Cyrix 6x86 too hot to touch but stable

MoJo-chan · Post by **MoJo-chan** » Wed Dec 28, 2005 2:10 pm

Oh, my poor burnt fingers

I have been trying to make a silent but reasonably powerful machine to run pfSense, a software firewall. I have a Cyrix 333MHz 6x86 with 128MB RAM, 10Gb HDD and two NICs. It has a virtually silent FSP power supply with a 120mm fan.

I've been trying to run the CPU fanless, and it works! pfSense was running for 24 hours, admittedly with no traffic (only some web interface action) and didn't crash. The CPU heatsink was far too hot to touch though...

Is it OK? I wonder... The mobo has an odd layout with the RAM on the left, so the CPU isn't under the PSU fan and would be hard to shroud. I could add an 80mm fan, but...

How hot is too hot? If it's stable...

dukla2000 · Post by **dukla2000** » Wed Dec 28, 2005 2:29 pm

MoJo-chan wrote:The CPU heatsink was far too hot to touch though...

I feel that is too hot. Those old Cyrix were pretty tough but there is no need to abuse them.

IIRC probably under 20W, which should be pretty easily manageable with a modern/large heatsink and a small amount of airflow (either case or on the sink itself). Hell, an 80mm 'flo @ 5V should keep it well under control, so why push your luck?

MoJo-chan · Post by **MoJo-chan** » Fri Dec 30, 2005 7:40 am

Fair comments there dakla2000. I tend to agree for the most part.

How hot is too hot? I have a Pentium III 500Mhz without a fan at work, and it doesn't get too hot to touch either. Maybe a K6 or something would be a better bet for me?

Post by **MikeC** » Fri Dec 30, 2005 10:09 am

Too hot to touch means different things to different people. Some who have very temp-sensitive skin on their fingertips might feel 50C is too hot; others may not have the same reaction till say ~65C, or even higher. Ask most chefs -- they can handle pretty high temps on their fingers because they're so used to it.

To get a better handle on the real temp, get a film thermal probe (the kind that comes with most cheap front panle temp monitor accessories) and try to stick it right at the edge of the CPU die. Wedge it in there if you can, or at least tape it in place temporarily. Whatever reading you get will probably be lower than the core temp; add ~10C to adjust, I'd guess.

MoJo-chan · Post by **MoJo-chan** » Fri Dec 30, 2005 11:17 am

Thanks MikeC. I did that and got a reading of 62C, so the core is probably 72 or there abouts.

Will running it this hot be a problem? I can't see why it should be really. CPUs rarely fail, even if pushed hard. In fact, a lot of Dell machines seem to rely on the P4 termal cut-out to keep from crashing, and presumably if you ran Folding or something that would max out constantly.

It's only an old machine, I don't really care if it dies I guess. I wonder if it will be OK in the summer though.

darthan · Post by **darthan** » Fri Dec 30, 2005 12:10 pm

72C is in potential danger territory for a lot of CPUs (especially since you don't know exactly how hot it is). Spending a few bucks on a panaflo or some recommended fan and running it at 5V is your best bet. It should make a big difference in temperatures with very minimal noise.

Elixer · Post by **Elixer** » Fri Dec 30, 2005 1:38 pm

Meh, I really wouldn't worry about it. I ran a Pentium 2 without a heatsink for 2 days straight and it was fine (I was hoping it would do something neat and than I forgot about it). Those old chips are tough. Most likely it will run in its current state for years without failing. If it does fail it's not like you're losing anything real expensive. Computers which will outperform that one in every way can be had for around $200. Just make sure it's stable by running prime 95 for a day straight.

MoJo-chan · Post by **MoJo-chan** » Sat Dec 31, 2005 3:39 am

One other reason for running fanless is lack of maintenance. No fan = nothing to get clogged up. Of course, there is the PSU, but oh well...

I'll probably just try it, see how it goes.

Anyway, I'm looking for a good way to stress test a machine for a long time without installing an OS. So far, I have come up with using Damn Small Linux and "cat /dev/urandom | gzip > /dev/null". Or, use DSL and download Prime95. It's a shame there is no DOS version of Prime95, or that lmsensors is such a pain to set up.

Here's a thought: What is needed is a bootable floppy or CD with a number of test tools, like Prime95, CPUBurn etc, and the ability to at least read the CPU temperature. I have searched extensively but nothing like that seems to exist.

jaganath · Post by **jaganath** » Sat Dec 31, 2005 5:30 am

What is needed is a bootable floppy or CD with a number of test tools, like Prime95, CPUBurn etc, and the ability to at least read the CPU temperature. I have searched extensively but nothing like that seems to exist.

