Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror

Slashdot videos: Now with more Slashdot!

  • View

  • Discuss

  • Share

We've improved Slashdot's video section; now you can view our video interviews, product close-ups and site visits with all the usual Slashdot options to comment, share, etc. No more walled garden! It's a work in progress -- we hope you'll check it out (Learn more about the recent updates).

×
User Journal

Blue Shifted's Journal: CPU performance doubling times are slowing down!

Journal by Blue Shifted

...on the desktop PC anyway.

i hate to say it, but it seems that the rate of increase for desktop cpu performance has been slowing down since the P4 hit 2.0 Gigahertz....

over the years, solely out of curiosity, i've been taking note of benchmark scores for many different benchmarks (mostly application-based "real world" benchmarks), and they were taking around 30 months to double their scores. sometimes faster, sometimes slower, depending on the benchmark, of course.

since 2001, though, the time to double desktop cpu performance is taking longer and longer, now i am seeing scores taking three to four years to double....

i'm sure it's common sense and probably even common knowledge, so i probably don't need to state this, but i will anyway: performance increase is not keeping up with transistor increase, same as we saw with frequency during the great Mega/Gigahertz race. now we are getting multicores, that not many applications can take advantage of....

(when i look at how many scores i've found and recorded just for Business Winstone 2004, i realize that this might have become an obsession...)

------------------------
------------------------

in another arena, which i don't follow near as often:

Trends in supercomputers:

19.2 months is how long it takes for supercomputers to double number of simple "adds"

in 192 months (sixteen years, ten doublings) they will do adds 1,024 times as fast, IF the pace keeps up, of course...

(sorry, i know that simple arithmetic adds may not to translate to real world performance; but i haven't taken the time to note and calculate Linpack scores over time. Maybe someday i will!)

-------------------------

these are all just fevered musings on my part, and just as they say about Slashdot polls, "This whole thing is wildly inaccurate. If you're using these numbers to do anything important, you're insane."

This discussion has been archived. No new comments can be posted.

CPU performance doubling times are slowing down!

Comments Filter:

[Crash programs] fail because they are based on the theory that, with nine women pregnant, you can get a baby a month. -- Wernher von Braun

Working...