Poster | Thread |
wawa
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 19:10:07
| | [ #101 ] |
|
|
 |
Elite Member  |
Joined: 21-Jan-2008 Posts: 6259
From: Unknown | | |
|
| @iggy
Quote:
How long has Gunnar been working on CPUs? |
point me to someone else around here who has developed own amiga cpu, in average faster than anything previously available, implementing the whole instruction set and an fpu, however long it would have taken. if you had found someoone like that you might now have a point. |
|
Status: Offline |
|
|
pavlor
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 19:20:26
| | [ #102 ] |
|
|
 |
Elite Member  |
Joined: 10-Jul-2005 Posts: 9660
From: Unknown | | |
|
| |
Status: Offline |
|
|
Anonymous
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 19:48:51
| | [ # ] |
|
| @wawa
Quote:
wawa wrote: @iggy
Quote:
How long has Gunnar been working on CPUs? |
point me to someone else around here who has developed own amiga cpu, in average faster than anything previously available, implementing the whole instruction set and an fpu, however long it would have taken. if you had found someoone like that you might now have a point. |
I think he should take a long, hard look in the mirror and ask "How can I be as brilliant as Iggy?"Last edited by clebin on 08-Dec-2015 at 07:51 PM.
|
|
|
|
|
AmigaOldskooler
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 21:21:32
| | [ #104 ] |
|
|
 |
Regular Member  |
Joined: 7-Mar-2015 Posts: 286
From: Unknown | | |
|
| |
Status: Offline |
|
|
iggy
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 21:28:21
| | [ #105 ] |
|
|
 |
Super Member  |
Joined: 20-Oct-2010 Posts: 1175
From: Bear, Delaware USA | | |
|
| @wawa
Quote:
point me to someone else around here who has developed own amiga cpu |
Seriously? How many 68K cores do you want a reference to?
And he hasn't implemented everything yet, but just like in past endeavors he constantly proposes new tangents to consider.
And I wouldn't reinvent the wheel (and doubt Gunnar has) I'd use someone else's work as a starting point.
As far as spending an enormous amount of time to get results that will still only be a fraction of what can be achieved with simple emulation? I wouldn't do it.
I'd use a more modern processor.
After all, if Freescale thought a more powerful 68K could compete with current CPUs, they would have kept developing it.
As someone that built and sold 68K based systems from the mid '80s to the '90s (and loved it), I learned to let it go and reserve my hobbyist time for fun.
I didn't decide to spend years trying to create a slightly faster computer.
BUT, really, BUT if you want to...have at it. I did mention something about fun earlier, didn't I? ;)Last edited by iggy on 08-Dec-2015 at 09:40 PM.
|
|
Status: Offline |
|
|
wawa
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 21:47:45
| | [ #106 ] |
|
|
 |
Elite Member  |
Joined: 21-Jan-2008 Posts: 6259
From: Unknown | | |
|
| @iggy
Quote:
iggy wrote: @wawa
Quote:
point me to someone else around here who has developed own amiga cpu |
Seriously? How many 68K cores do you want a reference to?
And he hasn't implemented everything yet, but just like in past endeavors he constantly proposes new tangents to consider.
And I wouldn't reinvent the wheel (and doubt Gunnar has) I'd use someone else's work as a starting point.
As far as spending an enormous amount of time to get results that will still only be a fraction of what can be achieved with simple emulation? I wouldn't do it.
I'd use a more modern processor.
After all, if Freescale thought a more powerful 68K could compete with current CPUs, they would have kept developing it.
As someone that built and sold 68K based systems from the mid '80s to the '90s (and loved it), I learned to let it go and reserve my hobbyist time for fun.
I didn't decide to spend years trying to create a slightly faster computer.
BUT, really, BUT if you want to...have at it. I did mention something about fun earlier, didn't I? ;) |
im really curious of your link to a 68k compatible core or chip that is more performant than what gunnar proposes in its current state. im fine with you not spending your time to develop slightly better 68k processor. but i wonder why you are angry at someone, who did, maybe he knows what he doues whith his time, especially that he actually achieved his goal. |
|
Status: Offline |
|
|
iggy
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 22:09:47
| | [ #107 ] |
|
|
 |
Super Member  |
Joined: 20-Oct-2010 Posts: 1175
From: Bear, Delaware USA | | |
|
| @clebin
Quote:
"How can I be as brilliant as Iggy?" |
Oh, even I would fail at that. And, the real question should be 'Can you go as over board?'
Really, if I can accept the idea of expensive video upgrades for 2MHz Ataris, what is wrong with pushing the 68K, for fun.
But I can be downright testy about the 68K, as I used to derive my income from those systems. It still galls me that Intel won that one (or that I would be advocating solutions based on that stuff).
|
|
Status: Offline |
|
|
megol
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 22:09:59
| | [ #108 ] |
|
|
 |
Regular Member  |
Joined: 17-Mar-2008 Posts: 355
From: Unknown | | |
|
| @iggy
Quote:
iggy wrote: @wawa
Quote:
point me to someone else around here who has developed own amiga cpu |
Seriously? How many 68K cores do you want a reference to?
And he hasn't implemented everything yet, but just like in past endeavors he constantly proposes new tangents to consider.
|
Really? AFAIK we have benchmarks run on a real FPGA based accelerator.
Quote:
And I wouldn't reinvent the wheel (and doubt Gunnar has) I'd use someone else's work as a starting point.
|
There is no soft core processor that is even close in goals/architecture/compatibility. It's from scratch. I can't honestly think of one thing that could have been reused from another processor.
Quote:
As far as spending an enormous amount of time to get results that will still only be a fraction of what can be achieved with simple emulation? I wouldn't do it.
|
Then this is not for you. Just ignore the whole thing.
Quote:
I'd use a more modern processor.
After all, if Freescale thought a more powerful 68K could compete with current CPUs, they would have kept developing it.
|
Bullshit. Motorola canceled their 88100 series RISC and their 68k series processors as they had found the new thing, Power PC! This processor would be able to compete with the already strong x86 family processors by combining the power of Motorola, IBM and Apple and gaining market share by showing that the x86 CISC was a dead end and RISC was the future!
Only problem was that CISC wasn't a dead end and PPC didn't gain enough market to make a difference.
There is no technological problem in scaling the 68k to the same level of performance as current x86 processors. Motorola just didn't want to.
Quote:
As someone that built and sold 68K based systems from the mid '80s to the '90s (and loved it), I learned to let it go and reserve my hobbyist time for fun.
I didn't decide to spend years trying to create a slightly faster computer.
BUT, really, BUT if you want to...have at it. I did mention something about fun earlier, didn't I? ;) |
|
|
Status: Offline |
|
|
iggy
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 22:12:41
| | [ #109 ] |
|
|
 |
Super Member  |
Joined: 20-Oct-2010 Posts: 1175
From: Bear, Delaware USA | | |
|
| @wawa
Quote:
And proposes, and proposes...
Quote:
he actually achieved his goal |
NOT yet.
Could happen, I won't discount it, except that its 25 years too late.Last edited by iggy on 08-Dec-2015 at 10:13 PM.
|
|
Status: Offline |
|
|
iggy
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 22:22:46
| | [ #110 ] |
|
|
 |
Super Member  |
Joined: 20-Oct-2010 Posts: 1175
From: Bear, Delaware USA | | |
|
| @megol
Quote:
Motorola just didn't want to. |
And IBM didn't want to build faster 970MPs (AKA G5) cpus.
I have heard this all before.
But outside of IBM's Power designs, no one else has really ever been able to challenge Intel. (except maybe AMD during Intel's Netburst period - but that is still X86).
Fact,, compare a 2.7 GHz G5 to a Xeon from the same period and the G5 will outclass its performance.
RISC WAS a good idea, one that still lives on in ARM.
As to the 68K, that don't scale.
But it did outclass Intel in its time.
BTW - I'll butt out of this one (at least for awhile - see you in 'two weeks'). Gunnar might think I don't like him anymore (when its more like 'Is the f'ing thing finished yet?')Last edited by iggy on 08-Dec-2015 at 10:25 PM.
|
|
Status: Offline |
|
|
wawa
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 22:32:57
| | [ #111 ] |
|
|
 |
Elite Member  |
Joined: 21-Jan-2008 Posts: 6259
From: Unknown | | |
|
| @iggy
im waiting for your link. |
|
Status: Offline |
|
|
cdimauro
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 22:51:43
| | [ #112 ] |
|
|
 |
Elite Member  |
Joined: 29-Oct-2012 Posts: 4127
From: Germany | | |
|
| @iggy
Quote:
iggy wrote: @megol
Quote:
Motorola just didn't want to. |
And IBM didn't want to build faster 970MPs (AKA G5) cpus.
|
They did it, but Apple wasn't interested anymore, because they draw too much power and generated a lot of heat. Quote:
But outside of IBM's Power designs, no one else has really ever been able to challenge Intel. (except maybe AMD during Intel's Netburst period - but that is still X86).
|
Digital/Alpha, Silicon Graphics/MIPS, and HP/PA-RISC. Quote:
Fact,, compare a 2.7 GHz G5 to a Xeon from the same period and the G5 will outclass its performance.
|
Not really: Athlon 64 vs. Apple G5 Systems: Not Even Close
It isn't about the 2.7Ghz G5, but it clearly shows that such processor cannot compete with the existing x86 designs, even if we scale the performance to 2.7Ghz (and we know that scaling isn't linear with the clock frequency).
Last but not really least, the x86 were underpowered, because they ran only in 32-bit mode, whereas in 64-bit mode (x64) the performance gains around 10-15% (thanks to doubling the number of GP and SIMD registers). Quote:
RISC WAS a good idea, one that still lives on in ARM.
|
It was a good idea when you had few transistors, but when the transistors packed became more and more, the savings of RISCs became less and less important. Negligible from very long time. Quote:
As to the 68K, that don't scale.
But it did outclass Intel in its time.
|
Not really, because Intel and the other x86 vendors continued to scale their chip, whereas Motorola wasn't able to follow them.
Giving an example to be more clear, Motorola stuck at a maximum of 33Mhz with its 68040, while Intel et co. reached 100-120Mhz. Yes, clock for clock Motorola's chip was more efficient, but it wasn't absolutely enough to fill the huge frequency gap. |
|
Status: Offline |
|
|
OlafS25
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 23:43:07
| | [ #113 ] |
|
|
 |
Elite Member  |
Joined: 12-May-2010 Posts: 6472
From: Unknown | | |
|
| @iggy
do I need to remind you on recent benchmarks Odyssey on Aros X86 versus PPC AmigaOS and MorphOS. PPC even is beaten by average smartphones. PPC is today as retro as 68k just with a fraction of the software base.
Regarding Gunnar, everybody can preoorder one of the Vampire 2 cards and use the core and make his own judgement. It is basically a toy, but a nice toy. It will make 68k mainstream again but it is geeky and offers at least a chance to build up a niche. |
|
Status: Offline |
|
|
OlafS25
|  |
Re: News about Vampire and Apollo Posted on 8-Dec-2015 23:46:26
| | [ #114 ] |
|
|
 |
Elite Member  |
Joined: 12-May-2010 Posts: 6472
From: Unknown | | |
|
| @iggy
Too late for what? To win the universe? 
Nobody seriously believes that such a solution will beat the big platforms. Impossible. It is all about creating a niche and how big such a niche becomes. |
|
Status: Offline |
|
|
Zylesea
|  |
Re: News about Vampire and Apollo Posted on 9-Dec-2015 0:16:05
| | [ #115 ] |
|
|
 |
Elite Member  |
Joined: 16-Mar-2004 Posts: 2264
From: Ostwestfalen, FRG | | |
|
| @OlafS25
Quote:
OlafS25 wrote: @iggy
do I need to remind you on recent benchmarks Odyssey on Aros X86 versus PPC AmigaOS and MorphOS. PPC even is beaten by average smartphones. PPC is today as retro as 68k just with a fraction of the software base. |
PPC is not retro. But it is a dead end. The raw performance of some ppc chips is not that bad. x64 is definitely faster, but ppc has some juice, too. The odyssey issue is not primarily of lacking raw cpu power, but no JS JIT and being stuck in some endian issues.
That is bad enough and reason enough to jump ISA to mainstream (x64), but it doesn't make ppc retro.
_________________ My programs: via.bckrs.de MorphOS user since V0.4 (2001) |
|
Status: Offline |
|
|
CodeSmith
|  |
Re: News about Vampire and Apollo Posted on 9-Dec-2015 1:14:58
| | [ #116 ] |
|
|
 |
Elite Member  |
Joined: 8-Mar-2003 Posts: 3045
From: USA | | |
|
| @Zylesea
Quote:
Zylesea wrote: @OlafS25
Quote:
OlafS25 wrote: @iggy
do I need to remind you on recent benchmarks Odyssey on Aros X86 versus PPC AmigaOS and MorphOS. PPC even is beaten by average smartphones. PPC is today as retro as 68k just with a fraction of the software base. |
PPC is not retro. But it is a dead end. The raw performance of some ppc chips is not that bad. x64 is definitely faster, but ppc has some juice, too. The odyssey issue is not primarily of lacking raw cpu power, but no JS JIT and being stuck in some endian issues.
That is bad enough and reason enough to jump ISA to mainstream (x64), but it doesn't make ppc retro.
|
Yep, in the progression Useful Device -> Obsolete Crap -> Retro Cool, PPC is not currently in a good place 
I wish Gunnar the best of luck, but he has a pretty bad problem with feature creep. I already warned him back in the natami days that an average design today is better than a perfect one next year (assuming you can even hit perfect), and here we are. I have no doubt that the new design is objectively better than the natami one, but if he never progresses beyond prototypes this is just his personal hobby and nothing for the rest of us to get excited about.
|
|
Status: Offline |
|
|
retro
|  |
Re: News about Vampire and Apollo Posted on 9-Dec-2015 2:02:24
| | [ #117 ] |
|
|
 |
Super Member  |
Joined: 16-Dec-2003 Posts: 1050
From: Unknown | | |
|
| |
Status: Offline |
|
|
cdimauro
|  |
Re: News about Vampire and Apollo Posted on 9-Dec-2015 6:24:02
| | [ #118 ] |
|
|
 |
Elite Member  |
Joined: 29-Oct-2012 Posts: 4127
From: Germany | | |
|
| @OlafS25
Quote:
OlafS25 wrote:
PPC is today as retro as 68k
|
What's the last 68K processor produced? And the last PowerPC one? Quote:
just with a fraction of the software base.
|
That has nothing to do with the "retro" label. And we already discussed it, and you had nothing to say about it. Quote:
Regarding Gunnar, everybody can preoorder one of the Vampire 2 cards and use the core and make his own judgement. It is basically a toy, but a nice toy. It will make 68k mainstream again
|
ROFL. Do you really believe on what you says?  Quote:
but it is geeky and offers at least a chance to build up a niche. |
geek and niche are the opposite of mainstream...
@CodeSmith
Quote:
CodeSmith wrote:
I wish Gunnar the best of luck, but he has a pretty bad problem with feature creep. I already warned him back in the natami days that an average design today is better than a perfect one next year (assuming you can even hit perfect), and here we are. I have no doubt that the new design is objectively better than the natami one, but if he never progresses beyond prototypes this is just his personal hobby and nothing for the rest of us to get excited about.
|
That's exactly what I repeat from years. But it seems that mules don't want to understand... |
|
Status: Offline |
|
|
wawa
|  |
Re: News about Vampire and Apollo Posted on 9-Dec-2015 7:04:37
| | [ #119 ] |
|
|
 |
Elite Member  |
Joined: 21-Jan-2008 Posts: 6259
From: Unknown | | |
|
| @Zylesea
Quote:
PPC is not retro. But it is a dead end. The raw performance of some ppc chips is not that bad. x64 is definitely faster, but ppc has some juice, too. The odyssey issue is not primarily of lacking raw cpu power, but no JS JIT and being stuck in some endian issues. |
then it will be good news to you that deadwood proposes a bounty to fix odyssey java script for big endian. but this is another matter. head to aros-exec for details. |
|
Status: Offline |
|
|
wawa
|  |
Re: News about Vampire and Apollo Posted on 9-Dec-2015 7:08:08
| | [ #120 ] |
|
|
 |
Elite Member  |
Joined: 21-Jan-2008 Posts: 6259
From: Unknown | | |
|
| @retro
Quote:
apollo is going to be super scalar, which means multiple execution units, so this comes a bit towards your proposal and still remains functional with the amiga system. in theory multicore should be possible, but what system would drive that hardware on 68k? aros? edit: one of typos..Last edited by wawa on 09-Dec-2015 at 05:11 PM.
|
|
Status: Offline |
|
|