| 68K=Rules / PPC =Shit | page 1 2 3 4 5 6 7
|
|---|
|
|---|
| | 
Samuel Devulder
Posts 248
19 Sep 2017 07:44
| John William wrote:
|
You are wrong. There is never need for FPU to watch movies.
|
Well I said *might*, but strictly speaking you only need one *eye* to *watch* a movie (and sometimes an eyeglass in front of it :) )
| |
| | 
Gunnar von Boehn
(Apollo Team Member)
Posts 6219
19 Sep 2017 08:44
| m rickan wrote:
|
How do you think the 080 in ASIC form would stack up against the PPC if clocked at the same rate?
|
Yes this is the point. Today the PPC have the major advantage of ASIC == high clock rate. You will of course expect that a 800 MHZ ASIC beats an 68K@80Mhz in every possible task - this as we see is not the case.
The 800 MHz PPC is in a number of areas not that much faster than the 80MHz 68K. And in some areas the 68K runs circles around the PPC. I see this only as beginning.
Our goal is to work towards 68080 ASIC in the future.
If all goes well we can have GigaHerz 68080 in some years. I think then it will be "GAME OVER" for PPC AMIGAs.
As the 68080 not only runs the huge AMIGA software base so much better, and also plays the original AGA games - the 68080 will then also outrun all high end PPC AMIGAs in pure horse power. And even if we not get the $ for doing the ASIC run -
FPGAs get cheaper and faster every other year.
So the performance of "Pegasos 1 or SAM" class PPC systems will be reached simple by waiting some years for next gen FPGA.
| |
| | 
Mr Niding
Posts 459
19 Sep 2017 09:00
| Intresting discussion. Its appears to be highlighting the importance of synergy and fundemental design decisions. If done incorrectly, can cripple your ,on paper, better hardware even if each individual area is supposedly populated with superior pieces.
And that doesnt matter IF the synergy doesnt exists. I guess the headline is a bit inaccurate..? More like "PPC=Shit as applied"....?
| |
| | 
ExiE CZEX
Posts 48
19 Sep 2017 09:25
| Steve Ferrell wrote:
|
I think English is getting in the way. In English, "invaluable" means "very valuable".
|
Yes my english probably suxx. Despite this I am well aware what invaluable means and I still don't think this test presented with incomplete information is invaluable. I would call it interesting.
Steve Ferrell wrote:
|
Yes, but it seems as those who are complaining most loudly have financial interests in A-EON, and/or Hyperion. It's interesting that they want to point out all the "flaws" in Gunnar's benchmarks yet it's nearly impossible to find benchmarking data on A-Eon's PPC offerings. At least Gunnar is posting claims. What little data I can find on the PPC chips used in the latest OS4 systems puts their performance on par with PC's from the early 2000's.
|
I am just not starry-eyed after all these years. And before we get deeper into A-EON/Hyperion conspiracy theories, only PPC I got is Cyberstorm PPC and I am waiting for V4 A1200. Steve Ferrell wrote:
|
And I disagree, his benchmarks say quite a lot about the power of the Vampire, if you choose not to see it then you're just being obstinate.
And no, I don't think people make their decisions on just one test. For me, this test and other benchmarks posted here are very helpful and it has helped me make a final decision against buying any more PPC hardware. I can't justify the cost of a PPC system when it's outclassed by an FPGA.
|
I always preferred real-case tests over synthetic benchmarks. That's it.
| |
| |
Samuel Devulder
Posts 248
19 Sep 2017 09:37
| In any case, in the future, the PPC-based amigas market will likely be dead. PPC-architectures for the Amiga are deadend solutions since they are based on an obsolete CPU. There is not need to worry about them I think. Their fate is sealed.
ExiE CZEX wrote:
| | I always preferred real-case tests over synthetic benchmarks. That's it. |
Good point. I'd like to see real use-cases and demos pushing the FPGA to the limits!
| |
| |
Gunnar von Boehn
(Apollo Team Member)
Posts 6219
19 Sep 2017 10:19
| Samuel Devulder wrote:
|
In any case, in the future, the PPC-based amigas market will likely be dead. PPC-architectures for the Amiga are deadend solutions since they are based on an obsolete CPU. There is not need to worry about them I think. Their fate is sealed.
|
True, PPC is dead.
POWER is still alive - but POWER is a complete different kind of animal - playing in a different price and performance league.
Looking back historically..
RISC CPUs did had an advantage for a few years.
RISC are easier to pipeline, so RISC chips were fully pipeline before CISC chips did the same. This gave RISC for some year an advantage in the 80th.
In the early 90th CISC chip manufacturers found it very difficult to decode more than 1 instruction per cycle.
At these years RISC had an advantage - as RISC instructions have the same length. Therefore decoding several of them is much easier.
After INTEL, AMD and others found good solutions to this problem - this advantage of RISC is gone.With todays technology both CISC and RISC can decode the same amount of instructions per cycle. As CISC can do more work per instruction - RISC has since many years a clear technical disadvantage.
