Page 1 of 4

XScale's Problem

PostPosted: Dec 6, 2002 @ 9:22pm
by James S
Very interesting article on PocketPCThoughts.com about why the XScale is seemingly so slow. And it explains why it seems that 300MHz won't be slower, if only a tiny bit, than 400MHz.

http://www.pocketpcthoughts.com/article ... 925947e8a5

PostPosted: Dec 6, 2002 @ 9:39pm
by Quantum
it seems so far that is is not an internal performance problem but the cache thus since the FSB speed is very close to the same for both 400 MHz and 300 MHz they will be very close in performance... I was think about this a few months back. I am curious to see some actual tests on this but I don't have the XScale devices to write the tests.

PostPosted: Dec 7, 2002 @ 1:35am
by Goosey

PostPosted: Dec 7, 2002 @ 1:49am
by James S

PostPosted: Dec 7, 2002 @ 1:54am
by eminefes

PostPosted: Dec 7, 2002 @ 2:28am
by damian

PostPosted: Dec 7, 2002 @ 11:34am
by Goosey

PostPosted: Dec 7, 2002 @ 3:28pm
by damian
Yeah, I know, but I have a 128 meg CF card so that's not really an issue.

This time, though, I actually read the article. From what I can deduce, it's impossible for a 300 MHz XScale to be any slower than a StrongARM than a 400 MHz XScale is. I guess that's good news; looks like my Dell will be fast enough.

I'm 99% likely to order my Axim today :D

PostPosted: Dec 8, 2002 @ 2:30pm
by Deje
If you read , you will see one point was missing in the article at PocketPCThoughts.com.
The XScale CPU could use a 32Bit memory bus! If hardware designers would use this 32bit interface on PPC, it will nearly double the performance for memory accesses!

PostPosted: Dec 8, 2002 @ 4:23pm
by James S
That was in the article.

PostPosted: Dec 8, 2002 @ 5:39pm
by Deje

PostPosted: Dec 8, 2002 @ 6:37pm
by MiniMe

PostPosted: Dec 8, 2002 @ 8:12pm
by James S
Ya, I remember that. I thought that at 300MHz the bus speed of XScales increased to 133MHz :/ Oh well, I'm neither engineer nor programmer.

PostPosted: Dec 8, 2002 @ 9:51pm
by Digby

PostPosted: Dec 8, 2002 @ 10:05pm
by Kzinti