Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
Please visit our Apollo-Discord Server for support.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Documentation about the Vampire hardware

Vampire Hardware Featurespage  1 2 3 4 5 6 7 8 

Wawa T

Posts 695
26 Apr 2017 01:13


its "lose", not "loose". the application data originates form a mass storage, no matter whether you pre-load it in ram or load it dynamically while the execution. its the speed of he mass storage interface that matters. if its slow and you have a lot of ram you may pre-load that data, but it will take time and delay the app anyway. if you have a fast interface, you dont need to pre-load, so you dont need that ram to cache data.

in conclusion, adding insane amounts of ram like you imply, can only be considered a workaround for slow i/o. better to speed up that i/o than work around that.


Thierry Atheist

Posts 644
26 Apr 2017 01:42


wawa t wrote:
in conclusion, adding insane amounts of ram like you imply, can only be considered a workaround for slow i/o. better to speed up that i/o than work around that.

You do things your way, some of us want it working better = more RAM!



Wawa T

Posts 695
26 Apr 2017 02:28


Thierry Atheist wrote:

wawa t wrote:
in conclusion, adding insane amounts of ram like you imply, can only be considered a workaround for slow i/o. better to speed up that i/o than work around that.

  You do things your way, some of us want it working better = more RAM!
 

okay. do your own better board with more ram, then. none is stopping you, except yourself.


Mark Smith

Posts 30
26 Apr 2017 06:05


bit late to the conversation but ...

Just simplify the process, put the max memory that will go on the board and not faff around with multiple configurations, more choice means more scope for accidently shipping the wrong size to someone, building lower spec units that don't sell because people want the full fat version and for apps to be written that need config-x and others that need config-y.

Just my thoughts on keeping it simple :-)


Uros Vidovic

Posts 31
26 Apr 2017 07:49


About the amount of RAM.
2GB except of using it as RAM Disk has no real sense. You can dream, but making computer even for a Cent more expensive with a feature almost nobody can use it, is expensive.
Yes you can use it just as proof something is possible but for that you can use WinUAE and it will be much faster.
3D rendering. I do that on my Atari clones, but just for fun. The quality is the same as on the PC but instead of a second it get rendered in hours :)

And having OS without memory protection it means you easyly crash OS while scene is rendered in background.

First think what computer is capable and dont make computer more expensive just because of your dreams as this will mean less new buyers = users.

But as an Atari user I actualy dont care. If standalone will become once usable Atari clone I will consider buy it :)


Ian Parsons

Posts 230
26 Apr 2017 09:34


512MB or 1GB would seem to be sensible amounts for a system based on the current FPGA hardware and other factors. If you need more than that then this probably isn't the right platform for your tasks (yet). I think the vast majority of potential customers would be more than happy with this much RAM. Without knowing more about the financial implications of the decision I'd be reluctant to recommend a choice. If the extra cost isn't too high 1GB does have a wow factor for a "classic Amiga" though.


Michael R

Posts 281
26 Apr 2017 11:41


Ian Parsons wrote:

  512MB or 1GB would seem to be sensible amounts for a system based on the current FPGA hardware and other factors. If you need more than that then this probably isn't the right platform for your tasks (yet). I think the vast majority of potential customers would be more than happy with this much RAM. Without knowing more about the financial implications of the decision I'd be reluctant to recommend a choice. If the extra cost isn't too high 1GB does have a wow factor for a "classic Amiga" though.
 

 
  We still have to realize what we are talking about though to put it into perspective. The design team will have to test future versions of accelerator cards (v1200, etc) for power constraints. Maybe only up to 256MB will work with a basic power supply.
 
  The standalone motherboard will be a different configuration because it will have a separate (laptop?) External power supply. Which also means 512MB or 1GB or even 2GB is possible. But these options will be for a future version of the standalone board most likely since Gunnar has said the design is finished for the v1200 and the standalone and the full specs and photos of those designs will be available in the coming days. So the max ram issue is for next year, or after for maybe Generation II Motherboard (Mainboard V2?).
 
