Pentium 4 Reaches the Speed Limit

[Steve Mehs]

As expected, Intel has introduced a 3.8GHz Pentium 4 processor in what could represent the end of an era for the chip maker.

The Pentium 4 570 processor at 3.8GHz will have the fastest clock speed of any processor available from Intel for an indefinite period. Intel has decided to cancel a planned 4GHz Pentium 4 processor and improve the performance of its desktop chips by adding cache memory.

MORE (http://www.pcworld.com/news/article/0,aid,118603,00.asp)

[Mike123abc]

The Pentium line needs a vast overhaul. AMD has been doing so well since they are able to have more processing power per clock cycle, giving them more speed at lower clock rates.

[HappyGoLucky]

Intel has simply come around to the same facts as everyone else: raw clock speed is not a valid indicator of processing power. AMD has proven that a slower clock in an efficient processor can and does outperform a faster clock on a less-efficient unit. New Intel chips are being marketed not by clock speed but by an essentially meaningless number that indicates their relative performance to other chips in the same family. Thus we now have the Pentium 520, 530, 540, 570, etc. The numbers do not correlate to clock speed, just that a higher number is a faster processor. In fact, some of the new Intel chips will actually be faster than the top of the line 3+GHz Pentium 4s but will be operating in the 2+ GHz range, just with more efficient cores built off of the Pentium-M line.

The "hot" thing in processors coming up is the dual-core models, for both Intel and AMD (as well as IBM in the PowerPC chips). Having two distinct processors sharing a chip, using an ultra-fast interconnect and thus sharing all the rest of the computer system invisibly will make for a new leap in performance. A Dual-Core P4 with HT Technology will have an operating system "think" it is running in a 4 processor computer. IBM is already building Power Processors with dual cores, AMD is forthcoming with Dual-Core 64-bit processors.

[Mark Holtz]

I'm holding off on upgrading now until PCI Express and serial ATA becomes more commonplace on motherboards and more video cards support PCI Express. It appears that PCI Express can handle more current that AGP.

I figured we would hit a speed wall sooner or later. How will the average consumer benefit from a speed increase nowadays?

[cdru]

I remember once upon a time when processors were hitting 100-133 Mhz that some industry analyists said this is the end of the road...we can't go any faster. We now have 3.8Ghz processors.

When 12x and 15x CD-ROM drives came out, I read an article that said CD-ROM drives just wouldn't spin any faster reliably. 40x drives are very common on up to the 72x Kenwood drive (yeah I know it cheats).

When multi-GB hard drives started to become commonplace on mid-range PC, I overheard someone say that they had no idea what people would do with so much space...there just wasn't that much stuff people would need to store on their hard drives. Now, a few years later, I'm pondering how I want to soon lay out a TB array for multimedia storage.

Whenever anyone says we've reached the end of the road regarding computers, I don't beleive them. Something will change and speeds will increase. It may not be the same type of measumrement a few years from now, but it will still be around.

[SimpleSimon]

cdru: What you're referring to "how can we ever use more than xxx", started 40 years ago when Dr. Eugene Amdahl tried to get IBM to build faster versions of his 360 architecture. Long story short, he struck out on his own when the stupid executives didn't listen. Dr. Seymour Cray went through a mirror image of the same story with CDC.

The bigwigs simply couldn't comprehend that someone could possibly want a computer with, for example, 1MB of main storage (RAM), or more than 4MHz clock speeds (slower than the first IBM PC!). I was there and knew better, but even people like me are still amazed that a frelling VIDEO CARD could have hundreds of megs of RAM :shocked:

One thing everyone agrees with - the speed of light has become a real problem. Even 25 years ago it was an issue to be dealt with. We synchronized the clocks in a mainframe that was 25 feet long - that's 32 nanoseconds (speed of electricity in a wire) - by changing the length of a wire (9 inches per nanosecond). :)

[HappyGoLucky]

cdru: What you're referring to "how can we ever use more than xxx", started 40 years ago when Dr. Eugene Amdahl tried to get IBM to build faster versions of his 360 architecture. Long story short, he struck out on his own when the stupid executives didn't listen. Dr. Seymour Cray went through a mirror image of the same story with CDC.

The bigwigs simply couldn't comprehend that someone could possibly want a computer with, for example, 1MB of main storage (RAM), or more than 4MHz clock speeds (slower than the first IBM PC!). I was there and knew better, but even people like me are still amazed that a frelling VIDEO CARD could have hundreds of megs of RAM :shocked:

One thing everyone agrees with - the speed of light has become a real problem. Even 25 years ago it was an issue to be dealt with. We synchronized the clocks in a mainframe that was 25 feet long - that's 32 nanoseconds (speed of electricity in a wire) - by changing the length of a wire (9 inches per nanosecond). :)

Do you have anything to back all that up? I'm not sure we should simply trust your word for it and all.... :sure: