[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [avr-libc-dev] more accuracy...
From: |
Joerg Wunsch |
Subject: |
Re: [avr-libc-dev] more accuracy... |
Date: |
Mon, 24 Apr 2006 16:06:52 +0200 |
User-agent: |
Mutt/1.5.11 |
As David Carr wrote:
> AVR
> float: 2476.24536132812500000000
> error: 23.75463867187500000000
Which is fairly well in line with the 32-bit float results
from the i386.
> Is there a way to have sizeof(double) == 8?
Currently not. Volunteers are needed to hammer all this
out, in particular to establish enough library support
for this to ever become a useful option.
> Considering that the AVR is 8bit and has no hardware floating point,
> being only 12x slower than the 386 is impressive especially
> considering it runs at 1/5 the clock speed.
Yes, indeed. I think the key point here is that the avr-libc math
libraries are carefully manually tuned. OK, but so is the FPU
hardware implementation of the i387...
--
cheers, J"org .-.-. --... ...-- -.. . DL8DTL
http://www.sax.de/~joerg/ NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)