Main Menu

Recent posts

#1
New Technology / Designing a 90's Apple Look-a-...
Last post by Nitro - Today at 04:54:55 AM
#2
NeXT Work Logs / Re: BlueSCSI v2 Time Trials
Last post by cooltr6 - Today at 01:54:32 AM
BlueSCSI Time Trials - The 'Not-so-Loaded Cube' and a Standard Hard Drive

25MHz 68040
16MB RAM/100ns
No Dimension board or other upgrades.
Verbose POST

...really a plain Cube overall.

DEC RZ26L 1GB SCSI Hard Drive [SCSI-2, 5400± rpm, avg. seek time 9.5ms] with a full install of OPENSTEP 4.2 [including Developer Tools].  There is only one partition on this drive.  No mountable CD's.  This differs from the BlueSCSI setup.

This computer can connect to the Internet and the local NFS share.

Test Results

Platter Drive

    From bonk at POST to Drive Ready: 14.4 seconds

    From Drive Ready to Login Window: 76.3 seconds

    Overall from bonk at POST to Login Window: 90.7 seconds

    From Login Window to Workspace loaded: 25.5 seconds

    Launching
        WordPerfect 1.0.1: 14.2 seconds
        FrameMaker 3.2: 13.9 seconds
        Concurrence: 2.77: 20.2 seconds
        Diagram 2.5: 16.8 seconds
        OpenWrite 2.1-8: 14.6 seconds
        ParaSheet 1.7: 20.0 seconds


    Return to Login Window from Workspace: 7.5 seconds

Analysis

The overall startup process was an extra 10 seconds longer when factoring in the POST to drive ready part of the startup, however if we just look at drive ready to Login Window, it's about 4 seconds shorter than the entire Internal or External BlueSCSI startup on the same machine.

Launching applications were nearly the same, with the exception of one - Concurrence, which overall was 7 seconds longer than BlueSCSI.

When we consider that these are times with a mechanical hard drive and not solid state technology, the times are not unacceptable.  Just think, platter drives with 9ms access times were fast and we just 'sat and waited' for things to complete.

I consider it somewhat impressive that the hard drive kept up in most things against new technology.  Remembering the fact that the drive chosen for these tests, a drive that would be found in a DEC workstation computer, 1990 technology held its own, in its own respect.  My 'new' money is on replacement technology, even if the past technology is still nice to see working, after all this time.
#3
NeXT Work Logs / Re: BlueSCSI v2 Time Trials
Last post by cooltr6 - Today at 12:41:13 AM
BlueSCSI Time Trials - NeXTstation Mono Turbo and a Standard Hard Drive

33MHz 68040
16MB RAM/70ns
No other upgrades.
Verbose POST

DEC RZ26L 1GB SCSI Hard Drive [SCSI-2, 5400± rpm, avg. seek time 9.5ms] with a full install of OPENSTEP 4.2 [including Developer Tools].  There is only one partition on this drive.  No mountable CD's.  This differs from the BlueSCSI setup.

This computer can connect to the Internet and the local NFS share.

Test Results

Platter Drive

      From bonk at POST to Drive Ready: 15.5 seconds

      From Drive Ready to Login Window: 58.3 seconds

      Overall from bonk at POST to Login Window: 73.7 seconds

      From Login Window to Workspace loaded: 20.9 seconds

      Launching
        WordPerfect 1.0.1: 17.8 seconds
        FrameMaker 3.2: 11.2 seconds
        Concurrence: 2.77: 13.6 seconds
        Diagram 2.5: 14.6 seconds
        OpenWrite 2.1-8: 10.5 seconds
        ParaSheet 1.7: 17.3 seconds


    Return to Login Window from Workspace: 5.9 seconds

Analysis

When comparing startup, the standard hard drive does tend to add some extra time to the process.  An extra 15± seconds were added, waiting for the hard drive to actually come ready and begin the actual startup process.

From Login Window to Workspace is nearly the same as the Internal BlueSCSI setup and half of External BlueSCSI.

Launching applications are nearly double the time of Internal and External BlueSCSI.  This is not unexpected, as we're dealing with a mechanical drive in this battery of tests.

I added another quick check of times, logging out of Workspace.  For the platter drive, 5.9 seconds was not unacceptable.  I'll follow up with the other published tests with this same time check.
#4
NeXT Work Logs / Re: BlueSCSI v2 Time Trials
Last post by cooltr6 - February 25, 2024, 06:00:13 PM
BlueSCSI Time Trials - The 'Not-so-Loaded Cube'

25MHz 68040
16MB RAM/100ns
Normally has a 1GB DEC SCSI Hard Drive inside; disconnected for this testing.
   I will load the same applications onto this drive, so I can get the 'spinny drive' numbers
No Dimension board or other upgrades.

...really a plain Cube overall.

BlueSCSI v2 [internal and external], using a 32GB SanDisk High Endurance Card [100MB/s read, 40MB/s write] and a full install of OPENSTEP 4.2 [including Developer Tools]

There are 3-2GB drives setup at SCSI ID's 1, 2 and 3. CD5 [SCSI ID 5] is loaded with 5 Peanuts CD's.  Tests were performed with the same SD card, but using the internal and external versions.

Since I'm using the same card as the Mono Turbo Slab, this computer can connect to the Internet and the local NFS share.

