Intel vs AMD on Toms HW

[Mikewarrior2]

AGodSpeed,

AMD PR was spreading the 20% heat reduction before the palomino's were released. NOw that the MP is out, and teh white papers for both processors are out, we can check the numbers for ourselves.


T-bird 1ghz: Max Thermal load: 54W.
Palomino 1ghz: Max Thermal load: 46.1W

Which equals a ~15% heat reduction from a 1ghz T-bird to 1ghz palomino. Good, but nowhere the 20-25% that you claim. NOw, again, this doesn't matter what websites say, because this is referenced from AMD white papers.

T-bird 1.2ghz: Max Thermal load: 66W
Palomino 1.2ghz: Max Thermal Load: 54.7W

~18% drop. Future Palomino heat still isn't known, but i'd surmise that it will be in teh same range as the 1.2 and 1ghz have shown, which is still under the prerelease PR numbers that were out there.

As far as power requirement, it may well be a 25% drop. But no way for heat output.

For reference, the two pdf files are available at amd.com. They are 24685.pdf and 23792.pdf



Mike

P.s. If you take AMD's typical power output, the drops are ~17% for both processors.

 

[Mikewarrior2]

AgodSpeed,

IN addition, the internal diode has nothing to do with palomino's reduced heat output. That is the result of chip optimization, the diode will do nothing itself for heat. It was included so that a user could read the internal core temp, something missing from t-birds, but it appears that more work on the internal diode should be done.



Mike

 

[AGodspeed]

the internal diode has nothing to do with palomino's reduced heat output.

You're right, I misspoke before.

EDIT: Your comment of "only 12-18%" still isn't accurate. Your last post listed 15-18%. But we'll see anyway once they hit desktops.

EDIT2: I looked at your last post again and saw that the heat production you listed goes down from 15 to 18% going from 1 GHz to 1.2 GHz. So actually, it's logical to say that as clock speed increaes, heat production will start going down into the 20 percentile range.

 

[Dark4ng3l]

Hmmm eve never heard of anyone frying their cpu because of heat, or having their heatsinc fallig off or anything like that. P2's and k6's had no thermal throtling things and nobody has ever complained about it, nobody has ever complained about the old slota athlons or tbirds for this reason either, so why would you now?? Just because the cpu cant take comditions it was not made to handle doesnt mean it's not realible.

EDIT

How long would you car run without oil and prestone in it??

 

[grant2]

Interesting to observe how all of a sudden oppinions on thermal throttling (after Van Smith's rather bad & highly inaccurate report) have gone 180 degrees & it is accepted as a good idea now.


Actually "most" opinions were that clock throttling was a good idea, but it should never be needed with a correctly applied HS+fan.

You have to admit that it's a pretty reasonable viewpoint, *if* the p4 were going at 1/2 speed right when the most demand was placed on it that would peeve off a lot of buyers!

 

[fkloster]

<< Hmmm eve never heard of anyone frying their cpu because of heat... >>



Ummm, hello? Where have you been?



<< You have to admit that it's a pretty reasonable viewpoint, *if* the p4 were going at 1/2 speed right when the most demand was placed on it >>



Good thing people who buy P4's realize that they will never have to worry about that *if*

 

[AGodspeed]

Hmmm eve never heard of anyone frying their cpu because of heat...

Ummm, hello? Where have you been?

I think what grant2 was trying to say is that you will never overheat and fry your processor if you have all the standard cooling. You'll only fry it if you do what Dr. Pabst did, or if your HSF randomly falls off (highly unlikely).

Heat itself causes frying however, so it was just worded poorly.

 

[Wingznut]

You don't think that if the cpu fan were to fail, the cpu would cook?

 

[AGodspeed]

You don't think that if the cpu fan were to fail, the cpu would cook?

Of course I would expect it to cook. It just doesn't happen all that often if you carefully install it. Or if you buy your computer from an OEM.

 

[Macro2]

RE:"
You don't think that if the cpu fan were to fail, the cpu would cook? "

The temp would climb slowly until the computer locked. A new fan and restart should work.

OTOH, if you expect the HS to fall off I'd get a Pentium...

Mac

 

[damocles]

The temp would climb slowly until the computer locked. A new fan and restart should work

Maybe with some heatsinks, but with most i'd be surprised if the processor didn't cook within an hour or so.

 

[wesman]

What I didn't like about the article was that he admits that most heat related failure is due to fan failure, but his test has nothing to do with fan failure. I bet for every 1 CPU that had the HSF fall of there are 20 that had the fan die. I know I forgot to plug the HS Fan in when tweaking my Duron several times. It never fried the CPU. It wouldn't boot, so I checked and found the problem. So, Why not disconnect the fan as a test? This would give us some valuable information. I would like to know if the fan dies on my new AMD chip, will the new thermal protection work? Tom's test may give him some dramatic video, but it doesn't give me much real information.

