48GX computing delay

[Steve Jirsa]

I have a 48GX which often has a several second delay when performing basic arithmetic. Has anyone else experienced this and is there anything that can be done?

[Han]

I have a 48GX (several actually), a 48G (again several) and 48G+ (just one). I also have several 48SX models. I would be willing to test out your calculations on these models and time them for you for comparison. Could you give us some details as to what it is you are calculating?

Han

[Steve Jirsa]

It's usually just basic multiplication or division and it seems to happen pretty randomly.

[Arnaud Amiel]

Wouldn't it be garbage collection? It would especially happen if you don't have much in memory.

Arnaud

[Han]

I believe you are correct. During garbage collection, the calculator will be seemingly slower. The goal of garbage collection is to get rid of temporary objects in memory which are no longer referenced. If you are low on memory, then the HP 48 will try to garbage collect often in an attempt to get more working memory. Is this the case? Do you have very little free memory?

[Arnaud Amiel]

If you have little memory, the garbage collection will be quite fast as it does not need to tidy much.
However, if you have lots of free memory, it will not happen so often but will take much longer.

Basically, it works like this, the hp leaves all its temporary calculation, objects, rubbish in memory until it feels the memory is too full. It then cleans everything and starts again. If less space (memory) is available, it will have to clean more often but will have less work to do each time and you would not notice it.

When the 49 was introduced with 256 K of free RAM there was lots of people complaining about the garbage collection delay. The solution was to store a huge string in memory to have only say 64K left and adjust the size of this string to your memory requirements. Garbage collection was later improved to cope with the 256K.

Arnaud

[Han]

Ahh... I truly blundered then. I had assumed that with such little memory, gargabe collection would occur so often that the frequency of gargabe collection was what was noticable. It makes more sense the way you described it.

Just out of curiosity: how did they make the fix for the 256kB of memory?

Edited: 2 Dec 2005, 4:27 p.m.

[Arnaud Amiel]

I believe that the Garbage collector of the 49G was rewritten completely in a ROM update. It still happens but it is much fatser. On the 49g+ I believe it was directly implemented in ARM

Arnaud