# Whats an easy way to compare computer speeds **URL:** \ **Category:** Factual Questions\ **Created:** [December 10, 2009, 8:49am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712 "2009-12-10T08:49:58Z")\ **Posts on this page:** 9\ **Page:** 1
**Author:** ![Mr.Lahey](https://avatars.discourse-cdn.com/v4/letter/m/9d8465/32.png) [@Mr.Lahey](https://boards.straightdope.com/u/Mr.Lahey)\ **Post date:** [December 10, 2009, 8:49am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/1 "2009-12-10T08:49:58Z")
Say I want to know which supercomputer had the same power as the machine I’m on now, or how todays supercomputers compare to my pc. Is there a good site or something that lets you put the computer speeds in perspective relative to other machines? ---
**Author:** ![HorseloverFat](https://avatars.discourse-cdn.com/v4/letter/h/8e8cbc/32.png) [@HorseloverFat](https://boards.straightdope.com/u/HorseloverFat)\ **Post date:** [December 10, 2009, 1:43pm UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/2 "2009-12-10T13:43:24Z")
FLOPS, specifically gigaflops for supercomputers: > **[FLOPS](https://en.wikipedia.org/wiki/FLOPS)** > > In computing, floating point operations per second (FLOPS, flops or flop/s) is a measure of computer performance, useful in fields of scientific computations that require floating-point calculations. For such cases, it is a more accurate measure than measuring instructions per second. > Floating-point arithmetic is needed for very large or very small real numbers, or computations that require a large dynamic range. Floating-point representation is similar to scientific notation, except everythin... ---
**Author:** ![Francis\_Vaughan](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/francis_vaughan/32/3093_2.png) [@Francis\_Vaughan](https://boards.straightdope.com/u/Francis_Vaughan)\ **Post date:** [December 11, 2009, 1:09am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/3 "2009-12-11T01:09:28Z")
It is hard. The standard measure of supercomputer speed, at least for the last 15+ years has been the Linpack benchmark. This is a test of floating point speed doing large scale matrix solving. (Linpack is the LINear algebra PACkage). However, something that is widely acknowledged, even this benchmark is reasonably bogus. It favours large clusters over more technicaly advanced systems, and does not balance other importnat aspects of the supercomputer’s performance in the metric. But it has taken hold as the metric everyone quotes, and is the basis for the Top500 ranking. [www.top500.org](http://www.top500.org) The Top500 has historical rankings, and is probably the best place to look for a perspective on how performance has changed over the years. However, don’t kid yourself that a PC with a similar Linpack rating to an old supercomputer makes the machines equally fast. Many early machines had highly tuned architectures that were very very good at sustained performance, and had very high bandwidth IO and internal communications systems. Linpack can be tuned to fit the memory/cache hierarchy very well, and can provide a pretty good fraction of peak performance even in systems with poor interconnect performance. Many other codes don’t have such a luxury and thus the more esoteric architectural features become very important. There is a reason SGI still sells Altix 4700 systems, and has just released the Altix UV system. ---
**Author:** ![Chronos](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/chronos/32/134_2.png) [@Chronos](https://boards.straightdope.com/u/Chronos)\ **Post date:** [December 11, 2009, 1:16am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/4 "2009-12-11T01:16:43Z")
Really, the best benchmark for how a computer will perform any given task is the task itself. If you really, really want a whole bunch of digits of pi, then the fastest computer is the one that has the best score on a pi-calculating benchmark. But it might not be very good for playing Crysis on the highest quality settings. ---
**Author:** ![Fubaya](https://avatars.discourse-cdn.com/v4/letter/f/c2a13f/32.png) [@Fubaya](https://boards.straightdope.com/u/Fubaya)\ **Post date:** [December 11, 2009, 1:37am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/5 "2009-12-11T01:37:05Z")
If I’m not mistaken, an average computer can get a few gigaflops. The Cray Jaguar clocks in at 1.75 petaflops per second and that’s currently the fastest computer in the world. It’s also just down the road from me, but they won’t let me borrow it for porn. ---
**Author:** ![Chronos](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/chronos/32/134_2.png) [@Chronos](https://boards.straightdope.com/u/Chronos)\ **Post date:** [December 11, 2009, 3:42am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/6 "2009-12-11T03:42:36Z")
Why would you want something with a lot of flop for porn, anyway? ---
**Author:** ![bobalink](https://avatars.discourse-cdn.com/v4/letter/b/71e660/32.png) [@bobalink](https://boards.straightdope.com/u/bobalink)\ **Post date:** [December 11, 2009, 11:58am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/7 "2009-12-11T11:58:52Z")
> [@Mr.Lahey](#): > > Say I want to know which supercomputer had the same power as the machine I’m on now, or how todays supercomputers compare to my pc. Is there a good site or something that lets you put the computer speeds in perspective relative to other machines? ``` the power is the same do you mean the speed ``` ---
**Author:** ![Oslo\_Ostragoth](https://avatars.discourse-cdn.com/v4/letter/o/a9a28c/32.png) [@Oslo\_Ostragoth](https://boards.straightdope.com/u/Oslo_Ostragoth)\ **Post date:** [December 13, 2009, 4:39am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/8 "2009-12-13T04:39:04Z")
> [@Fubaya](#): > > If I’m not mistaken, an average computer can get a few gigaflops. The Cray Jaguar clocks in at 1.75 petaflops per second and that’s currently the fastest computer in the world. It’s also just down the road from me, but they won’t let me borrow it for porn. Dude, if you are going to crack me up like that, you have to become a member. 😃 ---
**Author:** ![Voyager](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/voyager/32/133_2.png) [@Voyager](https://boards.straightdope.com/u/Voyager)\ **Post date:** [December 13, 2009, 8:07am UTC](https://boards.straightdope.com/t/whats-an-easy-way-to-compare-computer-speeds/520712/9 "2009-12-13T08:07:45Z")
The [Wikipedia entry on Benchmarks](http://en.wikipedia.org/wiki/Benchmark_%28computing%29) has a nice list. What benchmarks make sense depends on the kind of computing work you have to do. Linpack is fairly useless if you are Google and want to do lots of transaction processing. A benchmark that does well for a CPU with multiple cores is not too useful if you want raw floating point performance. Lots of results are suspicious, because it is common practice for compiler writers in a company to optimize specifically for certain benchmark sets - or used to be, at least. As for porn, maybe we need a thrusts per second benchmark or something.