So yes, there was a time period in the late 80th and early /mid 90th where using RISC for max performance was the correct decision. So looking back the movement to PPC for APPLE and AMIGA was technically a reasonable move at that time.
Since many many years the technical basis is different.
Now RISC chips are outperformed by CISC ISAs.
IBM POWER lives in a market where $ does not matter.
Their customer are Banks and other major companies.
IBM POWER can counter the technical disadvantage of RISC - by making very expensive chips - running at ultra clockrates and having extremely huge caches.
In fact, IBM POWER chips today use 100 MB CPU cache! to keep their market position.
| |
| | 
Martin Soerensen
Posts 232
19 Sep 2017 10:19
| Samuel Devulder wrote:
| | In any case, in the future, the PPC-based amigas market will likely be dead. PPC-architectures for the Amiga are deadend solutions since they are based on an obsolete CPU. There is not need to worry about them I think. Their fate is sealed. |
I doubt that the PPC market will take a major hit. Yes it is dead but so are most other retro-platforms and people still use those. Although it is not for me, I'm sure that some people like to use AmigaOS4.1 or MorphOS on their PPC machines. If the 68080 eventually beats it in raw performance, it is quite irrelevant here since it won't run PPC software.
| |
| |
Samuel Devulder
Posts 248
19 Sep 2017 10:28
| I tend to think that when 680x0 will reach (in a very near future) and overtake PPC performances by large margins, PPC-only softwares will be easly ported-back to 68k by their authors since PPC softwares are written in C/C++ and seldomly use ASM (if any), making porting them to other architectures a peace of cake.
| |
| |
Gunnar von Boehn
(Apollo Team Member)
Posts 6219
19 Sep 2017 10:30
| Martin Soerensen wrote:
|
If the 68080 eventually beats it in raw performance, it is quite irrelevant here since it won't run PPC software.
|
It is not the same situation.In the original AMIGA times people wrote 68K software.
Many developers write it in 68k ASM.
Of those applications written in compilable high languages most sources are lost now.
So all this software will forever stay 68K.
All the software running today on e.g AROS.
Is written in "C" - so you can compile all of it for 68K.
Mostly the same is true for OS4 and MOS application.
The big majority of software is "C" written, or even taken from LINUX or GNU or other opensource pools. Compiling it for 68K is not problem at all.
| |
| | 
Wawa T
Posts 695
19 Sep 2017 11:12
| Martin Soerensen wrote:
| Yes it is dead but so are most other retro-platforms and people still use those. Although it is not for me, I'm sure that some people like to use AmigaOS4.1 or MorphOS on their PPC machines.
|
considering "retro", the target audience constitutes itself of people who once had some platform or heard of it. no comparison between amiga and any of the followup projects here. be it aros, os4 or morphos. about vampire, we will see how it establishes itself in this context.
| |
| | 
Nixus Minimax
Posts 416
19 Sep 2017 13:16
| Gunnar von Boehn wrote:
|
In the early 90th CISC chip manufacturers found it very difficult to decode more than 1 instruction per cycle.
At these years RISC had an advantage - as RISC instructions have the same length. Therefore decoding several of them is much easier.
After INTEL, AMD and others found good solutions to this problem - this advantage of RISC is gone. |
I wonder if this is entirely true and what these good solutions are. I understand that decoding x86 instructions in parallel is even more difficult than anything else because x86 instructions can be multiples of single bytes while cleaner architectures such as 68k are restricted to multiples of two bytes. While the Intel's i7 clearly is a beast, AMD seems to have been struggling with decoding many instructions in parallel for a considerable amount of time. It still seems a good approach when designing a new ISA to keep the number of possible instruction lengths limited. E.g. there are some new RISCy architectures that use instruction words and double instructions words and that's it. So the problem is not solved completely. However, without complex address modes and in-memory operands there is always the difficult task of analysing bundles of RISC-instructions that can be executed as one while a typical CISC-ISA gives you the equivalent of one such bundle in the form of a CISC-instruction.
Btw, since you mention the POWER chip, I wonder about another IBM monster processor whose name I forgot. But it's a CISC architecture running at comparable clock frequencies as the latest POWERs and seemed to be quite a beast. That should give pretty good performance levels...
| |
| | 
Thierry Atheist
Posts 644
19 Sep 2017 18:03
| Anthony Jacques wrote:
| | That is actually the reason why he should be being more careful. Claims like these on a commercial website can count as advertising and therefore legally should not be misleading. There are rules about comparative claims such as these, that the advertiser is usually much much more careful with the wording of claims that is is "the best" or when comparing to other products. I'm sure that in reality nobody is going to complain to the ASA or other equivalent bodies. Also here it is more just the lack of caveats and detail that are the issues, but that doesn't mean it is OK to repeatedly post such bold claims with misleading details which are only highlighted when queried. As has been said before, post honest, transparent benchmarks without the unnecessary digs, from what I've seen, it's good enough that it's unnecessary and just gives a bad impression to those who see the flaws in the tests. |
Flaws? HARDLY! Gunnar is VERY conservative when making claims of any kind, as I've observed posts made by him for over 4 years! The PPC Amigas are GARBAGE compared to these results (is MY opinion), because when extrapolated to better hardware specifications it is in fact, BLATANTLY OBVIOUS!!! I mean, I see that data is pulled 16 bits x 2 RAM chips =32 bits at a time into the FPGA. Imagine if it was set up to draw 64 bits at a time!!!! And, Gunnar, didn't even go to "imagine the AmigaOne XE with NO graphics card (ATI9200)".... WHAT WOULD IT BE ABLE TO DO THEN??? This is a ONE CHIP CPU/FPU/AMMX/graphics AND audio mechanism!!!!! Vampire's video output (no need for RTG) and audio output (no need for AHI) CRUSHES... I repeat CRUSHES almost ALL PPC Amigas*!!!!!!!! AmigaOne XE audio was a non-starter!!!!!! * When using UAE in 1920x1080. NEED I CONTINUE TO CODE BLOAT?!?!!!!! This is the WORLD'S BEST COMPUTER! <--- Unsubstantiated claim? I THINK NOT!!
| |
| |
Mr Niding
Posts 459
19 Sep 2017 18:23
| @Thierry Atheist The NG machines are able to use 3rd parties graphicscards, so making "what if they didnt use it!!" as a point is silly. That aside, Vampire/Apollo Core is impressive, but I think you need a few more exclamations and capital letters to get that across.
| |
| | 
John William
Posts 563
19 Sep 2017 19:06
| Mr Niding wrote:
|
@Thierry Atheist
The NG machines are able to use 3rd parties graphicscards, so making "what if they didnt use it!!" as a point is silly.
That aside, Vampire/Apollo Core is impressive, but I think you need a few more exclamations and capital letters to get that across.
|
You forgot question marks. Thierry have a point however. Vampire is indeed better than any PPC Amiga I have seeing so far.
The future for Amiga is the 68k/Vampire and not PPC Amiga. The future OS is AROS also. This will open a huge game like we have never seeing before. I want 68k ASIC Amiga mind you! That would kick even my 8 quad core AMD processor in the butt hard!
| |
| |
Samuel Devulder
Posts 248
19 Sep 2017 21:21
| Notice however that AROS is damn slow on 68k. It doesn't equates with OS3.1 speed. This may due to an os written in portable C versus an highly ASM-optimised os. Time will tell if the bottlenecks can be eliminated.
| |
| |
Gunnar von Boehn
(Apollo Team Member)
Posts 6219
19 Sep 2017 21:27
| Samuel Devulder wrote:
|
Notice however that AROS is damn slow on 68k.
|
Speed looks OK to me.
With the new SAGA driver, AROS GUI is very fast now.
| |
| | 
Nicolas Sipieter
(Needs Verification)
Posts 115/ 1
19 Sep 2017 21:53
| i've seen aros getting a lot 'faster than before' videos lately.
but i wouldn't call that being 'very fast' just quite yet.unless things improved quite a bit again these last 3 days ?
if so i'll be interested to watch a new video that demonstrate it. i'm not minimizing the progress and achievements made.
aros certainly look more usable now. it's a more than welcome progress, for sure. but 'very fast' seems a bit of an overstatement, no ?
| |
| |
Samuel Devulder
Posts 248
19 Sep 2017 23:49
| Indeed AROS is unable to work correctly on unaccelerated amiga, contrary to the original OS. This is possibly the drawback of being developped on much faster platforms, henceforth loosing the target of basic amigas. That might not be such a problem since by today's standards basic amiga is also a dead platform. Future is accelerated amiga (e.g. vampirized)
| |
| |
John William
Posts 563
20 Sep 2017 01:19
| Samuel Devulder wrote:
|
Indeed AROS is unable to work correctly on unaccelerated amiga, contrary to the original OS. This is possibly the drawback of being developped on much faster platforms, henceforth loosing the target of basic amigas. That might not be such a problem since by today's standards basic amiga is also a dead platform. Future is accelerated amiga (e.g. vampirized)
|
I would love to cut the rope on the standard classic Amiga also. Already whdload brought all the classical Amiga games we need...so why do I need to linger into the classical hardware configuration or AmigaOS again if AROS can run all whdload games for example and provide modern up to date applications?
I see no lose here, only gain. Also it have RAM disk and the icons behave identical to classical Amiga icon system and have a better error trapping mechanism verses putting the entire Amiga down...and it is open source! Meaning improvement is possible!
| |
| | 
Kresimir Lukin
Posts 65
20 Sep 2017 04:07
| Correct, there is no need for AROS to work on unaccelerated Amigas at all. As long it is compatible with old software on Vampire.
New machine, need OS what develops constantly.
| |
|
Performance and Benchmark Results!