Slashdot Mirror
AMD's "Frantic Price Cuts" May Pressure Intel
kog777 writes in with news of a Needham analyst report alerting their clients to a possible price war between AMD and Intel. Analyst Y. Edwin Mok notes that AMD has cut its prices three times in three weeks. He says that Dell has been playing off the two chipmakers against one another to drive costs down. He suggests that bargain-hunting clients avoid both AMD and Intel stock for now. As an aside, Mok notes that so far Vista is not causing a spike in demand for chips. This story hasn't been picked up very widely; other coverage is at Seeking Alpha.
shares of AMD rose 3.17 percent, or 46 cents, to $3.17
:P
Maybe he should check his math processor
Monstar L
to see how many suppliers they can drive out of business before they drive themselves out of business.
Vista aint creating any spikes in purchases since the people that usually upgrade are the Gamer and we all know that Nvidia has serious driver issues, not sure if that affects ATI as well but as far as i know i will not buy a new sata drive and install vista anytime soon just to get my framerate lowered.
--A great company said "framerate is life"--
...is totally OT, but it's where he says that a seasonal dip is occurring in PC sales in spite of the release of Vista, which is not causing a rise. In other words, people are either not buying Vista, or are successfully (?) running it on their existing computers. I suspect it's more the former, since Vista is reputed to run slowly on even the latest equipment.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
It is the utterly lackluster performance of older hard drives and less than 2 gigs of memory that is killing most of the Vista installs I have seen. 5400 rpm drives are constantly seeking even for the most mundane tasks in Vista. A clean install of XP x64 pro on my old AMD 3000 1 gig ran gloriously for 2 years. Tried the Ultimate version last weekend on the same machine. Besides the Nvidia driver problem in Vista I have come up against a brick wall trying to get my old PCI IDE card working.
An Education is the Font of All Liberty
I wonder how much of the price cuts have to do with the fab costs. Intel has pretty much completely transitioned to 65nm fabs for their new chips, while AMD is still in the middle of the transition and just launched retail 65nm chips at the beginning of the year. Perhaps AMD is dropping their prices to get rid of all of their 90nm chips, and/or they are getting good deals from the 90nm fabs as they drop prices to compete with the 65nm fabs (I believe AMD outsources a lot of their fab work.)
Isn't that what competing companies supposed to do? This has been happening for a long time. During the 1990s AMD was selling their chips for cheaper prices then Intel. Then around the Early 2000s AMD finally got a good reputation and better then Intel's so Its prices went up (Increase in demand). Now with Intel Core Chips which perform very good and are relatively inexpensive Intel Chips are getting more demand. So in order to keeps AMDs line selling they will Lower the prices on their chips. Now Intel will choose wether the demand for their chips at there prices will still work with the market or they will need to lower the chip prices. Now a word of waning about Price Wars, The consumer usually wins at first then they they slowly get screwed as the war lingers. Lower Price Chips means less R&D and Less Good Improvements and More Quick Patches and Fixes. So quality will drop. I know people want to think of a perfect world where we get Top Quality Products at Discount Products, But in reality that is not the case, I am sorry but the $400 Dell Laptop is Lower Quality then the $2500 MacBook Pro. There may be a feature that is better but overall you are getting less.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
Please, mr dell, start a price war between RAM manufacturers next! I live in perpetual obsolescence thanks to the dramatic cost of DDR and DDR2! Won't someone think of the child processes!
What are pre-prepared , planned advertisements, allowed? ...
This is MORE than hollow
There has been a "possible price war" for the better part of a year with these two companies. AMD has cut prices a couple times now and Intel has responded similar moves and with new chip technology that proved to be a large, significant advancement. I am not sure what we are looking for to confirm a price war, but as far as I can tell, these companies have been going at it for some time now. With the industry changing every year it seems, it might be difficult to classify this as a price war. Is this simply strong competition in a large market that effects both business and individual consumers?
For those looking for a "price war" you do not need a confirmation. It has been going on for over 7 years now. This article dated Feb 28, 2000 details price cuts by AMD in response to Intel cuts. Then, look who is still at it 6 years later - Price Wars Intensify as Intel Slashes Chip Prices. It is a seesaw game that, hopefully, will not end any time soon. The more they go at it, the more the consumer stands to gain.
Now a related question... Do you think consumer demand or competition with each other is causing the rapid advancement in chip design and architecture.
Invexi - a Phoenix, AZ based web design and web development company.
the analyst industry is quite amazing - all you have to do is repackage common knowlege as something special and people will pay you for it!
seriously - AMD and Intel are normally out-of-phase in product intros. it's been this way for many years, so we have to assume it's deliberate. Intel made a major improvement by souping up the Pentium-M line into Core2, and has gained a nice lead in some, even most, benchmarks. mainly due to some fairly narrow improvements that AMD hasn't yet answered, like 1-cycle throughput SIMD operations. AMD's current offerings are largely unchanged since the original Opteron intro (2003?), except for smallish tweaks like bigger caches, faster memory, doubled cores. AMD still does well for applications which are sensitive to memory bandwidth, for instance - part of the original technological jump of the K8.
AMD is about to introduce their response to Core2, and it seems quite promising based on the hints AMD has provided. Intel's not in a position to respond immediately, since 45nm production is some way off, and it (Penryn) will apparently be just a shrink of the current Core2 design.
in short, it's only sensible, sound business practice for AMD to drop the prices of their mature, high-yielding, partly-outsourced half-gen-old products. performance is still competitive with Intel's products - at a time when Intel's yields are probably not yet mature. in a way, this sets the stage for AMD to introduce its next-gen parts at a more comfortable margin.
There's no sudden demand for CPUS for Vista for 2 reasons. 1) people aren't rushing to buy vista, and 2) if you can run w2k or xp on a system you'll almost certainly be able to run Vista on it with no changes. If you need a change, it's probably a slightly more recent graphics card, or another 512mb of ram.
Price wars (or marginal return on investment) are always going on for the products of fabs still producing older technology. It is just more noticeable when the old technology is still highly desirable. From a business view point, it is desirable to get every last dollar return for the multi-billion dollar investment made in the original technology as long as the marginal cost of production is less than the revenue obtained. When the curve inverts then the fabs get taken off line, or upgraded. AMD has next to nothing to lose on the price drop of outsourced fab product except cannibalized sales from the new 65nm. Since supplies are limited and selling, cannibalized sales has to be a zero quantity at present.
ATI has not released its upcoming DX10 graphics card yet, so the only available DX10 card is the Nvidia 8800 with lousy drivers. ;-)
Vista drivers for older (DX9) cards also suck, both for Nvidia and ATI. But for DX9 you can stay with XP anyway
C - the footgun of programming languages
Why keep bashing Microsoft, calling it evil etc? It is the consumers who should wake up. Let us say I give these companies big discount so that they can "make the numbers" for this quarter. But that would force them to give all their data to me and they have to pay me every quarter to access their own data. In a rational world, I would be laughed out of the business meeting in no time. But that is precisely what is happening in sales meetings between MS and the fortune 500 companies.
When it comes to the chips Dell is able to play AMD against Intel. It is in Dell's own interest to have a competition in OS/Office market so that it can play one against another and reduce the cost of computing to its customers so that it can sell more. But Dell buries alternatives deep, makes it difficult to buy the alternatives. Why? Why? Isn't there anyone who can break through the non-disclosure agreements and the secrecy and shed light on why corporations are acting seemingly irrationally? Sunlight is the best disinfectent.
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
http://www.smarthousenews.com.au/Computing/Industr y?Article=/Computing/Industry/B7M4N2K4
I'm sick of seeing these chips at outrageous prices. Who other than the most rabid gamer is going to be willing to fork over $500 - $1000 (US) for the latest processor? The worst part of it is, the processors are starting out so overpriced, that when they start to drop, it takes over two YEARS before they become reasonable. I don't know about the rest of Slashdot, but I'd like to be able to get something less than 4 generations old at a decent price point.
