@pavlor

No matter :) close enough.

Any way there is HUGE performance leap between 68040 and 68030. And smaller between 68040 and 68060

Something like 4x - > 2x

That would make a 800 MHz 68030 equalent to 100mhz 68060

Last edited by utri007 on 03-Dec-2015 at 11:57 PM.

Quote:

Vampire V3 FPGA(68k) + PPC http://phytec.com/products/system-on-modules/phycore/mpc5200b-io/

Apollo Phoenix Card +FPGA module
https://youtu.be/YdxOQr4itns

FPGA Cyclone5 with 128MB module
http://www.apollo-core.com/bringup/card2.jpg

Apollo Phoenix Card for FPGA module
http://www.apollo-core.com/bringup/apolloPhoenix.jpg

_________________
A1200 BlizzardPPC BVision FastATA Eth 3Com, EFIKA5K2

@SKOLMAN_MWS

When I buy my apollo card, does that mean I will get two cards. ?

@pavlor

Quote:

They published enough benchmarks for my estimate. In one of them (resource demanding demo) 68060 80 MHz was even faster. Sure, core may improve in the future.

Gunnar kindly posted an update on the forums (thanks!) with some more metrics. It appears they have far exceeded 100Mhz 68060 performance. :)

More info here: http://www.apollo-core.com/knowledge.php?b=4¬e=20

Snipped from his note:
----------------------------------------------------------------------------

AIBB MATRIX
68060 @ 80 Mhz = 60.44
APOLLO = 162.64
APOLLO equivalent to 68060 @ 216 MHz

AIBB IMATH
68060 @ 80 Mhz = 98.88
APOLLO = 357.14
APOLLO equivalent to 68060 @ 290 MHz


These are just 2 examples, in all AIBB tests and all other benchmarks that we ran so far Apollo scores better than 68060.
--------------------------------------------------------------------------

Last edited by gregthecanuck on 04-Dec-2015 at 07:57 PM.

@gregthecanuck

I would rather see application benchmarks, not synthetic results like two Gunnar mentioned. Show me lame, ffmpeg, Cinema4D, dnetc, Quake or Doom benchmarks.

Quote:

These are just 2 examples, in all AIBB tests and all other benchmarks that we ran so far Apollo scores better than 68060.

68060 50 MHz, to be more precise. I really don´t like marketing hype Gunnar creates for his product. Yes, it is fast, but quoting two marginal benchmarks as
evidence of overall performance is quite lame.

@pavlor

Is he creating that much hype? I dont really see him spamming forums etc.
While I really appriciate the Individual Computers products, it seems like the price vs performance
ratio is stacked heavily in Apollos favour (naturally).

Will be intresting to see DosBox performance for example, or Cinema4d.

@gregthecanuck

Wow, that is just incredible. Thumbs up to the Apollo team for their awesome work! :)

Imagine owning an Amiga 600 this fast.. who would have believed that years ago? It is like a dream come true.

I'm sure some demo coders are interested getting their hands dirty with the new hardware and their favorite Assembler software. ;)

_________________
Old School Game Blog
Gaming on AmigaOS 4
Void - Amiga Demo Group

@pavlor

Why are you being so negative?

This Apollo core is the best thing to happen in years (for classic) its no perfect (yet) but they are working on it.

Majsta will soon start to send out Vampire 600 to customers, so we will start to see many videos.. things are looking good imo.

René

_________________
Debugging is a state of mind

@RWO

Quote:

Why are you being so negative?

I don´t like intentionally flawed benchmark comparisons. Simple.

Quote:

but they are working on it.

Without doubt and my record shows I applaud every progress of their project.

Good news about performance. Need to get one for A1200 when they are availlable

@OlafS25

All in all the most exciting thing to happen for the Classics really...

Congratulations to all involved...

Q!

"i am home"

@pavlor

Quote:

I don´t like intentionally flawed benchmark comparisons.

aibb is widely considered a dependable benchmark. and who would be showing off a choice of worse results if he can do that with the higlights. dont be impatient, at some point you will certainly have your wishes answered. none is forcing this development down your throat
;)

Quote:

Without doubt and my record shows I applaud every progress of their project.

sometimes i wonder if you simply score on generic positive comments en gross not to be accused of negativity.

Last edited by wawa on 06-Dec-2015 at 12:03 AM.

@wawa: he's too much unbalanced towards the PowerPC side.

However he's correct to ask for real-world applications instead of synthetic benchmarks.

The Apollo team did a great work, and according to such preliminary synthetic benchmarks and the explanation of the softcore features, IMO it's realistic thinking that it can reach double the 68060 performance, clock for clock.

Anyway, the core introduces new instructions, a different microarchitecture, and support for 64-bit. It means that if you want to fully exploit it, you need a compiler that targets it.

GCC's 68K backend sucks, and it's also very difficult to change it. CLang/LLVM is a much better candidate. But whatever is the choice, it requires a lot of work, and I don't see anyone which made a step ahead here.

Another problem which I already reported is AROS's lack of proper graphic drivers for ECS/AGA: it has to be addressed necessarily if you want to use this o.s.. That's because Amiga o.s. 3.x isn't a good solution for the application side, since it's not developed anymore, whereas AROS is active and brings new and updated features. Who will do this work?

Last but not least, the introduction of 64-bit poses other big problems. Primarily to Olaf , which has always though of 68K = immutable software ("application") base, whereas such new, big, feature requires NEW software too.

NEW software = NEW support from developers (and tools... written by other developers). And it also means problems with backward compatibility (I know that Gunnar reported that 64-bit are backward compatible, but there's no information about it in order to verify the impact at every level, from applications to the o.s.).

Anyway, I think that this new project already had one result: Olaf has stopped complaining other post-Amiga platforms because they are opting for radical changes (x64, memory protection, etc.).

@cdimauro

Quote:

he's too much unbalanced towards the PowerPC side.

PowerPC? What?! I´m great supporter of "x86" transition!

@wawa

Quote:

sometimes i wonder if you simply score on generic positive comments en gross not to be accused of negativity.

Don´t judge me by your standards, your overabundant presence in OS4 related threads really is not something to follow.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: he's too much unbalanced towards the PowerPC side. PowerPC? What?! I´m great supporter of "x86" transition!

I reformulate: you're an OS4 supporter, and OS4 currently targets only PowerPC.

@cdimauro

Quote:

I reformulate: you're an OS4 supporter, and OS4 currently targets only PowerPC.

Better, but not entirely true... I´m Amiga.Inc supporter and AmigaOS 3/4 and AROS 68k/x86 user.

@pavlor: yes, but OS4 is your favorite system.

@cdimauro

Do not worry

As I understand it it stays compatible to existing processors. They say it will be more compatible than all real processors ever was. So existing software will run. Who want to use the new added features will need to write new software and change compilers to support it. But that is not necessary to write software for it.

I repeat it ... 15

15x bigger codebase
15x more users

@OlafS25: I never had something to say about the existing code base.

But a "crystallized" code base is, de facto, legacy: retrogaming/computing is its doom.

So, for you any new feature of the Apollo core is useless, I suppose. Right?