Carrera040 Info / Hacking Thread

[Trash80toHP_Mini]

LOL! I'll bet that info's in the old thread, thanks much for explaining what you've done.

IIRC the SE/30 and IIsi use some control signals differently. Is Bolle sending you both GALs and do you have a IIsi for testing the differences between machines and adaptations?

 

[joethezombie]

The MicroMac IIsi 040 adapter doesn’t have a PAL/GAL, at least mine doesn’t.

 

[Bolle]

IIRC the SE/30 and IIsi use some control signals differently. Is Bolle sending you both GALs and do you have a IIsi for testing the differences between machines and adaptations?

Neither of both

 

[GeekDot]

LOL! I'll bet that info's in the old thread, thanks much for explaining what you've done.

IIRC the SE/30 and IIsi use some control signals differently. Is Bolle sending you both GALs and do you have a IIsi for testing the differences between machines and adaptations?

Sniff... I still miss that little thread  It's like loosing a good friend ;-)  (Well, at least a valuable notebook)

Yes, while Bolle will provide the GAL(s) he already "MicroMac'ed" it, i.e. no GAL, pads bridged accordingly [he beat me posting that by some seconds ;-) ].... if needed I can socket and plug/program them at (evil) will, but for now we hope to get it done like the original MicroMac IIsi 040 adapter did (and Joe rightfully pointed out). 

 

[Bolle]

So the IIx adapter did replace the onboard CPU completely? So if it works without a host CPU in the IIx, why shouldn't it work in the SE/30 if we shut down the 030 hard?

That way we would at least know if it does use the 040 at all (even if the CP and INIT still might not load to activate the caches and everything)

I know that PowerCache and Turbo040 work without onboard CPUs in the IIci and SE/30.

Curious and curiouser... can't wait for my Carrera to arrive to get back into playtime mode :evil:

 

[GeekDot]

So the IIx adapter did replace the onboard CPU completely?

Curious and curiouser... can't wait for my Carrera to arrive to get back into playtime mode :evil:

Well, the extremely verbose (NOT!) former product page @ http://www.micromac.com/products/carrera040.html says "...and the Carrera040 for the Macintosh IIcx and IIx installs with an adapter that plugs into the CPU socket." - I was not able to find any other trace of this anywhere :-/  That adapter could be anything, with or without logic... who knows.

We love playtime mode, don't we?! 

Cheers, Axel

P.S: Here's a Tidbit. While the product page only mentions IIci, IIsi, IIcx or IIx as being supported, the CP code also checks for two other 'groups' which have their slot in common: The V's (IIvx, IIvi, IIvm) and those with an LC-PDS slot (LC, LCII, Color Classic).

 

[joethezombie]

Coincidentally, I happen to have the IIcx adapter.  There is no onboard PLD, but it does have a populated passthrough CPU socket.  This weekend I shall engage in playtime mode myself.

 

[Trash80toHP_Mini]

I've been mulling this over at work with no notes on hand so bear with me:

The Carrerra flat out works in the IIci for which it was designed

-----same with a straight through adapter in the IIsi

-----after disabling Gestalt check for SE/30 you can begin initialization, but process hits a wall?

I take it you you're too early on in the boot process, so you can't use something like MacsBug to find out what's borken and at what point in that process?

This all sends me straight back to the differences in the IIsi and SE/30 PDS implementations. ISTR that the SE/30 generates memory accesses from a different signal set than the IIsi?

They're almost ertainly different from those used by the IIci?

The way the building blocks are piling up in my head leads me to believe that straight through on that adapter is gonna be a no-go right off the bat.

As I see it, the SE/30 PDS is the PROBLEM. The IIsi PDS is NO PROBLEM, even though it may be different in terms of control signal combinations than IIci Cache. The Carrera corrects fo that onboard at any rate. Converting SE/30 PDS into a IIci Cache Slot of any type doesn't seem to be what's needed to me because it's intentionally not supported.

Instead of using Bolle's adapter straight thru, you may need to use his formula research and the GAL to convert the SE/30 PDS into the what looks to the Carrera like the passive adapter supported IIsi PDS in terms of control signals and give that a shot.

Aim at something already supported in hardware/firmware/software. The Carrera is fine with the way the IIsi is using the control signals, no?

 

[joethezombie]

What do you mean by "straight through".  The IIsi adapter still does the crazy PDS30 to CACHE rodeo,  there just isn't a PLD to handle /BG, /BGACK, /CFLUSH, /CENABLE, and /CPUDIS.  Those signals are almost certainly generated onboard the Carrera itself-- CACHE signals used internally, and the /BG, /BGACK being fed back down the bus.

The signals ran up though the n.c. pins are most curious to me.  Obviously the Carrera must need them, otherwise why do that?

/RBV will most likely be used to stop from caching DRAM reserved for screen memory.  Shouldn't be a concern on the SE/30, there is no vampire video.

/FPU, can really find a definition, but probably has to do if using full 040 or LC040.

C40M is the most worrisome.  It means Carrera is looking for a 40MHz oscillator.

