WarpSE: 25 MHz 68HC000-based accelerator for Mac SE

[JTRetro]

I've been doing some more testing simple testing, such as running my "After Dark" fish screen saver run indefinitely, including an hour so far today-with no problems. Another interesting test, this:

This is an original 400k disk that my parents gave me back in the late '80's with, of course, the game "Transylvania"! And....it works just fine in my 800k SE with the WarpSE accelerator:

 

[Zane Kaminski]

Just bought the overclocking board connectors! The finer-pitch 2.0mm type are quite a bit less common than the standard 0.1inch/2.54mm ones. The finer pitch is required since they're shorter vertically too. With the 2.54mm pitch connector the board stack was too tall and may have collided with the Mac's chassis. Anyway, I think that high-quality connectors are really important for a reliable system, especially on the female socket side. I made sure to get Amphenol brand sockets:

I was surprised to learn that among different connector vendors (Amphenol, Molex, Sullins, etc.), the lengths of the male pins that go in the socket are slightly different! Well, I mean, they all offer various pin lengths but their closest ones differ by as much as 0.2mm. Not really a big deal but I was surprised. So the pin headers that go in here are also Amphenol brand to ensure optimal compatibility. Of course, a no-name vendor for the male pin side would probably have been fine too as long as it has roughly the right length and the right width for the pins. The socket side is the side that can wear out and is much harder to make reliable than the pins.

Very interesting. Even though the problem is apparently fixed, it seems clear that we beta testers should probably test the Prince of Persia demo on every new firmware version release!

Thanks for all your hard work testing! I think though that with the incremental changes in each version, it's best to try different stuff once a problem is resolved. Then of course right before release or after a major restructuring of the code, it'd be good to retest old bugs to make sure there are no regressions. But usually I try not to mess stuff up once it's fixed haha. Hardest part of all this is just finding the apps that don't work.


By the way, @ppuskari, I have been trying the SCSI Director 4.0 drivers and can't duplicate the shutdown issue although I have only tested with my recent "prince of Persia fix," so maybe that fixed SCSI Director too. Basically the fix just amounted to slowing down for longer after an access to the VIA. Previously the slowdown time after hitting the VIA was 14-28 microseconds (21 avg). That was enough to fix the crash in the sound control panel but not enough for Prince of Persia, which needed it increased to 28-42 microseconds. Anyway, at first I thought SD4.0 was not working, but then I realized that it just didn't like a ~1900 MB disk on my BlueSCSI v2. Downsizing the drive to 200 MB got it working well with the WarpSE and I can't find any issues when using the driver. I have been busy buying parts and tweaking the final WarpSE board design but I'll try to get 0.7e out soon with the Prince of Persia fix and maybe you can let me know if that fixes the issues you were having with the SCSI Director drivers.

 

[Zane Kaminski]

Attached is firmware 0.7e. It's very similar to 0.7d-fastscc but the slowdown period after a VIA access has been increased as I mentioned in my previous post. That fixes Prince of Persia. In addition, in this version, the WarpSE speeds up after a SCSI chip access if slowdown was pending before the SCSI access. This should make the speed consistently fast for SCSI drivers that use blind reads/writes, ensuring they work reliably.

 

[JDW]

@Zane Kaminski
I will try to make time to test firmware 0.7e and the Prince of Persia Demo on my lunch break today.

LocalTalk should be still working fine, I assume?

My last tests of LocalTalk involved opening Photoshop 4 on my networked SE/30, with the app being stored on the WarpSE machines drive. It took forever to open, but it did work. That's the closest I could come to the "streaming" test you mentioned here.

 

[JDW]

@Zane Kaminski
Lunch Break Test Completed, with GOOD & BAD news. (All testing was done with my SE Reloaded motherboard fitted inside the SE's metal chassis. SWIM chip installed.)

GOOD NEWS​

• Prince of Persia audio sounds great and the Demo was able to run all the way through with no freezing! Bug fixed! Great!
• Tetris audio too still sounds great.
• Speedometer 3.23 scores with my BlueSCSI v1 (slower than v2, mind you) show the same scores I've seen before with the previous firmware....

​

​

BAD NEWS​

I am getting the same error dialog problem in SCSI Director Pro 4.0 that I reported before here. Increasing memory for the app in Get Info only causes the error dialog to ask for yet more RAM the next time I test it.

As I later reported here, my fix at the time was to press INTERRUPT at power-on to disable WarpSE. I then booted and launched SCSI Director Pro 4.0, then successfully ran all tests. After that, I rebooted with WarpSE enabled and repeated the SCSI Director Pro 4.0 test to find the problem gone. Sadly, I tried that again today (while in 8MHz mode), but it DID NOT WORK (using BlueSCSI v1). I tried again with the same SD card in my BlueSCSI v2 but got the same error dialog.

Keep in mind that all 4 SEEK tests must complete. The error dialog only appears after those SEEK tests are done. (NOTE: For some reason, BSv2 takes longer to complete the SEEK tests than BSv1, but that has always been true and has nothing to do with WarpSE.)

I would like to ask my fellow WarpSE beta testers to please flash the newest 0.7e firmware, then boot into System 7.1 with WarpSE enabled, then launch SCSI Director Pro 4.0 and run the tests on your drive to see if you get the same error I do. If you have a BlueSCSI, try that. I lacked time to test using my spinner drive.

 

[JDW]

WarpSE Firmware 0.7e TESTING CONTINUES...

I confirmed the SCSI Director Pro 4.0 problem affects BlueSCSI v1 & v2 only, and not a spinning platter drive:

After that video, I put WarpSE on my stock Apple motherboard with IWM chip and it had the same problem with my BlueSCSI v1. So no one can argue its my motherboard or SWIM chip. Different motherboards, and one with SWIM and the other with IWM.

Fellow beta testers, could you please test WarpSE Firmware 0.7e with SCSI Director Pro 4.0 and also report what drives you are using?
@JTRetro @techknight @ppuskari

 

[JTRetro]

@JDW -what O.S. are you using with SCSI Director Pro 4.0?

 

[JDW]

@JDW -what O.S. are you using with SCSI Director Pro 4.0?

System 7.1

 

[JDW]

@Zane Kaminski
I am still testing WarpSE Firmware 0.7e.

TEST #1​

Using WarpSE in my stock Apple SE motherboard with IWM, I booted into System 7.1 using MacSD and tested with SCSI Director Pro 4.0. No problem!

TEST #2​

Powered off, removed MacSD, attached my BlueSCSI v2 to the same ribbon cable (SD card different from MacSD), and booted into System 7.1 and tested. Surprisingly, there is no out of memory error today! But why? (Stock SE motherboard with IWM)

TEST #3​

Powered off, removed the ribbon cable, and installed my BlueSCSI v1 internally (same SD card as used in BSv2) using my SCSI Rider 1, booted into System 7.1 and tested in SCSI Director Pro 4.0. All tests passed! No out of memory error! What's going on??!! (Stock SE motherboard with IWM)



True, I am still using my stock Apple SE motherboard and IWM chip, but as I said yesterday, I confirmed the out of memory problem when booted from my BSv1 yesterday. This is a mind-bogglingly inconsistent problem!

TEST #4​

Swapped out motherboards, putting WarpSE back onto my SE Reloaded motherboard with SWIM, and I booted once again from my BSv1 into System 7.1 and tested in SCSI Director Pro 4.0. This time, I got the out of memory error again, even though I had 1.4MB free memory. Ack!

TEST #5​

Powered down, removed BSv1 and installed my MacSD, and retested. SCSI Director Pro 4.0 completed the test without error! (SE Reloaded w/SWIM)

TEST #6​

Powered down, removed MacSD, and retested with BSv1, but this time with SHIFT key held down to kill all CP's and Extensions. This time SCSI Director Pro 4.0 completed all tests without error. (SE Reloaded w/SWIM) But keep in mind that I didn't need to do this when using my stock Apple SE motherboard!



It's all very strange.

 

[JTRetro]

So I jut got done installing SCSI Director on my SE running O.S. 7.1 off of it's original 20Mb drive. So far i just did a simple test: get SCSI director up and running. And, it appears to run well, instantly seeing the multitude of SCSI devices that I have currently attached to this machine:

Again, this is a machine with a stock 800k board with the warp SE..

 

[Kai Robinson]

@eric could there be anything in the bluescsi codebase that could cause wierd inconsistencies like this, with accelerators?

 

[JDW]

So I jut got done installing SCSI Director... So far i just did a simple test: get SCSI director up and running. And, it appears to run well, instantly seeing the multitude of SCSI devices that I have currently attached to this machine

Does that mean you clicked this button?

In my previous post where I show my 6 tests in SCSI Director Pro 4.0, it was that "Test" button which takes you to the actual suite of tests. It does the SEEK tests at the beginning. After that, I get the out of memory error in some situations (again see my previous post about that). But when the test proceeds without error, then it does the "graphing tests" at the end.

So basically I just wanted to confirm if you merely installed SCSI Director Pro and launched the app, viewing your SCSI drives but doing nothing else; or if you proceeded to click the "Test" button and ran all the tests until the graphing test completed.

 

[JDW]

Lunch Break Test of WarpSE Firmware 0.7e continues.

If you re-read my post from yesterday where I did the 6 Tests, you will see that in Text #6 I held down SHIFT while booting. That allowed SCSI Director Pro 4.0 to complete all its tests without error.

Today, I reduced the number of startup extensions, which gave me just over 2.5MB of free memory in System 7.1. But even that still resulted in the same out of memory error.

I removed even more extensions, giving me 2.9MB free. Then SCSI Director Pro 4.0 was able to complete the tests without error.

HOWEVER, this crazy error only happens with my BlueSCSI v1 or v2, not a real hard drive, AND it only afflicts my SE Reloaded board, not the stock motherboard, AND it does NOT happen when I boot using 8MHz mode (WarpSE disabled) even with my full suite of control panels and extensions loaded.

So in light of this, I would encourage my fellow WarpSE beta testers to leave SCSI Director Pro 4.0 at its default Get Info box RAM setting (increasing that doesn't resolve the issue anyway), and try different CP's and Ext's to decrease your available RAM. Try 2.3MB free, 2.5MB free and 2.9MB or higher free. So if you get the same out of memory error I do. Again, I'm not getting this on my stock Apple SE motherboard, so it could be you won't see any problems either, but the reason for you all to test is to discover if you see anything I am seeing. Thanks.

 

[Zane Kaminski]

@JDW have you tried eliminating the drive contents including the driver (which is stored on the disk) as a source of the problem? Maybe you can try using another SD card, formatting it as exFAT, and putting a blank 100-200 MB drive image on there. You can use the dd command on macOS to do create the image:

dd if=/dev/zero bs=512 count=390625 of=/path/to/output/file

That creates a 200 megabyte file with all zeros. Of course replace “/path/to/output/file” with the path to the new file on your SD card, which on macOS will be something like /Volumes/BlueSCSI/HD0-test.img

 

[JDW]

@JDW have you tried eliminating the drive contents including the driver (which is stored on the disk) as a source of the problem? Maybe you can try using another SD card, formatting it as exFAT, and putting a blank 100-200 MB drive image on there. You can use the dd command on macOS to do create the image:

I don't have the opportunity to do that right now, but I can try it, of course. Thank you for the suggestion and details.

But of course, this raises the question of why that SD card works fine when I test in my stock SE Motherboard. Again, this issue seems to be affecting my SE Reloaded board only. And, yes, I am following the direction of @iPhil64 by doing all my testing with the motherboards inside the SE's metal chassis. Hmmm...

 

[JTRetro]

@JDW -Yes, I ran the tests on a few of my drives, including my Iomega 100Mb:

And my external SCSI2SD drive:

And I tested other functionality within the program, including it's ability to mount drives, such as my Iomega 100Mb:

Seemed to all work well.

 

[JDW]

OK. I bought a brand new SD card (16GB Gigastone), formatted it ExFAT and created the 200MB image (formatted with the HD SC Setup 7.3.5 driver, which I know is SE compatible). I did a clean install of System 7.1 followed by the System Update 3.0 install. I copied over various files including SCSI Director Pro 4.0. Here's what I found:

1. The out of memory error is indeed related to RAM. It can happen even with the PDS slot open.
2. The problem is not specific to WarpSE.
3. You can easily reproduce the problem by reducing your Available RAM. Rather that mess around with CP's and Extensions to achieve that, just open your Memory CP and set the Disk Cache to 512K. Normally, with the default 128K disk cache setting, you'll probably have 2.9MB of available RAM free. But after changing that setting to 512K and restarting, your available RAM will drop just below 2.5MB, and then if you run SCSI Director Pro 4.0, you'll get the out of memory error after the SEEK tests complete.
4. Changing the Get Info box RAM setting for SCSI Director Pro 4.0 doesn't resolve the problem for some reason I don't understand. But reducing disk cache, not loading INITs, etc — doing something to increase available RAM — will resolve the out of memory problem in SCSI Director Pro 4.0.

So I consider this matter resolved. It's just something important to take note of.