling a holes and craving a screw-thread into the heatsing will be in order? But i definitively see why you need precise measuring. Whats the center-to-center distance of the two holes anyway?
About the CNPS6000 - well, thanks for the offer, dude. I would rather like to buy it myself in a online shop and then send it to you, perhaps with something (CDs? DVDs? to compensate the post price) to avoid transfering money trough post (wire transfer get more expensive that the item itself )
Any good sweden shop, in witch i could make this happen?

[silvervarg]

I would desolder it from the mainboard Anyway, even close is capacitor, what high it is? And how do you planing to hold the heatsing there? I see there only two holes, so perhaps driling a holes and craving a screw-thread into the heatsing will be in order? But i definitively see why you need precise measuring. Whats the center-to-center distance of the two holes anyway?

Nah, I don't really feel like desoldering components on my brand new motherboard.
I measure the capacitor, and it is the awfully tall kind. 24mm height.
The center of the holes around the NB is ~57mm apart.
Assuming I could center the heatsink on the NB chip I just have 30mm sideways from the center of the NB to the capacitor.
I took some measuremeants from the photos you put up of the CNPS3100. It seems that I will be able to fit the base in on the NB, but I will have 3 problems:
1. I think the base will be very close to the measurement from the NB hole-to-hole. So it will not be easy to attach using the holes. I will probably need to use glue only.
2. There is no way I will clear the 24mm cap. My measurements tell me I would probably clear ~15mm high, so I lack 9mm. Defenatly a job for the Dremmel
3. The full width (110mm) will give me other problems. I would rather have it a lot tighter. I expect to get lots of problems with the main power cable to the motherboard and I also would like it narrower, so I might need to hack it down to 75mm wide.

Despite these problems it really looks doable, so I plan to go ahead and get that cooler.

I'll give you a private post about the CNPS6000 deliviry to you.

[Trip]

Kostik ran the C3 pretty hard in this review

[silvervarg]

I remember that review from Kostik, and it is a really good review.
The sad part is that he did not have time to do all the tests he wanted.
Here is very brief from that review:
"The Zalman CNPS3100-Plus can fit this motherboard, and will probably provide better cooling performance, but we couldn't test it."

Instead Kostik had to use a small low end cooler with a 60mm fan.
Still he managed to run folding @ home fanless, but he could not run CPU-burn fanless for a long time without getting too high temps.
With a poor cooler and the 60mm fan @5V he could run CPUburn without any problems.

But Kostik did cheat a bit since he still used a fan in the PSU, and that is located near the PSU in his setup. With a slight duct to the PSU he could easilly have removed the CPU fan.

[trodas]

silvervarg - well, i like soldering I will desolder the capacitor and extend its wires to make it lie flat on the mainboard
But thats just me
I think you can diassemble the heatsing and add other "ends" what holding the cooper plates rather, that use glue.
Remember, the heatsink is, uhm, heawy
When you screw it to mainboard, it will hold - breaking mainboard is hard to do However if you glue it using ArcticSilver on the poor and innocent chip, you can get your NB damaged as the heatsink is HEAWY

The size are - well, adjustable... Its just matter it LITTLE press and you can adjust it to wheather you need. Believe me. The cable you can put even between the plates


Trip - nice link, thanks However the maker suxx. His idea that the CPU cannot be cooled pasively is simply wrong and i proved him wrong already There is a HUGE diference betwen Al and Cu heatsinks and he also - as silvervarg mention - cheated by the PSU fan bellow, pretty close to the CPU to make enought flow for cooling
Othervise neat review. The maximum temperature are amazing. I wish i have something to measure the temps...
And there is one more thing - with passive cooling it VERY MUCH matter on HOW LONG you run the application. Thats why i run my machine at MAXIMUM load for 3 days W/O single break There is no single word about how long he run the CPU Burn, for example.
From my experience at least 40 minutes are necessary to get close to the max temperature. Believe me or not, as i say - the temperatures rising VERY VERY SLOWLY. I was shocked by this fact as well!