 

[Trash80toHP_Mini]

OOPSIE! I was in a hurry at lunchtime  .  .  .  been dopey all day long.  :blink:

I looks to me like Bolle wired up his adapter to have the signals run "straight thru" the board without logical intervention.

For "straight through" on the IIsi I mean the Carrera adapter doing "IIci Cache Slot Contortion" without logical intervention.

Configure Bolle's adapter GAL logic so the signals are mapped such that they appear to the Carrera to be coming from a IIsi PDS on its its native adapter.

This approach would be a long shot as well. That IIcx adapter has the 68030 on board for a good reason. It sounds to me like the Carrera does the Rocket double boot startup without the WHOOOOSH!!!!!!!!!!!!!!!!! theatrics? The driver setup routines apparently run just fine on supported machine's CPUs at startup. But it sounds like the SE/30's CPU dies hard fast and ugly somewhere in that process.

The IIsi can do it because it's set up to use IIci type signalling(?), not the SE/30's peculiar signalling architecture. The PDS Slot pinouts/signals are "the same" but the way the signals are utilized by IIsi and SE/30 are different.

Haven't found the list of differences between IIsi and SE/30 in the docs, anybody got it handy?

We'll probably want to take look at the IIcx docs as well. The SE/30 is purported to be a Compact IIcx, but the way the control lines in question actually work on its CPU will tell that tale. Is the architecture of the IIcx more like the IIsi and IIci than that of the SE/30? If so I don't see much hope of getting this thing goping in the Carrera hostile SE/30.

 

[Von]

I believe it was GeekDot who hacked certain memory addresses in the control panel to provide compatibility with the SE/30.  I'm not for sure, as the old thread was lost.  There were two versions.  I didn't pay enough attention back then, because I didn't have the adapter card to go with the accelerator, but I have since acquired the one pictured.   Two versions were released, but I only kept the 2nd version in my archive.  I hadn't tested it, because like I said, no adapter card until now.  That's why it is "untested".

But, according to Von's post just prior to yours, it looks to be put through the paces. It should be noted that Von's adapter card may not be specifically for the Carrera.  For one, his is not marked "for 040" and has a PLD on it, typical of 030 adapter cards.  Mine is specifically marked 040, and has no PLD.  

Speedster does look to be identical to the Carrera.

My bad in my last post...it was GeekDot who provided the hacked bits, not Bolle (It seems the site's Edit function can't be used to fix such errors after some period of time).

As to my MicroMac adapter and it being different than joethezombie's...mine does register an 040 when installed in my IIsi.  This is one of the screens from my original post:

...I tested this card in my IIsi by first creating a clean install for all macs of System 7.1 on an empty partition on my hard drive. The drive is an IBM 7200 IDE drive connected to an ACARD AEC 7720U SCSI to IDE card that I have used successfully in both the IIsi and SE/30. Using system picker I set this partition to be the system that is booted to. I installed the Carrera040 1.8 control panel and its associated Startup Carrera init. When I booted the system all was well and the 040 was functioning:

I assumed that like the Daystar 030 and 040 accelerators that are interchangeable in a IIsi and SE/30 that this would be the case with the Carrera040....

 

[tt]

Speedster does look to be identical to the Carrera.

Cool, I'll try to give this a shot with the Mobius Speedster. Will it work with any IIci Cache to 030 PDS adapter? I am not sure if mine came with an adapter. I think if it did, it looked like the 2-slot adapter with no chips onboard.

What about the Sonnet 040 accelerator, is it the same pedigree? 

 

[Trash80toHP_Mini]

Did some research on the differences between the SE/30 and the IIcx: there are very few in the Block Diagrams.

Dunno if this is good or bad, but it does mean that the SE/30's pre-MDU setup for memory control shouldn't be a hardware roadblock to Carrera hacking the /30. The MDU ASIC is one differentiation between the IIsi and SE/30. The IIcx/Adapter combo is running on the same type of MUX setup as the SE/30.  Implementations, memory and control signal differentiation within the two MUX matrices might differ however.

One consideration might be that the Carrera/IIcx adapter interface is interwoven with the onboard 68030 at the socket level.

Joe, when you engage in playtime mode this weekend, try running your Carrera/IIcx adapter combo in your socketed (hopefully?) SE/30 board.  []

That there's no logical intervention on what appears to be a passive adapter doesn't necessarily translate directly into the traces themselves being 1:1. If that kluge should happen to work in your PowerCache benchtest setup, it'd be a whole new pickup team ballgame for development on a peewee/softball league playing field, as opposed to the major league field we're dealing with at this point.

Wish-I Were-in' the drivers into IIcx machine state might just start working for your socketed IIcx kluge if not the hacked drivers as they stand. Might this be an avenue of approach for GeekDot's driver hacking as well. Disabling Machine ID detection may only move outright rejection to a bombe yet to be discovered in a switching mechanism within the driver's support tree?

re above:  the Rocket double boot analogy isn't really appropriate. The native Carrera's not hot booting into ROM copied onto onboard memory like the Rocket. The native CPU appears to be setting the system up for a Carrera handoff and then sidelining itself.