  I hope, in addition to more onboard memory the team also considers a method for future expansion for bus boards, etc. A PCI express edge connector is quite feasible from a design perspective and very exciting as well. It might go on the left side since the board is so small (100mm x 120mm)? PCI express x16 edge connector measures 96mm. With such a connector only 128MB or 256MB onboard is necessary when a PCI express RAM expansion board allows us to add what we feel we need. A high speed, high bandwidth edge connector like that would also allow connection another 100mm x 100mm (4in x 4in) board with multiple x1's and maybe USB 3, SATA 3, mSata for small SSD's, and whatever else fits. The two boards together (mainboard + busboard) together may be less than a Micro ATX form factor but maybe the sizes can be ajusted to fit a standard computer case or better yet a new re-manufactured Amiga 1200 case! An onboard A1200 keyboard connector would be great.
 
  All the above including your RAM options are good ideas for a future standalone version but we should wait to see what Generation I looks like in a few days. There will be time later for modifications, but if this board has options for expansion the modifications may be available sooner rather than later. But how much RAM does Amiga OS need? Up to 2GB? But what about other OS'es?


Michael R

Posts 281
26 Apr 2017 11:52


wawa t wrote:

its "lose", not "loose". the application data originates form a mass storage, no matter whether you pre-load it in ram or load it dynamically while the execution. its the speed of he mass storage interface that matters. if its slow and you have a lot of ram you may pre-load that data, but it will take time and delay the app anyway. if you have a fast interface, you dont need to pre-load, so you dont need that ram to cache data.
 
  in conclusion, adding insane amounts of ram like you imply, can only be considered a workaround for slow i/o. better to speed up that i/o than work around that.

A faster I/O requires faster ram which Gunnar has suggested will increase the performance overall. That's a good idea too!



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
26 Apr 2017 12:27


Regarding faster IO.
 
The Vampire IDE right now does reach 11-14 MB/sec.
This is very good for an AMIGA.
 
The limit is not the SAGA IDE hardware.
Our IDE interface can actually reach a lot more.
We could reach about twice as fast IDE speed.
The main bottleneck is today the software of the SCSI.device in Kickstart.
 
So you do not even need a hardware upgrade - all you need is someone slightly improving the Kickstart SCSI.device and then you can get over 20 MB/sec IDE speed.
 


Michael R

Posts 281
26 Apr 2017 13:31


Gunnar von Boehn wrote:

Regarding faster IO.
   
  The Vampire IDE right now does reach 11-14 MB/sec.
  This is very good for an AMIGA.
   
  The limit is not the SAGA IDE hardware.
  Our IDE interface can actually reach a lot more.
  We could reach about twice as fast IDE speed.
  The main bottleneck is today the software of the SCSI.device in Kickstart.
   
  So you do not even need a hardware upgrade - all you need is someone slightly improving the Kickstart SCSI.device and then you can get over 20 MB/sec IDE speed.
 

An IDE speed of 20MB per sec is a very interesting possibility! In a related question, is the on-board Micro SD card slot fully enabled and bootable? What potential speed could we get from that? Is it faster?



Nixus Minimax

Posts 416
26 Apr 2017 13:36


Michael R wrote:
In a related question, is the on-board Micro SD card slot fully enabled and bootable? What potential speed could we get from that? Is it faster?

The Micro SD is slower and software support for booting from SD is missing.




Gunnar von Boehn
(Apollo Team Member)
Posts 6207
26 Apr 2017 13:40


Nixus Minimax wrote:

software support for booting from SD is missing.

Yes BOOTING from SD is just a driver question.



Michael R

Posts 281
26 Apr 2017 14:42


Gunnar von Boehn wrote:

Regarding faster IO.
   
  The Vampire IDE right now does reach 11-14 MB/sec.
  This is very good for an AMIGA.
   
  The limit is not the SAGA IDE hardware.
  Our IDE interface can actually reach a lot more.
  We could reach about twice as fast IDE speed.
  The main bottleneck is today the software of the SCSI.device in Kickstart.
   
  So you do not even need a hardware upgrade - all you need is someone slightly improving the Kickstart SCSI.device and then you can get over 20 MB/sec IDE speed.
 

How do we motivate creative people interested in improving Amiga software drivers to be used with Apollo products (including accelerator cards and mainboard)?