About the undervolted AMD - the speed comparsions between this and VIA could be interesting. My friend trying to do the same, with 1.4G Duron downclocked to 1.2Ghz and with Vcore 1.2V only, and he nearly managed it working, only during decoding it getting over 85 degrees, so he must put here a fan...
And he using same heatsink as im - CNPS3100-Plus.

Therefore the results are a bit flawed - you cannot compare the VIA with crappy s*itty Al fan and the AMD with something probably much better Thats simply not fair.

Same as the 3DMark test. Hell, what a crap! Ofcourse the integrated graphics will perform shitty. When there is a AGP and Radeon9700, things migth have looking pretty differently, so my point is, that my machine now feels and works notably faster that when i have Celeron II / 900Mhz installed - even VIA C3 1200Mhz into test matching about 700 - 800Mhz Celeron II speed (Celeron 2 is the same as Pentium 3 - only half L2 cache)
And trust me, i know this well, because i working with my VIA C3 everyday. The CPU is simply simple - it handle simple tasks very fast. Never expect that it have big raw cpu power. It don't. But most the windows thing is nowhere near hungry for raw cpu power and calculations - just the simple stuff - mode, add, copy...
...and this, believe or not, handle VIA very very well and SURPRISINGLY quickly

Hell, even my network performace if faster now, so what?

[Kostik]

Trodas, thanks for a very interesting report.

You mention that the temperature you read rises very slowly. The motherboard I was using in my review of the C3 Nehemiah was reading the temperature directly from the thermal diode in the processor, and the temperature would rise *very* quickly. That is 10°c in a second when launching CPUBurn, and *then* the temperature would rise very slowly. That's typical of readings from the internal thermal diode, and clearly shows that the readings I got cannot be compared to the ones you got.

I assumed that running this CPU fanless wasn't safe because I got very high temp readings, but the computer was perfectly stable. So maybe your CPU actually runs at the same temps as mine, it's hard to say if we don't use the same mobo. I think that if your mobo was reading 176°c as a CPU temp under CPUBurn, you would certainly not feel very comfortable. And I don't think that I "suck" because I didn't feel comfortable.

As for the VIA/AMD comparison, if you had read the review, you would have noticed that I used the same heatsink. So accusing me of comparing CPUs with different heatsinks when you didn't even read the review correctly, well, that sucks.

And I didn't "cheat" with a PSU fan. I ran the C3 in free air with no airflow, noticed that the temperature was too high, and then decided to try it in a low airflow case. I would have run it passively if the temps had been lower. I don't think that there's anything missleading in my review, and I didn't write that I had been able to run the C3 passively.

"It's the fate of heros to be persecuted" - Voltaire.

[trodas]

Kostik - first at all im sorry for some negative and even derogative comments and remarks - and especially sorry for not reading about the heatsinks/fans properly My bad.

Your review is pretty interesting too Even i did not agree with the results so much

However, let's skip insults and go directly to the core of the problem and to some constructive debate, shall we?
Yes, temperature for me rising VERY VERY SLOWLY. Even with CPU burn high priority and MMX test. I have only one answer for this - passive cooling by huge heatsink simply works differently that all other cooling types - it not just get fast to the temperature we know, it go slowly, because the balance (between how much heat CPU produce now and what heatsink/fan can cool it down - take it from the CPU) of temperature is not provided by fan, but by the whole heatsink temperature.
I tried evaluate this problem into my posts, however i might write it bad or just hard to understand? I don't know, but give this try, please.

One remark - on the very same mainboard, using either P3 600 or 700Mhz? and CeleronII 900Mhz the temperature rising very fast with CPU burn. So i don't think that diference in measuring is the reason, why the temp go so slowly up...
BTW, i just remember we once tried burn the CeleronII 900Mhz crap alive with my Abit ST6R mainboard and the behaviour is roughly the same, except the temperature rising constatntly and notably faster - however not very fast...!
So the reason is simple - the heatsink simply getting warmer and warmer very slowly, because it's big and huge and the CPU producing only 18W of heat, or so...

