NEWS FLASH - AMD changes mobo specs!!!

[bigben2k]

been surfing around today, ended up at HardForums (thanks to Joe!), followed a few links and I ended up there:
http://www.tomshardware.com/cpu/01q3/010917/index.html

In short, this guy ran AMD processors without any cooling, and they burned up in a sec, where the Intels survived very nicely... (It really made me think twice!)

He got AMD to fess up that the Athlons need some support so that they don't burn up so easily, he even got them to change the mobo specs!

...which means... that the Asus A7V333 is as of monday, June 10, the only AMD approved mobo for the Athlon 2200+ (I think).

(Like I said, you have to use an Asus board to appreciate how smart the design is).

[Joe]

Thats some old video if I remember right :)

[WebMasta33]

Originally posted by Joe
Thats some old video if I remember right :)

Yes, if you read the grey area next to the picture, it was:

Created:
September 17, 2001

So old would be a small understatement :p

[jtroutma]

TomsHardware has had that review up for a long time already. I think it has been almost 3 months (could be wrong though).

I visit his site and others (ie. Anandtech.com) daily for up-to-date details about technology in general and what is going on in the computer world. AS a matter of fact, he intoduced me to water cooling!!! :)

Both those sites are really good for details and hardware reviews.

[Brad]

toms is also particularly great for biased reviews.

soltek, iwill, asus, and several other brands now have ways to turn your computer off if you're dumb enough to run with no heatsink

[bigben2k]

Originally posted by Brad
toms is also particularly great for biased reviews.

soltek, iwill, asus, and several other brands now have ways to turn your computer off if you're dumb enough to run with no heatsink
I think the point was that if the HSF falls off (during shipping for example), then the computer is at risk. Of course we are a special group: we build it ourselves!

Asus and Soltek are the only companies, as of May 9th, 2002, that have a board that meets AMD specs: read the CPU diode and immediately shut-off when it overheats. The installation of a temp probe under the CPU does NOT meet AMD specs. The MSI KT3 Ultra-ARU (MS-6380E) is listed as having thermal overload protection, but I guess they do that through the probe in the socket, which won't work to prevent the CPU from dying. Many mobos have that probe (why?)

Yes, the video is dated September '01, but the new specs are effective June 10, 2002. Definitely worth a look though!

I think it also emphasized the need to bolt down the cooler, instead of relying on clamps.

I think I'm ready to go ahead with the Athlon 2200+, a 226W Pelt and the Dtek TC-4 waterblock. I just wish someone would review it!

[bigben2k]

Tom was also able to OC a P4 to 3.0 , and he says that most Northwood can be OC'd to 3.0, with water cooling, making the P4 womp anything AMD has in every way...

He supercooled the P4, to maintain 20C.

hum... Too bad the P4 is out of my budget!

[pHaestus]

Tom is also a tool.

[bigben2k]

Originally posted by pHaestus
Tom is also a tool.
tool?

I'm not very impressed with his 18 mobo KT333 roundup, he seemed to read past the numbers, but the review had a lot of information. I didn't even know that there were already 18 KT333 mobos on the market.

He did get me interested in the A7V8X, to the point where I am considering waiting for it...

[jtroutma]

I have to agree that Tom does seem to be biasas against AMD and more in favor of Intel products.

Generally though, he does give a fairly accurate idea of what is going on out there in the world of computers and gives some "decent" tips on what can and can not be done.

I also have to say that I would like to see him not bash AMD most of the time, especially when he of all people should see what is going on in the background. (ie. Why AMD is not pushing really hard anymore on their Athlon line)

Anandtech.com is my prefered site for accurate reviews and details however. I have yet to sense any bias reviews against anyone, to my knowledge at least.

[Joe]

Tom is worse than a useful tool, hes a tool without a purpose. hes like some weird screw driver that never fits any screws you find.... hes that kinda tool.

[jaani]

Originally posted by bigben2k
In short, this guy ran AMD processors without any cooling, and they burned up in a sec, where the Intels survived very nicely... (It really made me think twice!)

hehe. Can't help myself.

http://members.optusnet.com.au/01001010/images/procooling/last%20week.jpg

JK! I'm bad aren't I... :rolleyes:
Seriously, though..thanks for that BB2k, it certainly highlights a design issue we're only too familiar with. What's the saying? "From zero to fried in 0.8 seconds"? hehe. But THG has, in the past, shown a peculiar bias towards Intel - interesting that he was mentioned in that Reg article a while back about Intel marketing tactics...And his reviews are generally very average.

I think THG is kinda like the jack of all trades. He tries to be every tool but suceeds at being none. ;)

[Brad]

the only reason tom likes intel is because they pay him more than amd.

the other thing I don't like about him is that he thinks he is god, h knows everything, and that he does everything first and shows everyone how to do it

[bigben2k]

Originally posted by jaani


hehe. Can't help myself.

JK! I'm bad aren't I... :rolleyes:
Seriously, though..thanks for that BB2k, it certainly highlights a design issue we're only too familiar with. What's the saying? "From zero to fried in 0.8 seconds"? hehe. But THG has, in the past, shown a peculiar bias towards Intel - interesting that he was mentioned in that Reg article a while back about Intel marketing tactics...And his reviews are generally very average.

I think THG is kinda like the jack of all trades. He tries to be every tool but suceeds at being none. ;)
You're welcome, and I appreciate the crack, I really do.

I thought I saw bias too when I read that, but then I realized that what he's doing is taking a stand on an issue. Wether or not this issue is important to any of us, that's subjective, and I think that that's what he's missing.

For example, this whole thing about the CPU burning up because a HSF would fall off during shipping, doesn't apply to me (or most people here), but it's still a significant test, because now I know why I'd rather use metal bolts to clamp a waterblock, instead of nylon. His point was that the cooling device needs to be absolutely secured, so I'm going to add a drop of loc-tite.

Why he chose to put down AMD in the process is beyond me, but he did demonstrate that Intel had a better thermal management feature in its processors, and that in the (remote) eventuality that the cooling device falls of, the P4 will allow a user to recover, where AMD's solution won't. That's a nice feature.

Still he misses the point where he could have simply said that instead of spending an extra $300 for a P4, you could just get 4 bolt for what, $1.50 ? (It's more complicated than that, but the price difference is there).

The Athlon shouldn't be compared with a P4 solely on the performance: price is an issue too. I think he understands that, but he makes himself look like an idiot when he doesn't take the application of this computer gear into consideration. My wife would do very well with an AMD, to write essays and surf the net once in a while.

[pHaestus]

I think AMD systems are suited to a little more than "wife surfing the web" boxes. Now granted, they are more attractive to most of us at the moment because we jumped ship to SocketA when Intel was deeply inhaling ass. A year ago, Intel on the desktop meant the first gen P4, RDRAM, and the P3 wihtout clock boosts. The situation is different today for sure, although Intel mobos seem to be going through some growing pangs. Most Northwood users I know have had 2-3 boards in the last couple months AND ram swaps to find a combo that worked for them. I would buy a 1.6a or 1.8a and shoot for 2.7GHz if I were building a scratch computer today. The only AMD setup I would consider would be dual XP CPUs @ 1800MHz or so on a Tyan (the Tiger mp refurbs are dirt cheap now at newegg).

To me, a heatsink falling off in transit is still sorta a non-issue; even if the P4 saves itself, a 2lb hunk of copper/aluminum bouncing around inside the case in the back of a truck or plane is going to **** all manner of compnents up. I mean you KNOW it is gonna take some traces with it on vid card, mobo, etc. I can't imagine a manufacturer that wouldn't take a brand new computer back that had been subject to this.

Tom should have tried just turning off the heatsink's fan, or mounting the heatsink backwards and booting the system. Plenty of people have had this result in killing TBirds; does the new COP protect against that? Oh, wait. More relevant and less dramatic; that doesn't sound like THG at all to me :)

On a related note, I have been yelling into the void about internal diode monitoring on AMD boards for about 6 months now (since I made a MAX6657 reader and saw how easy it was). I didn't get a single response on this article:

http://www.voidyourwarranty.net/review/review.php3?r1=asuscop/index.php3

such is life.

[Cyco-Dude]

i think the point of him frying CPUs is to point out that AMD needs better thermal management. the new XPs have the internal diode, but the vast majority of boards fail to take advantage of it.

that'll all change soon tho with the next gen boards...

[bigben2k]

Of course the AMD can do more, I just meant that for small, general purpose, the AMD is by far a better choice. For a performance system, then it depends on what you do with it, and it all comes down to the FPU, and the memory bandwith you need, because that's the real difference between the P4 and the Athlon.