It used to be that you would spend, AT MOST, about $100 - $200 (US) for the latest AMD offering (usually much less, under $75.00 US). Intel was never considered for gamers or home-builders because they were overpriced and underpowered. Lately AMD has been pulling the same crap that Intel was pulling back in the 90's. End result? We now have two chip makers, both with overpriced CPU's, trying to compete. It's about time there was a price war! They are using smaller and smaller die sizes, and are thusly getting more and more out of each silicon wafer. The damn things should be getting CHEAPER not exorbitantly more expensive!
Bring back the sub-$200.00 bleeding edge CPU. It's well past time.
Official Heretic from the "Church of Global Warming". Proven right thanks to whistle blowers. AGW = Flat Earth Theory
I was really rooting for the latest Athlon/Turion 64 X2's, but my recent laptop has a Core 2 Duo. AMD was ahead for a while, but they're playing catch-up again, and I'm not really surprised if their prices reflect this. Intel has been 65nm for over a year now, and it shows in power usage at the very least. I'll admit this is an interesting war to witness, however.
Read: Rabbit Rue - Free serial nove
Careful with the "power usage" statement. While Intel certainly has lower power consumption in their Core2Duo processors, that's only in relation to the power-hungry Prescott-based processors. Intel's PR department has made a lot of hay out of their decreased power consumption. The fact of the matter, however, is that Core 2 Duo processors at 65nm now have about the same power consumption as their Athlon 64 X2 counterparts at 90nm--about 65W.
I highly recommend taking a look at processor electrical specifications. And keep in mind that Intel's power figures are more optimistic ("typical") than AMD's ("max").
Help find a cure for cancer. Join the [H]orde
Us gutter IT guys can grab up Sempron 64 3000+ chips for absolute dirt right now. I built my daughter a computer to play Quake4 on for less than $150.00 and she can play it at high detail settings with that low end Geforce 6800GT on that cheap motherboard and really slow processor like that.
Hell that setup has the power to record 4 NTSC tv channels and 1 HD channel at the same time. Makes a great cheap MythTV backend recorder.
Do not look at laser with remaining good eye.
Holy jebus! What do you want. - On sale today, Canadian dollars!! so it is actually well under $200
AMD Athlon 64 X2 4400+ Dual Core Processor Socket AM2 Brisbane 2.3GHZ 2X512KB 65NM 65W Retail Box $198.98
None of them can see the clouds; The polished wings don't care.
Core2Duo may have lower power use but chipsets still use more.
Intel's new exons may use less power for the cpus but when you add the chip set and the FB-Dimms it is about the same as amd cpus + chipset + ddr ecc ram.
Both articles linked mentioned that Vista just wasn't pushing PC sales as anticpated, but neither article shed much light on what set of numbers were used to determine what this push should be.
.." (well maybe not that simple, but you get the point).
.. I'm not a stock holder of either, or complaining :)
Is this a forecast that MS puts out for each release, or is it determined by historical data? Since there's nothing really historical about Vista's CPU demands for the average user (well, not much really), how the heck did they come up with any kind of number?
This would (I guess) have to be MS saying "This is what we expect people to do with it, this is what we expect businesses to do with it, and this is what we expect CPU demands will be in both cases, hence here's the data to forecast what you'll be selling, we expect to push xxxx copies per day
Another way of looking at this would then be (speaking as Intel or AMD):
"Microsoft sold us a load of fud, we need to keep focused on attacking the virtualization and server market, and the other guy already has a strong foot hold there." (as either could say that about the other).
So in short, it looks like both AMD and Intel learned nothing from Enron's "virtual asset" mindset, which was counting on money that wasn't in the bank yet, but you were *pretty* sure would be there. Typical, I'd say unless I'm way off on how these predictions come into play?
I also saw no data in either article about growth either company made which they now need to find another way of paying for, but I guess that's not going to be availble to sift through for a while.