Also Tom says in the report "In less than a second Athlon 1400 dies the heat death." but the video clearly shows the computer running for 5 or 6 seconds (the game stops and an error message pops up after 3-4 seconds) before a cut to a CPU close up. In fact, we never actually see the computer die. I'm not saying this is a big difference, but it just shows how Tom may have sensationalized the AMD's death in his reporting. The fact that the CPU and possibly the MB were damaged is exactly what the AMD website and manual have been saying for months. "Any attempt to operate the AMD Athlon? Processor without a suitable cooling solution will result in permanent damage to the processor and potentially other components within the system."

Of course they don't mention the fact that it could start a fire, unlike Tom. I guess if you buy AMD you are risking buring your house down, with your family asleep inside. Just because you wanted to save a few bucks. Your kids would still be alive if you had bought a Intel P4?, which you can learn all about by clicking the ad on Tom's front page.

That said, I too was impressed with the P4's abilities. But not enough to spend triple the cost for an equivilent CPU, just in case of the extremely rare event of my HSF falling off.

 

[sandorski]

copy/paste from my post at Ars:

tsk tsk tsk and we're supposd to be the geeks? Has anyone here *ever* had a heatsink fall off their cpu outside of shipping/moving it?

I've seen heatsink fans fail, but not once have I seen a heatsink just fall off unexpectedly. Most SocketA mobos already shutdown on heatsink fan failure. Their is no issue here folks!

What's next? Are we going to start judging cpus on which one can have the most pins broken off and still function?

 

[damocles]

To be honest I have seen a few people ( at least one on AT) who have been working on computers late at night and fired up their PC without putting the HSF on. Strange but true

 

[damocles]

A quick search of the forums found a few people that may have found throttling to be a very useful feature

#1 Upon opening the case, I discovered that the plastic tab on the chip socket had failed (broken off completely) and my ALPHA PAL 6035 had come off of the chip. There had been tremendous heat generated and the stickers on the PAL were melted and two of the pads on chip melted as well.

#2 just came home from a business trip, I had planned to sell my current 1.2ghz Athlon here and buy a new 1.4 to go along with a watercooler I have here in a box, I get home and go to my PC and it will not wake up from sleep, I am thinking ok I'll just turn it off and back on, no luck, So I then open up the case and I find my FOP 38 laying at the bottom of my case, Sometime during my trip one of the nipples that hold the fan on broke and it caused my HSF to fall of, So noy I have a dead T-Bird!!! Also I have to find some type of device that will hold my new waterblock on via the 4 holes on my A7V133, ,man this sux

#3 I used a Taisol copper-bottom heatsink, and when I took the clip off ( 0 - U - C - H!!.....I have a few new scrapes on my hands) there was a very, very thin layer of the Artic Silver that I put on the CPU - on the copper plate of the heatsink. The Artic Silver on the CPU didnt even look smudged in any way. I have heard that a CPU will fry in about 10 seconds if it is not properly cooled, and since the Artic Silver on the heatsink was so thin, I am wondering if maybe it didnt properly connect with my CPU - and fried the CPU. So my first question is: Is there any simple way to test a CPU to see if it is working? I have no idea how you would do that, but maybe there is a way...........anyone? And, how much Artic Silver would you expect to transfer from the CPU to the heatsink if there was a proper "connection".

 

[HereThereandEverywhere]

I don't suppose replacing the fan say...every year on an AMD would be too hard? And instead of getting new processors you just get new HSFs? Putting them on isn't THAT hard...

 

[RVachon]

Well, if you let your HSF degrade to that point you obviously shouldn't be a computer owner. Especially when your running a high speed CPU.

Ray

 

[shathal]

I think the major problem with that would be that AMD would begin wondering why they have to replace so many FHS's and also, it would involve people BOTHERING (or rather - successfully getting them to be bothered) to do this.

I think the chronic laziness of most people would prevent such a thing from happening.

Generally don't want to open my systems any more than necessary. This may come particularly maybe from my work with servers, but hey ...

 

[damocles]

I don't suppose replacing the fan say...every year on an AMD would be too hard? And instead of getting new processors you just get new HSFs? Putting them on isn't THAT hard...

Well, if you let your HSF degrade to that point you obviously shouldn't be a computer owner. Especially when your running a high speed CPU.

99% of computer owners wouldn't even think of opening up their PCs.

 

[Zelgar]

This is a repost of my message in the CPU forum, but it seems to be dead while this one is still going. Basically, I had a little problem with the article on Tom's site...

The basis of my problems with the article was how Tom did the test and his conclusions. He stated that the purpose of this test was to determine the effectiveness of the thermal protection of the CPU, but did it at the most extreme case (which is most unlikely to occur), and the result on the AMD - Palomino is questionable since the video went blank before the heatsink was totally removed (around 1:43 on video - possible damage to the chip).