Test Results

Internal BlueSCSI v2

      From bonk at POST to Login Window: 80.5 seconds

      From Login Window to Workspace loaded [including mounting the first CD - Peanuts 1]: 23.5 seconds

      Launching
        WordPerfect 1.0.1: 11.5 seconds
        FrameMaker 3.2: 11.2 seconds
        Concurrence: 2.77: 12.8 seconds
        Diagram 2.5: 16.1 seconds
        OpenWrite 2.1-8: 10.9 seconds
        ParaSheet 1.7: 18.9 seconds



External BlueSCSI v2

      From bonk at POST to Login Window: 80.1 seconds

      From Login Window to Workspace loaded [including mounting the first CD - Peanuts 1]: 48.2 seconds

      Launching
        WordPerfect 1.0.1: 11.2 seconds
        FrameMaker 3.2: 11.6 seconds
        Concurrence: 2.77: 13.2 seconds
        Diagram 2.5: 15.1 seconds
        OpenWrite 2.1-8: 11.4 seconds
        ParaSheet 1.7: 20.1 seconds



Analysis

Much the same as the first set of testing with the Mono Turbo Slab, nearly identical times from bonk at POST to the login window.  Same phenomenon of the slowness mounting the first CD in Workspace showed up.  I have a feeling this will be the same for the remainder of the tests, but time [no pun intended] will tell.

Interesting comparisons already.  In the Mono Turbo Slab testing, Concurrence and Diagram were the two that stood out, but with the 'not-so-loaded' Cube, the times were nearly identical, across Internal and External versions of BlueSCSi.

Nothing really jumped out at me with this set of testing.  Let's keep in mind that at the moment, we're comparing apples-to-oranges.  25MHz as opposed to 33MHz.  However, I also have a non-Turbo Mono Slab.  It too will be used in this range of tests.

Much more to come!
#5
Emulation / Virtualization / Re: What Needs to be done for ...
Last post by andreas_g - February 25, 2024, 09:10:47 AM
Hello all,

I'm still looking for informations to make the device registers more realistic. Does anyone have informations about the addressing capabilities of the device registers? It is not clear which registers are meant to react in which way on byte, word or long access.

For example it seems that system control register 2 is a long (32-bit) wide register. But sometimes it gets written as 8-bit register which is somehow weird. There are many similar cases. Any information about the low level architecture of this would be welcome.
#6
Emulation / Virtualization / Re: WebAssembly port of Previo...
Last post by andreas_g - February 25, 2024, 06:00:15 AM
Quote from: mihai on February 24, 2024, 11:38:51 PMandreas_g, is there a benefit to enabling ADB support for the Turbo machines?

It seems the ADB mouse is a bit less jumpy and the keyboard is a bit less susceptible to stuck modifier keys. But the benefits are minimal.
#7
The Lounge / Re: Floppy Disk Archiver and E...
Last post by don_apple - February 25, 2024, 04:47:12 AM
Sounds like they are trying to create something similar to the Greaseweazle: https://github.com/keirf/greaseweazle
#8
New Technology / Re: NeXT's spiritual successor
Last post by brunocampos - February 25, 2024, 04:06:03 AM
Quote from: zombie on February 23, 2024, 12:59:39 PMAs far as I can tell no one is actually trying to develop a new operating system. They are all unix variants.

It's a shame because we now have such fundamentally new hardware capabilities, that rethinking software fundamentally makes a lot of sense.

I hope I'm wrong, but I haven't heard of anyone trying to advance anything new in operating systems in over 30 years. Even NeXT was more just making Unix on more personal computers, and advancing an object layer above that. And in many ways, the SmallTalk object layer which inspired it, was more robust. The revelation of NeXT was bringing that all down from large mini computers down to personal computers, which was a big deal.

But the fundamental concepts are now really old. There was a lot of fundamentally great out of the box advancements from the 50-60s in computer science, and it seems ever since then, it's just been refinement of those big ideas.

Sad things are you can fundamentally do things differently now then before. Like before you had multiple memory address spaces for RAM, then out to disk. With modern tech, SSDs, you really dont need that. You could have one global private and even one global public memory space. You could really challenge and change computing at fundamental levels. But I just haven't really seen anyone pushing to advance things because I guess people are satisfied with things as they are.

This is a really interesting point of view. An "unfeasible project" in the views of most people, but NeXT itself had challenging years and plenty of investment rounds until they found the right way to succeed financially.

At some point, a group of researchers and/or entrepreneur will start this challenge - developing a new operating system.
#9
Emulation / Virtualization / Re: WebAssembly port of Previo...
Last post by mihai - February 24, 2024, 11:38:51 PM
I've incorporated the other disk images that mikeboss had gathered in this thread. There's a filtered view of just NeXT images at https://infinitemac.org/?filter=next. I would appreciate any corrections for release dates (I used the kernel build time shown in verbose boots for ones where precise dates were not available) and changes to highlight in the description.

I also updated to the 3.1 release of Previous. andreas_g, is there a benefit to enabling ADB support for the Turbo machines? It would require some changes on my side (for the alternate key mappings) so I've thus far not done it.
#10
NeXT Work Logs / Re: BlueSCSI v2 Time Trials
Last post by barcher174 - February 24, 2024, 09:59:46 PM
Very useful comparisons! Looking forward to more.