Computer tech blog

Smokey · April 06, 2009, 05:41:45 AM

Outch! man, that reminds me of the time i got to service williams PC after they had brought it in for servicing with their regular PC-guy for some ten times...

Turned out all he did was to reinstall Windows (not everytime a PC is running unstable is Windows' fault), so when i got it, the first thing i did is what i always do; Open the case up, and i saw dust... Lots of dust, and a dust cake about 5mm thick between the CPUs heatsink and it's fan... so i knew what the problem was... Unfortunately when i removed the heatsink i saw a perfect impression of the CPU burned on the bottom of the CPU... Long live thunderbirds lack of thermal protection...

Dr. Mario · April 06, 2009, 02:36:32 PM

Wow-whee! That's no wonder it's named Thunderbird. ;014

Luckily, AMD etched thermistor on-die inside their 64-bit CPUs, that is, if it gets so hot it shut itself down. But just blame on motherboard if you got a fried 64-bit CPU.

Also I noticed that as I corrected the strips in the fan power plug, the retail heatpipe cooled down quickly as the fan is running - now I'm wondering, do it contain R134a refrigerant or ammonia gas for the extent of cooling??

Smokey · April 06, 2009, 04:17:19 PM

well, heatpipes are usually filled with a cheap liquid, like water as far as i've heard... but then again, who knows what stuff they put in there...

Dr. Mario · April 06, 2009, 04:56:53 PM

I guess AMD may have known better not to put water in there, providing the retail heatpipe (that came with Phenom CPU) have a uncanny fast recovery - water don't do that.

I suspect there may be one of three things in AMD's latest heatpipe: Alcohol, R134a refrigerant, or Ammonia gas.

Smokey · April 06, 2009, 05:33:39 PM

Well, then i guess it would be alcohol, since R134a would be a bit expensive, ammonia would be a bit dangerous.... And, well:

Quote from: "Wikipedia"Some example fluids are water, ethanol, acetone, sodium, or mercury.

[/url]

Quote doubles as link

BTW mercury sounds a bit dangerous to, as far as i'm concerned...

Dr. Mario · April 06, 2009, 06:04:27 PM

Yeah. And, Mercury is only intended for use in a hermitically sealed computer, like a part of upgrade package in Hubble Space Telescope, for example. Also, AMD can choose to use R134a refrigerant if they want. I think it is very likely to already have R134a inside while taking Phenom's TDP into account.

Smokey · April 06, 2009, 06:31:25 PM

Yeah, but heatpipes work because a liquid evaporates because of the heat produced by the (in this case) CPU and then moves to the cooler part of the heatpipe where it condensates and moves back to the hot side... i don't think a substance wich is already a gas in room temperatures will do that very efficiently...

But i will have to look up asetec's new phase change cooling solution to get a definate anwser on that...

Added after 8 minutes:

Well, that search turned out inconclusive due to lack of practical or technical information about their newest cooler...

Dr. Mario · April 06, 2009, 07:20:01 PM

Well, at certain pressure, R134a can become liquid with just a set of radiator fins and a fan. But for a wick, it has to be made of ceramic fiber, which is made hollow - in order to suction the liquified R134a refrigerant back to CPU block, to be vaporized. It doesn't have to be a passive heatpipe, like VapoChill Micro.

Smokey · April 06, 2009, 07:44:15 PM

Okay, i have sent an inquiry to AMD with the request that they send the technical data on the substance used in their boxed heatpipe coolers shipped with their CPUs...

I really want to know, I've been watching too much House, probably, but i want a clear anwser instead of guesswork...

Dr. Mario · April 06, 2009, 09:33:12 PM

Yeah, I agree. Still, definitely not water, though.

Smokey · April 07, 2009, 03:11:10 AM

Oh, no... I guess it's alcohol, since it evaporates at ~70c, making it a quick responding fluid and it's cheap to come by...

Dr. Mario · April 07, 2009, 04:29:18 AM

I thought so. Also, you found that people say that XDR-DRAM is never good enough, right? For me, I disagree (because I have used it. Sure, I admit it's hard to program the XMC to get the main memory activated, but it's still similar to what's in few of AMD's programming manual.) So, what do you think about that?

Smokey · April 07, 2009, 02:30:37 PM

Well, although i find the name a bit uninspired... I do think it's the logical next step in RAM evolution, and i am also glad that the company Rambus isn't just a part of intel... wich i thought because i only saw RDRAM in intel machines...

Vegapunk · April 07, 2009, 03:06:48 PM

I agree Dr. Mario, although activating the main memory is QUITE a pain in the ass......

Dr. Mario · April 07, 2009, 04:30:26 PM

True. It's a pain in the ass - you (will) have the garden-vareity of XDR-DRAM memory, which you must program the CPU to collect the SPD data, the clock the pulse trains (just like what you would do away with DDR memories.) then map it after. But pains and tears will just be worth it, you will have an access to much faster main memory in either your homemade OS or Windows (with the Rambus XMC driver that you either made or download off the CPU maker's webpage.)

That's fine, Smokey.

Added after 10 minutes:

Not much peoples find it uninspired. And it's indeed a next step, XDR's next to be used by new computers. Also, some peoples said it's slow - they're wrong, I have actually used this kind of RAM, directly linked to a x86 CPU (this processor's still pretty new) AND ran SISandra software's RAM bandwidth test, it went off the chart (being #1 Better in this test)!