Don't you hate it when a web site ERASES the whole body of your comment when you hit the back button? To me its stuff like that which is far more bothersome than high temperature from an overclocked CPU. Besides we all know the solution for high temperatures. Its air conditioning or refrigeration or running overclocked computers only on planet Pluto. High temperatures a problem? Send me an email.
Instead of pumping toward 4(and possibly 5)ghz Intel should be looking to lessen the bottlenecks in the software and on the chipsets.
When I can spend $400 and get a Vapochill cooler, buy a 2.4ghz Northwood with HT for $189, match it with some properly cooled 533mhz DCDDR, and overclock it to close to 4ghz (if not higher) and bump the ram to around 750... 90nm doesn't sound too snazzy.
I currently have a 2.4 hitting 3.6 with air. For the price they're going to want for a Prescott, I could have at least another full system up and running.
With clock speeds increasing rapidly, the market is becoming extremely tight. RISC processors have just recently hit close to the 2ghz bar, and with the right operating system they can pummel a 3ghz CISC.
Intel should get the hint and move away from CISC, if they do, AMD will follow.
A reduced instruction set would allow for the processor to do more in half the time it would take a CISC, which would mean they wouldn't have to compensate for it with these monstrous clock speeds.
The reduction of the die size was meant to allow faster clock speeds with matched performance, not faster clock speeds with subpar performance.
What ever happened to Intels plan to get rid of the polymer all together?
I thought the first step they were taking was to move to a metal before trying 90nm.
What's important is the reason we are pressing the envelope for higher and higher CPU speeds. Especially since Pentium architecture the CPU speed has the greatest effect on system speed. However, the achievement of higher system speeds may now be achieved faster not by investing in a higher speed CPU but in multiple CPU's or in multiple systems. Many corporations are now taking advantage of peer to peer operating systems that taps into the unused capacity of terminals, laptops, and workstations alike. We all know about Napster and SETI. The same arrangement on a home system with only two to ten systems may not be that dramatic but hey, its better than not utilizing that capacity at all. (Hope the password I typed is right so I don't have to type this reply in again...)
Finally. Just what I allways wanted. A CPU that I can boil water too cook my eggs with.
Our new rig for gaming/records:
Thermaltake VA3000SNA case
Dual NEC dual layer DVD burners
ASUS A8V Deluxe Wi-Fi mobo
AMD FX-53 skt 939
2 GB Corsair XMS extreme series pc-4000
eVGA geforce 6800 ultra
Audigy 2 ZS Platinum Pro Sound Card
1 zalman li
Ummm.... Actually most gamers would happily take to extra frames per second that an extra 500MHz would offer. They would probably know how to use it too by doing things like *gasp* turning up the resoloution of the game! One could argue that the driving force behind the race for the fastest components on the market is games.
Yeah... I agree. Its certainly not inventory or a hundred other business applications that require faster CPUs.
The battle between sequential and parallel processing is a game who's score is measured in terms of greater speed.
There are limits to how many transistors can be crammed on a chip so there is a limit to how many parallel processes can be performed. However, sequential processing can simulate parallel processing if it occurs within the time constraints of the parallel events. The speed of sequential processing is limited by transistor switching constraints such as electrical resistance.
While smaller transistors means more parallel operations per chip faster switching means that more parallel operations can be simulated in a given time.
Transistors made from superconducting semiconductors would in theory have no intermediate electrical resistance (no momentary electrical resistance after the transistor was switched). With zero momentary electrical resistance there might be no limit on transistor switching speed.
Heres the Deal, Buy Northwoods if ur an Intel buff (run much cooler and are faster), Prescott is still in development i think as the release model sux, thats how it is in this world, just wait until they have all the bugs out n such, but i think they should just make like 20 2 stage pipes instead of a single 40 stage pipe.
Still I say that high temperature superconducting semiconductors are the way to go... even low temperature superconducting semiconductors could still be used by immersing the CPU (or for that matter the whole motherboard) in liquid nitrogen. Its not that expensive to run a cryogenics compressor in the back room. Its even possible that stack or stage thermocouples might handle CPU temperatures below ambient.
Then there is always the possibility that a light switch will be invented that will replace the electron switch (transistor) so there won't be any problem with heat.
Sure they could switch to other conductors than silicon, but when the circuits get below about 60nm there'll be so much noise in the circuit they'll have to put in filters.
Intel's stock heatsinks work fine on lower-end CPUs, but they should have made differnt designs for the higher-end CPUs (2.8+). They did release those B and C steppings, which seem to have lower temperatures, but I couldn't find any at Microcenter or Fry's. Newegg tells you they have Bs and Cs but then again Newegg also sent my friend the wrong heatsink.
Yep, ATI used to be in XGI's position, and now they make GPUs equal to and soon to be better than nVidia. AMD struggled to keep up with Intel's 286, and won the 486 race. Now look! AMD is winning again. ATI is winning again. Intel is doomed.