A common complaint for people with a lot of RAM in these old machines is how long the memory test takes on startup. I'm looking at possibly disabling that test.
However, I could use some testing help. My IIx only has 8MB of RAM in it, which isn't very effective for evaluating RAM test speeds on power on. Could someone else who has a machine with lots of RAM and is capable of running a modified IIsi ROM, test out an image for me?
The image is here: iisi+nomem1.bin
This is a stock IIsi ROM, hopefully with the memory test disabled.
The specifics of the change are:
I believe the start of the memory test is located at:
So I have replaced entry to the routine with the jmp back to (A6). So, at location 0x64A96, I have replaced 0x7000 with 0x4ED6. My IIx still boots with this change, but that's not a great indicator of anything. If someone with a machine with lots o RAM can test this and report back, I'd appreciate it.
I've got my IIsi out which has a wopping 16MB of RAM. With the stock ROM, it takes about 13 seconds from power on to video displayed.
With the following ROM it takes about 3 seconds. Let me know if it works for you guys too: iisi+nomem2.bin Stock IIsi ROM with the changes in this post.
It looks like there are 2 memory test related routines in the IIsi ROM located at 0x467E0 and 0x468F8, both returning with a JMP (A6) (opcode 0x4ED6). So I've just placed a 0x4ED6 at the above two locations. These are 1 function before the last patching attempt...
On the IIci the video is up in 2.2 seconds, that's great! I'm seeing 24.8 seconds from pressing the power button to the Happy Mac, also a huge improvement. So I would say it's cut down the startup time by about 15 seconds. Now my question is: what is it doing after the video has loaded that takes so long?
I see that the two routines you've patched are called MOD3TEST and REVMOD3TEST. I googled for MOD3TEST and found this...am I allowed to post this link? If not I apologize.
As best I can tell, the startup chime indicates successful checksum, then other tests like the memory test is done, then some initialization and the floppy, sound, etc. drivers are loaded, then the slot manager calls the primaryInit routines of the slots, which is where the video gets initialized. Then some more initialization, and finally start booting. I have seen that if you don't have a startup disk stored in PRAM (selected in the startup disk control panel), it can take longer to find a boot device.
Aside from that, I don't have a good lead at the moment.
It worked great for me. Test setup is diskless SE/30 with 128MB RAM. It took quite a long time to get to the "grey" screen. Typical sequence was after switch on: blank --> checkerboard/stripes for at least a minute --> "grey" screen. The checkerboard portion has been eliminated, so seems like orders of magnitude of difference. Great job!!
In the past I found the checkerboard not only annoying but potentially confusing, because boards that have leaky caps could display a similar pattern. With the memory test bypassed, now that source of confusion will be eliminated.
I can confirm this works on my 128MB IIci as well; the delay between power on display matches your results. Of course, doug already verified his IIci works, so I'm really just verifying that the SIMM and SIMM programmer works
Interestingly, I'm not seeing anywhere near a 24.8 second delay between video and the Happy Mac.
I am, however, seeing a very long delay when starting the Finder, which may possibly be related to this being a IIsi ROM?
My long delay might be related to the PRAM needing initialization since I have the PRAM battery removed. I'll try putting a battery in and retest. (Why, oh why, does the PRAM battery have to be underneath the drive carrier?)
I don't notice any long delays, but System 7.6 is slow enough to boot anyway that I may not be noticing it.
Good to hear your SIMM/programmer seem to be working!
Oh, one more thing....I notice when a IIsi ROM is used in a IIci, my hard drive does this weird click pattern (some kind of initialization?) twice. With the new faster startup test, the initialization happens so fast that the first click pattern gets interrupted before it's complete. I don't know what the deal is with it, just thought it was kind of weird.
I just posted an updated ROM image in the ROMDisk thread which includes this change.
FWIW, I finally broke down and pulled my 128MB SE/30 off the shelf for some testing. I'm not a huge fan of the difficulty of opening those things and getting to the motherboard. But.. Using the ROMDisk + RAM test disable, and my trusty iphone stopwatch, the SE/30 goes from flipping the power switch to 7.0.1 Finder desktop in 13.9 seconds, give or take my reaction time.
Thanks! I just tried this image and the delay is MUCH shorter. It was still long with the PRAM battery but booting off the hard drive.
I think my hard drive (Quantum ProDrive 210S) just takes that long to initialize! (This is actually kind of scary because the ROM drive overrides the hard drive even when the hard drive is plugged in -- the hard drive is still busy initializing, and the ROM drive wins because it's already ready) In fact, System 7.0.1 is completely booted and still doesn't recognize the hard drive, so I think it polls the SCSI bus before the hard drive is responding and assumes there's no hard drive. Gotta love it.
Looks like it's now 10.9 seconds to Happy Mac and 16.6 seconds to a completely loaded System 7.0.1. WOW!
Without the hard drive plugged in at all, it's 8.4 seconds to Happy Mac and 13.5 seconds to the desktop