Floating Point

[trinibwoy]

This 3dcenter article seems to be a pretty good overview of the difference between fixed and floating point storage formats. But unfortunately it is in german and translations leave much to be desired. Anybody know of a similar quality article anywhere in english?

 

[Simon F]

This may be a bit more advanced than you were looking for but how about: What every Computer Scientist should know about floating-point?

 

[trinibwoy]

Whoa!! I think I'm more comfortable with german than formula notation. Even sigma gives me the creeps

 

[Geo]

I have decided that I want nothing to do with relative error. Ulps only. Please. Relative error makes me go all. . uhh, wobbly.

 

[Dio]

I remember one particularly arcane bit of my physics courses which was about error calculus... well, I say I remember it...

 

[Reverend]

This 3dcenter article seems to be a pretty good overview of the difference between fixed and floating point storage formats. But unfortunately it is in german and translations leave much to be desired. Anybody know of a similar quality article anywhere in english?

Is there any reason why you would want to know the difference between the two? I haven't read (all) the links in this thread... but that's because I don't care to.

Anyone like to re-visit DX3?

PS. You're asking for a bettter translation of an article and nothing else... should this thread be in another forum? Please don't tell me you're asking for opinions on the difference betwen "fixed" (which, IMO, is ridiculous in the way this word seems to have been interpreted, according to my perception) and FP. Because such a question does not belong in this particular forum. Let's talk about FP64 instead

 

[Simon F]

I have decided that I want nothing to do with relative error. Ulps only. Please. Relative error makes me go all. . uhh, wobbly.

So if I said that, say, the mass of a car was 1 tonne, give or take 5%, and a train was 100 tonnes, give or take 10%, you'd collapse? :?

 

[Kristof]

Is there any reason why you would want to know the difference between the two? I haven't read (all) the links in this thread... but that's because I don't care to.

Most handheld hardware has fixed point only processors (ARM without VFP unit) with floating point being emulated at X performance cost. So maybe the question is linked to that ?

And fixed point can be

K-

 

[flick556]

This article helped me

http://www.codeproject.com/dotnet/ExtremeFloatingPoint1.asp

It's all pretty clear

 

[trinibwoy]

Because such a question does not belong in this particular forum.

I think it does.

 

[Reverend]

Is there any reason why you would want to know the difference between the two? I haven't read (all) the links in this thread... but that's because I don't care to.

Most handheld hardware has fixed point only processors (ARM without VFP unit) with floating point being emulated at X performance cost. So maybe the question is linked to that ?

Perhaps. But maybe Dave recently re-vamped the forums for a reason, right?

And fixed point can be

K-