@COBRA

So... got any memspeed test results at hand?

EDIT: ok, found some results:

Quote:

Pegasos 2 G4/7447-1000MHz RAM: 768MB Using OS4/os4depot memspeed binary INT READ: 224MB/sec INT WRITE: 237MB/sec INTMEM: 207MB/sec FP READ: 228MB/sec FP WRITE: 322MBs/sec FPMEM: 213MB/sec Peg2, 7447 1GHz, native MOS/PPC compile: INT READ: 226MB/sec INT WRITE: 466MB/sec INTMEM: 216MB/sec FP READ: 227MB/sec FP WRITE: 466MB/sec FPMEM: 213MB/sec

While write speed is much better it is not enough to explain difference.

Last edited by itix on 11-Aug-2009 at 08:55 PM.
Last edited by itix on 11-Aug-2009 at 08:43 PM.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@Krashan: if I were Madonna I would sue you right now :p

_________________
http://www.warpdesign.fr/

@COBRA

Quote:

There's a very simple explanation, actually: memory setup.

Thanks! Thyat was the kind of answer I was looking for

So, anyone up for writing SetPII? We have SetA1 which changes the memory setup on the A1 to speed up memory operation by atleast 10-20% on the A1. So how about a similar tool to setup and optimize the PegII's memory controller similarly?

_________________
- Don't get fooled by my avatar, I'm not like that (anymore, mostly... maybe only sometimes)
> Amiga Classic and OS4 developer for OnyxSoft.

OK, Krashan, what about swapping your statement for "MorphOS is better writen"?

Yes, this is even better than my previous one, as is based on facts more. I hope ssolie would not call it a "cheap personal attack at AmigaOS 4"

_________________
Reggae · MorphOS Files · DigiBooster 3

@COBRA
Quote:

You can run tests on OS4/MOS on classic PPC Amiga hardware.

Classic Amigas are limited to AmigaOS4.0 - does the last incarnation of MorphOS work on classic setups?

_________________
X1000|II/G4|440ep|2000/060|2000/040|1000

@Leo

Quote:

What if the explanation was even more simple: MorphOS is better written than OS4. Period.

Except that it isn't. It has better 3D drivers and probably better USB drivers, and it has Pegasos2-specific optimizations which OS4 does not have. OS4 on the other hand has faster file I/O (according to my tests here), better file systems (PFS2 and JXFS with support for >4GB files), better memory protection, support for more hardware. I have both of them running here on my Peg2 and both are nice OS'es with advantages and disadvantages on both sides. And perhaps most importantly, both are light years behind current mainstream OS'es when it comes to features and applications, and have a long way to catch up. The weaknesses of OS4 compared to MoprhOS (and vice versa) are minimal compared to the weaknesses of both OSes when compared to a modern mainstream OS.

Quote:

OK, Krashan, what about swapping your statement for "MorphOS is better writen"?

That statement could be better written. (check the spelling)

P.S I've probably shot myself in the foot with that poor joke due to my own crap spelling.

_________________
If your nose runs and your feet smell, you're upside down.

Thanks! Thyat was the kind of answer I was looking for

This answer has one flaw however - it is false one.

_________________
Reggae · MorphOS Files · DigiBooster 3

Quote:

The weaknesses of OS4 compared to MoprhOS (and vice versa) are minimal compared to the weaknesses of both OSes when compared to a modern mainstream OS.

While I 100% agree on that, MorphOS is lights ahead of OS4. In every area except maybe disk I/O indeed. And I don't think this has to do with any optimisation... It's just as it is.

_________________
http://www.warpdesign.fr/

@Daff Quote:

nd for 170 MB/10 MB memory access, you can re-do the test (with ARexx) mesure it with the command "avail" before and after the test.

The results of used/free memory from the Avail command (or anywhere else) mean almost NOTHING on OS4.1. Read-up on the Slab allocator for a start, and how stuff is not truely deallocated until the system really needs it.

It's certainly annoying that accurate memory reporting is no-longer possible, because now it's v.hard to tell if there is a memory leak.

@Cobra
I don't think you're going to get balanced & reasoned arguments from Leo about MOS vs OS4 . Of course, there could always be a first time...

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@itix

Left: MorphOS 2.2, Right: AmigaOS 4.1

--- Direct FastMem Read 32 bit ---------
8k: 1310.830 MB/s 1311.203 MB/s
128k: 1061.868 MB/s 1049.550 MB/s
8MB: 216.553 MB/s 136.949 MB/s

--- Direct FastMem Read 64 bit ---------
8k: 2075.239 MB/s 2079.268 MB/s
128k: 1591.044 MB/s 1602.452 MB/s
8MB: 216.876 MB/s 141.450 MB/s

--- Direct FastMem Write 32 bit --------
8k: 3252.464 MB/s 3261.453 MB/s
128k: 1601.137 MB/s 1603.566 MB/s
8MB: 426.954 MB/s 172.744 MB/s

--- Direct FastMem Write 64 bit --------
8k: 2533.457 MB/s 2540.363 MB/s
128k: 2027.582 MB/s 2028.873 MB/s
8MB: 427.283 MB/s 367.951 MB/s

--- FastMem to FastMem Copy 32 bit -----
8k: 1268.065 MB/s 1271.288 MB/s
128k: 1048.414 MB/s 1548.010 MB/s
8MB: 105.929 MB/s 62.860 MB/s

--- FastMem to FastMem Copy 64 bit -----
8k: 2532.501 MB/s 2539.638 MB/s
128k: 1688.223 MB/s 1548.010 MB/s
8MB: 108.932 MB/s 65.559 MB/s


As you can see, 8k and 128k blocks are almost the same (L1 and L2 cache), only very large blocks where cache is useless, is there a difference (which is why I think it is the setup of the memory controller that is different). The most surprising result (to me) is 32bit fastmem write, which is just as fast under MorphOS as 64-bit write, and 2.5x faster than under OS4. I used a binary compiled under OS4 and I used OS4emu to run it under MorphOS.

