But do you really believe they'll scale down CPUs for less heat when they dump silicon for diamond?
Refer to the principle of heat generation in fusing wires to each end of a battery i.e. the wire gets hotter. The electron leakage is basically runs on the similar principle.
Low-K Black Diamond promotes lower consumption, hence less heat generation.
Lest anyone be misled, note that 'Black Diamond' is not a diamond substrate itself, but rather a preparation of "CVD carbon-doped oxides."
http://www.chipworks.com/chipnews/2002_i02/interconnect_2.htmThere's some company that's supposedly found the trick to growing vapor-deposited diamond crystals (talking literal bricks, here) at intense purity and cost comparable to Si. Intel were said to have shot them down when they made their pitch, betting on Si processes to continue to scale for the forseeable future. ("Oops?")
http://www.wired.com/wired/archive/11.09/diamond.html...
Those Cyrixen were quite toasty for their performance at the time, probably owing to both process issues and decisions of design. (Similarly, has there ever been a K7 chip that honors HLT in the manner previously expected?)
Scary numbers from the slot CPU era:
http://grassomusic.de/english/amdk7.htm(In editing for width, I note that Xoops takes poorly to links containing '?' options.)
...
Ewww. This is going to hurt IBM in sales and status. I hope they fix it.
Imagine if it was an Intel bug and not IBM. How many of us would be slamming the "I told u they were crap" line.
I think everyone hopes they'll fix it, IBM included. However, while it certainly won't *grow* sales... you think PPC customers have ever really had a choice? ;-) (Seriously, it sounds like the post-spinoff Freescale just might warm up the competition, but if what you want is a GHz *G3,* it's not like there's really a second source.) Meanwhile, I'm lost as to how many units they actually sold before this discovery, and bugs in paper launches don't effect reality much.
...
As it happens, the technology for active cooling of CPUs has improved a great deal: bigger and more heat conductive heat sinks, bigger and quieter fans, etc. What this needs is unrestricted airflow in cases, and this is where the vendors cut corners. They like to use smaller cases, so they cram them more, but they don't take care to make sure there are no restrictions to airflow.
Well, you should be enamored with BTX, then, and even I'll give Intel some credit for flipping PCI and AGP dust-collecting-side up.
The P4 does remain in the unenviable position of being the first line to throttle for protection under
certified conditions. (I'd say Prescott in specific, but I can't remember if this was first noticed on one or a late-model Northwood?)
There is little excuse for not using tubular cables any more,
Except that most rounded cables are produced by vendors without a single EE on staff, and some actually show a negligible but worrying (as regards integrity) performance hit in practice. Luckily, the SATA/SAS phy rather solves for this anyway.
[...] and for not tidying up the cabling in such a way that it does not restrict airflow. All done to squeeze that little bit of extra profit in a cut-throat market.
Or to provide what consumers want, in which case the world might actually have some demand for (non-watercooled) PPCs.
Actually, it'd be interesting to know whether mainboard draws have dropped significantly on K8 designs; you lose a major portion of one big hot chip, and some of the transistor count that went with it... but, in early designs, do up the number of discrete bridge ICs on board.
