Slashdot Mirror
AMD And THG update
Mhrmnhrm writes "In the interests of responsible journalism, the gang over at Tom's Hardware has developed this article in the interests of setting the record straight about their original AMD burn-out video, and the new release (possibly from AMD) of this past week. It would seem that BOTH videos are correct, and that the question of whether or not somebody is hiding something depends entirely upon your own point of view."
"Nevertheless, it must be said that the facts as shown in the video remain plausible and can be reproduced at any time. At the moment, none of the mainboard manufacturers offer a solution to this. Even highly reputable companies such as Asus and Gigabyte offer no support for this particular problem. "
Cruise TT
So both articles are right adn wrong?
Ummm... like, read the articles dude.
Both are right, THW is more right, but neither are really wrong. TomsHardware used equipment that is currently available to people today. AMD responded by correcting the problem, and then making a new demonstration based on the corrections. So now, going foward, new motherboards are likely to be based on the corrections.
They made an indirect statement about the video.
The board used in the AMD video was modified with a MAXIM health monitor chip that shut down the power supply when the core reaches 85 degrees celsius.
I am sort of disappointed in AMD for not telling us that the board was in fact modified in the video, and that new motherboards don't have this sort of feature.
Toms hardware originaly covered this. Whilst the majority of problems you will experience is a failed fan and therefore a slower temperature rise. The issue of whole heatsinks falling off is becoming more of an issue as CPU's get faster they get hotter, hotter means bigger headsink/fan systems. Bigger means *heavier*, Heavier means the little plasic bits that the sink is clipped to can break!
-- Vagnerr - (www.vagnerr.com) Never attribute to malice that which can be adequately explained by stupidity.
AMD zone also has an update on this story. The most interesting thing (to me) is that they have a 2nd video that show the P4 shuts down. With possible data loss like the modified XP tom was shown in munich.
There is no stability problems for the AMD CPUs. RDRAM is prohibitively expensive. SSE2 is hardly supported anywhere - Hammer will support them. DDR has much better bandwidth utilisation specs than RDRAM, it can use 70% of the available bandwidth whilst RDRAM can only use 45% of the theoretical bandwidth.
Wooo, a 10% overclock. That is nothing to should about. I would worry about the cooling mechanism that has to be bolted onto the board, bending the board in the process. Looks at VansHardware.com for some more information about the problems with the Intel P4 and chipsets.
Actually, with the proper chipset, the thermal diode can be used to throttle the speed. Go look at the AMD 766 register programming manual, section 4.6.1.4 and register C3A50:TTH_EN and C3A50:TTH_RATIO. The southbridge accepts a signal from a chip like the MAX6512 and will throttle the CPU when it asserts. But only if the motherboard has this hardware AND the firmware configures the southbridge to do so
First of all, there is nothing wrong with AMD processors and never has been in terms of design and engineering (certainly from K5 upwards). Intel, however, have had to recall their chips on at least one occasion, unlike AMD.
It is true that once upon a time some chipsets (particularly VIA) for AMD processors were unstable. But this is simply not the case any more.
Before anyone thinks of getting a P4, they should check this out:
http://www.emulators.com/pentium4.htm
it shows how badly Intel broke the P4, and why it's so much worse both in terms of value for money and speed than the new Palomino core Athlons. And yes, this far outweighs the benefits of SSE2 instructions.
Yeah, that explains all the Pentium 4 ads that were on his page when he did the original article.
You probably don't want to change the clock speed in any way - you may want to HLT the cpu for a given %-age of time. There should be Suspend-Pins on the CPU that allow for such things to be controlled from the outside.
That's nice, but better engineering hasn't translated into significantly better performance, has it?
I Dont know of any hardware site that says that an AMD cpu + some AMD or VIA motherboard can even approach the stability of Intel CPU+chipset.
Even if that is true (and that's certainly debateable) I can tell you first-hand that all of my Athlon systems, and my friends Athlon systems as well (including my friends Athlon 1.4 GHz which currently acts as his main server) is rock-solid stable. At least as stable as any Intel platform. On Windows 2000/XP it has only hard-crashed a handful of times ... as far as i'm concerned, the superior stability of the Intel platform is a myth.
My current box is a P4 2GHz (socket 478), Intel 850MV motherboard (tho an Asus P4T is probably just as good), 512Mb PC800 DRDRAM. It runs Linux and FreeBSD very well.
My Athlon runs Linux just as well as any Intel chip. I'd take my AthlonXP 1.4 GHz over your system any day of the week.
I'd urge Linux users not to dismiss Intel chips out of hand. You do pay for quality. Remeber these chips have SSE2 instructions (which even XPs dont have and probably never will), the thermal diode *IS* a kill switch here -- you will not burn them up, much greater memory bandwith than with DDR, they overclock decently (you can usually get 200MHz above spec), and benchmark very well.
You pay for quality? Well, in some cases yes. But in my opinion if you buy an Intel platform these days you are throwing your money away. Let's pick apart the issues you've stated, shall we?
1. SSE2 instructions. A nice boost to performance, but can you name anything that actually supports them (hell, can you name anything which supports SSE?) which isn't called Quake 3? Very few applications support SSE2 or even the years-old vanilla SSE instructions. By the way, the Hammer line of CPU's will support SSE2 instructions, so it isn't totally out of the question that future Athlons may support them as well, though i'm wondering why they would bother if no developers optimize their applications for it.
2. Heat issues. I'll give you that one, at least. If you want to pay twice the amount of cash for better thermal protection (which will eventually be solved on the AMD platform in the future) by all means. Like others have pointed out, it is pretty rare that a HSF will fail or fall off. I've personally never had it happen, but that doesn't mean it never does.
3. Greater memory bandwidth. Is this really relevant if it doesn't really translate into better performance in most situations? It's like your previous comment about the Intel platforms superior engineering -- what does it really matter if real-world performance isn't as good as its nearest competitors offering?
4. Overclockability. 200 MHz above spec? That's probably the equivalent of a 75 MHz increase in clock speed on an Athlon CPU. Sorry, I don't believe that's worth paying the extra money for.
Definitely go with i850 and RDRAM, the 845 is meant for low-end situations and doesnt perform nearly as well.
Of course not. And do not settle for anything less than PC800 either because it will drastically cripple the performance.
The bottom line for me is stability -- I dont think AMD can match Intel CPUs and chipset for that. Look in the Linux kernel source code -- VIA chipsets have many PCI quirks that Intel boards just dont have.
It can match the Intel platform in terms of stability, and it does. I am not a programmer, but any quirkiness with the Via platform hasnt' affected my experience with the Athlon platform in any way, shape or form whether I am using Windows, Linux or whatever else.
-- Jim
On Ace's Hardware, this subject is discussed thoroughly the last days. An interesting article can be found here. Some thoughts from this article:
- the component needed for proper protection of the cpu costs $0.85 (in quantities of 1K).
- "Obviously, Siemens used an external temperature probe and tried to pass it off as using the internal diode."
- The reaction time of the thermal diode is certainly not 1 degree/second: "At our worst case rise of 300C/sec, that translates to a response time of less than 1ms. No way would this result in a fried CPU if power off is immediate upon the signal occurring."
For a matter of fact, an engineer told a friend of mine it would be very difficult, if not impossible, to develop a thermal diode with such characteristics (1 degree/sec response time).
You can find the main thread of this article here.
A dead fan will trap heat in the space between heatsink fans and effectively turn the heatsink into a heater. It will heat up your chip.
This is actually a specific point of failure that notebook manufacturers try to avoid. Rather than placing a fan ontop of a heatsink, they try to induce airflow across a heatsink. Thus if it fails, at least you aren't heating up your part, you just aren't providing much airflow.