HP 48G+ UI slower than 48G

[ettlz]

I've got two HP 48G+ (both made in Indonesia) and an HP 48G (Singapore build), all with ROM revision R. One thing I've just noticed is that the UI elements are much slower on the 48G+. For example, in CHOOSE boxes the highlight moves very rapidly on the 48G when the down key is held, but really plods on the 48G+.





No libraries installed on either calculator. I've tried resetting the 48G+, it's CPU speed looks correct. Any ideas?

[David Hayden]

What does MEM say? Maybe you have a bunch of variables defined, causing the 48G+ to run slower.

[ettlz]

(09-20-2021 09:31 PM)David Hayden Wrote:  What does MEM say? Maybe you have a bunch of variables defined, causing the 48G+ to run slower.

121458 bytes free, so it shouldn't be getting caught in garbage collection. In fact my other 48G+ which is in a factory-reset state does this too so that should rule out some errant flag state.

[Eric Rechlin]

As I recall, more of the ROM in the 48G+/48GX is "covered" so performance will be slightly slower than on the 48G because of the bank switching.

Also, there is some variation in clock speed among the G series, ranging from 3.7 to 3.9 MHz, so it's possible your 48G is one of the faster ones and your 48G+ is among the slower ones.

But neither of these should mean a significant speed difference. I'm surprised how different your two are.

[DavidM]

I just ran the same test comparing my 48G to my 48GX. Both calculators are Rev. R, and both were factory default after clearing with ON-A-F (no memory card installed on the 48GX, FWIW). The speed of cycling through the menu options looked to be the same as yours, with the 48G cycling significantly faster than the 48GX.

I would suggest a different test:
- Open the EQN LIB on both calculators
- Simultaneously, press and hold the down-arrow key on both calculators
- Hold the keys down long enough to go through the entire list of categories a couple of times, then release the keys at the same time

Depending on how long you hold the keys, I believe you will find better parity between the two, with the 48G accumulating a slight advantage over time. At least that's what I observed when comparing my 48G/48GX.

The menu-cycling speed variation in the original test is more likely due to the two calculators running different code in that specific scenario than any inherent CPU speed differences or memory/garbage considerations. I believe the alternate test above is more representative of the actual UI differences between the two calculators.

It is interesting, though, that there's such a difference between the two systems in the original test. I wonder if the underlying form-processing code is calling different subroutines on those two calculators, or perhaps the key-repeat logic is different for some reason. Regardless of the cause, it is still a curiosity.

[ettlz]

I wonder if the menu cycling involves data at different ends of the address space, resulting in bank switching overhead on the 48G+/48GX? The machines are otherwise identical in speed (tested using FFT on a 256-element vector).

Would be interesting to try this on a memory-upgraded 48G.