NeXT Computers Forum Index NeXT Computers
www.NeXTComputers.org
 
Log in to check your private messagesLog in to check your private messages

Log inLog in  RegisterRegister


Profile  Search  Memberlist  FAQ  Usergroups
What Needs to be done for a NeXT Emulator
Goto page Previous  1, 2, 3 ... 70, 71, 72 ... 80, 81, 82  Next
 
Post new topic   Reply to topic    NeXT Computers Forum Index -> Emulation / Virtualization
View previous topic :: View next topic  
Author Message
mmu_man



Joined: 23 Oct 2016
Posts: 8

PostPosted: Tue Oct 25, 2016 2:58 pm    Post subject: Reply with quote

Hi,
I also had an issue on Debian Sid (still at SDL2 2.0.4): display doesn't get any update.
I posted a fix in another thread.
It could be a regression in SDL though, but it's generally a bad idea to call things from different threads with SDL...
Back to top
View user's profile Send private message
andreas_g



Joined: 30 Jan 2009
Posts: 446
Location: Austria

PostPosted: Wed Oct 26, 2016 2:02 pm    Post subject: Reply with quote

Hello all,

i am happy to announce the release of Previous v1.5! It uses the latest version of SDL, which supports audio recording. So now we are able to emulate audio recording via soundbox with its awesome telephone quality sound Wink

Previous v1.5 also includes some bugfixes:
- SCSI disk images greater than 4 GB are now supported (thanks to dtaubert for the patch)
- The clock works now for dates after 1999 (don't forget to install Y2K patch for NeXTstep 3.3)
- A problem on Linux, where no display content is shown, is now fixed (thanks to new forum user mmu_man for the patch)
- A problem with stuck i860 thread which caused NeXTdimension to fail in some rare scenarios is gone

I hope the new version works for everyone. You can load a binary for Mac OS X v10.6.8 or later here.
Back to top
View user's profile Send private message Send e-mail
eagle



Joined: 29 Mar 2006
Posts: 456

PostPosted: Wed Oct 26, 2016 6:31 pm    Post subject: Reply with quote

Andreas, this is a nice update - thanks!

Unfortunately the network issue I reported persists. Copying files from an NFS mount eventually kernel panics the system, and copying files to an NFS mount immediately crashes Previous.app. I didn't expect any change to that functionality, but thought I would test it with the new build.
_________________
My NeXTs:
NeXT Computer prototype (68030-25 x2, 68040-25)
Two NeXTstations (68040-25)
All mono
Back to top
View user's profile Send private message
andreas_g



Joined: 30 Jan 2009
Posts: 446
Location: Austria

PostPosted: Wed Oct 26, 2016 11:09 pm    Post subject: Reply with quote

eagle, I did not make any changes to this part of the emulator. Fixing that crashing bug and the bug that causes problems with dual MO drives is on top of my list.

For fixing the slirp crashing bug I need to reproduce it and therefore need to set up NFS on my system (host and guest). My host runs Mac OS X v10.9.5.
For fixing the MO bug I need confirmation on real hardware. After lots of investigations I'm no longer sure if this is a bug in Previous or NeXTstep.
Back to top
View user's profile Send private message Send e-mail
eagle



Joined: 29 Mar 2006
Posts: 456

PostPosted: Thu Oct 27, 2016 4:23 am    Post subject: Reply with quote

That reminds me, I was going to email you with a set of instructions to reproduce the error. I'll set up a VM running 10.9 and get those instructions to you.

On the MO bug, my Cube only has one OD and it is no longer functional so I cannot test that for you.
_________________
My NeXTs:
NeXT Computer prototype (68030-25 x2, 68040-25)
Two NeXTstations (68040-25)
All mono
Back to top
View user's profile Send private message
ardi



Joined: 11 Jun 2014
Posts: 85

PostPosted: Thu Oct 27, 2016 2:17 pm    Post subject: Reply with quote

eagle wrote:
That reminds me, I was going to email you with a set of instructions to reproduce the error. I'll set up a VM running 10.9 and get those instructions to you.

I'm also very interested in this bugfix, as it's the only way of having shared folders between Previous and the host.
Back to top
View user's profile Send private message
Rob Blessin Black Hole
Site Admin


Joined: 05 Sep 2006
Posts: 690
Location: Ft. Collins, Colorado

PostPosted: Sat Dec 17, 2016 3:13 am    Post subject: Large Hard Drive support in Previous 1.5 Reply with quote

Hello: Just taking Previous 1.5 for a spin , I like it.

The large Hard Drive support patch does that translate to NeXT Hardware as a patch?

How Large of a hard drive can we now support and is it one partition ?

I found mounting a second drive image and then initialising it copying files into it in Previous , then dding that drive image onto an sd works and mounting it on a NeXT Computer works for actually get files out of Previous and onto a NeXT , whats nice is I can download and install apps then test them and move them over to my hardware ...... not lazy about the Networking but I'm up to my eyeballs in NeXT projects lots of fun these days, appreciate all your support and business! Still ironing out difficulties with the version 6 microsd cards but hopefully we will figure it out soon.
_________________
Rob Blessin President computerpowwow ebay sales@blackholeinc.com http://www.blackholeinc.com
303-741-9998 Serving the NeXT Community since 2/9/93
Back to top
View user's profile Send private message Send e-mail Visit poster's website
andreas_g



Joined: 30 Jan 2009
Posts: 446
Location: Austria

PostPosted: Sat Dec 17, 2016 7:56 am    Post subject: Reply with quote

Hello Rob,

no, that patch does not translate to real hardware. It was just a bug inside Previous (used 32-bit integer type for critical variable, which was too small), that prevented reading large files (on host side). So this patch won't change anything about NeXTstep's maximum partition size. That limitation is inside NeXTstep's kernel and probably other parts of the operating system.

Best wishes,
Andreas
Back to top
View user's profile Send private message Send e-mail
ardi



Joined: 11 Jun 2014
Posts: 85

PostPosted: Sat Jan 28, 2017 12:06 pm    Post subject: Reply with quote

Would it take too much work to emulate a dual head cube? What graphics options combinations would be possible? I always work with two displays on Mac OSX, and having two Previous windows, one in each monitor, would be really useful.
Back to top
View user's profile Send private message
andreas_g



Joined: 30 Jan 2009
Posts: 446
Location: Austria

PostPosted: Sat Jan 28, 2017 2:53 pm    Post subject: Reply with quote

This is already possible, if you use NeXTdimension. Just go to the graphics menu, enable NeXTdimension board and set "show display" to "both". As a result you will get two separate windows, one NeXTdimension color display and one monochrome display.
Back to top
View user's profile Send private message Send e-mail
ardi



Joined: 11 Jun 2014
Posts: 85

PostPosted: Sat Jan 28, 2017 4:49 pm    Post subject: Reply with quote

andreas_g wrote:
This is already possible, if you use NeXTdimension. Just go to the graphics menu, enable NeXTdimension board and set "show display" to "both". As a result you will get two separate windows, one NeXTdimension color display and one monochrome display.

That's great! Didn't know that! Thanks a lot!!!!
Back to top
View user's profile Send private message
ardi



Joined: 11 Jun 2014
Posts: 85

PostPosted: Sun Jan 29, 2017 6:01 am    Post subject: Reply with quote

BTW, one more question, this one about the 68040 FPU: How are transcendental operations being emulated? Do you trigger an exception because the 68040 doesn't support them, or do you just pretend all the 68882 instructions are supported by the 68040? I'm asking this just to know if when building for an emulated Cube in Previous I should link with the alternate transcendental source code which you can find in several 68k sites, or if I'm fine using the standard math functions. I've read transcendentals impose a big performance penalty on true 68040 systems, although I never tried it myself.

If you are emulating it accurately (issuing the exception), I might feel tempted to disable that in my personal build and pretend I have a Cube with an hypothetical "68040 Pro" Smile
Back to top
View user's profile Send private message
andreas_g



Joined: 30 Jan 2009
Posts: 446
Location: Austria

PostPosted: Sun Jan 29, 2017 11:48 pm    Post subject: Reply with quote

FPU emulation is the last major issue in Previous. Using native FPU routines for emulating 68k FPU instructions is very unreliable. At the moment I'm working with Toni and the Hatari community to implement arithmetic operations using SoftFloat routines. We already have the basic functions working. But emulating transcendental functions will be much more complex. Any help is welcome!

Now to your question: Like real hardware, we raise an exception if any unimplemented instruction is executed on a 68040. The instruction is then emulated by NeXTstep. I do not recommend to force emulation with native FPU routines because as described above this is very unreliable.

Can you point me to the alternative transcendental source code you mentioned? It might be useful for our SoftFloat implementation.
Back to top
View user's profile Send private message Send e-mail
ardi



Joined: 11 Jun 2014
Posts: 85

PostPosted: Mon Jan 30, 2017 9:16 am    Post subject: Reply with quote

Yes, look here: http://www.nextcomputers.org/forums/viewtopic.php?p=21059&highlight=#21059

And, for the code, here:
https://ftp.nice.ch/pub/next/developer/hardware/m68k/

Apparently this code was written for increasing the FP performance of transcendentals in the 68040. I'm not sure about their accuracy. BTW, you shouldn't need to add such code for the 68040, if you generate an exception the operating system should do the math call in software.

BTW, did anybody have any real-world experience with this FP performance penalty on the 68040? Was it really that bad?
Back to top
View user's profile Send private message
t-rexky



Joined: 09 Jan 2011
Posts: 285
Location: Snowy Canada

PostPosted: Mon Jan 30, 2017 8:39 pm    Post subject: Reply with quote

ardi wrote:
BTW, did anybody have any real-world experience with this FP performance penalty on the 68040? Was it really that bad?


I wrote a lot of engineering code on my cube and then on the Turbo Color back in the day, the "ultimate" being a program that optimized propeller geometry using formal numerical optimization methods with penalty functions. That code used lots of transcendentals.

I recall that the 040 was overall much faster running the code but I cannot comment on what the performance penalty was associated with transcendentals since I never profiled it. It run for approximately 10 minutes to fully optimize twist and chord distribution of a propeller on the 68040 Turbo.

On a side note, it takes under a second on my 2009 MacBook Pro...
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    NeXT Computers Forum Index -> Emulation / Virtualization All times are GMT - 7 Hours
Goto page Previous  1, 2, 3 ... 70, 71, 72 ... 80, 81, 82  Next
Page 71 of 82

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum



Powered by phpBB © 2017 phpBB Group