About temperatures - think about it. There is no forced airflow. So in this case, there NEVER be a big temperature diference between the internal and the external temperature readings. Why? Simply because the whole neatsingk AND cpu slowly got to some temperature, on witch it disipiate into air as much heat, as the CPU produce. When you use fan, you cooling the heatsink much much much more that passively and on termal camera you see, how much the temperature on the whole heatsink differ. It will be more that 20 degrees (between the part sticking to the CPU and between the part closest to fan), while on passive cooled heatsink it will be lover that say 5 degrees for sure. (maybe less, cooper rules)
Therefore i strongly believe that my temperature measurings are precise - or at least adequately precise and you simply CAN't convince me, that in fact, the CPU temp is about 176 degrees or what the hell - no way i say

Next to die - about heatsinks, well, i either read it wrong, or misinterpreted it. However the truth is, that you should use big cooper heatsink, like Zalman CNPS6000 im about to buy
And ofcourse AS3 under it, etc. Then your temperatures will be acceptable.

As for end, my machine is now running at 39-40 degrees iddle (just w2k server and few small programs...) and at 69 degrees (improvment ) on full load after 6 days of nonstop Prime95
With CNPS6000 i expecting even better temperatures
...not to mention im thinking about laping the top of the CPU, because i feel the painted logo should making my temps a bit higher, that they can be (i don't expecting to shave more that 2 or 3 degrees, but even this is worth it )


And now the gauntlet for you
If you don't trust my readings, what about send my the mobo, what using the internal sensor of the VIA C3 CPU and i try it then with my fanless case and my Zalman coolers? (i should finally got the CNPS6000 hopefully soon ...so i can test with CNPS3100-Plus and CNPS6000) I try it, measure (and ofcourse send them back) and we will know, who done it right and how VIA C3 can hapilly run fanlessly
Because i would bet that you can use VIA C3 cpu under maximal load and still passively cooled and stay with the 85 degrees limit VIA stated.
What do you think, worth the try, or you are just want to keep saying that im wrong and my temps are way of what you got?


And yes, im aware about that internal CPU sensor can measure up to 20 degrees higher temperature (and the emperature are simply more right, it's just that peoples don't like the idea how hot theri CPU's running and prefer the lower values ) that external one under the CPU, however the difference will NOT be same on passive cooling machines - i believe that it stay around 5 degrees

[Trip]

And yes, im aware about that internal CPU sensor can measure up to 20 degrees higher temperature

Trodas, you just need a infrared reading thermal gun

[trodas]

I already almost have it Thanks to one helpfull member of SPCR not affraid that i use his addy for sending antrax, nuclear dump or wheather and willing to help me, i already bought it and now just waiting to come from US to CZ

I hope this toy:
http://www.professionalequipment.com/xq ... efault.htm

...help me to prove my temps are not fake


Idea What if Kostik damaged the internal sensor by trying the ability by the not only fanless, but also heatsinkless operation as shown in the movie? I know, unlikely w/o damaging any of the CPU logic, but just cause a temperature measuring shift - that could happen?

[Kostik]

Trodas, I think you don't get it. I'm not trying to convince you that you are wrong. I'm trying to explain that we used different mobos, that the temp readings can't be compared, and that you cannot say that I "suck" because I freaked out when I saw a temp reading of 176°c for the CPU. Period.

Of course the heatsink I was using wasn't very efficient. But I could run an undervolted Athlon passively with that very same heatsink, so it's not that bad. I know that you managed to run the C3 passively, and I believe you, by the way I did too. As I said, it was perfectly stable, but the temp readings were high.

I cannot send you the C3M266 mobo, I had it for two weeks only. You can read a review by another forum member here, same mobo and same CPU as me, using a CNPS3100 heatsink. He got high temp readings too, 60°c idle, and decided to add a fan just like me.

