80 replies to this topic

[kucalc]

Yeah, so here it is, the first non-official fx-9860 SDK library in binary form that allows you to access the fx-9860 hardware: EPSILON. It's a powerful hardware library. You can download it right now: http://www.casiocalc...hardware-lib-fo

I was planning to release a new SDK, but this has to come first. Included in the archive is the library, header file and a example project using EPSILON. The RTC and port reading functions will even work under the emulator! However, USB and serial communications will not work in the emulator. Much more functions will be added very soon, I just didn't have time to create C functions for the routines.

Tell me what you think about EPSILON!

Please read the README file to see if it will answer any of your questions.

[Menno]

great

happy happy joy joy

i will check it out this afternoon



Kucalc you are a genius

[kucalc]

Thanks Menno. I'll add more by functions by tomorrow.

[Menno]

Heya, i am trying some things out with your library but i cannot seem to keep the time updated, when i start the application the time is displayed but it is not updated. Also i do not understand how to set the time

Would you be so kind to show the source code of your clock application (with the "press f1 to set time")

This is my code

#include "fxlib.h"
#include "stdlib.h"
#include "stdio.h"
#include "epsilon.h"

char buffer[52];

int AddIn_main(int isAppli, unsigned short OptionNum)
{
	unsigned int key;
	int year, month, day_of_month, day_of_week, hour, minute, second;

	Bdisp_AllClr_DDVRAM();

	while(IsKeyDown(KEY_CTRL_MENU) != 1){
		Bdisp_AllClr_VRAM();
		RTCReset();
		year = RTCReadYear();
		month = RTCReadMonth();
		day_of_month = RTCReadDayOfMonth();
		day_of_week = RTCReadDayOfWeek();
		hour = RTCReadHour();
		minute = RTCReadMinute();
		second = RTCReadSecond();

		sprintf(buffer,"%d-%d-%d %d:%d:%d",day_of_month,month, year, hour, minute,second);
		PrintMini(1, 58, (unsigned char*)buffer, MINI_OVER);
		Bdisp_PutDisp_DD();
		

	}

	return 1;
}

#pragma section _BR_Size
unsigned long BR_Size;
#pragma section


#pragma section _TOP

int InitializeSystem(int isAppli, unsigned short OptionNum)
{
	return INIT_ADDIN_APPLICATION(isAppli, OptionNum);
}

#pragma section

[Menno]

aha

removing

RTCReset();

did the trick

but i still would like to know how to set the timers

i mean, in your clock utility you can set the date and time, i would like that too

[kucalc]

Ah, so you were able to install EPSILON easily? That's good to hear.

Well, about setting the date and time (let me see if I remember correctly):

1. Stop RTC timer first:

RTCStop();

2. Reset RTC timer next:

RTCReset();

3. Set the time and date now by using these functions. Just split your desired value into BCD:

/*
Functions for accessing the RTC. The SuperH sees these values
as BCD (binary coded decimals).
Example:
thousands = 2;
hundreds = 0;
tens = 0;
ones = 0;
Will give you: 2007 in decimal
*/
void RTCSetYear(unsigned char thousands, unsigned char hundreds, unsigned char tens, unsigned char ones);
void RTCSetMonth(unsigned char tens, unsigned char ones);
void RTCSetDayOfWeek(unsigned char ones);
void RTCSetMinute(unsigned char tens, unsigned char ones);
void RTCSetSecond(unsigned char tens, unsigned char ones);

4. Start the RTC again:

RTCStart();

Hope this helps.

If you can, it would be nice if you can write some tutorial for EPSILON. When I release the functions for the USB, serial port and grayscales, a lot of people are going to be asking questions on how to use the functions.

[Menno]

BCD wehehehe ok.... help :|

BCD code

0=0000

1=0001

2=0010

3=0011

4=0100

5=0101

6=0110

7=0111

8=1000

9=1001


did i get this right?

but then 2007 would be

0010 0000 0000 0111

i will fiddle with it in the afternoon (have to work now)

don't you have the source of clock? It would be so helpfull

[kucalc]

but then 2007 would be

0010 0000 0000 0111

BCD is what the calculator understands, you don't actually use BCD with the EPSILON library. Let's say we want to change the year to 2007. 2 is the thousands place, 0 is the hundreds place, 0 is the tens place and 7 is the ones place. Here's the prototype of the function for setting the year:

RTCSetYear(unsigned char thousands, unsigned char hundreds, unsigned char tens, unsigned char ones);

Now, let's actually use it:

RTCStop(); // Stop the timer first before we can change it
RTCReset(); // Reset just in case so it doesn't get messed up
RTCSetYear(2, 0, 0, 7); // Now we actually set the year
RTCStart(); // Now we turn the RTC back on

You would do the same thing for the other functions. So for example, in 31, 3 is the tens place and 1 is the ones place. Hope this clears up the confusion.

[kucalc]

Ok, I made a new utility using the next release of EPSILON: CPUSPEED. Download from here:

http://www.casiocalc...cpu---double-sp

You can double the speed of your calculator. You will notice changes in how fast the fx-9860 does it's functions. It may look normal when you exit out of CPUSPEED after changing the speed, but go to GRAPH mode and graph some functions. Run some programs and notice how fast they go now.

I could tripple (ex: 20MHz->60MHz) and quadruple the speed (ex: 20MHz->80MHz) making your calculator go ultra (I mean ULTRA) fast, but it might damage the calculator. When I release EPSILON v0.2, you can do that yourself.

[eew]

I think overclocking functions are a bit overpowerd, I mean, if it could damage the calculator why would you use it, only to take the risk? How do I if know someone use it. How do I know iof add-ins are not destoying the hardware of my calculator? If you release that stuff?

[Andreas B]

It is up to people how to use it. If you can code C, you can be careful with this. I think it is good that more experimental things get out - CPU?clock?limitations belong in a program that is made for users (that don't need to know technical stuff), not in an SDK for developers.

[eew]

I know, white software it is possible, but you have to be very carfull whit hardware
Allow this kind of things and realease such a dangerous whitout placing a waring

[Menno]

sweeeeeet i will get some liquid nitrogen and have the first ghz calculator whehehehehe just kidding, but this is still awesome man, i gues some overclocking can't hurt the calc, we should investigate it some more using thermocouples placed near the cpu

you are fucking brilliant kucalc (you should rename yourself to kungfucalc )

[eew]

No but if this is the first step, what's comming next?

[Menno]

also a question, is it also possible to clock the cpu down? it could be a very nice feature for energy saving, for example in submenu's or while waiting (i write non cpu intensive programs for my job (simple calculations with many many parameters) and this could help the battery consumption)

No but if this is the first step, what's comming next?

whehehe i bet a casio doomsday device...I wouldn't be surprised kucalc turns out to be dr. evil opting for world domination...



kidding ku



i guess i smell burnt plastic

[Manuel Naranjo]

GOOD JOB
I would love to see how you did it. I think you must have found the Timers Watchdog Address.
I tried vanhoa 3d drawer with more speed and it is amazing.
Cheers,
Manuel

[Xerxes]

Very good work kucalc. I love your speedup program.

Why not an Addin with selectable frequency with warning of using on own risk? But I don't think that it is really a risk. For example it is possible to overclock the HP-50G by software to 150 MHz instead of 75 MHz without problems or to 233 MHz with problems.

[kucalc]

I think overclocking functions are a bit overpowerd, I mean, if it could damage the calculator why would you use it, only to take the risk? How do I if know someone use it. How do I know iof add-ins are not destoying the hardware of my calculator? If you release that stuff?

Don't worry about it, the CPU in the fx-9860 is able to handle up to 133MHz. The power consumption is consistent even if you change the frequency. I think the fx-9860 has long battery usage (If I remember, over 200 hours of non-stop use). I've been running my calculator at quadruple speed (if the oscillator is running at 40MHz, that means I'm running at 160MHz! If the oscillator is instead at 20MHz, then I'm running at 80MHz) for about a day now and nothing has happened yet.

Awesome ! I want more speed !!!

Thanks!

brilliant kucalc (you should rename yourself to kungfucalc )

That would be an awesome name!

also a question, is it also possible to clock the cpu down? it could be a very nice feature for energy saving, for example in submenu's or while waiting (i write non cpu intensive programs for my job (simple calculations with many many parameters) and this could help the battery consumption)
whehehe i bet a casio doomsday device...I wouldn't be surprised kucalc turns out to be dr. evil opting for world domination...

You can slow down the calculator down to a fourth of the oscillator. But as I said earlier, changing the CPU speed doesn't affect power consumption. The CPU is very consistent. The CPU only needs 2.0v to function even at 133MHz. We should do a test to prove this. Who is willing to leave their calculator running non-stop?

GOOD JOB
I would love to see how you did it. I think you must have found the Timers Watchdog Address.
I tried vanhoa 3d drawer with more speed and it is amazing.

Awesome, note that the USB speeds won't change. The USB uses a different clock source. I could change the frequency of the USB also if you want, however I don't think it will be able to communicate with the PC anymore.

Very good work kucalc. I love your speedup program.

Why not an Addin with selectable frequency with warning of using on own risk? But I don't think that it is really a risk. For example it is possible to overclock the HP-50G by software to 150 MHz instead of 75 MHz without problems or to 233 MHz with problems.

Well, currently I'm seeing no problems with the CPU speed quadrupled. I'll re-release it again when I have the time. The CPU speed is determined by multiplying the oscillator speed by a constant. If you changed the adjustable oscillator in your calculator and set it to 60MHz (the oscillator can be adjusted to 160MHz), then after running my CPUSPEED program, you CPU will run at 120MHz. LOL! If you changed the speed of the oscillator to 160MHz, then running my CPUSPEED program will give you 320MHz! Quadrupled: 640MHz (The fastest handheld graphing calculator in the WORLD!)!!!! But your fx-9860 will be absolutely fried up. About 100MHz is still fast though and you're not really overclocking when at 100MHz, since the CPU can take 133MHz.

Hey Xerxes, can you run your speed comparison program again (after running my CPUSPEED program) and give me the results? It would be nice if you could do that.

[Xerxes]

Casio-Basic:

17.3 sec / 11.6 sec = 1.49 times faster


SDK-C:

248 msec / 129 msec = 1.92 times faster

(Unfortunately I was not successful on compiling with optimization for speed.)

[Manuel Naranjo]

Awesome, note that the USB speeds won't change. The USB uses a different clock source. I could change the frequency of the USB also if you want, however I don't think it will be able to communicate with the PC anymore.

Don't even think about doing that. USB 2.0 is uses not only another oscillator, but also uses another stuff. It's not so easy to make the usb faster. Maybe Fa-124 runs faster with your tool.

Well, currently I'm seeing no problems with the CPU speed quadrupled. I'll re-release it again when I have the time. The CPU speed is determined by multiplying the oscillator speed by a constant. If you changed the adjustable oscillator in your calculator and set it to 60MHz (the oscillator can be adjusted to 160MHz), then after running my CPUSPEED program, you CPU will run at 120MHz. LOL! If you changed the speed of the oscillator to 160MHz, then running my CPUSPEED program will give you 320MHz! Quadrupled: 640MHz (The fastest handheld graphing calculator in the WORLD!)!!!! But your fx-9860 will be absolutely fried up. About 100MHz is still fast though and you're not really overclocking when at 100MHz, since the CPU can take 133MHz.

The only problem with making the speed faster is memory synchronization. If you make the speed to high you will start getting memory problems. That's why you can't overclock much more. Anyway the actual speed (after running kucalc tool) is quite good.

Cheers

[kucalc]

(Unfortunately I was not successful on compiling with optimization for speed.)

Thanks Xerxes, it's good to know! Anyways, it's okay, I don't think optimization for speed would have made really difference.

[Xerxes]

I am waiting impatiently for your adjustable Addin. I want to see the FX-9860G on the first place in my list.

[kucalc]

Ok, I released it finally. Let's see if it will work. When playing Space Invaders with the fx-9860 quadrupled, it was crazy! Note that the CPU Speed Changer affects the CPU (hence it's called CPUSPEED), not the rest of the hardware. That means the computational and processing power of the fx-9860 is increased, while the hardware stays the same (key input and LCD stay the same). Also, you can change the frequency only once before you have to reset again. Anyways, download it off here: http://www.casiocalc...cpu---double-sp

Xerxes, can you do a speed test again (I wonder what the results will be when the speed is quadrupled)?

