Interware Booster 30-SE50F Info Dump

zigzagjoe

Well-known member
Wow, how are you designing the build for this project? Very impressive! Are the pictures enough to go off of?

I do all my work in Kicad. I reverse engineered the board and dumped the GALs of the SE/30 version, then I can recreate the schematic and design a new board. After that cames the iterative changes (using an actually available PLL, introducing IIsi ROM in SE/30 support, etc) so that I could build my clones.

Now that I know the IIsi board was fundamentally the same as the SE/30, just different GAL code, I will need to figure out the equations for a 2x multiplier but in theory I could do something similar. I know what signals need special treatment since I've got the 3x multipler code to look at.

In the long run, assuming I manage to implement a better slowdown mechanism, i probably have enough IOs that I could ID the type of machine depending on the clocks (does CPUCLK == C16M) and have one board that works in both, albeit the top connector would need to be changed. But that's a ways off: first step is to move the existing equations into CPLD (and have it work), then I can start iterating on additional functionality as I've added extra pins to allow me to work in some additional goals.
 

Mk.558

Well-known member
I bought a MacSIMM from GGLabs, how would you rate the compatibility?

I suspect the ROM on my SE/30 is bad, but considering I get nothing but a black screen with the GGLabs ROM, I have no idea what's actually on it.
 

zigzagjoe

Well-known member
I bought a MacSIMM from GGLabs, how would you rate the compatibility?

I suspect the ROM on my SE/30 is bad, but considering I get nothing but a black screen with the GGLabs ROM, I have no idea what's actually on it.
Any ROM SIMM is compatible with the booster, but ROM images must be SE/30 or IIsi based. Others (IIfx, IIci) will boot but sound will probably be glitchy and you will not be able to use floppy drive.
 

Mk.558

Well-known member
I must have missed the memo on stacking SE/30 cards. Something wrong here.

se30dualcards.jpg

Those tabs are pushing the cards off vertical.

Sound Manager 3.0 (comes with 7.1 Update 3.0) causes the sound to be a little weird. Removing it causes the sound to return to normal, aside from ROM-sourced sounds like the bong.
 

zigzagjoe

Well-known member
Stackable cards violate apple's design recs unfortunately. The tabs must be carefully be bent inwards, and some electrical tape or heat shrink put on them.

Anything that bypasses the ROM routines will be problematic, for now. I believe @obsolete found a version of glider that had the same trouble as it writes the hardware directly. I'm working on a new hardware revision that may be able to address that.

It might be possible to modify the sound manager extension to fix also, but system 7.5 would be more difficult as it integrates sound manager. I have not had the time and energy to really dig into reverse engineering sound manager as of yet.
 

Mk.558

Well-known member
The sound issue is only a minor affair as far as I'm concerned, I mostly use System 6 and 7.1 anyways. 7.5 is too slow for me to really bother with even on the IIci, and on 68040s, I still prefer to use 7.1 unless I specifically need a later version for something.

Any idea on what's a good way to compare performance outside of a benchmark program? Re-sort a large spreadsheet? Apply a mask or filter on Photoshop 2?
 

zigzagjoe

Well-known member
I find doing productivity tasks like compiling code in codewarrior/MPW, boot time, stuffit, etc is a good (if hard to measure) method to quantify the "practical" benefit of an accelerator. As those tasks are what you, the user, are waiting for and may find to be pain points. So, I'd pick what comes to mind as "this is slow and annoying" and use that as a measurement.
 

zigzagjoe

Well-known member
Incidentally, here are Norton system info numbers for this card (and clones) along with some others. Second number in each name is clock frequency, caches (where applicable) are enabled.
1000015599.jpg
 

Mk.558

Well-known member
Yeah I had the accelerator in there briefly and didn't really notice much of a change. I'll have to fool around with it more.

A fast 68040 accelerator however, is noticeable almost immediately.
 

zigzagjoe

Well-known member
Yeah I had the accelerator in there briefly and didn't really notice much of a change. I'll have to fool around with it more.

A fast 68040 accelerator however, is noticeable almost immediately.

I think it does depend on the system you're using, too. The booster seems to "fit" best with system 7.1. Where 7.1 is getting a little pokey as compared to the stock CPU under system 6 (snappy), the booster brings 7.1 back to a similar level of responsiveness. 7.5+ is really heavy and needs a cached accelerator to feel more responsive, and there's several points at boot which are consistently slow regardless of acceleration.

Personally, as far as most "noticeable" accelerators go, I find the the Diimo and powercache make the UI quadra-levels of snappy. It seems they are happy to cache VRAM, so the common R-M-W operations used in quickdraw benefit massively. The Carrera is ostensibly faster, but in application I find it doesn't feel as quick outside of stuffit and specific games as it doesn't seem to do the same and/or is slower at accessing the 68030 bus.
 

tokenalt

Member
Hello all, posting to report my success and failures with a booster. This may sound crazy but I got a booster a while back when I don't actually have an SE/30, I actually got it to try on a sun 3/60 and sun 3/80.