To adequately address the thermal protection of the CPU's, I believe that multiple types of testing should have been done and on multiple CPU's of the same type. I think the test should of been done in the following matter:
Test #1 - Fan failure on cold system
Test #2 - Fan failure on hot system
Test #3 - Heat sink removal on cold system
Test #4 - Heat sink removal on hot system (this is the on Tom did)
(with "cold" system being one at room temp to start, and "hot" being one that has stabilized at full load)

I think a test like this would of been a lot more useful and unbiased, instead of the one in the video that has questionable results.

 

[Mats]

So you're a geek who buys a AMD OEM PC. You have it delivered to your door. The guy delivering your PC kicks the box with your PC in and the heatsink falls off. Or the guy that put your PC together maybe didn't put the heatsink on properly.

So you turn your PC on and start a small fire. You're frying marshmallows and having a great time because you have a new fireplace in your study. The smoke that gets onto the marshmallows might give you a headache though.

Anyway, once the fire dies down you call the delivery dude and tell him you want a new PC.

Intel has superior heating capabilities incase your heatsink falls off. This is pathetic.

I have a 1.33 Athlon. It's doesn't have SSE2, it's small, very hot, and everytime I change the heatsink I get worried that the core might crack. I love this processor!!!

 

[fkloster]

<< He stated that the purpose of this test was to determine the effectiveness of the thermal protection of the CPU, but did it at the most extreme case (which is most unlikely to occur), >>



What in the Hell? How else are we supposed to test thermal protection devices? What IS likely to accur that will enable thermal management?

 

[Sugadaddy]

I don't know if it's been mentioned (too lazy to read all these posts ), but he tested the Palomino with a KT266 mobo. Does anyone know if it supports the Pally's thermal management? That might be why it fried...

 

[SniperWulf]

#23 here says this:
http://www.tech-report.com/news_reply.x/2883/




<< According to AMD's data sheet on the Athlon MP, the circuitry to protect the Palominos from burning up doesn't exist on the CPU itself. It need sto be supported either in the south bridge, or the firmware, which can assert a Stop Grant pin to throttle the CPU a la Pentium 4 if the on-die thermal sensor signals a high temperature.

In the case of Tom's test, I have doubts about the efficacy of the VT8233 south bridge, as well as that of the firmware used.

In Forge's case, it appears that the AMD 766 southbrifge or Winbond thermal sensor (not sure which) did correctly sense a higher temperature and throttled the CPU down (well, perhaps made it unusable), unlike Tom's Palomino.

I suppose that Gigabyte's new consumer AMD760MPX motherboard (with the AMD 768 southbridge) also has this functionality.

Also note that the motherboard Tom used is not AMD-recommended for the Athlon MP.
>>



You think thats why the Palomino still fried? I mean if AMDs solution is implimented correctly and all....

Also Forge (#2) says this:



<< Well, here's my (very on point) two cents:

I move my case *a lot*. Every three weeks, all the wires come out, and it gets carted out to the car. No bracing or prep is done internally. IKt is then put in the trunk. Yes, I don't exactly *throw* my 5K$ dually Athlon into my trunk, but I'm not breaking out any pillows and yelling 'careful, careful' either. The heatsinks I'm currently using, the Thermalright SK6s, are quite heavy, and they use the socket clips. This last LAN party, unknown to me, one SK6 had popped off the socket while in transport, most likely when I hit a bump. I failed to make my customary check at the LAN, as I was running very late and had maybe 1/2 hour to go from the car to up and running. It was three hours later before I realized the system was very sluggish and single-minded..... almost like I was back on a UP system..... A quick check later showed Windows talking to only *one* CPU. A reboot later showed 'Athlon 1200Mhz'.... '1 Processor(s) detected'. I removed the side and immediately noted the SK6, laying on the bottom of the case. I almost cried, sure that my dually box.... wasn't.

In one of those empty gestures, like rolling a loved car down the street one last time before the junkyard wrecker comes for it, I smeared the thermal compound around a little and clipped the SK6 back on. I hit the power button and was greeted by 'AMD Athlon 1200 Mhz'........... ' 2 Processor(s) detected'.

My point: Maybe the Palomino's on die thermistor doesn't catch heatsink removal under heavy load in time. I cry a single tear. It *does* catch missing heatsink at startup, at least in my real world test. I'd also estimate the odds of (heatsink commits Hari-kiri and flies off towards your graphics card) versus (heatsink gets knocked loose by your klutzy self while moving the PC) to be somewhere in the neighborhood of 1000 to 1.


Final idea:

Don't jump up and down on your PC or sling it around the room while on, and your heatsink should stay on, and your CPU (not to mention the GF3 that heatsink would land on) will thank you.

Also, heatsinks with extremely sh!tty clips probably aren't a good idea, either. I can pick them out of a lineup with my naked eye. >>



So maybe it does work with the correct motherboard. BTW, what MB was Tom using?

 

[ST4RCUTTER]

If it does turn out that Tom's site used a non-Palomino supporting motherboard, his credibility as a hardware reviewer must be put to serious question... :disgust: