| Posted By
MCes
on 2022-10-21
09:06:57
 |  what about the "superCPU" for C16? can be a ... super TED for C16!
We recently had an interesting conversation with @MMS and @gerliczer about the CPU acceleration of the 264 machine.
forum/45509#45603
Everyone has different ideas about it, but I think the most appropriate superCPU for the C16 is .... the CPU that is already installed in the C16!
The machine will remain compatible with old software (no fast C-MOS cpu needed, that can't manage the illegal opcodes...), but analyzing the TED access to the @2MHz bus of C16 it can be found that at the end of an active frame the active cpu cycle are not much more than an equivalent machine that run at @1MHz without graphical/refresh accesses,
so the idea for a "simple" superCPU: the OLD cpu that ALWAYS run at @2MHz with a modified TED that use the bus dead times for getting the graphical information (from a Static RAM, obviously...).
Immagine:
A) at power on it is a common C16,
B) POKEing into a register the C16 switch to "turbo": @2MHz operation but with "bad lines" (the TED will stop the processor into DMA only for fetching more datas every 8 active lines)
C) POKEing into another register CPU run at @2MHz and will be not halted by DMA: special care for using the TED registers (not directly compatible with kernal routines)
A): standard behaviour (100% compatibility: you don't lose your C16....)
B): near double frequency with all sw compatibility but old/not specific SW that are time-dependent can't work properly
C): the maximum CONSTANT speed of operation (always @2MHz, not any TED access, never!), do you imagine the SW benefits? (price: accessing to TED registers need precautions that kernal routines don't have...)
Immagine a game that can detect this board and, if detected, double the graphical frame-rate...
It can be done with a board to plug on TED socket, the TED chip has to be fitted on the new board...
Is the advantage (always double speed) enough to modify your C16?
|
|
Posted By
SukkoPera
on 2022-10-21
10:55:13
| Re: what about the "superCPU" for C16? can be a ... super TED for C16!
Hey, what about this? It's a different approach and wouldn't work as-is on the C16/+4, but I guess we could work with the developer in order to achieve that!
|
|
Posted By
MMS
on 2022-10-23
16:49:06
 | Re: what about the "superCPU" for C16? can be a ... super TED for C16!
@MCes
Thank you for spending some valuable time on our "SuperCPU"  This idea sounds great. Double speed is nice
The c) version seems to be equivalent to the C128's Fast mode Certainly the SRAM really helps here.
Actually I have only one concern: I have a theory, that the original 7501/8501 ICs started to fail after they continouosly switched to >2MHz mode by the decompressors of the newer converted games. (some programmers stated that the hackers knew this trick far before us).
Certainly at 2MHz the IC generate more heat, but probably a proper bigger heatsink will do the job.
|
|
Posted By
MCes
on 2022-10-24
11:20:08
| Re: what about the "superCPU" for C16? it can be a.... "superTED" for C16!
My "JUGGLER" card was intended as the card that closes a personal circuit within the possible projects around the '264 machine, I'm not normally interested in upsetting systems modifying the basic behaviour of our Commodore, but the "superTED" solution intrigues me.
@SukkoPera
your proposed system , in my opinion, is too far from an original C16 to be called another time "C16", the motherboard will be transformed a few more than a power supply for the ROMs...
And... where is a 65F02 stock? at which price? wich availability?
Personally I don't like this solution.
@MMS
2 MHz make more heating than 1 MHz? yes, but not a lot more and normally the heating not more than 70° in electronics are considered not a problem (70°= you can't stay with finger on it).
From personal experience ESD breaks NMOS chips, ESD that arrive at the chips through the pins of integrated parallel port: the CPU is killed by tape+serial connector pins, the TED by joystick pins.
If you still want to lower the temperature inside you can use low power devices for the ROMS and for PLA, you can use a DC-DC converter instead the 7805 voltage regulator (and cut out the white resistor near it...).
The "superTED" I have in mind use the same strategy as the C64 but clocked at 2 MHz with memory running at 4 MHz (cheap and fast RAM was not available in the 80s ...)
Immagine a C16 bus that work always at 2 MHz but during the low state of the clock (CPU is not accessing to the BUS) an extra circuit link the memories at the TED chip
|
|
Posted By
MMS
on 2022-10-25
15:07:52
| Re: what about the "superCPU" for C16? can be a ... super TED for C16!
@MCes
Thank you for the information. I tried to catch up and read the latest information about the CPU deaths, and I see there is a consensus related to the external ports as the main reason.
The explained method has real potentials.
As you know I already rebuilt my C16 to have low power PLA and ROMs and I use stabilized 9V input with 1.5A limit.
Update:
But can it run Crysis? :-D
https://www.youtube.com/watch?v=g1leK1PJb9o
|
|
Posted By
unclouded
on 2022-10-27
03:09:23
| Re: what about the "superCPU" for C16? can be a ... super TED for C16!
@MCes
"you can use a DC-DC converter instead the 7805 voltage regulator"
Does a DC-DC converter introduce noise on the rails? I use a buck regulator to supply 7V to the power jack instead of the original C16 power brick but I keep the linear regulator inside for a smooth final stage. Using 7V instead of the 10V I get from my power brick means the 7805 has less voltage to drop -> less heat.
|
|
Posted By
MMS
on 2022-10-28
15:46:34
| Re: what about the "superCPU" for C16? can be a ... super TED for C16!
Thanks for the hint. I have a multivoltage psu brick, but always used the official 9V DC. I will make a try with 7V too.
|
|
| |
Copyright © Plus/4 World Team, 2001-2024 |