Last edited by COBRA on 11-Aug-2009 at 09:08 PM.

@Krashan

Quote:

Thanks! Thyat was the kind of answer I was looking for This answer has one flaw however - it is false one.

So in your opinion the memory performance results me and itix posted are false/faked/incorrect?

@COBRA

MorphOS memspeed is probably the same since version 1.0.

But really. Pegasos 2 is the fastest hardware OS4 can run on currently. It is not our fault if it is equalized to A1/SAM level performance

Last edited by itix on 11-Aug-2009 at 09:20 PM.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

First of all. Somebody should make a native English translation because that automatic (I assume) translation is partially hard to understand. Like for example quoting: "Copy of a file of 255 Mo, hard drive with the RAM", do they mean they're using a hard drive with internal cache RAM or did they copy from HD to RAM, from RAM to HD, from RAM to RAM or from HD to HD ? There's "post the files (mode icon)". Does that mean "display the files (mode icon)" or something else ? (I guess it means that.)

Also this subject is already discussed in a thread on Amigaworld.net. Still they made all the same mistakes, false comparisons and false conclusions in this comparison. (I come back to this when I have enough time.)

Also reading that article very quickly I didn't see any explanation how they shared those points they gave to the both platforms. For example one test where half of the tests were faster on one platform and the other half was faster on another platform. Still they gave 8/10 to one platform and 10/10 to the other. I would think if one was faster in half of the tests and the other on other half they should both get equal points.

Quote:

So why is everything so slow on OS4 then..?

Comparing apples to oranges and misinterpreting the results.

@ChrisH
Quote:

The results of used/free memory from the Avail command (or anywhere else) mean almost NOTHING on OS4.1. Read-up on the Slab allocator for a start, and how stuff is not truely deallocated until the system really needs it.

And in that particular comparison they didn't take it into account that AmigaOS4 caches many things. For example large 32bit icons takes some room in memory. (I come back to this later when I have time.)

Last edited by Tomppeli on 11-Aug-2009 at 09:25 PM.
Last edited by Tomppeli on 11-Aug-2009 at 09:23 PM.

_________________
Rock lobster bit me. My Workbench has always preferences. X1000 + AmigaOS4.1 FE
"Anyone can build a fast CPU. The trick is to build a fast system." -Seymour Cray

@Tomppeli

Quote:

And in that particular comparison they didn't take it into account that AmigaOS4 caches many things. For example large 32bit icons takes some room in memory. (I come back to this later when I have time.)

Do you think Ambient reloads an icon from the disk every time it is drawn? Why dont you say Workbench is slower because it caches more

(In fact Ambient caches the original unscaled image and scaled 1st and 2nd state images. And of course times include scaling overhead.)

Last edited by itix on 11-Aug-2009 at 09:38 PM.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix

Quote:

But really. Pegasos 2 is the fastest hardware OS4 can run on currently. It is not our fault if it is equalized to A1/SAM level performance

Actually, as I stated before, the same tests running under OS4 on AmigaOne hardware perform significantly faster:

Pegasos2 - G4 - 1Ghz:
--- Direct FastMem Write 32 bit --------
8k: 3261.453 MB/s
128k: 1603.566 MB/s
8MB: 172.744 MB/s

AmigaOneXE - G4 - 933Mhz:
--- Direct FastMem Write 32 bit --------
8k: 3033.710 MB/s
128k: 1491.687 MB/s
8MB: 275.227 MB/s

So it seems that using the same OS, despite the lower clock speed, the AmigaOne XE outperforms the Pegasos2 when it comes to memory operation

So much for the benchmarks... or is the AmigaOne really that much faster than the Pegasos2?

Last edited by COBRA on 11-Aug-2009 at 09:44 PM.

Guys, don't worry. Seriously.

While MorphOS 2.3 might currently be faster than AmigaOS 4, that doesn't mean it will always be.

AmigaOS 5 will probably be faster than both together!1!

Tomppeli : perhaps Google translation is a bit better than Babelfish. You can also learn 50-100 basic words in french so you will be able to understand my articles

For your first question, it's a copy from HD to RAM.

For the note 8/10 and 10/10 about test with RAM, it's because results are "balanced" (don't know how to say it in englsih, sorry). MorphOS wins two tests and AmigaOS 4.1 wins also two tests but with a more little difference.

Last edited by Daff on 11-Aug-2009 at 09:52 PM.

Quote:

So it seems that using the same OS, despite the lower clock speed, the AmigaOne XE outperforms the Pegasos2 when it comes to memory operation So much for the benchmarks...

OS4 on AmigaOne being slower than MorphOS on Pegasos 2, is it going to help much?

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

In an ideal world it would be nice to see a vast improvement to the USB stack as in my experience the native OS4.1 stacks is slower in copy transfers on an 800Mhz A1XE than 68K Poseidon 4.2 is, running under 4.0, on a 233Mhz 603 equipped A1200 with a subway.

/Dreaming
A port of poseidon to OS4.1 target hardware would be fantastic.

Last edited by JurassicC on 11-Aug-2009 at 10:01 PM.

_________________
A1200T 603e 330Mhz - Mediator TX OS4.1 F.E.
CDTV 8MB Fast, OS3.1, SCSI, MicroSD SCSI & CD32 FMV
X5000, X1000, A1XE with OS4.1F.E.