 

[GeekDot]

Speedster does look to be identical to the Carrera.

It indeed is. They officially shared the the Control Panel and INIT. 

I've documented that in the original thread which... oh my. We're all doomed!  :-D

Well, the CP's resources are full of mentions of "Mobius" and especially "Speedster".... I assume (aka 'WAG)' at one point MicroMac bought the design from them.

There's a leftover in the ReadMe:
"Applications that do not work with Quadra or Centris Macs are not likely to work on '040 accelerators, including the Carrera040. Generally, these incompatibilities are limited to the '040’s copy-back cache, or FAST mode on the Speedster."

 

[Trash80toHP_Mini]

When I researched differences between the SE/30 and the IIcx I didn't come up dry, I jest dedn't want to clutter up the thread with the usual drivel. I found key differences between SE/30 and IIsi directly related to the Bus Mastering conundrum that's stopping my NuBus in SE/30 project dead in its tracks. The info's related directly to differences between the IIcx and IIci adapter implementations as they appear to be reflected in adapter/driver assaults on Carrera's summit:





If ever there were a bus mastering problem it'd be the SE/30 setting the Carrera/driver up for handing off bus mastery.

WAG: GeekDot, hack the driver so it's hardwired to return a IIcx Gestalt to itself and you and the gang see if it will load at all. It may only work in the zombie's SE/30 testbed with whatever combination of socket elevator shoes/cable hotwiring/mis. klugery that might be involved in setting up that hardware mess.

Just a thought. []

edit: forgot to mention that the IIcx has full bus mastering compatibility due to its NuBus implemented Slot Manager Architecture as does the IIsi. The SE/30 setup's lameass by comparison.

 

[joethezombie]

So I attempted the Carrera in an SE/30 this evening with the MicroMac PDS adapter.  I received identical results as Von.

All of the control panel versions (original unpatched, and both patched versions) opened just fine, but the controls had no affect on enabling the 040.  System remained at 16Mhz 68030 regardless of selector switch positions.



Placing the extension to the system folder and rebooting causes a bomb whilst loading extensions, regardless of which control panel is accompanied:



So, the extension will need serious attention.

I was reading the Carrera040 Update Notes, and there is this:

STEP 3: Install a Universal System

Use your System installation disks and do a "Custom Install for any Mac". DO NOT install a minimum System OR the System for your particular Mac. With Installers prior to 7.5 you can shift click to select multiple options (such as printers and networking) at the same time.

STEP 4: For Mac IIx and IIcx only.

Install Mode32 Mode32 is located on the Utilities disk provided with the Carrera040.

So testing should be done in that environment.

Also, more problems are bound to be had.  Later on, the notes caution with this little gem:

◊    Graphics cards in slot E: If you have trouble booting up your Mac IIci with a Carrera040, and you have a NuBus graphics card installed in slot E (the NuBus slot closest to the power supply), relocate the graphics card into another NuBus slot.

Of course, the SE/30 native video is located at slot $0E.

Tomorrow I shall try the IIcx adapter in my socketed logicboard, but I seriously doubt we'll get any further.

EDIT:  Well, scratch that.  The IIcx adapter blocks the power connector.  I could remove it and solder an extension directly to the logicboard, but I’m not sure I’m that determined.  I’ll sleep on it.

 

[Trash80toHP_Mini]

Of course, the SE/30 native video is located at slot $0E.

Aha! Go stock ROM/MODE32, Pivot card in the PDS/Video ROM pulled. Shazzam, no more video mucking up $E for the time being, especially when you do the test below.

Tomorrow I shall try the IIcx adapter in my socketed logicboard, but I seriously doubt we'll get any further.

Don't sell it short just yet, it's an entirely different hardware ecosystem in that state, emulating a nice little IIcx home for the Carrera is going to be a big part of the puzzle.

Has anyone found docs on how the IIcx handles Bus Mastering as compared to the SE/30? I really don't think the host CPU is handing off bus mastery in hardware anything like the PowerCache just taking a mallet to its control lines. Smells like a driver kind of conflict given the bombs. I'm wondering if the native CPU ever even goes offline. Might the Carrera be using it as an I/O co-processor in the way the Rocket does under RocketWare in Accelerator mode?

Interesting mention of the IIx there!

 

[Trash80toHP_Mini]

Are you running MacsBug(?) to identify the addressing error?

 

[joethezombie]

Nope, never used it.  I’ll look into though!

 

[Von]

Very exciting to see the progress on this!

FWIW, when I was testing the hacked CP from GeekDot, I did have a IIsi ROM in my SE/30 w/ 80 MB RAM. I can easily switch to Stock SE/30 ROM if needed.

RE the slot E occupied by the SE/30 on-board video, would it help or hurt matters if there is an Xceed card in the mix? Below is the only screen I have handy from my PDS Extension Cable Thread that has Mac Envy showing both E and 9 being occupied with the Micron card installed:

Just under 11 years ago I started this thread on AppleFritter about trying to create a PDS extension cable for my SE/30....

...Tried again when plugged in and Shazam, it fires up! Screen of a few apps showing what is running: