-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
Wall Street Journal figures out "emergency edition" P4...
Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
jbond04 is spot on. As a TI employee I see, hear, and read all kinds of crazy stuff. I can't divuldge it and I can't allow it to impact by TXN/NOK/SUNW buy and sell decisions. But I can utilize that info to guide my discussions on the bounds of what is physically possible versus improbable given what I know being already reduced to practice.
Bottom line is the non-semi industry employee can be easily duped by anyone claiming to be "in the know" but for those few people you do trust to be legit and in the biz you should learn to have a little more faith and a lot less uneducated opinion.
Yeah...it wasn't until I started to work at HP that I realized how uneducated I was before. When I first joined these forums, I didn't know an ALU from an FPU. I would take every word a review site said as the unwavering truth, and I would harbor biases of the worst kind. Now, after a few years and a lot of learning, I realize that there are people out there who know more (a lot more) than I do. Respect these people. Basically, until I started work, I didn't know sh*t. It still amazes me how some people on these forums will come out and blatantly tell employees they're lying. Just because you're an EE or a CpE major doesn't mean that you know everything there is to know. I learned way more in 2 months at HP than I did in 1 year of college...there's a lot to be said for real world experience.
Employees are enthusiasts just like everyone else, and if I were to share information or give a strong hint, it's only because I want to help other people to learn and understand more, not to say "my company is better than yours". Wingznut and Sohcan don't make any money if you buy Intel processors; that's not why they debunk rumors or point out misinformation. Instead, they're trying to provide accurate information from the best source possible, the place where they work. Give these guys a break...they're not out to convert you...they're here to learn and teach and talk about the technology they love, not give inaccurate numbers for the thermal output of the P4EE.
First
Lifer
We do. Intel would have to sell the P4EE for a ridiculously small amount of money (say somewhere under a 2.4C) to actually lose money on the P4EE. Now, if WSJ meant to say that Intel wouldn't make as much, that's completely different. But WSJ certainly didn't word it clearly if that was the case, that's for sure.
Of course! Can I divulge any of it? Absolutely no! err, I mean maybe...
If you want to read up on what is publicly available knowledge then see: TI on R&D Requirements for 90nm and Beyond
Once you get past the usual marketing boilerplate look specifically to page 15 and 16 for process and numbers on the USIV. Not that anything has been officially announced but you'll notice comments about 35nm xtors for 90nm node. I work on the 65nm node, all I can say is things will be smaller and faster...maybe (disclaimer, opinions expressed here are mine and not reflective of my employer).
Edit: I can't spell!
Originally posted by: Sohcan
Yes, I do work for Intel; and no, I don't need to have worked on this particular processor to know that 117 watts TDP is out of the question.
You don't need me telling you this, there are more than enough data points from the publically available datasheets for the P4 and the Xeon MP to estimate power consumption.
I'm not about to start posting information I might have from confidential meetings and documents. I linked the Xeon MP datasheet (here's the P4 datasheet), there's more than enough public info you need.
Intel can't lie about power consumption, vendors rely on the datasheets to supply enough power to the processor and provide adequate cooling. If the numbers were intentionally lowered, systems would fail.
Right, im not arguing with you about the numbers, but the info that is posted only allows you to get and estimate of how it MIGHT work! not the way its gonna work upon release! point being that its just an estimate. The actual wattage could be +-5 watts or more, or less.
as for the NV comment, i was j/k, i know the cant lie about that. My point here is that they can market one thing while its still being developed, and then anounce that its another after the release. anyone with marketing sense would do that to create hype around the product! The thing is, you cant go too far with that, cause you could end up shooting your self in the foot like NV did!
Originally posted by: Goose77
Originally posted by: Sohcan
Yes, I do work for Intel; and no, I don't need to have worked on this particular processor to know that 117 watts TDP is out of the question.
You don't need me telling you this, there are more than enough data points from the publically available datasheets for the P4 and the Xeon MP to estimate power consumption.
I'm not about to start posting information I might have from confidential meetings and documents. I linked the Xeon MP datasheet (here's the P4 datasheet), there's more than enough public info you need.
Intel can't lie about power consumption, vendors rely on the datasheets to supply enough power to the processor and provide adequate cooling. If the numbers were intentionally lowered, systems would fail.
Right, im not arguing with you about the numbers, but the info that is posted only allows you to get and estimate of how it MIGHT work! not the way its gonna work upon release! point being that its just an estimate. The actual wattage could be +-5 watts or more, or less.
as for the NV comment, i was j/k, i know the cant lie about that. My point here is that they can market one thing while its still being developed, and then anounce that its another after the release. anyone with marketing sense would do that to create hype around the product! The thing is, you cant go too far with that, cause you could end up shooting your self in the foot like NV did!
Goose77, I'm not sure what you are trying to nit-pick about here. Even after release, the manufacturing processes will produce normal (as in Gaussian) distributions of chips and the "actual" heat dissipation numbers from any given Lot's worth of Si (or batch of wafers) will vary both in average and 6sigma.
