English Amiga Board Amiga Lore


Go Back   English Amiga Board > Main > Amiga scene

 
 
Thread Tools
Old 02 November 2016, 13:10   #41
Overflow
Registered User

 
Join Date: Nov 2014
Location: Norway
Posts: 304
I personally enjoy progress reports. If the team comes back 1 week later and says "sorry gents, we went into a deadend with the code, so we are starting fresh", thats fine too.

Given how much the team(s) communicate on IRC regarding direction and hurdles, I dont think they can be accused for false advertisement.
People are constantly poking them for any information regarding what they are working on, so if anything, they are "doing as they are told/asked"
Overflow is offline  
AdSense AdSense  
Old 02 November 2016, 13:40   #42
wawa
Registered User
 
Join Date: Aug 2007
Location: berlin/germany
Posts: 883
@overflow

Very right!
wawa is offline  
Old 02 November 2016, 13:45   #43
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,379
Quote:
Originally Posted by flype View Post
It belongs to the author(s) depending on the license form choice. So i respect first opinions of author(s) before telling irrelevant informations.
No ? Don't tell me that the team doesn't even know if they can release some source they wrote themselves ???


Quote:
Originally Posted by flype View Post
Ah, now it is forbidden. Didn't knew that.
not clever != forbidden


Quote:
Originally Posted by flype View Post
Again, i told that a Cyclone have limited ressources, you continue to ignore obvious things.
Resources are limited, ok : then there are better uses for them than ammx.
This recalls me my nice days within the team. For something the boss wants to add there is always space, but for things i wanted, never.

But in fact I didn't even say it was necessarily about adding something in the fpga itself and it's not me here who ignores obvious things.

Last edited by prowler; 02 November 2016 at 21:26. Reason: Cleanup.
meynaf is online now  
Old 02 November 2016, 17:42   #44
OlafSch
Registered User
 
Join Date: Nov 2011
Location: Nuernberg
Posts: 454
@meynaf

So Gunnar was not accepting your advices? I can remember discussions at that time betweend certain people and Gunnar, it was not making proposals and then accept whatever people doing real work decide, but it was "you have to do this or that otherwise it will not be successful and only I know what to do". It is a project done by some skilled and experienced hardware developers, perhaps you like direction or you like it not. If not simple stay away from it and any related discussion. It is their project so if it fails because they are not as genious as you and some others then it is their problem, not yours. At least up to now everyone I read from who really owns the hardware and uses the core seems to be happy so I assume that Gunnar did not everything wrong. You are free to buy a vampire and develop for it but it does not sound like you would be interested in it. So only badmouthing the project and the people behind it? Sounds boring... Or you develop your own core if you think you can do better. Talk is cheap.

Last edited by prowler; 02 November 2016 at 21:27. Reason: Cleanup.
OlafSch is offline  
Old 02 November 2016, 18:09   #45
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,379
Quote:
Originally Posted by OlafSch View Post
@meynaf

So Gunnar was not accepting your advices? I can remember discussions at that time betweend certain people and Gunnar, it was not making proposals and then accept whatever people doing real work decide, but it was "you have to do this or that otherwise it will not be successful and only I know what to do". It is a project done by some skilled and experienced hardware developers, perhaps you like direction or you like it not. If not simple stay away from it and any related discussion. It is their project so if it fails because they are not as genious as you and some others then it is their problem, not yours. At least up to now everyone I read from who really owns the hardware and uses the core seems to be happy so I assume that Gunnar did not everything wrong. You are free to buy a vampire and develop for it but it does not sound like you would be interested in it. So only badmouthing the project and the people behind it? Sounds boring... Or you develop your own core if you think you can do better. Talk is cheap.
Yeah, i know, you want yes men.
Now if you could stop personal attacks and talk about technical stuff, it would be nice.

Last edited by prowler; 02 November 2016 at 21:28. Reason: Cleanup.
meynaf is online now  
Old 02 November 2016, 20:49   #46
cmsj
Registered User

cmsj's Avatar
 
Join Date: Feb 2016
Location: London / UK
Posts: 146
Quote:
Originally Posted by meynaf View Post
Now if you could stop personal attacks and talk about technical stuff, it would be nice.
Demanding that people release their code, or that they wait longer before sharing progress updates, isn't technical stuff, but you're happy to post that.

