50g clock bug, what is it?

[HP67]

What is the "50g clock bug?"

Is it just when you set alarms? Because my 50g has been in use for a couple of months and the clock remains at the correct time. I haven't used any alarms though.

[DavidM]

(05-21-2014 01:50 PM)HP67 Wrote:  What is the "50g clock bug?"

Is it just when you set alarms? Because my 50g has been in use for a couple of months and the clock remains at the correct time. I haven't used any alarms though.

Probably depends where you saw that, but it most likely refers to the haphazard nature of alarms actually occurring when they are supposed to on the 49g+/50g.

Most of the time, they go off as planned. But you simply can't depend on them, because there's no set of conditions that can guarantee either success or failure. When they fail, they may either occur much later than scheduled, or simply not at all.

This bug only affects the ARM-based calcs (49g and previous aren't hampered by it). Speculation has been that it's related to the fact that these systems don't have interrupts that can be easily matched to the original hardware, so the Kinpo O/S simulates them for the Saturn environment as best it can.

JYA at one point indicated that he thought he knew what the cause was, and communicated a potential fix to Kinpo. But unfortunately no solutions have ever resulted from that action.

[HP67]

I wonder if it's flashable.

[DavidM]

(05-21-2014 03:51 PM)HP67 Wrote:  I wonder if it's flashable.

The Kinpo OS? Yes. Claudio, Dave and friends are actively working on something that will presumably negate the need for it entirely (if I understand the NewRPL project correctly).

But I wouldn't hold my breath waiting on a fix from HP/Kinpo. This has been a well-known problem for a long time. It appears that Alarm functionality was never a high enough priority to sink funds into fixing it. And something tells me that we won't see any more updates for a product that is on its way out the door anyway.

While some here will say "good riddance", I still have a fondness for the 48-50 series and an admiration of its underlying software design. It has proven to be extremely flexible and adaptable over time.

[HP67]

No, I mean if a fix was submitted (I don't know who JYA is) then presumably the ROM could be patched and/or reflashed.

I'm aware the problem is an old problem but I don't know the scope of it. I expected the clock to be inaccurate but mine seems to be very stable. I will avoid using alarms though

[Tim Wessman]

The suggested fix was analyzed way back (long before my time) and actually determined that it didn't really fix anything - it just shifted it to another location. There were many people who put a lot of effort into trying to find that and nobody ever was able to identify what was the exact cause. Most likely it was a combination of several that caused some interdependent behavior.

[DavidM]

(05-21-2014 06:51 PM)HP67 Wrote:  (I don't know who JYA is)

"JYA" = Jean-Yves Avenard, one of the authors of the MetaKernel, eventually hired by HP in Australia to work in the ACO group which produced the 49g and updated several other systems. Let's just say that he was in a better position than the rest of us to know something about the internals of the 49g+/50g.

(05-21-2014 06:51 PM)HP67 Wrote:  I expected the clock to be inaccurate but mine seems to be very stable. I will avoid using alarms though

From my own experience (and others that I've heard from), the clocks on 50g systems tend to be off by 1-2 seconds/day. I have both a 49g+ and 50g, and they are both slow by about 1.5 seconds/day. That's good enough for my purposes.

(05-21-2014 07:31 PM)Tim Wessman Wrote:  The suggested fix was analyzed way back (long before my time) and actually determined that it didn't really fix anything - it just shifted it to another location. There were many people who put a lot of effort into trying to find that and nobody ever was able to identify what was the exact cause. Most likely it was a combination of several that caused some interdependent behavior.

Thanks for the explanation, Tim. I may have missed it before, but that's the first time I've seen that there was actually some serious work put forward to analyze JYA's suggested fix. Glad to hear that it was given some serious consideration.

I will confess that I used to use alarms on my 48sx, but that was at a time when I needed them more (and had fewer options from other devices as a replacement). As such, I haven't really missed them so much on the 50g. The part that bothers me most about it is simply that I can't schedule activities to happen on a regular basis and still count on them actually happening. There's workarounds, sure. But I haven't had a great enough need to warrant putting the energy into programmable fixes -- I just do what I need manually on an occasional basis.

[HP67]

Thanks, Tim. This is good info.