Should I Stay or Should I Go?

[ARM Hacker]

The bug in the B2 PXA250 isn't as bad as it sounds. But the 'workaround' does kinda hurt performance. The main problem is there was a very very slight (and very rare) occurence where a bit in the L1 datacache that said whether or not the cacheline was modified would get changed. If the bit did get flipped (presumably by some random EMI or something), the processor would think whatever garbage in the cache line was actually valid data and it would write it out at some point. Since the 'tag', or the address of this line would most likely also be garbage, it would result in a crash of the handheld.

It occurs proportional to heat also, and it only seems to occur once in a long while. I.E. you could run a pocket PC for a week or more without it happening. I'd almost venture to say the OS & software is buggy enough you wouldn't know it was hardware. Anyways, the only totally safe workaround for this was to TURN OFF the 'write-back' portion of the cache. So Microsoft decided to do that.

Since there are tons and tons of the B2 parts (thats all that is being sold now) out there, its safe to say that even PocketPC 2003 will probably run without the writeback cache 'just to be safe'. So for a lot of reasons, you wouldn't even see the gains of having a 'bug-free' chip. From what I can gather, you might be looking at next xmas time before you see a handheld with one of these new chips. There are tons of the B2 parts waiting to be sold. The PXA250 was 'available' and being put into reference designs and development boards as early as January 2002, and there weren't any XScale PDA's in stores until July.

As for this bug, the ASUS pda actually lets you turn on the WB cache, at your own peril. I belive ExtremeTech benchmarked it as a 20% increase in some benchmarks! not bad. I can explain what kinds of operations the bug (and having the cache off ) hinders if anyone is interested.

[Racer X]

This is a ridiculous thread. What does anyone plan on doing with thier Axim where they will notice the performance difference b/w the old and new Xscale chip? Oh wait...the scores on some artificial benchmark will be higher with the new chip. OK. I get it now ;)

I use my Axim to keep my appointments, contacts and notes. I listen to music, browse sites, read e-books, carry around family pictures, play games and impress my friends. I don't edit movies, compile code or disarm countries with weapons of mass destruction with my Axim. I am perfectly happy with my pokey 400Mhz legacy chip.

BTW, none of the above rant applies to cpu chips or video cards, all of which must be the latest/greatest. B)

[andsoitgoes]

Originally posted by Racer X@Jan 30 2003, 11:51 PM
BTW, none of the above rant applies to cpu chips or video cards, all of which must be the latest/greatest. B)

Brilliant! I love it, so true.

[Beowulf]

I am using my Axim to power my 12000 lb SUV as oppossed to using gasoline. When not driving my SUV. it is borrowed by a constuction company to power their 5 bulldozers....seems to work fine.

Oh wait a minute, I use my Axim as a traveling unit that compliants my laptop...contacts, small excel spreadsheets, etc. etc.


Stay with it....wait for the Intel brain implant expected out in the 3rd quarter of 2005.

[danterner]

Originally posted by Emonair@Jan 30 2003, 07:14 PM
So how do I answer this poll?

Should I Stay or Should I Go?
Yes or No?

Yes - I know - the poll is screwed up. I thought that the poll question was going to be "Return my Axim due to new XScale Rev.?" Instead, it was given the title of the post itself, resulting in a silly question that doesn't really make sense. (Much like the body of my post itself, as some may say). I didn't catch it while previewing the post, but saw the problem immediately after it posted. I tried to edit the post, but unfortunately could make changes only to the text of my post and not to the poll itself.

My "maybe Dell will issue a recall" comment *was* tongue-in-cheek. I don't think it's even a remote possibility. At best, maybe there'll be something similar to the way Microsoft has been handling the problems owners of first-generation Xboxes have been having with "dirty disk errors." No formal acknowledgement of there being a problem, but all newer Xboxes have been shipping with a different DVD drive and when first gen 'pioneers' call to complain Microsoft is making repairs at their own expense.