Quote:

Just remove it and plug in another card then.

Quote:

Runs as you are used to from MorphOS. Not too spectacular from that perspective. But If you consider how fast it went from getting one board to this state where it runs just like MorphOS runs - wow, that is pretty impressive!
Not finished though, but seems pretty far already. Hats off!

Amazing progress in such a short notice! I would like to see some OpenMOHAA and Wayfarer usage next time, to get a better idea of the board. I am too mostly interested for the version with Altivec support.

Quote:

Just now I tested Dhrystone v2.1 on my Mac Mini with 7447B / 1.83 GHz = 1009 DMIPS
According to picture from link above Mirari with T1042 / 1.6 GHz = 2436 DMIPS
Not bad! T1042 is really nice.
P.S. My Powermac Quad is now without MorphOS and need cooling repair, so I cannot compare it.

Addendum:
Quote:

according to this, it looks that my Mac Mini should have much higher result.
Please, can somebody with Mac Mini try Dhrystone v.2?
( sorry for a little offtopic )

Update:

> Complete MorphOS-related entries from the "blog": […]

2nd June: "Into the Tomb…
With MorphOS 3.20 running on the first prototype of the Mirari board , it was time to test a golden oldie…. Tomb Raider V1. So far we are happy with the result already. Thanks to MorphOS beta testers BeWorld and BSzili for the port."
https://www.youtube.com/watch?v=ERk7nOWThig

BeWorld and BSzili being MorphOS team members is news to me, though, and not reflected on the official members list.

Speaking of which, I notice the MorphZone member named Jonathan Schleifer as recent addition to the "Core Development Team". Nice :-) He has published two blog posts mentioning MorphOS:
https://blog.nil.im/?76
https://blog.nil.im/?8 (translated from German)


Edit: reflect correction of "team members" to "beta testers"

[ Edited by Andreas_Wolf 03.06.2025 - 13:19 ]

News to me as well, I think this is a simple misunderstanding.

> News to me as well, I think this is a simple misunderstanding.

Has been corrected to "beta testers".

Something is probably very wrong with my PM.

It's a dual 2.5Ghz, and my Dhrystone 2.1 results are 1028 MIPS (DMIPS )

Quote:
Interesting, I'm beginning to suspect that my Mini and your Powermac are fine.
Our results are relatively consistent. G4/1.83 GHz ( 1009 ) vs G5/2.5 GHz ( 1028 ). Some 32-bit programs runs a little more efficiently on a G4 processor.

I will test powerbook too and let know.

Just where did you get your dhrystone 2.1 benchmarks?
There are two versions. One (from the Aminet, the WhetDhryStone benchmark - apparently compiled without optimisation), and a second one (compiled with -o2 switch). The "optimised" version was used in "our" tests vs Mirari because geennaam used the optimised one when testing Mirari on Linux.

Here it is for you to try. It's the same code from Aminet, just compiled with -o2.

Ahh, thanks koszer!

With your version, I'm getting 3513 on the PM 2.5Ghz

Yes, I get about the same with mine 2.5 (3498).

Quote:

Am getting 2500 on my iMac G5. Pretty much on par then with the Mriari.

@koszer
Thanks for dhrystone. We used Whetstone+Dhrystone package from morphos-storage.

So with your -O2 optimized dhrystone Mac Mini 1.83 GHz has 2457 DMIPS, so it is on pair with Mirari 1.6 GHz.
Nice result!

> Mac Mini 1.83 GHz [...] is on pair with Mirari 1.6 GHz. Nice result!

From suspected and "really nice" 2.4× performance to just "nice" 1.0× performance... ;-)

> From suspected and "really nice" 2.4× performance to just "nice" 1.0× performance...

Someone as well-read as you will know that it is better to avoid repeating the same phrases in the text

Ok than I summarize it: very nice!

> Ok than I summarize it: very nice!

Okay, then 1.0× obviously evokes the same reaction 2.4× would have done. Strange …or nice ;-)
My point being: If 2.4× is considered very nice, then 1.0×, i.e. 42% of that, is usually not considered nice, but rather disappointing. Also, nothing of this comes as a surprise (see comment #9), as it has the same core as the known and thoroughly benchmarked X5000. 1.6 GHz e5500 delivers 80% of 2.0 GHz e5500 performance and 73% of 2.2 GHz e5500 performance, no matter if the chip is called T1042, P5020 or P5040. The smaller node size of T1 vs. P5 just makes it faster per watt, not faster per clock. What'll be a little more interesting to see is the e6500/T2081 performance (as is and per clock) as there's so far been no MorphOS on this core. Scalar performance-per-clock increase compared to e5500 may not be significant unless the benchmark executable will be compiled with the right flags to leverage the e6500 ISA and microarchitecture improvements.

Plus you also have to consider data throughput- DMIPS set a cut off for calculations, but often the data throughput is the effective bottleneck.

Video playback and browsing are faster on my iMac G5 while jpeg decompression is faster on my G4 machines. The Mirari will hopefully/probably deliver good throughput.. I guess in MorphOS land there is virtually only demand for the T2081 version as this will probably at least be close to the G5 powermacs.And with our capable but demanding available softwares (wayfarer!) we need as much grunt as possible..

I think in MorphOS land the most attractive thing about the Mirari board is age and energy uptake.combined with a hopefully not insane pricetag. I still haven't bought a G5 powermac as it is a nasty energy hog and too bulky (albeit still a nice design IMHO).
Gladly all my old Apple kit is still working, but it won't last forever...

I have a Mac mini PPC with 1.25 GHz. It sucks, haha! If this new board gives me new hardware that turns MorphOS into a real modern OS, then I will definitely go for it

Hey, quick question: What would be the fastest CPU we could hope for here?

@pikewerfer

That 1.8GHz T2081 would be the best drop in option. Might have some overclocking headroom.

Everything better (if it even exist for sale) would require a new board design.

> I have a Mac mini PPC with 1.25 GHz. It sucks, haha! If
> this new board gives me new hardware that turns MorphOS
> into a real modern OS, then I will definitely go for it :-)

Compared to your Mac mini, you'd be able to use a better GPU, up to 256 MiB VRAM and up to 1.7 GiB RAM. You'll have to decide whether this is enough to turn MorphOS into a "real modern OS" for you.

Quote:

Plus: several times faster NIC, SATA II HDD ( T1042 ver ), NVME, USB 3.0 and two PCIe x1 expansion slots. For me another advantage - much faster linux on four cores.
Minus: version with T1042 has no AltiVec ( T2081 has ).
Of course it is enough reasons

> several times faster NIC

Doesn't really make a difference with MorphOS' old TCP/IP stack and ancient SANA-II interface, especially for Internet download/upload speed. Or how much faster is that on the X5000 using similar (same?) NIC than on the Mac mini?

> NVME, USB 3.0

MorphOS drivers?

> and two PCIe x1 expansion slots.

Yes, but at the expense of the special form factor of the Mac mini.