Re: long double data type
Re: long double data type
- Subject: Re: long double data type
- From: Jonas Maebe <email@hidden>
- Date: Mon, 14 Jun 2010 21:22:07 +0200
On 14 Jun 2010, at 20:47, Alastair Houghton wrote:
> On 14 Jun 2010, at 19:32, Jonas Maebe wrote:
>
>> "long double" has 128 bit precision on PPC and 80 bits of precision on Intel (in both cases the precision is the same in 32 and 64 bit mode).
>
> Erm, no.
>
> Precision means something very specific in this context, and it *isn't* the same as the size of the type,
The size of the type is actually something else altogether. The size of long double variables on Mac OS X Intel is defined as 128 bits. Of those 128 bits, only 80 bits are actually used by the FPU. And of those 80 bits, only part is used as mantissa, obviously.
> For instance, 64-bit IEEE floats have 53 bits of precision, while 128-bit floats have 113, I believe. Intel's 80-bit format has 64 bits of precision.
Yes, I was using contracted forms of the more correct expressions "128 bits quadruple precision" and "80 bits extended precision", which while technically probably wrong, I thought to be quite well understood. It's true that the significant part is smaller.
> SSE definitely supports 64-bit floats (doubles) as of SSE2, though of course with a 128-bit register you can only pack two per register.
And the 80x87 fpu supports an 80 bit extended precision type.
Jonas _______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden