12 replies to this topic

[algailani]

  In 2005 I got my fx-9860G SD, one of the earliest releases of the model. It contains a coin cell battery for memory backup (never needed). I decided to try the new fx-9860GII PowerGraphics 2. As a first impression I didn't notice a performance differences between the two, so I wrote a simple nested loop program to test the processor speed independent of any algorithm.

For 0 -> X To 100
For 0 -> Y To 100
Next
Next

Locate 7,4,”DONE!”

Note: fx-9860G SD OS version is 02.01.0200, while the fx-9860GII PowerGraphics 2 OS version is 02.04.0200.

 

  The results were surprising to me. My 9 year old model wins the race every time. Tried different values 100, 200, …, 1000 and as I increase the loop number the difference between the two calculators to reach the end increases  .

 

I was expecting an increase in performance for the new model, or at least same performance for both models  . Could someone explain why this degrade in performance for the new Power Graphic2 model?

[TeamFX]

In 2005 I got my fx-9860G SD, one of the earliest releases of the model.

 

Could you tell me the product number on the back?

Something like this:
G363-74
A118195 8Q0128 R/H B
 

Could someone explain why this degrade in performance for the new Power Graphic2 model?

 

You probably see the same difference between OS 2.04.0200 (SH-3, 2009 models) and OS 2.04.0201 (SH-4A, 2011 models). The SH-4A software branch is a lot more bloated than the SH-3 branch. Mostly because of backports from the fx-CG 10/20.

Edited by TeamFX, 24 October 2014 - 01:50 PM.

[algailani]

Model #:G363-51

[TeamFX]

And the production date?

(the remaining product codes)

 

That would be interesting.

[algailani]

G363-51
A138924
5Q10602A

What information can we get from this code?

[TeamFX]

G363-51 (hardware identifier):
- Model GY363
- Revision 2005-1

A138924 (device number):
- We assume production started with A100000

5Q10602A (production date):
- Date 2005-06-02

We know that OS 1.00.0000 was compiled on 2005-05-18 13:31 and the oldest boot code on 2005-05-11 17:48
You can check this by turning off the calculator followed by pressing OPTN+EXP+AC/on, which should display a diagnostic mode window, followed by F1, 9 and 4. DateO is the OS timestamp, DateA the boot code timestamp.

That machine is quite old, but I don't think there is anything special about it.
 

Edited by TeamFX, 25 October 2014 - 12:46 PM.

[algailani]

Interesting information.
This old machine is faster than my 2014 one! If casio upgraded the 2005 hardware (SH3) to 2014 (SH4), then the performance should be improved, even if the OS became bigger. Do you have a test in mind that verify the new product is superior compared to the old one?

[TeamFX]

The SH-4A CPU uses a superscalar architecture which allows execution of two instructions in parallel if there is no dependency between the two.

Some time ago, I created two assembly programs where one can be optimized by the processor and the other one cannot. I could confirm that the fx-CG 10/20 frequency is around 58 MHz.

I did the same test on an fx-9860GII with an SH-4A CPU (which is identical to the fx-CG 10/20) and I could confirm that these models are a bit slower than their predecessors in normal operating mode, but they are clearly faster in superscalar mode. The frequency is around 29 MHz.

So, the CPU got a lot more complex but in general it should still be faster or at least similar fast compared with older models. But it is not and I can only blame badly optimized software for this.

Edited by TeamFX, 26 October 2014 - 02:26 PM.

[sentaro21]

Hello everyone.

 

I got the GII-2 model before a little too.

I met with the same problem.

 

I do not know whether it is the problem only for this model,

backlight on and it will slow down about 20%,

but, it becomes fast as about 30% when it connected the USB.

This is unrelated in the add-in by the problem only for internal interpreters.

 

Then, I got this result after variously investigation,

I found that memory access speed was considered to be considerably slow setting in comparison with SH3 model.

I think that this is to satisfy speed balance between Prizm.

If the same settings as the previous model , does not have to be slower than before.

 

I developed addin tool to liberate native power of GII-2 model now.

Please try it if you'd like.

 

http://www.casiocalc...casm#ftune2-100

[TeamFX]

I can now confirm that the fx-CP400 operates at 116 MHz.

Some FRQCRA values set by the OS:

CP400: 0x8F001102 (116 MHz)
Prizm: 0x8F102203 (58 MHz)
CP330+/FD10/9860/9750/7400: 0x8F202203 (29 MHz)

Edited by TeamFX, 17 November 2014 - 11:49 PM.

[sentaro21]

fx-CP400 exceeded by default 100MHz over!!

 

Thanks very much always for the deeply interesting information!

[TeamFX]

fx-CG50: 0x8F001102 (116 MHz)

[sentaro21]

fx-CG50: 0x8F001102 (116 MHz)

I understood about the same specifications CG50 and CP400(CG500)

I am always grateful to your information.