We could set up Software Bounties open to all Amiga 68k developers to help motivate them to revise the SCSI.device in kickstart to enable faster IDE. In a related bounty maybe revise AROS IDE code to allow AROS 68k and AROS Vision (68k) to boot from IDE on Vampire cards. In a separate bounty we could request a revision of the Micro SD driver for booting from Vampire. How much is that worth to us? I would contribute $50 or $100 for each project but someone has to write the bounty requests.

The Apollo Team can come up with great hardware! They know no limits! (Good Motto).

But at some point it's up to us, the Amiga Community, to provide the motivation and creativity to do the rest. Let's keep up that spirit of creativity and cooperation that they have started here! Keep the Amiga spirit of invention alive!



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
26 Apr 2017 14:47


Michael R wrote:

    In a related bounty maybe revise AROS IDE code to allow AROS 68k and AROS Vision (68k) to boot from IDE on Vampire cards.
   

Actually the AROS IDE problem is NOT related to the VAMPIRE.
AFAIK the reason is that AROS does not correctly support all AMIGA IDE versions.
So finally fixing this would benefit all AMIGAs

Actually tweaking the original SCSI device would not be such a big task. Could be done on a weekend literally.
I agree with your thinking that improving software is really what the community does need.


Michael R

Posts 281
26 Apr 2017 15:01



 
  I agree with your thinking that improving software is really what the community does need.
 

 
  We also could take advantage of the six pin Arduino style connector on the V500 V2+ cards by providing some custom made modules if necessary based on Arduino module designs at first but designed to be completely Amiga and Vampire compatible. Those might provide Wifi, Bluetooth, Micro USB, SATA...
 
  We would need Amiga/Vampire drivers and config software for those too.
 


Michael R

Posts 281
26 Apr 2017 15:17


Gunnar von Boehn wrote:

 
  Actually tweaking the original SCSI device would not be such a big task. Could be done on a weekend literally.
  I agree with your thinking that improving software is really what the community does need.

That's an easy $100. Do you accept checks? :-)

I look forward to seeing the first 20MB per sec benchmarks for Vampire IDE. Then it will be FAST IDE.


Wawa T

Posts 695
26 Apr 2017 15:41


Gunnar von Boehn wrote:

Nixus Minimax wrote:

  software support for booting from SD is missing.
 

 
  Yes BOOTING from SD is just a driver question.
 

a device to bootable needs to be either in kickstart or in its own autoconfig rom afair. sd has probably just a device driver. is it romable?



Wawa T

Posts 695
26 Apr 2017 15:43


Gunnar von Boehn wrote:

Actually the AROS IDE problem is NOT related to the VAMPIRE.
AFAIK the reason is that AROS does not correctly support all AMIGA IDE versions.
So finally fixing this would benefit all AMIGAs

exactly. have not been able to achieve this together with toni and it has been shelved.. sad truth.



Wawa T

Posts 695
26 Apr 2017 16:11


wawa t wrote:

Gunnar von Boehn wrote:

  Actually the AROS IDE problem is NOT related to the VAMPIRE.
  AFAIK the reason is that AROS does not correctly support all AMIGA IDE versions.
  So finally fixing this would benefit all AMIGAs
 

 
  exactly. have not been able to achieve this together with toni and it has been shelved.. sad truth.
 

actually it must be some lapsus. maybe it s even been corrected in meantime. i have not been testing so much with real hardware lately and there were some commits to interrupts and timers handling. one way or the other if anybody has any proposal i can compile a test version of kickstart here in a few minutes. 



Nixus Minimax

Posts 416
26 Apr 2017 16:29


wawa t wrote:
a device to bootable needs to be either in kickstart or in its own autoconfig rom afair. sd has probably just a device driver. is it romable?

Heiroglyph was working on the driver and got pretty far. I think the last status was that he was stuck on some undocumented early-boot stuff that happens somewhere very early in the kickstart initialisation. So yes, the driver was already in the rom but was ignored or didn't work the way it was intended to. Unfortunately Heiroglyph resigned from the project due to some interference from the real world.

I'm not sure it's worth the effort anyway. Onboard IDE works fine and the IOs for the SD Card can be used for other less redundant stuff like networking adapters.

posts 142page  1 2 3 4 5 6 7 8