3 more NatAmi's - Yippee!

[Piru]

Gunner has been playing/tuning with the 060 card and says it can keep up with 600mhz PPC Amiga in mem copy.
 
MEMORY READ 145 MB/sec (Optimised Code)
MEMORY READ 114 MB/sec (Average Code)
MEMORY WRITE 155 MB/sec
MEMORY COPY 140 MB/sec
http://www.natami.net/knowledge.php?b=1¬e=39600&order=&x=3
 
He also says he isnt finished tuning it yet which is cool, I wonder how the N050 FPGA and the DDR2 memory will stack up to these results ;-)

Memory interface of this "600MHz powerpc amiga" is the partcularly weak spot in it. Beating it is hardly worth cheering about.

Comparing the results against 7447 with 166MHz bus and proper chipset would have given a slightly different picture:

write: 774 MB/s
read: 375MB/s

(Actual figures from Mac mini G4 1.5GHz, unoptimized C code)

And if we go to anything computational the G4 crushes classic, and even natami, regardless of if it runs real 060 or some FPGA 68k emulation. Some figures here: http://via.i-networx.de/bench_en.html

[Piru]

Yes Piru MOS in a Mac is impressive, I am sure Gunner was refering to Pegasos with OS 4.x if you read the whole natami thread.

And specifically G3 which had abysmal performance. Peg2 with G4 is a lot faster. I wonder why he didn't compare it against that?

But dont you think those figures are impressive concidering how old the 060 processor is ?

They're not. They're to be expected with proper bus management. Atari accelerators have had this for ages for instance.

[Piru]

I think

You can think what you want.

Piru feels threatened by the NatAmi

No I don't.

simple reason being that when it becomes available we will truly have a "new" Amiga that can run at speeds unheard of before (even with a PPC board), ease of use for adding things like HD's DVD writers etc... but most of all the threat comes from the fact that it's 68K (so to speak) and will run 3.x at speed we could only ever dream about before...

No, it won't beat even moderate PPC systems, and it isn't even trying to do so. Not that I would worry even if it did. NatAmi is targeting a different audience (it may well overlap with MorphOS user base of course). Running 68k ultra fast isn't anything spectacularly interesting. It has been done before with UAE and the like.

As for superaga, I don't feel it will make much impact. Amiga devs are few and far between already, and going even more niche with expensive hardware won't help a bit.

Rest of your nonsense isn't worth commenting.

[Piru]

I did noticed that Piru left out the memory copy comparison.

Of course that was a deliberate, I was just hoping no one would notice!

Or not. The only reason is that the benchmark doesn't provide those numbers (actually it's no benchmark at all, those numbers are from a tool that checks memory for errors.. thus it doesn't do memory copy). But if anyone is interested memory copy is multiple times faster than natami, too.

[Piru]

And wy you do not post values of the G3 600 ?
Gunnar write that natami is faster in memory transfer as a PPC with 600 MHZ.So you should test it with a PPC run at 600 MHZ, if you want something usefull post.

You missed my point. It's common knowledge that Pegasos2 G3 sucks in memory performance department. Pegasos2 G3 is particularly slow in memory operations, slower than some other G3 systems with some other chip sets. So it does not give you G3 600 memory performance benchmark to compare against.

[Piru]

You're missing the point about the statements that have been made in this thread. The people who have commented aren't against Natami per se, but rather have disagreed with distorted statements about the Natami. It's the statements that are the issue, not the hardware.

Just like them, I like the Natami, and am interested in seeing it released, but get pissed off when people make stupid statements about the power it'll have, as it makes the project seem more ridiculous than it actually is. Tone down the hype and all will be well.

++

[Piru]

I have big hopes it'll get used in demo competitions.

I find that unlikely. It couldn't compete with classic amigas since it'd be unfair. So the only category would be some "wild demo" or so. Also, more than often demo coders want to release their demos for platforms that are widely available.

At least one high profile & active amiga scener has already voiced his disinterest.