Q over at Hold and Modify on YouTube just posted a Lightwave Raytrace Scene benchmark and I absolutely love this type of stuff so if you're interested please follow the link below to follow the simple steps to run the benchmark and post your results & system specs

Hold and Modify Lightwave3D Benchmark Raytrace Scene!

Last edited by klx300r on 22-Jan-2023 at 06:41 AM.

_________________
____________________________
c64-2sids, A1000, A1200T-060@50(finally working!),A4000-CSMKIII
! My Master Miggies- Amiga 1000 & AmigaOne X1000 !
mancave-ramblings
X1000 I BELIEVE

A4000, CS-MKIII, 060@50 (Rev1), OS3.2.1- 47m21

A1200, TF1260, 060@50 (Rev1), OS3.2.1- 51m48

_________________
____________________________
c64-2sids, A1000, A1200T-060@50(finally working!),A4000-CSMKIII
! My Master Miggies- Amiga 1000 & AmigaOne X1000 !
mancave-ramblings
X1000 I BELIEVE

@klx300r

https://drive.google.com/file/d/1_VxGqSt2fMQbDxdZNO-SvA5Th0gfgp3n/view?usp=sharing

Pi Storm is dropping the mic there...

Emu68K ran in 5m 51s for Pi 4B 2.2GHz

About Twice the speed as the MorphOS / G4 1.76 GHz. Although that user doesn't know how to benchmark as it reports running various other potentially CPU intensive stuff in the background

Last edited by Karlos on 28-Jan-2023 at 09:14 AM.

_________________
Doing stupid things for fun...

@Karlos

Yeh, old ram and memory controller.

I See X5000 scores conveniently left out.
I guess it does not fit the narrative.

Last edited by NutsAboutAmiga on 28-Jan-2023 at 02:18 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@NutsAboutAmiga

Or nobody has tried it. Occam's Razor. If you have that machine, why don't you try? Or does that not fit the narrative?

Last edited by Karlos on 28-Jan-2023 at 02:41 PM.

_________________
Doing stupid things for fun...

@Karlos

On Amigans, there is an X5000 doing 3m 27sec.
Interesting point is that the Morphos Trance 68K emulator is so much slower. Factor 3.

@bennymee

Chances are the results haven't been shared with whomever is maintaining that spreadsheet.

Based on that result, the x5000 is about 70% faster than the PiStorm form this task.

Last edited by Karlos on 28-Jan-2023 at 03:27 PM.

_________________
Doing stupid things for fun...

@Karlos

Quote:

Based on that result, the x5000 is about 70% faster than the PiStorm form this task.

But at what cost?

You can write out "Emulated 68k trounces native PowerPC" a 100 times across the city walls, and the "Romans" are likely to focus on correcting your syntax rather than alter their strategy to one that makes sense.

_________________
All the way, with 68k

@agami

That was my point. Perhaps it's too subtle.

_________________
Doing stupid things for fun...

today arm is also pc. there is no fun in it.

@ppcamiga1

Fun is subjective. There's plenty of fun to be had with PCs thanks to the huge amounts of relatively inexpensive processing power on offer. If you're struggling to think of something to do there, I suggest you check your imagination isn't at fault first.

_________________
Doing stupid things for fun...

@bennymee
Quote:

On Amigans, there is an X5000 doing 3m 27sec. Interesting point is that the Morphos Trance 68K emulator is so much slower. Factor 3.

Not from what I've gathered

Last edited by pixie on 29-Jan-2023 at 04:09 PM.

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

@bennymee

No, this isn't true. Check the link that pixie provided. On the X5000, Trance appears a bit faster than Petunia, doing the same benchmarks under the same LW version on the same system.

@pixie

No, the benchmarks described on the link are being run under different settings (via cli, using the default options). By using low resolution and square pixels on the camera panel as the YT video creator suggests, I am pretty sure that the value reported on the YT should be pretty much close to the truth.

@agami

Petunia and Trance translate 68k commands to native PPC while retaining a transparent interface with rest of the system components (libs, datatypes, ahi, dynamic ram allocation, whatever else it is needed) and are close to conception to what MS did with Office 32bit when running into a 64bit CPU/Windows version. Have you ever herd any user stating: let me emulate just a sec my Outlook and drop you a mail, when he opens the application?