Doesn't the Ultimate Boot CD have what you need?

jb_ · Post by **jb_** » Sat Dec 31, 2005 6:05 am

Elixer wrote:Meh, I really wouldn't worry about it. I ran a Pentium 2 without a heatsink for 2 days straight and it was fine (I was hoping it would do something neat and than I forgot about it). Those old chips are tough. Most likely it will run in its current state for years without failing. If it does fail it's not like you're losing anything real expensive. Computers which will outperform that one in every way can be had for around $200. Just make sure it's stable by running prime 95 for a day straight.

Disposable consumerism at its finest.

jaganath · Post by **jaganath** » Sat Dec 31, 2005 7:34 am

Disposable consumerism at its finest.

<off-topic>

Australians can hardly lecture others on environmental sensitivity given their refusal to sign up to Kyoto and gas-guzzling suburban lifestyles.

</off-topic>

mb2 · Post by **mb2** » Sat Dec 31, 2005 8:20 am

"Australians can hardly lecture others on environmental sensitivity given their refusal to sign up to Kyoto and gas-guzzling suburban lifestyles. "
to be fair, jb_ isn't the PM of Australia so its not really his fault is it.

as for the 'eco' aspect, it would do more damage to the environment by buying a new fan, than by burning out a CPU and replacing it with another, which would most probably be destined for the scrap-heap anyway if
MoJo-chan didn't have it. it would also be a lot cheaper than a panaflo or nexus too i imagine.

i think CPUs are pretty heat resistant anyway, i've had my tbred at upto almost 85*C and its still fine, and ran p1s and cel 2s with no airflow atall ok.*

its probably possible to remove the psu fan too, certainly if u move it out the case. having said that if ur cpu is 'on the edge' of heat then it probably will do the cpu harm to remove the only airflow out of the system.
i dont know how u have it housed but if it is enclosed then its probably going to help by opening it up to a bit more air.
*all the times when i've run cpus 100% without airflow they have been open or no cases.

having said all that, since u have a 10gig hdd, its probably quite loud, so there wouldnt be much noise advantage gained by not using a 5v'ed fan blowing across the HS, even if its an old one lying around.
if u got somethin CD booting that loaded in ramdisk, or similar, then i'd ditch the fans.
i dont know if ur able to underclock it a little either?

edit: as for the stress testing, you could try using a tiny modified windows98 from winimize.com on a boot CD. or just use any type of windows-based (w9x i'm guessing due to the specs) boot CD á la bart PE, and add prime95.exe to the cd.. it runs without install. i think most similar programs can do the same.

qviri · Post by **qviri** » Sat Dec 31, 2005 9:02 am

mb2 wrote:i think CPUs are pretty heat resistant anyway, i've had my tbred at upto almost 85*C and its still fine, and ran p1s and cel 2s with no airflow atall ok.*

*all the times when i've run cpus 100% without airflow they have been open or no cases.

My demolished Dell laptop (tiny CPU heatsink, but in free air), with a P2-Deschutes in it, is programmed to turn on its fan when CPU hits 85*C. As it's 233 MHz, I quite frequently max the CPU out (maybe not to Folding/Prime95 levels, but still). I've had the chip running for 14+ hours daily for at least two weeks: average temperature was well into 70s, the CPU is unscathed to this day.

MoJo-chan · Post by **MoJo-chan** » Sat Dec 31, 2005 9:35 am

jaganath: The UBCD is interesting, certainly... unfortunately, there is no way of monitoring CPU temperature in real-time from the diode on the chip in DOS, only via the BIOS. I'm going to look into porting CPUBurn to DOS too, I can't see any technical reason why it shouldn't work.

For now, the best bet seems to be DSL, because you can then download Prime95 and CPUBurn from the net, and try to make lmsensors work if you are feeling particularly masochistic.

jb_: Well, at least it's not going in the bin. Britain is well on it's way to meeting what it agreed at Kyoto, and it isn't ruining our ecconomy. Take note America, and please stop pissing in your end of the pool.

mb2: I hadn't see that Win98 thing before, it looks fantastic... ten minuites to install! I agree that older CPUs seem to be very heat resistant. I had an Amiga with a 68060 clocked at 50MHz, without even a heatsink and only a single fan in the PSU for the whole case. It too was always too hot to touch, but never crashed. Heat was never an issue in those days.

I'm interested in this as an experiment, because you can pick up computers of this spec for free quite easily, and make them into excellent routers with wireless access and traffic shaping. Better to recycle them that way than to bin them I think. But I want them to be as reliable as possible, and zero fans (OK I have one) means a lot less dust and noise.

Cyrix 6x86 too hot to touch but stable

Cyrix 6x86 too hot to touch but stable

Re: Cyrix 6x86 too hot to touch but stable