SparcStation Core

[Alkadian]

Guys,
Did you figure out how to mount the .vhd file?
Thanks.

[jca]

I only tried linux and it booted fine, if a little slow. What is the root password for Linux Debian/Etch?

Read the Note.txt which comes with it, I think it is user/user.

[jca]

So far I have been unable to boot anything, I always get "No valid state has been set by load or init-program".
Before I give it a rest and finish my new MISTer setup I have a few question?
What is the version and date displayed by OpenBios? At this time I have 1.1 from 4/30/2020. At one time during my experiments I think I saw 1.4 but I am not sure.
So far it looks like only NeXTSTEP is able to run and may be Etch as per mapf.
For NeXTSTEP:
Which settings did you change in the OSD?
The note says to type -v at the boot menu. What does this mean? -v is not a command and boot -v gives the same error as mentioned previously.
For Etch:
What did you do to be able to boot this OS?
Thanks.

[mapf]

I un-xv-ed and untarred the Etch disk image on my PC and put it onto the MiSTer sdcard. Left everything default in the core, selected the HDD image file and did a reset using the osd. It takes quite some time until the bios boot message appears. Bios finds the disk and I pressed return when the boot prompt was waiting for input.

[thera34]

So far I have been unable to boot anything, I always get "No valid state has been set by load or init-program".
Before I give it a rest and finish my new MISTer setup I have a few question?
What is the version and date displayed by OpenBios? At this time I have 1.1 from 4/30/2020. At one time during my experiments I think I saw 1.4 but I am not sure.
So far it looks like only NeXTSTEP is able to run and may be Etch as per mapf.
For NeXTSTEP:
Which settings did you change in the OSD?
The note says to type -v at the boot menu. What does this mean? -v is not a command and boot -v gives the same error as mentioned previously.

If you get "No valid state message", mount a raw image and issue "boot" on the prompt

For NeXTSTEP
On OSD: L2TLB from ON to OFF and IOMMU rev from 26(Default) to 11(Next)
After that, loaded OpenBios v1.1, mounted next.raw and after few mins a message will appear to press "-v"
I hited that, and when prompted for a "Ctrl+C" for networking part, do so and it's good to go

[Alkadian]

So far I have been unable to boot anything, I always get "No valid state has been set by load or init-program".
Before I give it a rest and finish my new MISTer setup I have a few question?
What is the version and date displayed by OpenBios? At this time I have 1.1 from 4/30/2020. At one time during my experiments I think I saw 1.4 but I am not sure.
So far it looks like only NeXTSTEP is able to run and may be Etch as per mapf.
For NeXTSTEP:
Which settings did you change in the OSD?
The note says to type -v at the boot menu. What does this mean? -v is not a command and boot -v gives the same error as mentioned previously.

If you get "No valid state message", mount a raw image and issue "boot" on the prompt

For NeXTSTEP
On OSD: L2TLB from ON to OFF and IOMMU rev from 26(Default) to 11(Next)
After that, loaded OpenBios v1.1, mounted next.raw and after few mins a message will appear to press "-v"
I hited that, and when prompted for a "Ctrl+C" for networking part, do so and it's good to go

I confirm those are the same steps I made yesterday to get NeXTSTEP to work with the new .rbf

[jca]

Many thanks. I made 2 mistakes:
I did not use this core in a very long time (the old core, not the current one) and forgot how to proceed. I first made all the changes in the OSD, including mounting the drive, saved the settings and reloaded the core. I did not realize that the drive has to be mounted after OpenBios started.
Thing got better and NeXTSTEP started to boot, finally saw the boot menu, typed -v and waited for the network message which never came and got a panic.
That was my second mistake: going back to the beginning of the thread I saw that a new SS5.rbf fixed the problem. I missed that one!
I just booted the OS, first time ever on MISTer!
Before going back to setting up my MISTer I will play around a little more with this core.
Thanks again to all.

[jca]

Whoa! Linux running on the FPGA side of MISTer.
I tried the Etch disk image: IOMMU back to 26 and L2TBL back to ON (panic if set to OFF).
Mount etch.raw, type boot, press enter at the boot menu, go out for lunch, come back and type user/user for user name and password, dinner time and after a coffee you have a Linux graphical desktop. One application which autos-tarted crashed and shows a popup with several choice, one being to Inform the Developer, I did not dare to click on that one and choose Close. Now I am starting Gimp, I will report tonight or tomorrow.
Note: in the old days I had a chance to work on a pizza box, at the time these things were blazing fast, no so much nowadays

[jca]

Faster than I expected.

[Alkadian]

I am glad that you got it to work at the end!

[jca]

Thanks.
Some notes for whose who want to try.
It is a little glitchy, not sure if it is the core or the linux. There are many settings in the OSD which I did not touch so it could come from that.
When booting etch for the first time it does an fsck as the system as not been started for a long time (I noticed that the core does not support the RTC), at the end it does a sync which fails and try to reboot the kernel but nothing happens. Reloading the core, remounting the HD while OpenBios is on and rebooting works.
The application which crashed was nm which I suppose is Network Manager and just closing it allows the system to work. On subsequent reboot it looks like nm is not restarted so no popup.
After shutting down properly and reloading/rebooting etch it does an fsck again and you have to reload/reboot to make it works.
During the boot process there are some errors but I did not check what they were. As they do not prevent the system from booting I did not yet try to check what these errors are.
I tried unsuccessfully to load an ISO. I made it using ImgBurn.
I tried to load it at the same time as I mounted the HD, the options for the CD Rom are None, 512 and 2014. I chose 512.
I also tried to mount it when Linux was running. No success so far and each try takes forever.
Young people have no idea of what were the Ferrari of the day:
The first program I wrote was in Fortran on an IBM 1130. Imagine a huge room full of huge racks, raised floor and airco.
Max memory size 32K words of 16 bits, clock in the 3MHz range ..
What you have in your pocket is a gazillion time faster with a huge amount of memory but at the time it was like Nirvana using these beasts.
I did not have a chance of seeing/using a Cray supercomputer:
The Cray-1A, weighed 5.5 tons including the Freon refrigeration system. Configured with 1 million words (64-bit) of main memory, the machine and its power supplies consumed about 115 kW of power; cooling and storage likely more than doubled this figure and had a whooping clock of 80MHz.
Those were the days.

[jca]

Urgent news:
Do not leave your Linux sitting for too long. While typing my previous post the Linux screen turned of and so far I am been unable to revive it. I guess I will end up with a destroyed Hard Disk. I will have to check if the timeout can be disabled.

It also be due to the fact that my HD timed out and stop spinning, I made it spin again via WinScp but it did not help.
Normally I run a script on MISTer boot up which keeps the drive spinning but did not do it on this new setup. More testing to do.

[LamerDeluxe]

I did not have a chance of seeing/using a Cray supercomputer:
The Cray-1A, weighed 5.5 tons including the Freon refrigeration system. Configured with 1 million words (64-bit) of main memory, the machine and its power supplies consumed about 115 kW of power; cooling and storage likely more than doubled this figure and had a whooping clock of 80MHz.
Those were the days.

The Cray was used a lot for 3D animation back then, would have loved to see one IRL. It should actually be possible to run a Cray core on the MiSTer, someone already made an FPGA version.

[jca]

The Cray was used a lot for 3D animation back then, would have loved to see one IRL. It should actually be possible to run a Cray core on the MiSTer, someone already made an FPGA version.

I remember reading about that, the big problem was that there is no software archive for it. The developer finally found a hard drive and extracted stuff from it but I did not follow up. The core is for a Cray-1A.
I just looked on the internet:
The actual design was implemented in a Xilinx Spartan-3E 1600 development board. This is basically the biggest FPGA you can buy that doesn’t cost thousands of dollars for a devkit. The Cray occupies about 75% of the logic resources, and all of the block RAM.
It runs at 33MHz and is missing some features.
Would that fit in a MISTer?

[jca]

Back to Linux Etch: the main problem was my HD not spinning but during my experiments and after the screen turned off I lost all the entries in the "taskbar", I had to "power off" the core and destroyed the disk.
I also tried SunOS and Solaris on SS5 and indeed they do not work. They work on the old SS.rbf. One problem with these OSes is that I have no idea on how to shut them down properly. I can of sort of shutdown one: when I reboot is it always has to do an fsck. For the other one I always destroy the disk. If anyone has any idea regarding the proper shutdown procedure I would greatly appreciate.
I don't know if the old SS.rbf is still available but I can post it in exchange for the shutdown info

[LamerDeluxe]

The Cray was used a lot for 3D animation back then, would have loved to see one IRL. It should actually be possible to run a Cray core on the MiSTer, someone already made an FPGA version.

I remember reading about that, the big problem was that there is no software archive for it. The developer finally found a hard drive and extracted stuff from it but I did not follow up. The core is for a Cray-1A.
I just looked on the internet:
The actual design was implemented in a Xilinx Spartan-3E 1600 development board. This is basically the biggest FPGA you can buy that doesn’t cost thousands of dollars for a devkit. The Cray occupies about 75% of the logic resources, and all of the block RAM.
It runs at 33MHz and is missing some features.
Would that fit in a MISTer?

110K gates on the Cyclone V versus 1600K on the Spartan, so sadly it seems not at all.
Sad to hear the software wasn't archived. Hopefully enough of it will be recovered.

[Alkadian]

Back to Linux Etch: the main problem was my HD not spinning but during my experiments and after the screen turned off I lost all the entries in the "taskbar", I had to "power off" the core and destroyed the disk.
I also tried SunOS and Solaris on SS5 and indeed they do not work. They work on the old SS.rbf. One problem with these OSes is that I have no idea on how to shut them down properly. I can of sort of shutdown one: when I reboot is it always has to do an fsck. For the other one I always destroy the disk. If anyone has any idea regarding the proper shutdown procedure I would greatly appreciate.
I don't know if the old SS.rbf is still available but I can post it in exchange for the shutdown info

I have found out the correct way to safely power off the system with NeXTSTEP

From the terminal:
su
# /usr/etc/shutdown -h +1

(it will shutdown in 60sec with +1)

and then bingo! I have rebooted 3-4 times without that annoying fsck!

[jca]

I had no problem with NeXTSTEP, I did it from the GUI, I have to check where it is but it is there. The same with Linux but it is difficult to find it, I have to re-upload the disk as I destroyed it last time I used it. For Linux it can also be done from the terminal but there is a little trick: the shutdown command is in sbin which is not accessible from a regular user and sudo has to be used, I did not try it but I wonder if it is really possible as it will ask for a password, not sure user will work.
I will also double check NeXTSTEP to be sure the shutdown from the GUI works properly as you mention having to run fsck.

[jca]

Thanks, it gave me an idea for SunOS: there is an /etc/shutdown command and the man page give all the info needed. The simplest form is /etc/shutdown now to do an immediate shutdown. It seems like it will always do an fsck on reboot but there is a flag to prevent it. I will check.
Getting there, slowly.

[Alkadian]

I had no problem with NeXTSTEP, I did it from the GUI, I have to check where it is but it is there. The same with Linux but it is difficult to find it, I have to re-upload the disk as I destroyed it last time I used it. For Linux it can also be done from the terminal but there is a little trick: the shutdown command is in sbin which is not accessible from a regular user and sudo has to be used, I did not try it but I wonder if it is really possible as it will ask for a password, not sure user will work.
I will also double check NeXTSTEP to be sure the shutdown from the GUI works properly as you mention having to run fsck.

I see. I have had problems in shutting down the system from the GUI with NeXTSTEP. The usual fsck issue. I don't know if I have been lucky till now but using the terminal did the trick, so fingers crossed!

Hahaha, I have lost count of how many times I have managed to corrupt the .raw files

From the SPARCstation 5 Service Manual, it says that for Solaris 2x (SunOS 5.x):

/usr/sbin/shutdown -y -g60 -i0

to shut down safely. It shouldn't ask for the password. In fact it didn't ask for it with NeXSTEP. Anyways, I wouldn't swear on that as I am still exploring it.

I can't test the command above as I can't get either Solaris or SunOS to work yet!

[Alkadian]

We wrote to each other almost at the same time

[jca]

When I was experimenting in the old times I tried shutdown but it was not in the PATH and I did not find anything on the internet.
The GUI I was using was SunView, I ran /etc/shutdown now from a terminal window. Upon reboot fsck found some problems, fixed them and did a software reboot which does not reboot on this core. After reloading the core and starting again SunOS it told me that it was doing an automatic reboot (which is manual in this case). No filesystem problem this time so I tried SunView one more time. I exited SunView (right click on the desktop and choose Exit SunView. At the console I tried again /etc/shutdown now. After reboot I once again got an fsck error but this time more begnin: Free Block Count in Superblock.
One more manual automatic reboot , no fsck error.
Now testing with OpenWindows on the same SunOS.

[Alkadian]

Hahaha...the manual automatic reboot does always the trick

[jca]

Thanks for the info on Solaris which will be my next step (not NeXTSTEP).
For OpenWindows I just started it and exited it right away with a right click on the desktop and choosing Exit.
I did the same shutdown as before and as expected I got an fsck Superblock error. I leave it like that knowing that the next time I use SunOS I will have a manual automatic reboot.
When I discovered the core some years back Solaris was the big headache: destroyed disk after each use. Solaris on MISTer: the one time use OS.
Now I am hopefull thanks to you.

[jca]

Whoa! First time I use Solaris and do not destroy the disk.
In the old days I did not realize there was a shutdown command as typing shutdown gives unknown command because it is not in the PATH. If I had tried man shutdown I would have known everything about shuting down SunOS and Solaris.
When I used man shutdown on Solaris I found out more about the init state:
0 -> Stop the oS.
5 -> Shut the machine down so that it is safe to remove the power. Have the machine remove power, if possible. The rc0 procedure is called to perform this task.

