Is Intel telling the truth about thier processors heat and power REQ?
- Thread starter Zebo
- Start date
Interesting read. wants me to FORCE them to go dothan path rather than keeping on their prescott path.. Dual core prescotts, or dual core dothans... I would rather se a 2.4GHz dual core dothan which will be just as fast and less heat..
No offense, but it's ZDNet.com
I'd rather even THG did a test than ZDNet numbers.
Everyone know Prescott's put out more heat though, hardly a shock, and given the temps they reach, the amount of extra heat isn't too suprising either.
What I love the most is the 1x512MB stick for AMD tests, and 2x256MB sticks for Intel tests.
And they don't list the power supply, so what are the chances of using different PSU's too?
hmm....
maybe i don't understand this but how does a 350w power supply vs. a 400w power supply effect the wattage output by a cpu?
does it matter?
i understand for overclocking it would.. but for running stock tests???
with the ram thing...
i would think they were running 2 x 256 on the intel setup to take advantage of dual channel while using only a 1 x 512 due to the lack of dual channel on the amd board...
if this were an overclocking test, then this may have unfair because overclocking with 2 sticks is harder than one... but as stated above... these were all stock tests...
- Aug 22, 2001
- 32,112
- 32,672
- 146
Viditor brought this up a while back. He linked to this article which in turn links to others, that show Intel is not providing accurate TDP figures. Meanwhile AMD seems to be providing higher than actual TDP for most their CPUs=erring on the side of caution?
zephyrprime
Diamond Member
- Feb 18, 2001
- 7,512
- 2
- 81
Power supply capacity wouldn't have a big effect but some power supplies are more efficient than others.
And the different ram configurations matter since higher density chips are sometimes of newer process generations that use less power at a given speed. And even if the ram is of the same process generation, having more IO interfaces would increase power consumption. More drive current would also be needed for more dimms.
Thats hardly going to make that much of a difference. Having seen my prescott drop the 12v line by .5v underload I can well believe it. The same PSU had no such problems with 2 x amd mobile@ 2.4ghz.
In terms of PSU's, Enermax's can be 80% efficient (according to SPCR, IIRC) but the ATX spec calls for something like 50%~70% depending on load situations (lower load should have higher effciency, IIRC).
50% vs 80% efficiency is a lot of difference (comparing worst to best scenario).
I was also just saying, because they don't list everything they used, and they also varied some parts, how can we say their test is reliable?
If they vary RAM in the two systems, something they did NOT need to do (RAM is a component which is the same for both, only mobo and processor should differ), then what's to say there were not other differences.
It makes me doubt the reliability and consistency, and consequently the whole article if they cannot do the simple things like keep the setups as similar as possible.
Is ZDnet telling the truth?
They show 60w more for Prescott vs Northwood CPU's.
This shows just a 20w difference between a Northwood and Prescott system.
Who to believe, when ZDnet have already managed different RAM between their conficurations, and do not state what was running when the systems were tested.
Under idle, the german site gives a difference of NW vs P as 30w. Still nothing near the sensational 60w of ZDnet.
So, ZDnet telling a story, or Zdnet reporting something vaguely true? I tend to believe the German site myself. What they show is shown to be true (more heat from Prescott systems), but it's not sensationalist, unlike ZDnet. They also seem to have a standard test setup, unlike ZDnet.
In a world where there is no standard for TDP, and there is protection for overheating processors, as long as Intel states how they made their measurements, I see no problem with what they list as TDP.
Like with most things in computers, to take things at face value is, to be honest, stupid.
And with Celeron's as a history, can anyone be suprised at Intel?
ZDnet definately seems to be attempting sensationalism though, from what I see.
They show 60w more for Prescott vs Northwood CPU's.
This shows just a 20w difference between a Northwood and Prescott system.
Who to believe, when ZDnet have already managed different RAM between their conficurations, and do not state what was running when the systems were tested.
Under idle, the german site gives a difference of NW vs P as 30w. Still nothing near the sensational 60w of ZDnet.
So, ZDnet telling a story, or Zdnet reporting something vaguely true? I tend to believe the German site myself. What they show is shown to be true (more heat from Prescott systems), but it's not sensationalist, unlike ZDnet. They also seem to have a standard test setup, unlike ZDnet.
In a world where there is no standard for TDP, and there is protection for overheating processors, as long as Intel states how they made their measurements, I see no problem with what they list as TDP.
Like with most things in computers, to take things at face value is, to be honest, stupid.
And with Celeron's as a history, can anyone be suprised at Intel?
ZDnet definately seems to be attempting sensationalism though, from what I see.
First, there is no TDP standard, so it is up to each company to define their own. As long as they document it and are consisent, then I see no problem with the two companies doing their own thing.
Second, from each company's documentation, we see that AMD takes the max power of their fastest chip in the series, and then uses that as the value for all chips in the series (not sure what happens if they release a faster chip much later than the specs are published though). Sometimes you'll find that when they make a new "low power" version, not only do they lower the voltage but they also make the fastest chip in that series a bit slower than the fastest chip in the "high power" series. This way they can quote a much lower number for the "low power" series than they otherwise could. However, it is safe to say that the AMD ratings are the max (or at least extremely close to the max) that the chip will produce.
Intel on the other had uses more of an "average" number for their TDP, and they set each rating based on each specific processor. This makes their number more useful in some cases, and less useful in others. It all depends on what you want to know (but with the Voltage and Current ratings, you can easily calculate the max number that AMD gives you, and therefore Intel's method gives you a bit more info when you can dig up those V and I ratings).
Finally, power supplies become more efficient as they become more fully loaded. Therefore, running a very large power supply on a small system is less efficient than running a small power supply on a small system (unless there are great differences in the overall efficiency of the supply). In either case though, the power supply should be exhausting air right out of the computer and should not affect the overall temps too much.
-D'oh!
Second, from each company's documentation, we see that AMD takes the max power of their fastest chip in the series, and then uses that as the value for all chips in the series (not sure what happens if they release a faster chip much later than the specs are published though). Sometimes you'll find that when they make a new "low power" version, not only do they lower the voltage but they also make the fastest chip in that series a bit slower than the fastest chip in the "high power" series. This way they can quote a much lower number for the "low power" series than they otherwise could. However, it is safe to say that the AMD ratings are the max (or at least extremely close to the max) that the chip will produce.
Intel on the other had uses more of an "average" number for their TDP, and they set each rating based on each specific processor. This makes their number more useful in some cases, and less useful in others. It all depends on what you want to know (but with the Voltage and Current ratings, you can easily calculate the max number that AMD gives you, and therefore Intel's method gives you a bit more info when you can dig up those V and I ratings).
Finally, power supplies become more efficient as they become more fully loaded. Therefore, running a very large power supply on a small system is less efficient than running a small power supply on a small system (unless there are great differences in the overall efficiency of the supply). In either case though, the power supply should be exhausting air right out of the computer and should not affect the overall temps too much.
-D'oh!
RalfHutter
Diamond Member
- Dec 29, 2000
- 3,202
- 0
- 76
Lonyo -
Well, they do and they don't...
They never really test TDP at Max (that we know of), and the average that they DO publish is based on a software suite that they won't release...
Well, they do and they don't...
They never really test TDP at Max (that we know of), and the average that they DO publish is based on a software suite that they won't release...
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter