View Single Post
Old 04 July 2014, 13:54   #177
Registered User
Join Date: Jun 2010
Location: NL/PL
Posts: 1,430
Originally Posted by FrenchShark View Post
Interresting stuff about brigette.
In the case of an accelerator + AGA in an FPGA, the whole chip RAM bus and the CPU/Chip bridge is anyway integrated into the FPGA.
Moreover, it is easier to stay with 16-bit Chip RAM bus and to increase the bus speed than to use a 32-bit bus.
Yes, in theory but going this way - it is best to hook ECA/AGA directly to FPGA and reverse situation - with fast RAM (controller and arbiter in FPGA) almost all bandwidth can be dedicated to CPU, fraction of bandwidth used by chipset - im thinking currently about such machine.

Originally Posted by FrenchShark View Post
As for the faster 8520 hack, if it does not work you will end up with a Guru right away
My question was realted to two things - first SYBIL (mentioned by JimDrew) - never saw this device but based on functionality seems it is controlled clock source for whole Amiga - based on this it was possible to achieve variable datarate to read Macintosh floppies. But this is my speculation as i never found anything on SYBIL.
Second of course 68000 in pre Gayle systems where E clock is produced by CPU - if timer.device is capable to sense E clock value then no problems with 14MHz MC68000 (yes, i know - i can build clock generator as in Gayle and modify timing for Gary but then why not build remains things).

Originally Posted by FrenchShark View Post
I have one question about Kickstart 3.1 : does the A500/A2000 3.1 ROM contain AGA routines in the graphics library ?

No clue - not my area...


Originally Posted by Mrs Beanbag View Post
The first part about 16 bit samples, an imaginary Amiga with better sound capabilities.
Ordinary Paula can be modified - there currently rarely used functionality where one channel can modulate amplitude/frequency or both for other channel but for amplitude modulation only half of samples is used - should be possible for example in new Paula to have possibility to combine two channels to form 16 bit samples.

Originally Posted by Mrs Beanbag View Post
The second part about using the Copper to poke data into Paula, can you do that on any real Amiga?

Not analyzed code but perhaps other possibility is put reasonable high sampling frequency then set pointer in memory to 1 word filled with null data - audio should be looped and played as usual however with CPU or more convenient with Copper (as timing is critical) feed from time to time real samples - this should work quite well also. Side effect is that probably audio level must be set as maximum to avoid distortions produced be volume adjust mechanism (seem it is PWM like).

Last edited by pandy71; 04 July 2014 at 14:02.
pandy71 is offline  
Page generated in 0.05583 seconds with 9 queries