Let's all just be honest here. The AMMX stuff is developing quickly, is pretty cool, has lots of potential, isn't finished yet, and you have a personal axe to grind with Gunnar, so you're biased against it.

Is that so hard?
cmsj is offline  
Old 02 November 2016, 21:27   #47
gulliver
BoingBagged

 
Join Date: Aug 2007
Location: The South of nowhere
Age: 39
Posts: 1,489
I am a little bit worried about the future of the apollo core now that I have witnessed many amiga coders and programmers have "mixed" feelings about it.
gulliver is offline  
Old 02 November 2016, 21:40   #48
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 364
Quote:
Originally Posted by gulliver View Post
I am a little bit worried about the future of the apollo core now that I have witnessed many amiga coders and programmers have "mixed" feelings about it.
Many? I think some narcists are afraid that their arcane wisdom and prominent position are at danger because less optimised code will be able to work well or even better than their cycle counted legacy code. And that they might be forced to learn something new.

Whatever they code, the 080 will happily execute it. And even if nobody wrote any AMMX code, what's the damage? Then there are a few new instructions that nobody uses. "You come from nothing, you're going back to nothing, what have you lost? Nothing!"

Another point to add is that the coders I know will always try to write the fastest code possible. In the above example 48 instructions were replaced by only three. Obviously this will speed up execution massively (up to factor 16). And as you know, programs typically spend most of the time in small loops. Multimedia codices will usually work on lots of data like pixels or audio samples. Being able to work on 8 of them per cycle (!) is a huge speed boost. Thus, I think that less dogmatic coders will embrace the AMMX instructions. Has anyone in their right mind ever rejected AltiVec, MMx, SSE or NEON?

Last edited by prowler; 02 November 2016 at 21:54. Reason: Back-to-back posts merged.
grond is offline  
Old 02 November 2016, 22:11   #49
Overflow
Registered User

 
Join Date: Nov 2014
Location: Norway
Posts: 304
I think what Gulliver alludes to is threads like over at http://ada.untergrund.net/ where Britelite and Corial highlighted that they prefered the oldschool 060 limitations, and if they wanted beefed up hardware they might as well go all out on Wintel hardware.
Since Im a big fan of their demos, Im obviously a bit sad to see them dissmiss the Vampire, but on the positive side, I can atleast watch their productions on either my a1200/30 (altho it suffers when exposed to demoes destined for 060) or V600.

Ive been giving getting a1260 some thought, but given the price and age of the hardware, I decided to go for the Vampire, which is activly supported.

While they dont support (for now...? ) the Vamp, hopefully we can enjoy their future aga/060 productions via the V1200.
Overflow is offline  
Old 02 November 2016, 22:27   #50
gulliver
BoingBagged

 
Join Date: Aug 2007
Location: The South of nowhere
Age: 39
Posts: 1,489
Thank you for clarifying Overflow.

I was also referring to, for example, UAE (WinUAE, FS-UAE, etc.) and vbcc specific support whose authors are reluctant to provide for one reason or another.

If apollo core does not get willing programmers/coders then all these extended opcodes and special mmu modes will not get much use.
gulliver is offline  
Old 02 November 2016, 22:32   #51
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 364
They don't like to see the rules to the game they've been playing for so long change. With a more powerful hardware you can do more complicated effects. Or you can do what they did on a less powerful hardware more easily. That's why they dislike progress.

Personally I'm tired of the same silly tricks and pointless chunky-to-planar conversion. They sit on a library of code that will just become worthless when coding for the vampire. Anyway, let them do as they wish. Their code will just run fine on the 080 even though they waste some of its potential.
grond is offline  
Old 03 November 2016, 01:12   #52
wawa
Registered User
 
Join Date: Aug 2007
Location: berlin/germany
Posts: 883
Quote:
Originally Posted by grond View Post
They don't like to see the rules to the game they've been playing for so long change.
thats their earned privilege, but they are secure, at least as long as vampire hasnt been introduced as a valid platform in the demoscene compos. but with this attitude they would simply be against any new accelerator faster than 060/50, so nothing personal.
wawa is offline  
Old 03 November 2016, 02:25   #53
kolla
Registered User
kolla's Avatar
 
