View Single Post
Old 08 June 2017, 20:53   #142
matthey
Banned
 
Join Date: Jan 2010
Location: Kansas
Posts: 1,284
Quote:
Originally Posted by OlafSch View Post
my critisizm regarding how you tried to "convince" gunnar stays. You have nothing to offer to him except some vague promises what you "might" do. Very propable basic decisions were done long time ago, they are now finishing their vision step by step. It might be that your vision would have been different but he (and a small group of people) are doing the work, not you, not me, not kollar. There is even a open source core like TG68 out there, so everybody with enough skills can do something. I could not and as far as I understand it neither you nor kollar. But still you all are requesting this or that in every thread, kollar even accusing me to do a shitstorm what is funny, I must think of "if you sit in glass house you should not throw with stones" there. I think the basic problem with you and gunnar and some time ago thor and gunnar are different goals of the project and people are not really caring what the other side thinks.
I didn't have any promises. I was still trying to figure out what everyone needed and wanted and trying to put it together. It didn't matter as it became all about what one person wanted when he chose to take a left turn, head in another direction and ignore everyone else.

Quote:
Originally Posted by OlafSch View Post
Thor talked against the project how it is done because he thought it should best integrate in existing amiga hardware universe but that is not the goal, it is defining a new compatible but different platform replacing many existing extensions. Thor did not understand this so there were long partly emotional debates. You I think have the idea of a general purpose 68k core but Gunnar wants a new amiga 68k platform mainly so he does not care which requirements might be f.e. in embedded market. Am I right?
Gunnar ignored ThoR when he should have been working with him to design a new MMU with a compatibility layer for the 68040/68060 MMU which didn't scale well (as ThoR proposed). Of course you need a virtual address capable MMU but Gunnar couldn't afford the performance hit in his FPGA only CPU. So Gunnar's new standard he wants to create is neither 68k FPU or 68k MMU compatible and there is no allowance in his ISA for compatibility because it is not scaleable or designed for the future. He attempted to make his ISA high performance but the lack of scalability means his SIMD is stuck with 64 bit integer only registers when high performance SIMDs have at least 128 bit integer+FPU capabilities. I wouldn't invest money in creating an ASIC with these limitations. I would like to know what market he targets as it appears he is ignoring a larger market than he appeals to with an old outdated MMX SIMD unit and no 68k compatible FPU or MMU.

Quote:
Originally Posted by kolla View Post
My issues are:
* pointless compatibility problems should be avoided, not embraced
* compatibility with existing operating systems and software should have much higher priority than utopian visions about making Amiga... eh, no... m68k great again!!11 (that one goes to Matthey too btw)
* the War on FPU and anyone who may have need for FPU, should stop. Notice how _every_ posting regarding software not working is left uncommented, or even removed, from apollo-core.com?
* The apparently existing fully working and functional 68882 softcore _should_ be made available, for two reasons - firstly, there are Vampire owners who have use for it. Secondly, it would be waste to not make it available.
The ISA I had documented were ideas the "team" came up with before Gunnar took a left turn by himself. Integer compatibility would have been practically perfect while FPU compatibility would have been high (practically perfect for the Amiga). The FPU would have been based on the 68060 FPU as it is not practical to bring back all the 6888x instructions. I wanted to add the FPU before the SIMD. As you can see, I do strongly support keeping a 68k compatible FPU. I am open minded enough to look at less compatible opportunities but they need to be compelling. Some compatibility probably necessarily needs to be broken where poor decisions were made or there is a lack of scalability. For example, a TLSFMem memory handler may not be 100% compatible but fast bit field instructions could make dynamic memory allocations/deallocations several times faster. There may be a few programs which would need recompiles or patching if deemed valuable enough.

Quote:
Originally Posted by kolla View Post
* marketing should be a lot clearer about what one get - especially now with retailers shipping it.
* name calling and behaviour:
- Just because people don't agree with your CPU design choices does not justify calling them "idiots".
- Just because people find the AMMX a wasted effort does not justify calling them "idiots".
- Just because people question how the Apollo Core is superior to ARM does not make them "idiots".
- When your only argument against ARM is "it sucks!!", it does not make the Apollo Core superior.
- Letting over-enthusiastic groupies state just about _anything_, no matter how over the top bat shit crazy, rant about how insanely awesome the Apollo Core is and how awesome the future will be, while at the same time engage in name calling and mud slinging contests when skilled people (not me) who do disagree with decisions dare to speak up.
- I could go on here, but what would be the point...
I agree on both points. The marketing is deceptive and the personal attackers with any connection to the Apollo Core are doing a disservice to the product.

Quote:
Originally Posted by kolla View Post
As I see it, Gunnar does not really even care about Amiga, his love is for 68k and he wants to push the 68k architecture to wherever his whim brings it. What OS it ends running, or what community he ends up with doesn't really matter, what matters for him is that he can have a following who takes part in his quest for what he sees as the ultimate 68k design. Software exists merely to showcase the superiority of his design, any software that does't is pointless. He knows the Amiga community, he knows its mix of "desperate" and "gullible", and he knows he can play this.
I disagree. I believe Gunnar cares about the Amiga community and believes he is doing good by leading us away from what he considers outdated and slow hardware. This isn't the first time he lost touch with reality and tried to sacrifice everything, especially compatibility, for maximum performance. I hope the community and his business partners can bring him back from his obsession once again. I believe he would be a valuable employee if he just wasn't in charge.

Quote:
Originally Posted by Akira View Post
No, you can't! No cheer and good vibes in this thread!
Only bile, bickering and pointless resentment over a dead platform people keep wanting to zombify
It isn't that bad is it? Much of the discussion is pointless if there is no attempt to bring the 68k Amiga back but if there is I can't emphasize enough how much bad standards will make it harder.

Quote:
Originally Posted by Open Sources: Voices from the Open Source Revolution
This is somewhat analogous to the concept of creating and maintaining standards with open-source software. It is very expensive to create a standard and get it right. But it is far more expensive to work without standards or to try to maintain a standard if the standard is bogus. There is great value in having good people working on software whose precedents will set the standards of tomorrow. We believed at the beginning that people would understand this value proposition, and would value the opportunity to pay us to create high-quality, open-source programs that would become the de facto standard of the software world.
http://www.oreilly.com/openbook/open...k/tiemans.html

This is from Michael Tiemann who was a major contributor in making Linux and GCC portable and popular. His book also talks about bad interfaces and avoiding them later. There is an interesting section called "Amiga and the Motorola Port". Their business model was unique at the time and many people told them it wouldn't work. Could the Amiga borrow some ideas from this business model? Does anyone read anymore (the complete book is at the link above)?

Last edited by matthey; 08 June 2017 at 20:59.
matthey is offline  
 
Page generated in 0.06965 seconds with 9 queries