Poll: What KS would you prefer in a Vampire... Aros or KS3.1 - Page 5

gulliver · 17 February 2016, 18:00

Quote:

Originally Posted by Romanujan

1. All the ROM patches for the Vampire (as far as I understand, currently the Vampire uses patched 3.1 ROM) should be available in a Remus compatible format for anyone wanting to create own custom ROM

This is very important. Without having the patches to the rom modules, no one can build custom kickstart roms and you can bet you wont get Aros support too easily.

These rom module patches should be publicly released ASAP.

NorthWay · 18 February 2016, 16:09

Quote:

Originally Posted by Romanujan

as far as I understand, currently the Vampire uses patched 3.1 ROM

I missed this one - does it???

If so, why?

I still fail to see why the solution is not to map to the real rom chips and have a tool that is MapRom compatible that will re-map it to fast ram. Solves all kinds of problems.

Amiga1992 · 18 February 2016, 16:25

Yes the Vampire needs special patches to be applied to ROM and it needs this ROM to run. It's not like a MapROM feature, it's different (and ROM will NOT be in Fast RAM)

I also don't understand why the ROM cannot be copied form your machine, patched and uploaded to the Vampire. Well, nobody said it cannot be, but is the solution I would have taken instead of dealing with CLoanto or whoever the hell.

IanP · 18 February 2016, 17:06

Another reason for only allowing the 3.1 ROM as patched by them currently may be because the Apollo core is a work in progress. If they allow any ROM to be used there's the potential for issues where the board could have problems booting up and it's difficult for an end user to fix the issue. As the Apollo core matures and more details of the patching become public such restrictions may be eased.

kolla · 18 February 2016, 21:25

Quote:

Originally Posted by Akira

I also don't understand why the ROM cannot be copied form your machine, patched and uploaded to the Vampire. Well, nobody said it cannot be, but is the solution I would have taken instead of dealing with CLoanto or whoever the hell.

They may want to replace large parts of the kickstart in order to achieve the speed they want, and certainly a lot of A600s do not even have 3.1. The alternative would be to patch the whole range of kickstarts people may have on their A600 systems.

wawa · 18 February 2016, 22:22

Quote:

Originally Posted by gulliver

This is very important. Without having the patches to the rom modules, no one can build custom kickstart roms and you can bet you wont get Aros support too easily.

These rom module patches should be publicly released ASAP.

today i have met grond with his vampire2/64 and we wasnt able to get it to boot aros off my cf card, which is expected to be working universally. i wasnt able to get any serial debug out of it, but i have not tried very hard, as there were other matters to discuss. i suppose to get aros booting an early startup on an almost bare a600 (with 2mb ram) without a vampire would be a first step i would start with (if i still had an 600).

gulliver · 18 February 2016, 22:52

Quote:

Originally Posted by wawa

today i have met grond with his vampire2/64 and we wasnt able to get it to boot aros off my cf card, which is expected to be working universally. i wasnt able to get any serial debug out of it, but i have not tried very hard, as there were other matters to discuss. i suppose to get aros booting an early startup on an almost bare a600 (with 2mb ram) without a vampire would be a first step i would start with (if i still had an 600).

Life would be a lot easier for you if the Apollo Team could just hand out the patches, along with information on what do they patch so that you could actually build an Aros68k compatible rom for the Vampire2. Hacking thru it, will require a lot of effort, and then you probably may leave some unexpected code unpatched, which will result in a randomly bad behaving port. Hacking should be a last resort. The Apollo Team´s developers should assist you ASAP.

grond · 19 February 2016, 10:12

The Apollo team has sent a vampire-equipped A600 to Jason McMullan. I think that we will see AROS boot on the vampire pretty soon.

Regarding the state of AROS on 68k the impression I got is that there are quite a lot of bugs to fix. A lot of stuff that works on x86 breaks on 68k. However, most of these issues are things that a normal C programmer can fix. Just somebody has to get down and finally DO it. It shouldn't be too difficult to get AROS 68k to the level of AROS x86. With regard to speed, I think that there probably is a lot that can be improved. AmigaOS programs that run on AROS are not slower than when running on AmigaOS 3.1 (except perhaps for some library function calls but if you are a cycle counter, then portability and clean code probably don't mean anything to you).

In my opinion the 68k is where AROS belongs and where AROS should have its home. Now with the vampire there is 68k hardware available for little money that has the power to run AROS well. It will take some work to get AROS to the level of just AmigaOS 3.1 but in contrast to everything else it is the only way to IMPROVE the OS beyond what we got in AmigaOS 3.1. (Yes, 3.5 and 3.9 blabla...)