First on the 3/80 the booster unfortunately does not fit directly, not surprising since it was designed for another machine. I had to use an extra socket to raise it up to clear the memory and P4 socket. This means it's not a practical upgrade for the machine but I thought I'd test it anyway. On my first test with the booster it did not work, I believe because the machine runs the CPU at 20MHz so the booster would be running at 60MHz, well out of spec. So I removed the original crystal and put in a 33.33MHz crystal. I tested it first with the original CPU and it did not work in this config, it just output garbage on the serial port, but I tested the booster with this crystal anyway and it also output garbage on the serial port. So I believe the booster could work on this machine with a 2x multiplier and a 40MHz crystal. Here's some pics with it in the machine.

IMG_1774.JPGIMG_1775.JPG

On the 3/60 I ran into an unexpected problem. For those that don't know the 3/60 originally shipped with a 68020 but had solder holes for a 68030. I've heard reports that an 030 will work in the machine but haven't tested myself. When I tried to put an 030 socket on the board I found it was missing 2 pins. They are, I believe, pins F4 and F10 in the middle labeled as NC. I will have to see about modifying the socket and booster to fit in this machine.

Just sharing my experiences trying to get upgrades on machines that aren't normally upgraded.
 

zigzagjoe

Well-known member
Hello all, posting to report my success and failures with a booster. This may sound crazy but I got a booster a while back when I don't actually have an SE/30, I actually got it to try on a sun 3/60 and sun 3/80.

First on the 3/80 the booster unfortunately does not fit directly, not surprising since it was designed for another machine. I had to use an extra socket to raise it up to clear the memory and P4 socket. This means it's not a practical upgrade for the machine but I thought I'd test it anyway. On my first test with the booster it did not work, I believe because the machine runs the CPU at 20MHz so the booster would be running at 60MHz, well out of spec. So I removed the original crystal and put in a 33.33MHz crystal. I tested it first with the original CPU and it did not work in this config, it just output garbage on the serial port, but I tested the booster with this crystal anyway and it also output garbage on the serial port. So I believe the booster could work on this machine with a 2x multiplier and a 40MHz crystal. Here's some pics with it in the machine.

View attachment 78232View attachment 78233

On the 3/60 I ran into an unexpected problem. For those that don't know the 3/60 originally shipped with a 68020 but had solder holes for a 68030. I've heard reports that an 030 will work in the machine but haven't tested myself. When I tried to put an 030 socket on the board I found it was missing 2 pins. They are, I believe, pins F4 and F10 in the middle labeled as NC. I will have to see about modifying the socket and booster to fit in this machine.

Just sharing my experiences trying to get upgrades on machines that aren't normally upgraded.

Oh, nice experiment! Truthfully, I've run those CPUs up to 60mhz - so that alone isn't necessarily a disqualifier. But, the CPU does have a few straps set low that might differ from what sun wants. Those NC pins should be able to just be removed from your socket and work just fine. I've actually started to not populate those on newer cards since not all sockets have the keying pins.

Here's the schematic of the original design. My cards use a different PLL design, and some changes in the GAL, but the CPU is set up identically.

Interware Booster 30-SE50F Schematic.png
 

Mk.558

Well-known member
The booster seems to "fit" best with system 7.1.
...
Personally, as far as most "noticeable" accelerators go, I find the the Diimo and powercache make the UI quadra-levels of snappy.

Yeah, most of the time I don't even install 7.5 on the SE/30 as if I need to access 10.4, 7.1 with OT can do that already (ish, but usually I just use MacTCP for less system overhead) or I just use FTP which works fine. 7.5 is rather overrated on most 68030s, even faster ones like IIfx, you'd be surprised how much 7.1 can actually do. On a 68040, sure, but ...

I was reading that wiki on peterhuman about the PowerCache and I believe it. Your accelerator doesn't make aggressive use of caching, and that's fine. Drop-in driverless accelerators have some value in their own. I don't know what PowerCaches cost back in the day but I imagine they weren't that cheap. Apparently Turbo 601s were cutting it close to costs, which the company does need to make enough to survive the dips and downs of computing back then. Rockets, retrospectively, must have been somewhat expensive, but good value in that they were the only accelerator I know of that could function as either a hardware VM or an accelerator, which isn't that bad for roughly 1200$ back in the day for the 33MHz version.
 

zigzagjoe

Well-known member
Yeah, most of the time I don't even install 7.5 on the SE/30 as if I need to access 10.4, 7.1 with OT can do that already (ish, but usually I just use MacTCP for less system overhead) or I just use FTP which works fine. 7.5 is rather overrated on most 68030s, even faster ones like IIfx, you'd be surprised how much 7.1 can actually do. On a 68040, sure, but ...

I was reading that wiki on peterhuman about the PowerCache and I believe it. Your accelerator doesn't make aggressive use of caching, and that's fine. Drop-in driverless accelerators have some value in their own. I don't know what PowerCaches cost back in the day but I imagine they weren't that cheap. Apparently Turbo 601s were cutting it close to costs, which the company does need to make enough to survive the dips and downs of computing back then. Rockets, retrospectively, must have been somewhat expensive, but good value in that they were the only accelerator I know of that could function as either a hardware VM or an accelerator, which isn't that bad for roughly 1200$ back in the day for the 33MHz version.
Having designed and built socketed versions of the Diimo and Carrera, yeah, I can confirm the costs are easily an order of magnitude (or two) greater to design, consistently source obsolete parts, build cards, and QC each. I've flirted with the idea making more socketed cards but currently I don't think I could do it without both a disproportionate amount of time and frustration, nor a halfway attractive price point. I assume the Rockets had a different value proposition due to their unique capabilities.

Personally, I run 7.5.5 and A/UX 3 on my socketed diimo@58mhz machine. With 128K cache, it's pleasant to use under both OSes and extremely snappy. The socketed carrera@45mhz gets 8.1 (040) and 6.0.8 (030). 8.1 is still pretty bad. It's just not meant for 68k. 6.0.8 is, of course, just fine.
 

David Cook

Well-known member
I purchased a @zigzagjoe Interware Boost Clone from eBay. It comes with a nice 3D printed case.

I dropped the booster into a newly recapped IIx today. Snug fit.

Performance is immediately noticeable in System 7.1. Norton shows a 67% overall performance boost (640x480 256 colors). Slightly better than the IIci.

1726365814232.png

The only problem is that sound is too fast. My test was to play alerts in the Sound control panel. I thought this would only be an issue is later systems. Maybe something about the IIx ROMs?

All in all, a really nice performance improvement. Keep up the good work!

David
 

zigzagjoe

Well-known member
I purchased a @zigzagjoe Interware Boost Clone from eBay. It comes with a nice 3D printed case.

I dropped the booster into a newly recapped IIx today. Snug fit.

Performance is immediately noticeable in System 7.1. Norton shows a 67% overall performance boost (640x480 256 colors). Slightly better than the IIci.

View attachment 78337

The only problem is that sound is too fast. My test was to play alerts in the Sound control panel. I thought this would only be an issue is later systems. Maybe something about the IIx ROMs?

All in all, a really nice performance improvement. Keep up the good work!

David
Glad to hear it's working well for you!

If you have Sound Manager in your extensions folder, pull that out and try again - it should sort that out. Unfortunately, anything that runs out of RAM rather than the ROM sound code will have this issue. I suspect a patched version of the sound manager could be made in order to slow down the rate of sound buffer updates, which seems to be the root cause here. This would at least help 7.1, but 7.5's integration of the sound manager would make that rather difficult to patch.

Just as a note though, I sell everything I make cheaper via the forum... I can't absorb the ebay fees, so everything I list on ebay is more expensive :(
 

Melkhior

Well-known member
First on the 3/80 the booster unfortunately does not fit directly, not surprising since it was designed for another machine. I had to use an extra socket to raise it up to clear the memory and P4 socket.
For those not familiar with Sun terminology, the "P4" is that familiar DIN connector you can see on the right underneath the booster, where the framebuffer would go [1]. From the schematics of the 3/60, it's basically a PDS for the MC68020 (so no /STERM). From some tracing and discussion for the 3/80, they are identical between the two - it seems the only differences are physical and minor (basically the bracket).

I've wanted to do a P4FPGA for a while, but unfortunately my 3/80 motherboard has no sign of life and has blown a capacitor.

Another funny detail on the 3/80: there's a jumper to switch the MC68882 from the 20 MHz clock to 40 MHz. With an appropriately rated PGA '882 in the socket, it did improve things for running POVRAY back in the day!

[1] useless trivia: it's called P4 because the original VME spec (electrically based on the 68k bus) uses two DIN connectors named P1 and P2, and Sun had already expended VME with its own P3 connector...
 

zigzagjoe

Well-known member
Something interesting!! With the new CPLD based hardware.

1726714143151.jpeg1726714156052.jpeg1726714171500.jpeg

Comparison. This is using a 30Video with a FPU (20mhz) installed, as the IIsi is on loan to me, unrecapped, and onboard video has just flaked out.

1726714518446.png

I also have an ultra small socketed version in build at JLC now! This is mostly a birthday present to myself, but if there's interest I may make more.
 
Top