The heatsink falling off is definitely a non-issue, and the scenario that THG advances really doesn't make much sense, or more specifically, is so remote a possibility that it almost invalidates the entire article, but there is a point, and it's a good one, within the realm of what is realistic: the P4 will not leave you with nothing if/when your PC shuts down. Also, for newbs or the occasional accident, the P4 is forgiving because it is effectively idiot proof.

The other point was that MBM is virtually useless, and will not always work. I tend to agree with that, but I haven't used MBM, and I don't know how far the refresh rate can be pushed, but it makes far more sense to have a built-in circuit to protect the CPU, instead of using the CPU's precious processing time to watch over itself.

Why didn't any mobo manuf think of this before? I guess you'd have to ask AMD, since they didn't require it. This is an age old debate that's very similar to what airlines face when dealing with the FAA. Airlines rarely do more than what the FAA requires them to do: that way, the liability rests solely on the FAA.

It's a finger-pointing exercise... (don't ask me, ask them)

[bigben2k]

Originally posted by Cyco-Dude
i think the point of him frying CPUs is to point out that AMD needs better thermal management. the new XPs have the internal diode, but the vast majority of boards fail to take advantage of it.

that'll all change soon tho with the next gen boards...
It will change, since it's now an AMD requirement (as of June 10, 2002)

[pHaestus]

when their overclocked and overvolted XPs are running 50C under load with high end WATER cooling? You see this in the forums a lot already; Epox 8k3a users getting told to upgrade this and that and remount their blocks again and to buy new heatsinks etc. Everyone's CPU has been running this hot all along but no one believed it. The immense heat generation of AMD CPUs is their single biggest drawback. The wool has effectively been pulled over o/cers eyes about this though to the point that when they see realistic numbers then they won't believe it.

[bigben2k]

It makes a good case for cooling the core with water directly...

[pHaestus]

I thought we were talking reliability? I can't see a situation where that would be a logical choice for mission critical cooling :)

[bigben2k]

Originally posted by pHaestus
I thought we were talking reliability? I can't see a situation where that would be a logical choice for mission critical cooling :)

Then let's figure out a way to make it a mission critical choice!

I was just going over the bong/vortex thread, that's why I brought it up. seems that core temps remain 20degF over the water temp, for this guy Pepsi, who just glued an acrylic block right over his CPU. Water flows right onto the core.

But to get back on topic... (we can start another thread with that later)

I was surprised to see that the temp went as far up as 360C. I've always wondered what temp the CPU would reach, if it could. It's nice to know what power the CPU emits, but the temp is also a factor which unfortunately, can't be measured. So for now, we know that without any cooling, the Athlon is attempting to reach at least 360C.

[pHaestus]

It's nice to know what power the CPU emits,

I have a pretty good handle on CPU temps, but power is a black hole. There is a ton of bad info out there. Check this out:

http://phaestus.procooling.com/watts.jpg

Take temperature readings online with a huge helping of salt. Direct die coolers usually continue to use the under socket probe. You are cooling the base of the processor (where heat has to pass through to get to the thermistor) and so the temperatures look exceptionally low. Forced convective cooling is related to water velocity and surface area (among other things). You are getting rid of a big barrier to cooling (the TIM junction) but you are also sacrificing a huge amount of surface area. I would suspect that moving to designs completely different from traditional waterblocks (nozzle?) would be needed to truly get good cooling from direct die approach. And unless this setup comes integrated with the CPU from AMD I can't imagine it ever being implemented on any sort of large scale.

[bigben2k]

I see what you mean. That's a 20% margin there...

So the internal temp diode of the Athlon, is it usefull? Is it accurate? Where is it anyways?

[schoolie]

It was packed with lots of useful information.
As an encore, maybe try hooking up multiple MAX6659 chips to the same smBus header. I think Maxim might also make a thermocouple chip to interface to the SMBus too! I'm think about all sorts of cool stuff to connect to the SMBus. You can control, potentiometers and elays through SMBus too!

It's too bad that the thermal diode reader is so sensitive to cpacitance and resistance in the leads. I was also wondering if anybody got the Winbond chip bypass to work correctly with the thermal diode socket pins.


Your fan,

Rob