As for the price, that's the current requested amount for having available 4K video player, OpenGL, more apps, games, a semi - modern web browser and not the need to alt+tab on an alien OS every now and then. Because, even if the PiStorm costs 150 eur. + the cost of the host Amiga, that's still 150 times more + the cost of the host Amiga to get zero result when browsing to the internet due to lack of software.

_________________

@Karlos

That was my point, it was not correct.

Waiting for Winuae on Windows results :)

Last edited by bennymee on 29-Jan-2023 at 08:54 PM.

@Cool_amigaN

Quote:

As for the price, that's the current requested amount for having available 4K video player, OpenGL, more apps, games, a semi - modern web browser and not the need to alt+tab on an alien OS every now and then. Because, even if the PiStorm costs 150 eur. + the cost of the host Amiga, that's still 150 times more + the cost of the host Amiga to get zero result when browsing to the internet due to lack of software.

Classic Amiga host is 150x more than €150? If that were true I would be selling my A4000 and A1200 immediately.

But beyond the very specific example of PiStorm, or PiMiga, I am advocating for 68k emulation on high performing commodity x86 hardware, in a way that Amithlon did it. Where the emulated 68k would be equivalent to 2GHz+ on the latest AMD Zen 3/Zen 4 CPUs.

While currently OS 3.x or AROS 68k do not have a Wayfarer port, in large part due to limitations of existing 68k hardware, it should have no problem running on an emulated system approximating 2GHz (effectively) with upwards of 2GB of fast RAM.

Everything AmigaOS 4 has, can be ported to 68k. Even if it weren't all that optimised, the minimum spec could be above that for OS4 on PPC, being:
- 68k @ 1GHz
- RTG graphics with 64MB VRAM
- 256MB RAM
- 16bit audio
- etc

Even emulating on budget PC hardware can exceed the above specs.
Now, would it be wasteful to have a 64-bit 6-core/12-thread machine with 8GB+ of DDR4/DDR5 RAM, NVMe SSD, and Radeon RDNA GPU, emulate a 32-bit 1-core/1-thread, max 2GB Z3 RAM, with emulated Zorro III or PCI graphics?
Yes. For a while. Though no more wasteful than the waste of unused hardware in the X1000/X5000, and at a much lower price tag and greater availability.

And I say "for a while" because the likely larger user-base could encourage more development in targeting the use of the unused host hardware. Which in turn could further grow the user-base, and then perhaps move over to some 68k with 64-bit extensions a la x86-64, which in turn unlocks more of the host's RAM, which gives developers more resources, and so on.

Last edited by agami on 30-Jan-2023 at 12:07 AM.
Last edited by agami on 30-Jan-2023 at 12:01 AM.

_________________
All the way, with 68k

@bennymee

Quote:

Waiting for Winuae on Windows results :)

54s on my 5800x. If you check the comments, someone had run it natively on windows in a 24 thread computer about 1.3s. doing the math, you would get 31.2. About 57% of native speed, which is a bit nuts in itself.

Last edited by pixie on 30-Jan-2023 at 01:22 AM.
Last edited by pixie on 30-Jan-2023 at 01:14 AM.

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

Duplicate

Last edited by pixie on 30-Jan-2023 at 01:14 AM.

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

@bennymee

Quote:

Waiting for Winuae on Windows results :)

21 seconds on my i9-13900k

@pixie

It's (WinUAE CPU performance) less than %20 native, single threaded speed.
The renders on Windows versions of Lightwave are rendering at higher resolutions with higher precision and greater color depths, so its not a like for like. There's many multiples of the amount of processing done for renders on Windows versions of Lightwave. There's also a bit of time used in the initialization not used for rendering the scene.

Last edited by fishy_fis on 30-Jan-2023 at 09:13 AM.
Last edited by fishy_fis on 30-Jan-2023 at 09:11 AM.
Last edited by fishy_fis on 30-Jan-2023 at 03:29 AM.
Last edited by fishy_fis on 30-Jan-2023 at 03:26 AM.

@fishy_fis

Any half modern native version for windows will likely make use of multiple cores and SIMD units on them too. Unless you are using some ancient version that works the same way as the Amiga one?

_________________
Doing stupid things for fun...

@pixie

1.3 sec on native Lightwave if I am correct.
No comparison.

The 21 sec results in this topic for the 13900K are more usefull, comparing them with 68K emulation on Amiga platforms like Petunia, Trance and Emu64.

Last edited by bennymee on 30-Jan-2023 at 08:32 AM.