19-Apr-2024 17:36 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 23 items in your selection
[Forum] MorphOS 68k speedANN.lu
Posted on 03-Nov-2002 12:34 GMT by Anonymous (Edited on 2002-11-03 18:27:14 GMT by Teemu I. Yliselä)23 comments
View flat
View list
At http://www.amiga-news.de/forum/thread.php3?id=4353&start=21&BoardID=4 (german) two Betatesters make statements regarding MorphOS' 68k-emu speed. No exact numbers, but for example some rendering with Cinema4D on a Pegasos-600MHz (and debug-output on) seems to be at least as fast or even faster as with WinUAE-JIT on a 1-GHz-PC.
MorphOS 68k speed : Comment 1 of 23ANN.lu
Posted by Alkis Tsapanidis on 03-Nov-2002 11:43 GMT
And this is without the JiT afaik...
MorphOS 68k speed : Comment 2 of 23ANN.lu
Posted by Alkis Tsapanidis on 03-Nov-2002 11:45 GMT
In reply to Comment 1 (Alkis Tsapanidis):
I was wrong.
MorphOS 68k speed : Comment 3 of 23ANN.lu
Posted by Rob on 03-Nov-2002 12:35 GMT
About the same as OS4 Petunia then?
MorphOS 68k speed : Comment 4 of 23ANN.lu
Posted by Alkis Tsapanidis on 03-Nov-2002 12:43 GMT
In reply to Comment 3 (Rob):
I wouldn't say so...
MorphOS 68k speed : Comment 5 of 23ANN.lu
Posted by Anonymous on 03-Nov-2002 12:44 GMT
Careful, musn't look too desperate for publicity.
MorphOS 68k speed : Comment 6 of 23ANN.lu
Posted by Anonymous on 03-Nov-2002 12:44 GMT
In reply to Comment 4 (Alkis Tsapanidis):
Because you don't know.
MorphOS 68k speed : Comment 7 of 23ANN.lu
Posted by Alkis Tsapanidis on 03-Nov-2002 12:52 GMT
In reply to Comment 6 (Anonymous):
The speed, approximately has been posted, so yes, I do know.
But in the end we'll see after both products have been released
and tested.
MorphOS 68k speed : Comment 8 of 23ANN.lu
Posted by Anonymous on 03-Nov-2002 13:12 GMT
In reply to Comment 7 (Alkis Tsapanidis):
So comparing an approximate speed ( posted ) on one processor with a subjective feeling with debug on with a completely different processor you are
able to indicate which one will be faster?
Wow, can you give me some tips for the Epsom races?
MorphOS 68k speed : Comment 9 of 23ANN.lu
Posted by Alkis Tsapanidis on 03-Nov-2002 14:25 GMT
In reply to Comment 8 (Anonymous):
On a different processor? What are you talking about? I'm talking
about the estimated Petunia speed on the PPC750CXe vs the approximate
Mos JiT speed on the same CPU, and the 604e@233 Petunia benchmarks with
the MOS JiT speed on my 603@240. It shows that Petunia still has room
for improvement. On an AmigaONE it won't matter, the speed would be
enough most of the time, but on a 603, the difference is huge.
MorphOS 68k speed : Comment 10 of 23ANN.lu
Posted by Kjetil on 03-Nov-2002 14:42 GMT
In reply to Comment 8 (Anonymous):
Don’t se where you are going with you’re comment,
Rendering requires a grate deal of CPU speed and sense this is the same program and the same OS, it is basically like compare JIT against JIT. Take in count that the i686 have to do an byte switch for every DWORD or LONG WORD operation, And that an hole Windows system even share some of CPU restores with UAE, this result only shows the that i686 is more power full then PPC to day.
The result do not show that PPC is faster then i686 1Ghz is only show the benefit of having the same endien on 68000 and PPC Cpu’s, this is way I prefer an PPC over an I686 when is comes to running AmigaOS and I’m not to worried about the MHZ, Motorola will catch up any way. If they don’t well way not AROS.
MorphOS 68k speed : Comment 11 of 23ANN.lu
Posted by Chris on 03-Nov-2002 15:59 GMT
In reply to Comment 9 (Alkis Tsapanidis):
I thought the speed tests that are listed on the Petunia web site are
old and do not reflect FPU emulation?
MorphOS 68k speed : Comment 12 of 23ANN.lu
Posted by Ben Hermans/Hyperion on 03-Nov-2002 16:06 GMT
In reply to Comment 9 (Alkis Tsapanidis):
Alkis, please stop spreading nonsense.
The only available Petunia benchmarks are taken from the current crippled Phase 5 hardware running on WarpOS (not Exec SG - context-switches!) and with the 68K CPU happily stealing cycles from the PPC.
It's complete BS to try and extrapolate those benchmarks to an architecture with differs substantially like the AmigaOne with 600 Mhz (different CPU with on die L2 cache, much faster memory access, no dual CPU architecture etc.)
It's simply impossible to do come up with anything more than an educated guess.