If I were either company, I'd be treating Vista like Bob until some longer range (real) predictions could be made. But hey, cheap servers
AMD market's a typical "optimistic" spec. Where have you been? AMD's marketing trickery w.r.t. power numbers has been exposed on /. for some time. Both sides have their own "optimal" conditions.
1 20864
I refer you to this thread to see how AMD markets power. They use barrels of hype.
http://slashdot.org/comments.pl?sid=210098&cid=17
https://www.accountkiller.com/removal-requested
Just to expand on the point I think you're making, companies generally go out of business when the profit they generate is less than the amount of money that a similar amount of capital could make, if invested elsewhere.
I.e., if your semiconductor business, which has physical and cash assets of $1B USD, is generating less than $1B invested directly in the stock market, then it probably doesn't make sense to keep going, unless you expect that you can turn the company around and get it more profitable.
In real life, many companies shut down (or get shut down by their investors) when the price per share * shares outstanding is less than the net real assets of the corporation. That's basically saying that the stock value, which is sort of a prediction of the company's future performance and overall "market value," is worth less than the assets that it's using. Thus, it's liquidated. (However, there were some exceptions to this; there were companies in the 1980s that were basically ravaged by "raider" investors and sold off piecemeal, who probably could have been turned around under better management.)
"Ladies and gentlemen, my killbot features Lotus Notes and a machine gun. It is the finest available."
Yea, I did RTFA, but I still have no idea what these 3 price cuts written about are or what I should now expect to pay for various AMD cpus. So much hype, so little information.
I'm an American. I love this country and the freedoms that we used to have.
Info here.
Intel transfer the difficult from Hadware to software, for get more power, programmer need more technology. -- chinaitn
Amd is at it's lowest for like 6 months, it had reached a $42 maximum and plunged to $14.40 three days ago. Now is floating around $15... The only way is up ? I guess so.
Most people considering Vista on their existing machine will be considering a RAM upgrade not a CPU one as their first priority. It is reassuring that after Xmas 2006 RAM prices have been dropping (as I and other predicted). This is for two main reasons: 1) There was a spurt in RAM buying my PC makers to fulfil holiday new PC orders which is now quiet 2) Vista adoption has not been very high so far.
I wonder when the Vista version of LitePC http://www.litepc.com/ will be available. Once one can successfully remove DRM and other cruft from Vista at the click of a button, it should become more popular.
Excuse me, but please get off my Pennisetum Clandestinum, eh!
And your proof is a link to your own post which makes the same unsupported accusation? Sweet, I'm convinced.
Anyway, we're not talking about marketing about which uses less power in a common day's usage and will cost you less on your electric bill. We're talking about how their chips are rated for power consumption. This is the TDP, Total Design Power, the number that computer makers will need to use to design their cooling solutions. You can't market your way around TDP; if you try then the cooling solution will be inadequate and the computer will overheat.
The difference in TDP between Intel and AMD is that AMD gives MAX power, which isn't actually achievable in real software, while Intel gives TYPICAL power, which is a number that is around a standard deviation or so away from the average in the bell curve of power consumption for various programs. They use clock throttling to guarantee that no app will ever use more power than that. Both are equally valid as measures of TDP, since in either case it is simply a promise to the HSF designer that the chip will not use more than the stated power. The catch is that Intel's number comes with a caveat that some software may not run at maximum speed because clock throttling is required to keep the power down. This was noticeable with Pentium 4; I haven't heard of Core 2 having this problem, which makes sense because it is in general less power hungry than P4 and probably has a much nicer looking power curve.
The enemies of Democracy are
First of all, you don't even understand the terms:
2 ,00.asp
e _papers_and_tech_docs/30430.pdf
TDP = Thermal Design Power. Not "total". Get your facts right.
You are totally wrong about TDP: TDP is entirely a marketing construct. It is based on a typical scenario. They choose a point arbitrarily that doesn't cause too much perf. loss from PowerNow!, but still reduces thermal solution cost. It's market pressure.