Idea What if Kostik damaged the internal sensor by trying the ability by the not only fanless, but also heatsinkless operation as shown in the movie? I know, unlikely w/o damaging any of the CPU logic, but just cause a temperature measuring shift - that could happen?

Who said I did that ? And what if you were an alien disguised as an SPCR forum member, and you were here to steal my psychic energy or something ? See ? I can make unfounded accusations too.

Again, I'm not saying that you are lying, you do not need to "prove that your temps are not fake" as you said. I'm just saying that I'm not lying either, and that our experiences are different, probably because we didn't use the same mobo. That doesn't mean that one of us is wrong. If you cannot understand that, and if you continue attacking me like you do because you feel that you need to prove you are right (when we are both right), we are not going anywhere.

[trodas]

Okey, dokey - we still not going anywhere

Because i believe that the temperature readings even from different mobos should be comparable.

Any it seems to are comparable anyway - you get 60degrees iddle and 90 degrees on load, i got 40 degrees iddle and about 69-73degrees on load.
The difference is pretty obviously in the Zalman cooler versus - let's use polite words - low-end aluminium something, errr, heatsink
You are of course innocent, as the mobo have only a very little space around the CPU - i know why i bought P6STMT rather that VIA

About the Athlon, well, my friedn failed to run even the lastest Duron pasively with CNPS3100-Plus and he tried hard. The difference is into the Forton PSU with 120mm bottom intake fan, witch providing the low, but sufficient airflow There is the diference between your test and my machine. My machine is fanless

About the mobo - well, when i can't use there CNPS3100 or CNPS6000, then is useless and i probably get similar temperatures as you - likely higher, because i run not 1Ghz, but 1.2Ghz VIA and also i lack of airflow - however seen the two big holes in my case? It works! The hot air really coming out from the top exhust even w/o any fan, just using funnel effect - hot air moving up

And when the mobo preventing usage of decent heatsinks, then no-wonder that people adding fans OTOH, most peoples are very sensitive to higher temperatures, witch could play a important part too Im also concerned about my temps, however as long i don't hiting temps over 85 degrees im comfortable with them

And yes, i say that it might happen - you canot argue with this Just a pure speculation. And yea, of course im alien and of course im going to suck all your life energy
But this is fun.
Idea, that driving CPU to 176 degrees might damage it sound logical, at least to me. I prefer not go over 85 degrees

And sorry for my attackings and sometimes pretty confrontative way of writing - i just keep saying what i like ony our review and what i don't like You just picked the bad mobo, and then you failed to get fanless and still bellow the max operational temperature VIA C3 machine. It don't happened for me, so im happy
But i still try lower the temperature using CNPS6000 instead CNPS3100-Plus and also i thinking about laping the top of the CPU to ge rid of the color logo and maximize the heat transfer

[silvervarg]

Hey guys, don't start a flame war. Try to stick to the facts.

One big difference is that Kostik had a very limited time to test, and he had to return the stuff afterwards, so he could not do any non-reversable mods.
Also Kostiks main point was comparing VIA C3 with other processor, and I think he did a good job in this comparison.

Trodas had much more success with a silent system. In fact I think he has done amazing stuff with the few mods he has done. Much of the success comes from picking the right components, and he defenatly has done that.
Picking the right components is very tricky, so either you spend lots of time and effort to find out what the best stuff are, or you just get some stuff and hope to be really lukcy.
I think Trodas spent at least a few days on research, and Kostik did probably spend a lot less time on research.

[Trip]

the C3M266 has enough room for an SLK800A (or an SLK900A if the PSU is not too close, ie. Sonata case) A small clip of the very edge of the retention clip is all that is needed and it fits very tightly. My 24.5W PIII runs passively though not under load - which is why I dropped a 5V Papst on top. Is there another reason the board is bad?

[trodas]

Let's start a flamewar

...ok, just joking There is not much to be "flaming" around, IMHO Temps, mobos, coolers and even CPUs used do difer and no wonder, that difer also results
silvervarg has nice point in the fact, that Kostik did good job on the CPU comparsion stuff - wich i skiped alltogether, because i know what i bought (nothing fast, just simple average cpu for simple tasks, never good at any computing power ) but CPU that allow me run fanlessly for sure
OTOH i maybe can consider more expensive mobo that can do undervolting and using a latest Duron, undervolted and underclocked However in not sure it it handle it. As i mentioned, friend tried this with the very same CNPS3100-Plus and failed - he is unfomfortable with 84 or 85 degrees after booting into windows - no wonder
Maybe he should underclock/undervolt even more, maybe if he added intakes/exhusts to the case... maybe... etc.
For me it serving perfectly and i also see a little point of plaing DivX movies as CPU speed testing, because everyone (at least there) using FFD show decoding filters for all the DivX, Xvid, WMA and such codecs - and they don't only allow breathless number of settings, but are also that fact, that using them one can play on crappy slow computer with only 300Mhz Cyrix CPU big DivX movie - over network!
However if there is interest, i can do the performance CPU graph as did Kosik Not a big problem. Instead of Kosik, i have the big advantage of almost unlimited time to play with this fanless machine
But i see a little point of comparing such cpu-easy thing as DivX playback - when i did DVD playback and it worked pretty well and OS still feeling very snappy during it, so Im more that satisfacted
Also mySQL db performance compared to "big-ass" company hosting server on net are pretty close - witch is a wonder, because win+php+mysql is not recommended and the server i comparing my home one to is more that 10x expensive and of course running linux, witch is recommended configuration.


Trip - if the board did not fit Zalman CNPS 3100/6000 coolers, then from my point of view (it might be pretty deformed or perverted point of view, bevare! ) the mobo is bad. Considering the wanted goal - fanless machine - i need the Zalman coolers to fit. That's very important for me. Also i did not trust the VIA chipset speed/features/functionality.
As i mentioned earlier, i have a VERY good experience with the SiS 630 chipset, so i bought it again. This might be also the reason, why the DivX playback feels so "light" for my machine - because i swear to god or evil, anger or devil - the SiS 630ET is one hell of the chipset!
Ha, you tried the same as me - i tried run fanlessly with the CNPS3100-Plus my CeleronII 900Mhz. (no idea what heat output this thing have, but the Vcore are shamefully high 1.75V... ) It worked well (over 50 degrees or such on iddle, tough...), except when i fire up CPU burn MMX hi-priority test, the temperature did not stop climbing even over 77 or so degrees, where i lost my "don't care about temp" point of view and i "pulled the plug" and stoped the torture
Fan is sadly necessary.
What wattage have the mentioned cpu anyway?
BTW, to demonstrate how the board are cool, there is my curent uptime:

...i try update the image regulary (hit F5 sometimes )

[alglove]

The Celeron II 900 (SL5LX, SL5WY, SL5WA, SL5MQ) has a thermal guideline of 26.7 W and a thermal spec of 77 degC, except for the SL633, which is 30.0 W and 72 degC.

http://processorfinder.intel.com/script ... orSpd=5852

[the_smell]

Hi trodas

good to hear that your fanless PC has come together well without too much heating . sorry I may have used a few too many smilies in this message

But hey! Fanless PC's are cool


Seriously I may be missing the point but even on a via 85 degrees is a bit high How often does it get that high in real life applications (basically I'm thinking about upgrading my C3 800 to 1.2 or 1.4 GHz but still using passive passive cooling) Is this system of your on 24/7 and running stable?

Appologies for the extreme sarcasm before I couldn't resist

[trodas]

alglove - thanks for the info Mine is SL5LX and i driven it - seems - to it's limits pretty hard Well, it survived my fanless test

the_smell - smileys are cool And thanks, yup, come together and working pretty well The fanless PSU now workign more that year w/o issues and the VIA workign fanlessly now for 3 months almost non-stop I torturing the poor, innocent chip hardly
Into real-life applications the CPU are around 40 degrees only. With higher usage it might be 50-55 degrees, but hardly more. Mine never reached 85 degrees, max i ever saw was 77 degrees before i do some modding and this was also after so many hours of Prime95 running. Now as you can check, my machine running for 13 days non-stop with Prime95 on it, and i checked and ooops, a bit high - 76 degrees
I have to get the CNPS6000 finally and lap the CPU core to get better temps

And yes, stability of the system is great In fact, it's most stable PC i ever seen Never having anz problem with it, never crashed, never restarted... Just great system and even running winblows 2000 server (SP3) and it's still stable Linux will be probably better, however i must use winblows because this is application and file-server
Used for long-time running apps as well and night downloads and such Stable like a rock

[davide_casarin]

I have 2 questions for Mr Alien disguised as a SPCR forum member:
-do I have to type or can you just read my mind ?
-do you look like the aliens in the Simpsons ?
Do not steal my psychic powers as it would be wasted time.

[trodas]

ROFL ...my camouflage is gone

First answer - yup, you have to type, i got lazy recently

Second answer - nope, they spilling too much of their liquids

...and no, no stealing will happen, im got Earthsick now
However i have one question - why you can't run you machine fanlessly after all? With CNPS6000 and nice holes like in my case are done, it could become possible quite easily!


PS. i fixed some bad galery links



...looking nice, is not it?

[trodas]

Bad news - my fanless server machine died on me It could be attributed to the extemly hot days this summer (many temperature records fall and at one point I was having 33°C in my room - folding with 6 computers just in my room are torture sometimes...) however up for direct blame are ECS company that used known bad brands of caps on the mobo, witch failed miserably.



G-LUXONs are right on the list http://capsmod.net/forum/viewthread.php?tid=19 of the bad caps that fail even if they aren't used - not to mention run nonstop two and half year
And since failed the primary Vcore powering caps right after the PSU, then my suspiction made me check the PSU and sure, another brand of "great caps" Hec is used there.

So, let's take a look. My first fanless PSU modification lasted 16 months. Then the PSU fried itself... (not mosfets, but logic and something on the PCB...) Now this one did not died yet, however it pretty well could attributed much for the failure of the mainboard... Then let's say 14 months it lasted for sure. That is not THAT bad, considering the Hec caps and the fanless usage of low-cost Eurocase PSU design...

But I don't give up so easily!

So, I replaced ALL the caps on the mainboard with quality Panasonic FM ones from digi-key.com. The smallest 10uF ones aren't awailable in super low ESR FM line, so I used the FC line of Pannyes there + added some ceramic caps for even better filtering. I bridged all primary and secondary Vcore caps with them as well, as bridged the PSU connector power imputs with bigger ceramic SMD ones from the back of the mobo... For more info and more pictures you can look there: http://capsmod.net/forum/viewthread.php?tid=179



After all this work I realized that the mobo still don't kicking in, so I checked more and both the Vcore mosfets are dead. After big help from Hong Kong friend "Big Pope" who can find these prehistoric mosfets and send me them I finally managed the mobo to work, so it is kicking well and strong again now

ATM running with Zalman 400W PSU, before I win the fight over caps from USA send directly to me (company - person = problem, person - person = no problem) with customs...

And when I have them, I'm going to recap the PSU - it still working, but I made some improvent ideas, changes in caps and mainly use good caps and we see, how long it last then
I think 5 years is not unrealistic, but we see

Much more technical info about the PSU recap are there: http://capsmod.net/forum/viewthread.php?tid=169

...and yes, it is a old thread and stuff, however I think that SPCR readers deserve to know what happend to this project and if it will remain functional for some time...


PS. the server just once crashed on BSOD and week later it stoped working completely. So, not much warning there...


edit: fixed pictures (one seems lost, fu*k imageshack!) and links

[trodas]

Well, it is a year 2008 and my fanless server is still working pretty well and entierly stable

Recapping is a good thing and it is also necessary to recap the PSU as well