One week you might have 3.2GHz P4 EEs with average 100W +/- 10W (6sigma) and the next week you may have a batch of 3.2GHz P4 EEs with average 90W +/- 20W...the specific chip you might purchase will most likely operate at the average heat dissipation (under given load) for that batch of chips but all Intel or any IDM can tell you is what the worst-case power usage will be allowed to pass through some validation program and into the consumer market.
So what does someone telling 117W is too high mean to you? You've got to get beyond your thirst for technically correct semantics and learn to distinguish between someone "estimating the error of the mean (accuracy) for the heat dissipation distribution within some statistical distribution assuming 3 or 6 sigma (precision)" and someone "estimating the most-likely maximum allowed (6sigma) bin for heat dissipation " Both the mean and distribution about the mean will vary from wafer to wafer, fab to fab, and lot to lot within the same stepping. And guess what, not only would a "this is the watts" value be meaningless when the chip is sold to you, but it would be meaningless everyday thereafter as thermal profiles of a chip are also a function of the chips age and usage.
From what I've read here it would appear that most people are trying to clue you in the fact that 99% of the chips will not likely come even close to 110W heat output whereas you appear to want people to believe you are fixated on estimating the global mean-value of all chips to be produced.😕 Stop biting the hands that feed you and let some of this soak in, most of us don't post for practice. 🙂
Edit - Me can't spell or paste good 🙂
Originally posted by: Idontcare
Originally posted by: Goose77
Originally posted by: Sohcan
Yes, I do work for Intel; and no, I don't need to have worked on this particular processor to know that 117 watts TDP is out of the question.
You don't need me telling you this, there are more than enough data points from the publically available datasheets for the P4 and the Xeon MP to estimate power consumption.
I'm not about to start posting information I might have from confidential meetings and documents. I linked the Xeon MP datasheet (here's the P4 datasheet), there's more than enough public info you need.
Intel can't lie about power consumption, vendors rely on the datasheets to supply enough power to the processor and provide adequate cooling. If the numbers were intentionally lowered, systems would fail.
Right, im not arguing with you about the numbers, but the info that is posted only allows you to get and estimate of how it MIGHT work! not the way its gonna work upon release! point being that its just an estimate. The actual wattage could be +-5 watts or more, or less.
as for the NV comment, i was j/k, i know the cant lie about that. My point here is that they can market one thing while its still being developed, and then anounce that its another after the release. anyone with marketing sense would do that to create hype around the product! The thing is, you cant go too far with that, cause you could end up shooting your self in the foot like NV did!
Goose77, I'm not sure what you are trying to nit-pick about here. Even after release, the manufacturing processes will produce normal (as in Gaussian) distributions of chips and the "actual" heat dissipation numbers from any given Lot's worth of Si (or batch of wafers) will vary both in average and 6sigma.
One week you might have 3.2GHz P4 EEs with average 100W +/- 10W (6sigma) and the next week you may have a batch of 3.2GHz P4 EEs with average 90W +/- 20W...the specific chip you might purchase will most likely operate at the average heat dissipation (under given load) for that batch of chips but all Intel or any IDM can tell you is what the worst-case power usage will be allowed to pass through some validation program and into the consumer market.
So what does someone telling 117W is too high mean to you? You've got to get beyond your thirst for technically correct semantics and learn to distinguish between someone "estimating the error of the mean (accuracy) for the heat dissipation distribution within some statistical distribution assuming 3 or 6 sigma (precision)" and someone "estimating the most-likely maximum allowed (6sigma) bin for heat dissipation " Both the mean and distribution about the mean will vary from wafer to wafer, fab to fab, and lot to lot within the same stepping. And guess what, not only would a "this is the watts" value be meaningless when the chip is sold to you, but it would be meaningless everyday thereafter as thermal profiles of a chip are also a function of the chips age and usage.
From what I've read here it would appear that most people are trying to clue you in the fact that 99% of the chips will not likely come even close to 110W heat output whereas you appear to want people to believe you are fixated on estimating the global mean-value of all chips to be produced.😕 Stop biting the hands that feed you and let some of this soak in, most of us don't post for practice. 🙂
Edit - Me can't spell or paste good 🙂
You have a good point here, and im sorry for any frustration that i have caused, just that heat production does matter to those that overclock. The more a proc produces heat the less likely you are to overclock it well. Plus, this is one factor that some people base there purchasing decisions on. I am starting to inclue the amount of heat production in my decision on which proc to buy. The one in my system is hell. In the summer, it get so hot in my room that i cant sleep at night! this is with ac on. The other floors are ice cold and my room is blazin hot! (note: halogen light doesnt help either 😉 ) maybe to joeschmoe it might not really matter, but to fanatices everything matters! 😉
Originally posted by: Goose77
Originally posted by: Idontcare
Originally posted by: Goose77
Originally posted by: Sohcan
Yes, I do work for Intel; and no, I don't need to have worked on this particular processor to know that 117 watts TDP is out of the question.
You don't need me telling you this, there are more than enough data points from the publically available datasheets for the P4 and the Xeon MP to estimate power consumption.
I'm not about to start posting information I might have from confidential meetings and documents. I linked the Xeon MP datasheet (here's the P4 datasheet), there's more than enough public info you need.