[vanhoa]

Great! It cant crash the calc any way, hope i have a 9860 now

[vanhoa]

[b]Increasing Instruction Execution Speed[/b]
To improve instruction execution speed, consider the following when programming:
? To prevent contention between MA and IF, locate instructions that have MA stages so they
start from the longword boundaries of on-chip memory (the position when the bottom two bits
of the instruction address are 00 is A1 = 0 and A0 = 0) wherever possible.
? The instruction that immediately follows an instruction that loads from memory should not use
the same destination register as the load instruction. This will avoid causing contention with
the memory load triggered by the write back (WB).
? Locate two instructions that do not read SR immediately after any instruction that overwrites
the M, Q, S, and T bits of SR. This will prevent contention with SR update instructions from
occurring.
? Locate instructions that use the multiplier nonconsecutively (excluding PWULS).
? Immediately following a data operation using the DSP unit, do not use an instruction that
transfers data to memory or the CPU core from the register where the operation result is stored.
By placing some other instruction in between, contention can be avoided.
? Do not use MOVX.W, MOVS.W, or MOVS.L to perform a memory store immediately
following a PLDS or PSTS instruction using the DSP unit. Also, do not specify a PLDS or
PSTS instruction in parallel with a memory store instruction using MOVX.W.

Key Features:


Operating frequency
100/133/167/200MHz
...
Other features
High performance, Low power voltage (200MHz/167MHz)
High-speed version of SH-3

Wow, CP can be speed up to 200MHz!!!

[kucalc]

I think CPUSPEED is the first speed changer through a program for a CASIO calculator without having to open up your calculator and changing the oscillator. I'm patiently now waiting for Xerxes results. Right now, I'm working on a music player now that I found the Digital to Analog converter. This should be interesting.

Wow, CP can be speed up to 200MHz!!!

Umm... Let's see you do that. For that to happen, since the oscillator can be max only quadrupled, the oscillator in the ClassPad would have to be 50MHz, but I think it's actually less than that. But as of right now, the fx-9860 is WAY faster than the ClassPad 300 thanks to CPUSPEED.

[vanhoa]

No, I dont do that (but i have the cpu speed up source code from Hitachi support) since it will lower the live-time of ClassPad (for example, a classpad speed x2 will 4 time shorter life than the original x1).

The only way to faster your calc safely is replacement! (if you speed the cpu x2, I doubt that you will buy a new calc in < 2 years)

[Xerxes]

@kucalc:

Basic:

speed   sec	speed up
1x	  17.3	 1.00
2x	  11.6	 1.49
3x	   9.9	 1.75
4x	   9.1	 1.90


C:

speed  msec   speed up
1x	  248	  1.00
2x	  129	  1.92
3x	   88.8	2.79
4x	   69.1	3.59

Its interesting to compare the speed up factors of C and Basic. The C program runs in RAM and the Basic interpreter is executed in ROM. So the RAM access seems to be much faster than access to ROM.

There is another Casio with software speed up, but only 20% (PB-2000C with Pascal-Compiler-ROM).


@vanhoa:

I don't think that it's a problem if you only use speeding up if necessary and not all time. I am using some speeded up calculators since many years without problems.

[eew]

But if you run this you have to reset you wole calc. Or is point P enogh?

[Xerxes]

I turn the power off and turning the calc on by pressing (P).

[kucalc]

The only problem with making the speed faster is memory synchronization. If you make the speed to high you will start getting memory problems. That's why you can't overclock much more. Anyway the actual speed (after running kucalc tool) is quite good.

Yes that is correct. That's why CPUSPEED only changes the speed of the CPU alone, the rest of the hardware is left unaffected. Forcing the hardware to go faster is when the calculator will crash.

No, I dont do that (but i have the cpu speed up source code from Hitachi support) since it will lower the live-time of ClassPad (for example, a classpad speed x2 will 4 time shorter life than the original x1).

The only way to faster your calc safely is replacement! (if you speed the cpu x2, I doubt that you will buy a new calc in < 2 years)

@Manuel Naranjo : I dont think overclock whithout replace the chip is a good way. Casio told they used the best speed...

First you say that it is impossible to change the speed of the CPU on MSN Messenger...

Dude, look up overclocking in a dictionary. Overclocking is when you force the CPU to go faster than the CPU's manufacturer rating. Max speed of the CPU is 133 MHz. Quadrupling the CPU oscillator takes you to only 80MHz. I still have 53MHz to go before I reach the point where I am now overclocking. That's why CPUSPEED is safe, it doesn't overclock and affects only the CPU. Most people are confused, because they think the default speed of the fx-9860 is the max speed (100%) of the CPU. In reality, the fx-9860 without CPUSPEED, is using only 15% of the processing power of the CPU. CASIO hindered the fx-9860 on purpose (like how they use a 512kB RAM chip and only allow the fx-9860 to use 64kB). My CPUSPEED allows you to use max of 60% of the processing power of the CPU.

The calc freezes if cpu speed has already been changed and you try setting another frequency.

You have to reset the calculator before changing the frequency again. Press the RESET button in the battery compartment, don't worry, it only reboots the system, it doesn't clear any memory.

@kucalc:

Basic:

speed   sec	speed up
1x	  17.3	 1.00
2x	  11.6	 1.49
3x	   9.9	 1.75
4x	   9.1	 1.90
C:

speed  msec   speed up
1x	  248	  1.00
2x	  129	  1.92
3x	   88.8	2.79
4x	   69.1	3.59

Its interesting to compare the speed up factors of C and Basic. The C program runs in RAM and the Basic interpreter is executed in ROM. So the RAM access seems to be much faster than access to ROM.

Wow! It's nice to see that the fx-9860 is now in first place!: http://www.hpmuseum....es.cgi?read=700

But if you run this you have to reset you wole calc. Or is point P enogh?

Only pressing Point P, don't worry, it doesn't clear the memory, it only reboots the system back to it's default settings.

Ok, I have to work on my music player for the fx-9860 now.

[Xerxes]

Sorry kucalc, but your link is not correct: http://www.hpmuseum....es.cgi?read=700

I like it to have a Casio on the first place.

Overclocking means a higher power consumption of the CPU that results to more heat and a shorter life of it. But as you wrote, the power consumption dosn't changed. So I also think that the CPU isn't really overclocked.

[The_AFX_Master]

Good job dude!!!!, it's amazing!!!

Can you code it for the Classpad?, some CPLua users would appreciate it!

[kucalc]

Sure, I'll see what I can do. I'll go download the ClassPad SDK. Can you test for me once I finish writing the add-in?

[The_AFX_Master]

For sure!.. i can be your beta tester, don't worry.. only give me 2 weeks (i'm on exams right now )

[kucalc]

I know this isn't a classpad forum, but how do I get the source. doesn't the 9860 use the same bios? I want this for the classpad.

AFX_Master, can you check the IP address of the user who posted this and check if it has been used before? I think I have a good idea who posted this, lol ....

[TyYann]

I got a bad side effect with CPUSpeed : when the calc uses it, it can't communicate with another calc via 3-pin cable any more... Even if both cals are set at the same speed. We had to reset our calcs to restore communication...

[Menno]

Hey Kucalc,

I think i found a problem in the Epsilon library,

I am trying to make a clock and when i try to set the timers i get an error:

** L2310 (E) Undefined external symbol "_RTCSetHour" referenced in "C:\Program Files\CASIO\CASIO FA-124\clock\Debug\smlclock.obj"

so i opened the epsilon library and

void RTCReset();
void RTCStart();
void RTCStop();
void RTCSetYear(unsigned char thousands, unsigned char hundreds, unsigned char tens, unsigned char ones);
void RTCSetMonth(unsigned char tens, unsigned char ones);
void RTCSetDayOfWeek(unsigned char ones);
void RTCSetMinute(unsigned char tens, unsigned char ones);
void RTCSetSecond(unsigned char tens, unsigned char ones);

there is no RTCSetDayOfMonth and no RTCSetHour

that explains i guess the error i get :|

could you please upload an updated version of Epsilon?

[Menno]

when the set hour problem is solved i will upload my leet digital clock


it supports zero's like 01:01:01

in the settings it detects the maximum days of the month including february leap and non leap

but best of all, the large digital clock, resembling a real 24 bomb-like counter....now you can be Jack Bauer yourself