Login

Login

Need to register? Lost password?

Options

Advertisements

Advertisements

 

10.5 billion instructions per second

Hardware, Software & Internet

Minitokyo » Forum » Computers & Internet Fora » Hardware, Software & Internet  10.5 billion instructions per second

page 1 of 1

What it the minimum processor speed need if one would like to perform say 10.5 billion instructions per second? Just curious.

well look up a supercomputer and see what they do, it would give you an idea then

I gues thats about 10teraflops. Not sure if thats true.
I don't know how mutch calculations / second a teraflop is.
Their aren't many comps on the earth with that mutch teraflops.
http://top500.org/lists/plists.php?Y=2005&M=06

Woah, 10 teraflops is absolutely insane.

No 10.5 billion instructions per second would be a lot less. For example my computer benchmarks at about 32 billion per second (using 7-Zip) so I'd say a 2Ghz would be able to handle it.

Don't blame me if I'm wrong though because I don't really know what type of instructions you are talking about or what 7-Zip actually does.

  • Zareste
  • Restricted Member
  • 2y 44wk ago

Teraflop - a billion floating point operations in a second. It's the standard measure of speed for supercomputers, but instruction speeds differ according to variable type and the equations used. I think floating point calculations are just the most challenging for the processor.

I think IBM and a few other companies have supercomputers that can do over 10.5 teraflops. Not sure though

Sorry, just to clear thing up, where I come from a billion is a 1000 million, not a million million like in some countries. So the above poster is refering to the million million type of billion.

BTW, Gigaflops are the standard unit, because most PCs don't even his one Teraflop. And I'm not sure if floating point operations and instructions are the same, so if anyone woul care to clarify that would be great.

IBM's blue gene puts out 16.5 ish Teraflops, which is absolutely insane.

one way to try out, get linpack and measure it yourself how mutch tera or rather megaflops your computer has. I once found the instruction files of linpack and i can tell you that they are amazing heavy.

I was going to do that, but it looked like it would take more than 5 minutes to set up so I couldn't be bothered.

Quote by TSPhoenixWoah, 10 teraflops is absolutely insane.
No 10.5 billion instructions per second would be a lot less. For
example my computer benchmarks at about 32 billion per second (using
7-Zip) so I'd say a 2Ghz would be able to handle it.
Don't blame me if I'm wrong though because I don't really know what
type of instructions you are talking about or what 7-Zip actually does.

So a mere 2 GHz could run something churning 10.5 teraflops? o_0 I thought it would have to be much faster that. By instructions, what do you have in mind?

Noooo, 10.5 teraflops and 10.5 billion instructions are verrrrry different things.
A TeraFlop is 1000 billion instructions a second. so 10.5 billion instructions is only 0.0105 TeraFlops!

Quote by WikipediaComputing devices exhibit an enormous range of performance levels in floating-point applications, so it makes sense to introduce larger units than the flops. The standard SI prefixes can be used for this purpose, resulting in such units as the megaFLOPS (mflops, 10^6 flops), the gigaFLOPS (gflops, 10^9 flops), the teraFLOPS (tflops, 10^12 flops), and the petaFLOPS (pflops, 10^15 FLOPS).


So in instructions a second say:
A teraflop = 1,000,000,000,000 instructions
10.5 billion = 10,500,000,000

Also, bluegene is 136.8tf, not 16.5 as TSPhonenix mentions:

Quote by WikipediaAccording to Top500.org, the fastest computer in the world as of June 2005 was the IBM Blue Gene/L supercomputer, measuring 136.8 tflops, with a peak performance of 183.5 tflops. Blue Gene currently has 64 racks, with each rack holding 1,024 processors, yet the chips are the same as those found in mid-performance computers (PowerPC 440 700 MHz (2.8 GFlops each)).

But yes, 2GHz processor can run 10.5 billion instructions (or 0.0105 teraflops or 10.5 gigaflops)

Quote by TSPhoenixI was going to do that, but it looked like it would take more than 5 minutes to set up so I couldn't be bothered.


I did but i came to the conclussion that my linux distro didn't realy work well with it :(

I'm kinda curious why you'd want to know?
since performance rather depends on what you want to do how... a CISC processer will usually take more cycles per operation so less operations per second, but will be able to do more complex things within an operation, where as with a simple RISC processer will be able to do 1 operaion per cycle, and thereby have more operations per second, but will be able to do less complex thingis within a cycle (thereby CISC being more optimized where as RISC is more flexible.)
most x86 (Intel, AMD, Crusoe cpu's are CISC btw, some of the more rare cpu's like ARM's and MIPS cpu's are risc)

Forgive me, I am not much of a computer person, as such, I should have had a more definite idea what I am seeking to know before I should have asked....

Ah, I have found the specs for the processor I was thinking, (Made by Hughes). Its general processing capacity is 700 MPIS with growth to 2000 MPIS. Signel processing is on the order of 20 BOPS with the expansion capacity of 50 BOPS. I have learned this is for military applications, but compared to what is out in the civilian market is that pretty decent performance?

page 1 of 1

Only members can post replies, please register.