Samurai_Crow · 19 February 2016, 10:43

I agree with Grond. The legal stipulations on the Skylake core Intels and equivalent AMD processors regarding Win10 makes x86 a no go for me. The high performance ARM Snapdragon could be at risk also since Qualcomm has a similar thing going on. We need a new CPU and the Apollo core is it.

jbenam · 19 February 2016, 13:04

Quote:

Originally Posted by Samurai_Crow

We need a new CPU and the Apollo core is it.

...Well, as long as you don't need to do anything that needs the computing power from the last 15 years or so

Samurai_Crow · 19 February 2016, 13:42

Gunnar has expressed interest in making an ASIC when the bugs are worked out of the softcores so another 10+ times speedup is still possible if it clocks up to 1GHz.

jbenam · 19 February 2016, 13:46

Quote:

Originally Posted by Samurai_Crow

Gunnar has expressed interest in making an ASIC when the bugs are worked out of the softcores so another 10+ times speedup is still possible if it clocks up to 1GHz.

Nice - has someone done some kind of benchmarks which can tell us roughly how the current core compares to an existing architecture in terms of raw grunt? (x86 would be a good reference point, since almost everyone is familiar with them)

Also, I fear we're steering a bit off-topic with this

Maybe a mod can split up those posts in another thread?

It's quite an interesting topic of its own!

Samurai_Crow · 19 February 2016, 15:16

Quote:

Originally Posted by jbenam

Nice - has someone done some kind of benchmarks which can tell us roughly how the current core compares to an existing architecture in terms of raw grunt? (x86 would be a good reference point, since almost everyone is familiar with them)

Also, I fear we're steering a bit off-topic with this

Maybe a mod can split up those posts in another thread?

It's quite an interesting topic of its own!

Let's wait until the gold version of the CPU is out for the speed comparisons with x86. For now it's much faster than an 80 MHz 68060 on the silver core on many benchmarks despite similar clock speed. (Okay, It's about Pentium 3 performance in the FPGA.)

wawa · 19 February 2016, 15:38

Quote:

Originally Posted by grond

The Apollo team has sent a vampire-equipped A600 to Jason McMullan.

hope he wont break it. he sometimes seems to mess with hardware a bit too much, as result we dont have mediator supported, as he broke it

grond · 19 February 2016, 16:00

Sam, have you got an A600? Have you ordered a vampire? Do you want one?

Or in short: have you read my pm?

Samurai_Crow · 19 February 2016, 18:13

Quote:

Originally Posted by grond

Sam, have you got an A600? Have you ordered a vampire? Do you want one?

Or in short: have you read my pm?

I read the PM a few seconds ago. I don't own an A600 however. I hope when the A500 version or A1200 version comes out I can upgrade them with a Vampire 2 for either one! (But probably not both. I'm rather short on money right now.)

chaos · 19 February 2016, 22:02

Quote:

Originally Posted by Samurai_Crow

Let's wait until the gold version of the CPU is out for the speed comparisons with x86. For now it's much faster than an 80 MHz 68060 on the silver core on many benchmarks despite similar clock speed. (Okay, It's about Pentium 3 performance in the FPGA.)

~100MIPS is Pentium I territory, Pentium III had ~2000MIPS, which is 20x more. Current CPUs have ~100000MIPS. (MIPS is of course not an absolute measure of CPU performance, but it is a pretty good indicator). Apollo still seems to be the fastest 68k processor around

IanP · 20 February 2016, 01:53

It's not that important if a modern Intel or AMD processor is 1000 times faster than the Apollo core if they are both capable of doing the tasks they need to. Why does it matter if it takes 1 microsecond or 1 millisecond for the processor to render a page of text in a word processor, you're not going to notice the difference. Most x86 software runs fine on a 10 year old PC. The Apollo core increases the processing power of a vanilla A600 or A500 by HUGE amount. If it ever does become an ASIC using modern processes it will blow single core ARM away.

Samurai_Crow · 20 February 2016, 07:35

If future Apollo cores go multithreaded, we'll need the Arix kernel which, in turn, is based on Aros.

IanP · 20 February 2016, 11:08

I remember some talk of Arix a couple of years ago but a quick google doesn't turn up much since.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
What Crunchers do you guys prefer?	Plagueis/KRX	Coders. General	67	30 July 2014 23:20
Which would prefer handheld or laptop	Vars191	Amiga scene	5	06 June 2011 23:13
KS3.0 + Deneb = KS3.9??	alexh	support.Hardware	10	04 January 2010 18:36
Which version of Afterburner do you prefer?	paul773car	Retrogaming General Discussion	3	10 September 2009 05:28
pcengine cdrom which game do you prefer	turrican3	Retrogaming General Discussion	10	12 June 2008 16:21

18 February 2016, 16:25	#83
Amiga1992 Registered User Join Date: May 2001 Location: ? Posts: 19,645	Yes the Vampire needs special patches to be applied to ROM and it needs this ROM to run. It's not like a MapROM feature, it's different (and ROM will NOT be in Fast RAM) I also don't understand why the ROM cannot be copied form your machine, patched and uploaded to the Vampire. Well, nobody said it cannot be, but is the solution I would have taken instead of dealing with CLoanto or whoever the hell.

18 February 2016, 17:06	#84
IanP Registered User Join Date: Mar 2015 Location: Bristol/UK Posts: 166	Another reason for only allowing the 3.1 ROM as patched by them currently may be because the Apollo core is a work in progress. If they allow any ROM to be used there's the potential for issues where the board could have problems booting up and it's difficult for an end user to fix the issue. As the Apollo core matures and more details of the patching become public such restrictions may be eased.

19 February 2016, 10:12	#88
grond Registered User Join Date: Jun 2015 Location: Germany Posts: 1,918	The Apollo team has sent a vampire-equipped A600 to Jason McMullan. I think that we will see AROS boot on the vampire pretty soon. Regarding the state of AROS on 68k the impression I got is that there are quite a lot of bugs to fix. A lot of stuff that works on x86 breaks on 68k. However, most of these issues are things that a normal C programmer can fix. Just somebody has to get down and finally DO it. It shouldn't be too difficult to get AROS 68k to the level of AROS x86. With regard to speed, I think that there probably is a lot that can be improved. AmigaOS programs that run on AROS are not slower than when running on AmigaOS 3.1 (except perhaps for some library function calls but if you are a cycle counter, then portability and clean code probably don't mean anything to you). In my opinion the 68k is where AROS belongs and where AROS should have its home. Now with the vampire there is 68k hardware available for little money that has the power to run AROS well. It will take some work to get AROS to the level of just AmigaOS 3.1 but in contrast to everything else it is the only way to IMPROVE the OS beyond what we got in AmigaOS 3.1. (Yes, 3.5 and 3.9 blabla...)

19 February 2016, 10:43	#89
Samurai_Crow Total Chaos forever! Join Date: Aug 2007 Location: Waterville, MN, USA Age: 49 Posts: 2,186	I agree with Grond. The legal stipulations on the Skylake core Intels and equivalent AMD processors regarding Win10 makes x86 a no go for me. The high performance ARM Snapdragon could be at risk also since Qualcomm has a similar thing going on. We need a new CPU and the Apollo core is it.

19 February 2016, 13:42	#91
Samurai_Crow Total Chaos forever! Join Date: Aug 2007 Location: Waterville, MN, USA Age: 49 Posts: 2,186	Gunnar has expressed interest in making an ASIC when the bugs are worked out of the softcores so another 10+ times speedup is still possible if it clocks up to 1GHz.

19 February 2016, 16:00	#95
grond Registered User Join Date: Jun 2015 Location: Germany Posts: 1,918	Sam, have you got an A600? Have you ordered a vampire? Do you want one? Or in short: have you read my pm?

20 February 2016, 01:53	#98
IanP Registered User Join Date: Mar 2015 Location: Bristol/UK Posts: 166	It's not that important if a modern Intel or AMD processor is 1000 times faster than the Apollo core if they are both capable of doing the tasks they need to. Why does it matter if it takes 1 microsecond or 1 millisecond for the processor to render a page of text in a word processor, you're not going to notice the difference. Most x86 software runs fine on a 10 year old PC. The Apollo core increases the processing power of a vanilla A600 or A500 by HUGE amount. If it ever does become an ASIC using modern processes it will blow single core ARM away.

20 February 2016, 07:35	#99
Samurai_Crow Total Chaos forever! Join Date: Aug 2007 Location: Waterville, MN, USA Age: 49 Posts: 2,186	If future Apollo cores go multithreaded, we'll need the Arix kernel which, in turn, is based on Aros.

20 February 2016, 11:08	#100
IanP Registered User Join Date: Mar 2015 Location: Bristol/UK Posts: 166	I remember some talk of Arix a couple of years ago but a quick google doesn't turn up much since.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)