Join Date: Nov 2007
Location: Trondheim, Norway
Posts: 723
Aminet is getting "flooded" with EC/LC 040/060 stuff these days, lol
kolla is offline  
Old 03 November 2016, 06:29   #54
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 364
Quote:
Originally Posted by wawa View Post
they would simply be against any new accelerator faster than 060/50
The irony about it is that this "traditional" demo category was introduced because it was the fastest possible configuration allowing for the most advanced demo effects...
grond is offline  
Old 03 November 2016, 08:47   #55
buggs
Registered User

 
Join Date: May 2016
Location: Rostock/Germany
Posts: 44
Some thoughts to valid concerns around here

Quote:
Originally Posted by gulliver View Post
I am a little bit worried about the future of the apollo core now that I have witnessed many amiga coders and programmers have "mixed" feelings about it.
It has taken me some time to contemplate whether I'd keep away from this poisoned discussion (as implied by a previous post of mine). Your concerns, however underlined one truth: FUD out of whatever reasons has the tendency to stick. Especially, when personal animosity is hidden in techno-babble, it's hard to decide upon merit for those who have chosen to collect their share of experience in other areas of the vast knowledge available to us mere humans.

The future of the Apollo Core will not hang on a thread, that's for sure.

Contrary to some wild accusations I've been made aware of, the Apollo Team is not sitting in a bubble, scheming to destroy all that people were fond of when remembering the venerable 68k. Neither are they^H^H^H^Hwe (note to self: get used to it) a closed inner circle. The team members are on a constant lookout for people who are willing to lend a bit of their spare time contributing something to the project. To underline a point here: there are members like for example Jim Drew (Fusion, PCx) who keep Gunnar and Chris on their toes concerning compatibility to existing software, irrespective how badly the code in question was abusing the 68k.

The only true information you can take out of this thread is simple: Yes we are experimenting with SIMD, yes we've chosen a full motion video decoder/player* as live testbed and yes, initial results are encouraging.

@@wawa, zetro
The cute thing about Softcores is that _nothing_ is ever set in stone, unless of course published as "ready for use". What we are experimenting with is (to my knowledge) at least the third iteration of SIMD extensions for the Apollo. If and when it works out as we think it might right now, it could be made available for use after all. Then again, the functionality in itself is not the only concern. The tradeoff between possible speed gain and HW cost in the confined space of the current FPGA is something we constantly keep in mind. We are also aware of the necessity of operating system support before a putative general announcement.

Now some words concerning the thankfully still active Demo scene. There are two main active directions in my line of sight: OCS@68000 and AGA/RTG@060. Nothing changes for them. They found a consensus about target platforms and are competing (friendly, at that) for coding skill and style. Bad hacks will confine the audience to owners of the exacting platform, good code will run anywhere. So, nothing new in Amiga-land. That much is quite apparent when looking at the transition from 68000 to the later CPU's.

*My thanks and respects to Stephen Fellner at this point. His project was bold. He invested a lot of time writing an MPEG-1 Systems + MPEG-1 Video decoder with a myriad of video output options for a platform which was clearly below the minimum performance requirements for such an endeavor (or so, I initially thought).
buggs is offline  
Old 03 November 2016, 11:24   #56
britelite
Registered User
 
Join Date: Feb 2010
Location: Espoo / Finland
Posts: 242
Quote:
Originally Posted by grond View Post
They don't like to see the rules to the game they've been playing for so long change.
Well, yes, the platform has stayed static for so long which is pretty much the whole idea and what makes it fun.

Quote:
With a more powerful hardware you can do more complicated effects.
And as I've said, why limit myself to the Vampire/Apollo-core if I only wanted more power

Quote:
Their code will just run fine on the 080 even though they waste some of its potential.
Time will tell if the code will run on the 080, but then again that's not our problem. And I'm fine with wasting the "potential" of the 080, the added features are just not very interesting and will make stuff incompatible with everything else available.
britelite is offline  
Old 03 November 2016, 11:52   #57
Samurai_Crow
Total Chaos forever!

Samurai_Crow's Avatar
 
Join Date: Aug 2007
Location: Ft. Collins, CO USA
Age: 43
Posts: 890
Send a message via Yahoo to Samurai_Crow
Quote:
Originally Posted by britelite View Post
Time will tell if the code will run on the 080, but then again that's not our problem. And I'm fine with wasting the "potential" of the 080, the added features are just not very interesting and will make stuff incompatible with everything else available.
My thought is that if the AMMX code is encapsulated into only device drivers like datatypes, enhanced Kickstart routines, AHI and a SAGA graphics driver for AROS, for example, user code will not need to be altered to benefit from the performance boost. The original MMX for the Pentium Pro was hardly ever used by user code since it required code to be rewritten in Assembly and disabled the FPU but users still benefited from improvements in DirectX, for example.
Samurai_Crow is offline  
Old 03 November 2016, 11:56   #58
britelite
Registered User
 
Join Date: Feb 2010
Location: Espoo / Finland
Posts: 242
Quote:
Originally Posted by Samurai_Crow View Post
My thought is that if the AMMX code is encapsulated into only device drivers like datatypes, enhanced Kickstart routines, AHI and a SAGA graphics driver for AROS, for example, user code will not need to be altered to benefit from the performance boost.
I'm of course looking at this from a democoder point of view, and nothing of what you mentioned would give any benefit in demos as they usually access the hardware directly

But yes, it might give some boost to system friendly software in general.
britelite is offline  
Old 03 November 2016, 12:09   #59
wawa
Registered User
 
Join Date: Aug 2007
Location: berlin/germany
Posts: 883
Quote:
Originally Posted by britelite View Post
I'm of course looking at this from a democoder point of view
i think its all fine. the compo categories will simply stay as they are and apollo users might, just might, watch the demos at home with few frame rates more.

im also rather concerned about backwards compatibility and fracturing the scene even more, so i hope the extensions will primarily be used for drivers, not that we need to introduce another separate platform in aminet.
wawa is offline  
Old 03 November 2016, 13:09   #60
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,379
Quote:
Originally Posted by cmsj View Post
Demanding that people release their code, or that they wait longer before sharing progress updates, isn't technical stuff, but you're happy to post that.
This is demanding for technical stuff because there isn't much to discuss right now. Now as it seems i'm not allowed to ask, i'll wait until both the modified program and the doc about the extension are available.


Quote:
Originally Posted by cmsj View Post
Let's all just be honest here. The AMMX stuff is developing quickly, is pretty cool, has lots of potential, isn't finished yet, and you have a personal axe to grind with Gunnar, so you're biased against it.

Is that so hard?
My strong distaste for cpu vector extensions traces back a lot before i even joined the team.
They lack the flexibility of other instructions, are short-lived, and are superseded by gpus for heavy duty stuff such as video and 3d.
Of course they also have a non-negligible hw cost. (Btw. 64-bit misaligned dcache accesses seem excessive for a "small" fpga.)

***

Perhaps i'm just wrong and there is nothing bad in these extensions.
This is not a reason for having treated me with contempt (forcing the moderation to come and remove stuff).

Some people like their 060, if not their 030, because they like doing much with little. Nowadays this is what demomaking is all about.

I'm not against new instructions. But i see benefit only on these that would enhance the ease of programming, which that ammx clearly is not.

One could say it's just an fpga so things can be added and removed at will. This holds true until there is a release that gets in wide enough use.
If it becomes kind of a standard it also becomes legacy. And the cost of supporting legacy isn't zero. This is my main concern.

In fact i see more potential in this project than any other people here. For me it's not just another accelerator : it's about the eventual future of the one and only cpu that's reasonably good for direct asm programming of whole applications. This is why i pop up any time my guts tell me something wrong's being done.
meynaf is online now  
AdSense AdSense  
 


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

Similar Threads
Thread Thread Starter Forum Replies Last Post
RiVA mpeg player amiga request.Apps 18 25 February 2014 10:32
ACA1231 vs M-Tec 1230 vs A630 benchmarks alenppc support.Hardware 4 11 July 2012 18:39
Some benchmarks, and a request Damion support.Hardware 29 14 March 2011 16:10
riva amiga request.Apps 6 12 May 2008 18:56
Benchmarks. ECA support.Hardware 4 14 June 2002 15:14

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +2. The time now is 15:35.


Powered by vBulletin® Version 3.8.8 Beta 1
Copyright ©2000 - 2017, vBulletin Solutions, Inc.
Page generated in 0.24043 seconds with 11 queries