|
in reply to Re^17: Module for 128-bit integer math? (updated)
in thread Module for 128-bit integer math?
That's why I asked up front about a 128-bit math package, not an infinite precision math package
Understood. At no stage was it was my intention to suggest that you should use Math::GMPz instead of Math::Int128 - I was merely querying the speed of Math::Int128 (whose basic math operations seem to be about 3 times slower than need be), and using Math::GMPz for benchmark comparisons.
And why I tried really hard to avoid this debate
Apologies for my part in forcing that upon you - though that was never my intention. (I know very well how annoying it is to be taken to task over matters that are irrelevant to the task at hand.)
Cheers,
Rob |
Re^19: Module for 128-bit integer math? (updated)
by BrowserUk (Patriarch) on Feb 14, 2011 at 12:16 UTC
|
I was merely querying the speed of Math::Int128
It is early days for Math::Int128. Now salva's made it work, it's up to those of us using it to drive the process of making it work faster, as the need arises. Along with any enhancements we'd like. I have a few already.
(whose basic math operations seem to be about 3 times slower than need be
Where does the "3 times slower" come from? Looking back at your benchmark I see the functional interfaces of Int128 being 30% to 40% slower than those of GMPz. Or did I misread the numbers?
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.
| [reply] |
|
|
Where does the "3 times slower" come from?
The (unposted) code that I wrote lifted the particular M::I128 multiplication that I benchmarked from 371517 multiplications/s to 927835 multiplications/s.
That's a bit less than a factor of "3", admittedly. The rest was just my bad memory ... or perhaps deliberate exaggeration ;-)
Cheers,
Rob
| [reply] |
|