Intel can't lie about power consumption, vendors rely on the datasheets to supply enough power to the processor and provide adequate cooling. If the numbers were intentionally lowered, systems would fail.
Right, im not arguing with you about the numbers, but the info that is posted only allows you to get and estimate of how it MIGHT work! not the way its gonna work upon release! point being that its just an estimate. The actual wattage could be +-5 watts or more, or less.
as for the NV comment, i was j/k, i know the cant lie about that. My point here is that they can market one thing while its still being developed, and then anounce that its another after the release. anyone with marketing sense would do that to create hype around the product! The thing is, you cant go too far with that, cause you could end up shooting your self in the foot like NV did!
Goose77, I'm not sure what you are trying to nit-pick about here. Even after release, the manufacturing processes will produce normal (as in Gaussian) distributions of chips and the "actual" heat dissipation numbers from any given Lot's worth of Si (or batch of wafers) will vary both in average and 6sigma.
One week you might have 3.2GHz P4 EEs with average 100W +/- 10W (6sigma) and the next week you may have a batch of 3.2GHz P4 EEs with average 90W +/- 20W...the specific chip you might purchase will most likely operate at the average heat dissipation (under given load) for that batch of chips but all Intel or any IDM can tell you is what the worst-case power usage will be allowed to pass through some validation program and into the consumer market.
So what does someone telling 117W is too high mean to you? You've got to get beyond your thirst for technically correct semantics and learn to distinguish between someone "estimating the error of the mean (accuracy) for the heat dissipation distribution within some statistical distribution assuming 3 or 6 sigma (precision)" and someone "estimating the most-likely maximum allowed (6sigma) bin for heat dissipation " Both the mean and distribution about the mean will vary from wafer to wafer, fab to fab, and lot to lot within the same stepping. And guess what, not only would a "this is the watts" value be meaningless when the chip is sold to you, but it would be meaningless everyday thereafter as thermal profiles of a chip are also a function of the chips age and usage.
From what I've read here it would appear that most people are trying to clue you in the fact that 99% of the chips will not likely come even close to 110W heat output whereas you appear to want people to believe you are fixated on estimating the global mean-value of all chips to be produced.😕 Stop biting the hands that feed you and let some of this soak in, most of us don't post for practice. 🙂
Edit - Me can't spell or paste good 🙂
You have a good point here, and im sorry for any frustration that i have caused, just that heat production does matter to those that overclock. The more a proc produces heat the less likely you are to overclock it well. Plus, this is one factor that some people base there purchasing decisions on. I am starting to inclue the amount of heat production in my decision on which proc to buy. The one in my system is hell. In the summer, it get so hot in my room that i cant sleep at night! this is with ac on. The other floors are ice cold and my room is blazin hot! (note: halogen light doesnt help either 😉 ) maybe to joeschmoe it might not really matter, but to fanatices everything matters! 😉
I hear ya. One of the things I read recently that perked my ears was that Intel might begin locking in the default-operating voltage per each individual CPU's leakage characteristics at test.
This will quickly segregate the market into "O/C friendly" chips as we start questing over the low-voltage high GHz bins.
I haven't read the rest of the thread, but I thought that I'd mention that I disagree with you here.
Without trying to tell you what you probably know already, defects can be thought of a random scattering of sand on a wafer. They appear all over randomly. So hypothetically if you have 400 die per wafer and you randomly scatter 12 grains of sand over them to take out a few of them, you may find that you have lost 12 die per wafer. 97% yeild. But if you have 50 die per wafer, and you throw your 12 grains over the wafer, you find that you may have die big enough to grab 2 of these defects at a time, but overall you are now down 10 die. 80% yield. And if you somehow had 10 die per wafer, you might find that you are down to 1 defect-free die once you throw your defect scattering sand over the wafer.
True... But then you are talking about defects such as particles. If you think about others such as focus spots, spin defects, polish defects, etc, then my theory holds true.
Obviously we cannot get into which defects are more prevalent, or more specific examples... But I'd say that both of our statements are true. 🙂
What pm said is pretty much what I was referring to.
In the article I read, they interviewed some suit at an Intel fab, and he mentioned that the biggest cause of lost dies was due particles, though of course he didn't go into exact yield numbers or anything.
This was quite a few years ago though, so I guess maybe things are better on that front these days?
JustAnAverageGuy
Diamond Member
Just to skew the thread OT for a sec:
I never really understood why people always quoted 3 pages of text, but only follow it up with a paragraph, a sentence, or even just a word sometimes. Why not just quote the very last one you're responding to, and just reply. Or even if it's the very last post on the thread, not even quote it at all? Makes browsing and finding your old spot when you last read the thread
SO much easiar,
I never really understood why people always quoted 3 pages of text, but only follow it up with a paragraph, a sentence, or even just a word sometimes. Why not just quote the very last one you're responding to, and just reply. Or even if it's the very last post on the thread, not even quote it at all? Makes browsing and finding your old spot when you last read the thread
SO much easiar,
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: 25K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 24K
-
-