Second: RTFA.
http://www.extremetech.com/article2/0,1697,206825
"AMD's argument goes like this: Modern desktop and notebook processors constantly scale up and down between full speed and an idle state, which AMD has branded "Cool 'n' Quiet". At a given time, pushed to full load by an application, AMD's chips run hotter and consume more power. But across a typical computing day--where a user might check his email or surf the Web--the processor idles more often then not. At idle, AMD's 90nm Athlon 64 X2 consumes 7.5 watts. A 35-watt, 65-nm chip will idle at 3.8 watts, AMD said. By comparison, the 65nm Core 2 Duo idles at 14.3 watts."
Marketing drones on the prowl. AMD took Intel's spec, changed what is considered a "typical" day, and said, "Hey look, we win!!!" I'm sure Intel will respond in kind.
AMD does not give MAX, if they did, they wouldn't be able to sell their products to any platform vendor. No one has spec'd MAX since pre 1GHz days.
Repeat after me: AMD DOES NOT SPEC MAX POWER. AMD DOES NOT SPEC MAX POWER. AMD DOES NOT SPEC MAX POWER>
Don't believe me? Here is their POWER THERMAL DATA SHEET, and it only specs TDP.
http://www.amd.com/us-en/assets/content_type/whit
PowerNow! defines TDP as the max power of the CPU under TDP conditions. There's the rib. TDP conditions (see page 83 of aforementioned documents) are not explained, "please see your FAE for details." But we already know the answer to that.
Intel and AMD both spec to their own versions of what TDP means. Been that way since the 100W+ CPU appeared on the market.
https://www.accountkiller.com/removal-requested
Get SCSI and those problems go away.
It is one of the key reasons I can run new applications and games as well as many 'hard core' gamers with a 4 year old box. Yes, the video card is upgraded.
The same people who spends days trying to get every last clock cycle, and buy a 300 dollar box complain about price.
The Kruger Dunning explains most post on
TDP = Thermal Design Power. Not "total". Get your facts right.
Big whoop, an acronym wrong. Rote memory vs understanding? I'll take understanding, which you fail.
Marketing drones on the prowl. AMD took Intel's spec, changed what is considered a "typical" day, and said, "Hey look, we win!!!" I'm sure Intel will respond in kind.
I don't care about marketing spiels about how it costs less to own Brand X for Purpose Y. The article talks about situations where the processor is idle and consuming less power, in processor talk running at a higher P-state. It sounds like you think this "typical usage scenario" where the computer is idle most of the time has something to do with TDP. That has nothing to do with TDP. TDP is about the physical requirements to transfer power away from the silicon. If they bullshit their TDP numbers, then the HSF the OEMs use will be inadequate, and they will absolutely notice this in the form of chips failing from overheating.
Do you understand? You can't market-speak your way around thermodynamics.
No one has spec'd MAX since pre 1GHz days.
Utter bullshit. It was impossible for AMD to spec anything but max power until late in the life of K7 when they implemented thermal sensors and motherboard-impl. Intel specced max power until Pentium 4. The reason they do this is because unless you can guarantee that using the lowest P-state setting and running any arbitrary program that the TDP will not be exceeded then the cooling solution may fail. The only reason Intel started quoting a lower number than max power was because they could guarantee with their P4 clock throttling thermal sensor that the chip would never go above that number. AMD couldn't actually do this; they could only cause the chip to shut down if it exceeded the stated TDP number. That's not desireable, so they continued to have to issue TDP numbers for maximum power.
Did you even see how the thermal spec only listed TDP for a specific P-state? PowerNow! is how those P-states are managed; TDP is for one P-state, and the TDP for the lowest P-state is without any power saving chip features in use, because that's what it means. PowerNow! is not a method of reducing TDP; it cannot, because TDP is not defined that way.
Repeat after me: AMD SPECS MAX POWER BECAUSE THEY MUST.
PowerNow! defines TDP as the max power of the CPU under TDP conditions. There's the rib. TDP conditions (see page 83 of aforementioned documents) are not explained, "please see your FAE for details." But we already know the answer to that.
Uh, yeah, if you don't know what IDD_max is then I guess it's unexplained, then maybe you should see your FAE for details and I'm sure they'd be happy to explain it to you. I'll save you the trouble: maximum current draw. It is when the chip is sinking the maximum amount of current due to having the maximum number of transistors enabled. Being able to switch to a different P-state in the following few microseconds doesn't change that number. The only thing AMD may not be disclosing is whatever worst-case code sequence they use to determine the maximum number of transistors that can be enabled at once.
I can't believe you have confused a marketing press release about cost-of-ownership discussions of power consumption with an actual engineering document used for constructing cooling solutions that must actually work in situations nothing like the "typical" scenario of the marketing BS. If you want to talk shit about AMD's marketing, go ahead they're terrible and evil like basically all marketing, but don't act like that's the same as what engineers use to do their job.
The enemies of Democracy are
I highly recommend taking a look at processor electrical specifications. And keep in mind that Intel's power figures are more optimistic ("typical") than AMD's ("max").
I don't think we should trust the power "specs" from Intel, AMD, or any source that only focuses on the CPU. If you want to compare power consumption between Core 2 Duo and Athlon 64 X2, I think you should look at total system power consumption with whatever chipsets you plan to use.For example, on the "mainstream" P965 and nForce 590 chipsets, the Core 2 Duo systems consume significantly less power than the Athlon 64 x2 counterparts. Note that the nForce 590 chipset is power-hungry, but NVIDIA chipsets are more popular than ATI chipsets. I cannot find Core 2 Duo vs Athlon 64 x2 benchmarks that use a more power-friendly ATI chipset (I'm sure I've seen one before), but I guess that shows how unpopular ATI chipsets are. I'm sure AMD acquisition will change things.
TO START
PRESS ANY KEY
Where's the 'ANY' key? I see Esk, Kitarl, and Pig-Up...
Your link is to TDPs. Which are not representative for normal power usage, and hence pretty useless. You need to look at figures at http://www.silentpcreview.com/ or a similar place. Last I looked, the desktop Core 2 Duo is more efficient under load with Athlon 64 X2 being more efficient in idle (apparently because Intel's EIST is not really working).
TDP is about the physical requirements to transfer power away from the silicon. If they bullshit their TDP numbers, then the HSF the OEMs use will be inadequate, and they will absolutely notice this in the form of chips failing from overheating.
Half wrong.
If they lowball TDP, the OEM designs an inadequate solution, and both Intel and AMD cpus will throttle more, reducing performance. If the throttle cannot cool the CPU, the catastrophic diode triggers, halting the CPU.
TDP is selected based on performance/cooling cost tradeoff, which is entirely based on the distribution of power of typical applications, as both companies have admitted in print. This can be fudged both for bragging rights, and to reduce OEM thermal solution cost, but it cannot be set so low as to impair performance. Sure, AMD/Intel could spec 5W TDP, the OEM would design a 5W solution, and the CPU would run as fast as a 200 MHz CPU because it throttles all the time. See?
if you don't know what IDD_max is then I guess it's unexplained
IDD_max is y-coordinate right-hand endpoint of the VRM load-line specification. The X coordinate is the Vcc at that Imax, which is a Vmin that affects writability of storage or sequential elements. Instantaneous drops are handled by caps on the board and on the die. If the VRM cannot supply this voltage, the circuits cannot switch. How does this relate to TDP? IDD_max is TDP / Voltage. TDP comes first, then IDD_max is spec'd. There is no need to spec a beefier VRM requirement if the CPU will not exceed a particular power threshold, which is TDP.
Your point is that AMD spec's MAX power and Intel does not, so AMD is more honest. This is factually incorrect, as they both us TDP numbers. Hopefully I've convinced you, as I am clearly neither an AMD or Intel fanboy, I just hate people spreading misinformation.
https://www.accountkiller.com/removal-requested
If AMD hadn't rescued us with open x86-64 (yes, it's open, Intel was allowed to use it without asking AMD for permission. All of Intel's Core 2 Duos use AMD's x86-64) we would all be forced to run the desktop version of the Itanic processor and be married to IA-64 that no other company could use without getting permission/paying royalties to Intel. So, we clearly know what *WOULD* happen if Intel were a monopoly. They tried it already!