MorphOS 68k speed : Comment 13 of 23ANN.lu
Posted by Don Cox on 03-Nov-2002 16:13 GMT
In reply to Comment 12 (Ben Hermans/Hyperion):
"It's simply impossible to do come up with anything more than an
educated guess. "
And the best guess is that it would be in the same ballpark as
Amithlon on a comparable CPU. Probably slightly faster, but that would
depend on buffer settings, and you could argue for ever as to what is
a comparable CPU.
MorphOS 68k speed : Comment 14 of 23ANN.lu
Posted by Christophe Decanini on 03-Nov-2002 16:50 GMT
In reply to Comment 13 (Don Cox):
And anyway depending on what is running in the JIT the result may be very different.
I would test it with several applications (not benchmarks) to know how it is performing. It would be very easy to choose some apps that would be more accelerated on a JIT compared to another.
We are not yet done with biased "reports"
MorphOS 68k speed : Comment 15 of 23ANN.lu
Posted by Kjetil on 03-Nov-2002 17:18 GMT
In reply to Comment 14 (Christophe Decanini):
Well you’re argument is nonsense as long as we compare with the same 68k program, the test results should be about the same, after all what an JIT emulator is doing is translating 68k instructions to PPC instructions or to I686 instructions, If how ever the JIT emulator translate 68k instructions to PPC instructions that are slow compared to otter instructions it may use in that case it may differ some. Depending on the number of cycles the instruction take to execute, that there is the CPU load that translation takes to do, it all so may take way some cycles from the executing program.
MorphOS 68k speed : Comment 16 of 23ANN.lu
Posted by Alkis Tsapanidis on 03-Nov-2002 18:44 GMT
In reply to Comment 12 (Ben Hermans/Hyperion):
Well, I did say that they we'll have to wait for both products to be released
to do tests and compare them, I know that without that everything is a
speculation. I just talked about the estimation.
BTW, will OS4 be available for PUP and then for the AmigaONE, or at the
same time?
MorphOS 68k speed : Comment 17 of 23ANN.lu
Posted by Mike Veroukis on 03-Nov-2002 21:01 GMT
So who cares? The P4 will be clocking in at 3Ghz by the end of the year. The x86 platform will always be quicker for emulating 68K, plus it's way cheaper.
- Mike
MorphOS 68k speed : Comment 18 of 23ANN.lu
Posted by Christophe Decanini on 03-Nov-2002 21:10 GMT
In reply to Comment 15 (Kjetil):
I said that because someone will point to sysspeed or the gfx tunnel demo that was used in MorphOS.
There are some big chances that the different JIT act different with such particular tests and that results does not show the average CPU / JIT combo performance.
It is not because application X performs better in JIT A than in JIT B that application Y will perform better in JIT A.
MorphOS 68k speed : Comment 19 of 23ANN.lu
Posted by Ole-Egil on 04-Nov-2002 07:08 GMT
In reply to Comment 17 (Mike Veroukis):
Well, a P4 2.8GHz costs the same as an A1G3-SE here at the moment. Add another 200 euro for motherboard, I'm not convinced you can say it's _both_ faster and cheaper at the same time. You either get 2.5 times the speed, OR you get 40% of the price. NOT both ;-)
MorphOS 68k speed : Comment 20 of 23ANN.lu
Posted by Mike Veroukis on 04-Nov-2002 16:09 GMT
In reply to Comment 19 (Ole-Egil):
>Well, a P4 2.8GHz costs the same as an A1G3-SE here at the moment. Add another
>200 euro for motherboard, I'm not convinced you can say it's _both_ faster and
>cheaper at the same time. You either get 2.5 times the speed, OR you get 40%
>of the price. NOT both ;-)
Or you go with AthlonXP, which is what I did. :) The sad thing is the A1 lacks other things like DDR ram, a fast southbridge or even ATA133, so you can't just compare the CPU, you need to compare the entire system. So even a cheaper P4/AthlonXP would still benefit from a better motherboard.
- Mike
MorphOS 68k speed : Comment 21 of 23ANN.lu
Posted by Anonymous on 04-Nov-2002 16:21 GMT
In reply to Comment 20 (Mike Veroukis):
>the A1 lacks other things like DDR ram
DDR ram is around twice the price of SDR ram, though.
>or even ATA133
This is no real standard but a Matrox stopgap thing while the next step after
ATA100 should have been SerialATA. Even brandnew x86 board may not have ATA133.
MorphOS 68k speed : Comment 22 of 23ANN.lu
Posted by Anonymous on 04-Nov-2002 16:22 GMT
In reply to Comment 21 (Anonymous):
Not Matrox but Maxtor, of course.
MorphOS 68k speed : Comment 23 of 23ANN.lu
Posted by strobe on 04-Nov-2002 22:17 GMT
UGH!
Why does everything reduce to pie-in-the-sky arguments?
Let's get back to what was demonstrated at the show for all to see. What I got from it was the MorphOS product is potentially faster than the Amithlon product at running 3.x software due to a) better 68k emulation and b) PowerPC-native 3.x compatible libraries.
Anonymous, there are 23 items in your selection
Back to Top