Terrible Fire Accelerators

[alenppc]

I flashed the latest firmware on the a1000. The microbotics expansion card now works. I tried the memcheck utility from Aminet and the ram passes tests but chip ram and z2 ram fail address tests... not sure what that is about. Haven't had time to check the rest yet, the kids are demanding tonight... I also have kippers card to try

[dalek]

Have you seen Mike Sterling's SPI drivers for the "k1208"?

Looks like he already has the network module working. They are GPLv3 and as I understand it that board is a fork of the TF320 (so the cpld wiring/code is/was similar to the TF530 from memory?)

https://github.com/mikestir/k1208-drivers

[8 Bit Dreams]

Originally Posted by plasmab

But there are already cards that do this... cheap ones.

I really try to do things that havent been done before.

I'd like to know the timings of that 8Mb card.. is it the one with the GFX card on it? is there firmware code somewhere?

You are right, there are some cheap cards available, but mostly these are 4-8MB ZII cards, none of them has 64MB, and none of them allocates RAM in Zlll space, this is a real pain in the ass if we try to use it with TF530, and next problem is DTACK line, if this missing - card fails to work, right?
And third - these fastRAM are slow (7Mhz) atm, could be nice if RAM card can check what CPU is installed and try to sync at higher frequency if 030 is detected (since no one will run 030 CPU @ 7Mhz)..
There are two ways to go: try to solve the issue (im afraid there will always be RAM cards that doesn't work, or build one that will work with both accelerators (and has more RAM available)..
(Yes, i really want that persons owning TF530 continue to use them instead of buying new accelerators)..

In my opinion it's smarter to go that way instread of designing a new 8MB version.. Cause even 8MB isn't comfortable to use with WB 3.9, and im sure, that's the target

[plasmab]

Originally Posted by dalek

Have you seen Mike Sterling's SPI drivers for the "k1208"?

Yes that’s what I’m porting to TF this weekend

[plasmab]

Originally Posted by 8 Bit Dreams

next problem is DTACK line, if this missing - card fails to work, right?

Not exactly. I fixed the timing so that for the ZII area, DTACK is ignored until S7. This fixes the kipper card and makes the timing like a 68000. But without other cards to put on the LA I can’t do anymore.

(Yes, i really want that persons owning TF530 continue to use them instead of buying new accelerators)..

In my opinion it's smarter to go that way instread of designing a new 8MB version.. Cause even 8MB isn't comfortable to use with WB 3.9, and im sure, that's the target

The more you want compatibly and attention on the older cards the less new cards you will get. I’m not that keen on supporting old cards unless I can do it with no effort!

The idea with my cards is that they are so cheap they’re not going to break the bank to upgrade.

[8 Bit Dreams]

Ok, i understand what You mean
it is just like reinventing the wheel, TF530 is a great card, and so is TF534, all we need - is a stable FW and a bigger RAM, and that is what im offering to do..in the end, not only TF53X cards will profit, also other accelerator cards will

[plasmab]

Originally Posted by 8 Bit Dreams

Ok, i understand what You mean
it is just like reinventing the wheel, TF530 is a great card, and so is TF534, all we need - is a stable FW and a bigger RAM, and that is what im offering to do..in the end, not only TF53X cards will profit, also other accelerator cards will

I wouldn’t bother. The next generation of TF cards will be SDRAM. There is no point in doing anything with the 53x cards. They are prototypes... why modify the prototype when the main production version will fix everything and add more.

If I can get my hands on some ZII cards that fail
I’ll fix them but the TF530 was an 030 plus IDE experiment, the TF534 was a ZIII and burst mode experiment (and a failed SATA one).. both are experiments that added value to the Amiga community at the same time. That is my ethos.

As we learn we can back-port the fixable bits to the older cards. But I also make each revision cheaper.

[plasmab]

Another way to put this might be... it only makes sense to maximise existing boards if the new boards aren’t extremely cheap to reproduce/replace. I don’t want 30 forks of my designs.. the more permutations the more support is needed... the more support I have to give the less new boards appear.

My approach is agile for hardware.

[AJCopland]

I like that approach, it means we can all learn from what you're doing with these cards.

[project23]

Originally Posted by plasmab

I don’t want 30 forks of my designs..

+1

At this point given your ethos, any 'forks' should be for personal use only and kept private. Not that i'm attempting anything myself, but pissing about with the 530 to add more ram like has been done elsewhere is fine in my opinion if its a one off for experimentation/personal fun. But it is not okay to get the community all hopeful and riled up - so i'd keep it to myself really (closed), ya know?

In short: this aint my project - i might play with it, but I sure wont step on the toes of the project leader(s) (this applies to all projects) in terms of their plans/wishes.

EDIT: Unless the project is left unmaintained and considered dead, which I think has been the ethical approach for a good while. (Dead for years, not weeks)

[plasmab]

If someone makes an improvement on a private fork and hands it to me ill incorporate. But right now 10x10cm is the board size limit. For cost reasons

[alenppc]

A couple of pics from the A1000.

I am going to check if I have any ZII boards left at home and if I do they are yours, Stephen. Although I suspect I probably got rid of them years ago.

[plasmab]

Originally Posted by alenppc

A couple of pics from the A1000.



I am going to check if I have any ZII boards left at home and if I do they are yours, Stephen. Although I suspect I probably got rid of them years ago.

Closer... but minus the cigar. Few tricks left in the bag. The bus CPLD is now full

That looks like a missing refresh rather than DTACK timing

[alenppc]

You still made major progress here. Missing refresh on external memory shouldn't have anything to do with the TF534 though, or am I wrong? I am now getting the same type of errors on the chip ram.

[plasmab]

Originally Posted by alenppc

You still made major progress here. Missing refresh on external memory shouldn't have anything to do with the TF534 though, or am I wrong? I am now getting the same type of errors on the chip ram.

That’s weird. Truly. What clock speed? I’ve been testing around 33mhz

[plasmab]

Originally Posted by plasmab

That’s weird. Truly. What clock speed? I’ve been testing around 33mhz

I’ve just realised my Z2 area decode is a pile of hairy baws. More later

[alenppc]

Originally Posted by plasmab

That’s weird. Truly. What clock speed? I’ve been testing around 33mhz

40 mhz

[plasmab]

Originally Posted by alenppc

40 mhz

40MHz is where my 16Mhz rated 030 soldered into my breakout board starts to crap out.

:/

Was that the kipper board?

[alenppc]

No, that was the external Microbotics. I haven't had the chance to test the kipper board yet, will do it tonight

[plasmab]

Btw... just for amusement...

TF534 fast ram bandwidth 14Mb/sec
Fast ZII ram bandwidth 3.5Mb/sec
TF534 CF card access 4Mb/sec...

A proper MMU based virtual memory setup would beat zorro ii add on cards. Or be v close.