If they lowball TDP, the OEM designs an inadequate solution, and both Intel and AMD cpus will throttle more, reducing performance. If the throttle cannot cool the CPU, the catastrophic diode triggers, halting the CPU.
AMD doesn't have thermal clock throttling. They didn't even have an on-chip thermal diode to handle chip kill until K8 -- K7-based designs required a sensor on the motherboard. Clock throttling based on temperature (no OS control) was a feature Intel introduced in the Pentium 4, and which was kept in the change away from Netburst. So I repeat: If AMD under-specs their TDP, then their parts will fail.
TDP is selected based on performance/cooling cost tradeoff, which is entirely based on the distribution of power of typical applications, as both companies have admitted in print. This can be fudged both for bragging rights, and to reduce OEM thermal solution cost, but it cannot be set so low as to impair performance. Sure, AMD/Intel could spec 5W TDP, the OEM would design a 5W solution, and the CPU would run as fast as a 200 MHz CPU because it throttles all the time. See?
Yes TDP is "selected" based on power/performance tradeoffs, but that is nothing like it being a pure marketing decision. Marketing may say that market X has max power envelope Y, but after that it's the physical requirements of the chip that dictate the cross of performance/TDP.
AMD's TDP is based on the distribution of power of applications, but it uses the right-most end of the graph -- like I said, they must do this because they have no way of guaranteeing that the power won't go that high. If the chip -can- use that much power, then they must spec that much power even if marketing doesn't like it. The only performance/cooling cost tradeoff AMD can make is raise/lower VDD and raise/lower the frequency. They could produce a higher clocked part, but it would use more power. To make a 5W part, it would have to actually be 200MHz. This is why dual-core parts that are supposed to fit in the same socket infrastructure (== cooling design, == TDP) are a couple speed grades slower than the single core. If they could "fudge" the numbers so that their dual-core parts didn't lose to their single-core product in single-threaded benchmarks, they would. But they can't. They spec TDP at max power because they must.
Intel has more flexibility, in that they can also count on their thermal clock throttling to guarantee power stays in range and pick a spot in the application power distribution closer to the mean. In practice they pick a spot on the curve a few standard deviations out from the mean, where very few applications go beyond. When clock throttling is engaged, it is at a halving of frequency (actually a 50/50 full speed/completely off duty cycle), so it is very expensive performance-wise to rely on that feature. Chips that throttled a lot would be almost as bad as chips that failed; it would have made more sense to lower the frequency and avoid the throttling. Thus their main lever for adjusting TDP is still voltage/frquency, they just get some extra leeway from their methodology. This was important for Netburst, which had a very long tail to their power curve that they wanted to lop off. When they moved away from Netburst to a design with a narrower power distribution, this became less important.
There is no need to spec a beefier VRM requirement if the CPU will not exceed a particular power threshold, which is TDP.
And if you inadequately spec your voltage regulator then the part will fail. Which again directly contradicts the idea of TDP being a number made up by marketing.
Your point is that AMD spec's MAX power and Intel does not, so AMD is more honest. This is factually incorrect, as they both us TDP numbers.
Ah, see, there's the problem, you're arguing against an assumed but incorrect conclusion. I never said AMD is "more honest", if anything I'm saying both are completely honest, because even Intel guarantees
The enemies of Democracy are
Heh:
Yes TDP is "selected" based on power/performance tradeoffs, but that is nothing like it being a pure marketing decision. Marketing may say that market X has max power envelope Y, but after that it's the physical requirements of the chip that dictate the cross of performance/TDP.
vs.
You're making it sound as though TDP is like AMD's model numbers, completely arbitrary, and nothing could be further from the truth.
I guess we just see the same data completely differently.
https://www.accountkiller.com/removal-requested