So I used /usr/sbin/shutdown -y -i5, no need for -g60 which is the default.
Before shutdown it says to logout in order not to loose data but I did not do it as I did not do anything, just started the OS and shut it down.
The GUI closed, the screen went black for a while and white after that. I saw the disk activity blinking for a while after that (may be MISTer flushing buffers) but I waited until it stopped before rebooting the core and checking that everything was OK.
Note: in games/SparcStation I have a Backup subdirectory which contains immaculate disk images so I do not have to look for them in case of emergency.

Thanks and enough pizza for today.

[Alkadian]

Whoa! First time I use Solaris and do not destroy the disk.
In the old days I did not realize there was a shutdown command as typing shutdown gives unknown command because it is not in the PATH. If I had tried man shutdown I would have known everything about shuting down SunOS and Solaris.
When I used man shutdown on Solaris I found out more about the init state:
0 -> Stop the oS.
5 -> Shut the machine down so that it is safe to remove the power. Have the machine remove power, if possible. The rc0 procedure is called to perform this task.

So I used /usr/sbin/shutdown -y -i5, no need for -g60 which is the default.
Before shutdown it says to logout in order not to loose data but I did not do it as I did not do anything, just started the OS and shut it down.
The GUI closed, the screen went black for a while and white after that. I saw the disk activity blinking for a while after that (may be MISTer flushing buffers) but I waited until it stopped before rebooting the core and checking that everything was OK.
Note: in games/SparcStation I have a Backup subdirectory which contains immaculate disk images so I do not have to look for them in case of emergency.

Thanks and enough pizza for today.

Brilliant!

I am glad to hear that that command also worked with Solaris.

Just out of curiosity. Did you use the old .rbf to get Solaris to work? As of today only NeXTSTEP worked for me.

Also can you please confirm that you can visualize letters/characters correctly on your screen? I am currently away from my computer so I can't share with you a screenshot showing that. But all letters/characters appear in a sort of very low quality. In fact at times I can only guess what letters they are. Maybe something related to the video adapter resolution configuration?

Thanks.

[jca]

For SunOS and Solaris I use the old SS.rbf.
For NeXTSTEP and Linux Etch I use the New SS5.rbf.
SS20.rbf is waiting for anyone making a NetBSD image to test with 3 CPUs.
Until now I did not do much with the different OSes, I just test that they work, don't destroy the disks and can reboot. It takes so long doing that. For example this morning I wanted to see if NeXTSTEP found problems upon reboot as last time I used it I shut it down via the OSD. On the first try it stopped on the CD-Rom, before asking for Ctrl-C. When I tested Linux I had enabled the CD-Rom (512 setting) and I still had it enabled in the OSD. So I had to disable it and reboot so it took quite a while to test this simple thing.
Regarding the screen it would be better if you posted a screenshot, I did not pay too much attention to it, it is not very high quality but so far I did not notice anything horrible. You can see in the screenshot I posted earlier which was Gimp on Linux Etch, I just had a look back and for this OS it is very good.

[gklinger]

Thanks for sharing these. I tried them last year when I first saw your posts abou them and am looking forward to trying the updated cores. Will cores like these ever make it to the 'mainstream' and be installed by the update_all script?

[jca]

I guess it is possible. Grabulosaure had a few unreleased cores on his ftp server which I suppose are not finished but usable, he released some like Fairchild Channel-F, Intellivision and more. He still has more like the Thomson MO. I thinks he is the one who developed the scaler.
He also posted the following:
I've posted on GitHub the current sources of the SparcStation core with updated MiSTer framework.
I've wanted to cleanup that code for ages. And update support for the other boards I have (Xilinx SP605, Terasic C5G, MYIR FZ3...). Not done yet.

So I guess it will come to the official repository sometime in the future.

[Alkadian]

For SunOS and Solaris I use the old SS.rbf.
For NeXTSTEP and Linux Etch I use the New SS5.rbf.
SS20.rbf is waiting for anyone making a NetBSD image to test with 3 CPUs.
Until now I did not do much with the different OSes, I just test that they work, don't destroy the disks and can reboot. It takes so long doing that. For example this morning I wanted to see if NeXTSTEP found problems upon reboot as last time I used it I shut it down via the OSD. On the first try it stopped on the CD-Rom, before asking for Ctrl-C. When I tested Linux I had enabled the CD-Rom (512 setting) and I still had it enabled in the OSD. So I had to disable it and reboot so it took quite a while to test this simple thing.
Regarding the screen it would be better if you posted a screenshot, I did not pay too much attention to it, it is not very high quality but so far I did not notice anything horrible. You can see in the screenshot I posted earlier which was Gimp on Linux Etch, I just had a look back and for this OS it is very good.

From the screenshot you posted I can see that it is fine from your side. This is what I get: