Amiga 500 14 MHz accelerator with DRAM (phase 2)

The 14 MHz accelerator card project has reached a stage when real speed measurements are made possible. Based on Sysinfo speedtest results, it produces exactly twice the speed of a real Fast RAM equipped Amiga 500. See the pictures below for the wonderful, long awaited results. :)

This is not a simple 14 MHz hack, which is available on aminet. An accelerator without 0 wait state RAM worths nothing, as the CPU is always waiting for the Chip RAM and register access. We can talk about a real accelerator if there is some 0 wait state RAM available.

This card contains a 68HC000 16 MHz CPU which is running at 14 MHz, 8 MB 0 wait state RAM, which is the maximum size of the real Fast RAM in an Amiga with 24 bit address bus.

The next version will be built on a complex PCB which will contain:

• 14 MHz 68HC000 CPU
• 8 MB RAM
• IDE connector
• CF card
• Flashkick
• NMI button etc.  

A500 68k turbo rev 2. (top)

A500 68k turbo rev 2. (bottom)

A500 68k turbo rev 2. (top)

A500 68k turbo rev 2. (bottom)

FlashKick rev. 1 - A 256k x 16 bit EPROM / FlashROM adapter for Amiga

Finally I had a little time to complete the long awaited 16 bit Kickstart EPROM / FlashROM adapter card on a PCB. There is no much to say about it, look at the pictures below.

Amiga 500 68HC000 @ 14 MHz accelerator (phase 1)

A new month, a new project. This is the first phase of the 68k (maybe 020, 030 later?) accelerator card for the Amiga 500.

The roadmap is:

1. Getting the bus from the motherboard CPU
2. Resolving the E clock issue, as the MB CPU always generates the E clock, even when it is not a bus master
3. Designing an external 7 MHz CPU board, with asynchronous CIA access (using the MB E clock)
4. Accelerating the system to 14 MHz (sync mode) or 16 MHz (async mode) 
5. Fixing the DTACK issue when accessing the amiga bus (chips)
6. Adding own dynamic RAM running at 14/16 MHz

Steps 1-4 are complete, the system is running stable at 14 MHz. As the program is running from the Amiga's RAM, the speedup is very tiny (cca. 12%) as visible on the screenshot below.

Like my other projects, this card is also built on a home made 2 layer PCB:

And some CPU measurements at 7 and 14 MHz:

7 MHz normal CPU cycle (7M clock, /AS, /DTACK)

14 MHz CPU cycle with unmodified /DTACK (14M clock, /AS, /DTACK)

14 MHz CPU cycle with delayed /DTACK (14M clock, original /DTACK, delayed /DTACK, /AS)

Kefmania running at 14 MHz

Powerpacker measurements also confirm the 12-13% speedup, as expected.

More details are coming soon. Stay tuned! :)