I see the only version of Ultra Fractal available for macOS is still an Intel build. Are there plans to do a build for the new Apple M1 processors? I imagine it would be a lot faster for those devices.

I see the only version of Ultra Fractal available for macOS is still an Intel build. Are there plans to do a build for the new Apple M1 processors? I imagine it would be a lot faster for those devices.
 
0
reply

I'm looking into it, but from initial testing on an M1 Mac I found that 1) the Intel version works amazingly well and is still quite fast and 2) no huge speedups are to be expected from a native M1 version. It will be approximately as fast as on a recent Intel Mac, and the Intel version of UF6 with Rosetta is definitely not much slower. Nevertheless, it's an interesting technical challenge and I plan to eventually release a native M1 version of Ultra Fractal.

The main reason the M1 Macs perform very well on benchmarks seems to be that memory access is extremely fast, but that's something that doesn't really matter for most fractal calculations. For straight-on floating-point calculations, they are not faster than where Intel currently is.

I'm looking into it, but from initial testing on an M1 Mac I found that 1) the Intel version works amazingly well and is still quite fast and 2) no huge speedups are to be expected from a native M1 version. It will be approximately as fast as on a recent Intel Mac, and the Intel version of UF6 with Rosetta is definitely not much slower. Nevertheless, it's an interesting technical challenge and I plan to eventually release a native M1 version of Ultra Fractal. The main reason the M1 Macs perform very well on benchmarks seems to be that memory access is extremely fast, but that's something that doesn't really matter for most fractal calculations. For straight-on floating-point calculations, they are not faster than where Intel currently is.

Ultra Fractal author

edited May 8 '22 at 8:26 pm
 
0
reply

That's fair. One of my reasons for asking is not just speed, it is concern over UF6 no longer functioning on M1 devices once Apple inevitably drop support for Rosetta there, as they did in the past for PowerPC. I currently only have a license for UF5 but that stopped working in macOS 10.15 where 32-bit support was dropped. I'll probably upgrade if an M1 version of UF6 comes along since this will give me a bit more confidence that it is going to be usable for a decent amount of time.

That's fair. One of my reasons for asking is not just speed, it is concern over UF6 no longer functioning on M1 devices once Apple inevitably drop support for Rosetta there, as they did in the past for PowerPC. I currently only have a license for UF5 but that stopped working in macOS 10.15 where 32-bit support was dropped. I'll probably upgrade if an M1 version of UF6 comes along since this will give me a bit more confidence that it is going to be usable for a decent amount of time.
 
0
reply

That's exactly why an M1 version of Ultra Fractal is inevitable in the long run. But, at the moment Apple is still selling Macs with Intel processors so that cut-off will be years off.

That's exactly why an M1 version of Ultra Fractal is inevitable in the long run. But, at the moment Apple is still selling Macs with Intel processors so that cut-off will be years off.

Ultra Fractal author

 
0
reply

It will be approximately as fast as on a recent Intel Mac, and the Intel version of UF6 with Rosetta is definitely not much slower.

I'm finding motion blur is much slower. This 38 hour render took 1.5 hours on Intel (3.1 GHz i5 dual-core, 8 GB RAM). The animation is a basic Mandelbrot zoom with motion blur at minimum setting. Pixels/sec on M2 dropped below 700 during motion blur but stayed above 4000 on Intel. My other tests show similar results.

UF's motion blur looks really great, especially on 24 fps animations, which are what I make. It'd be super cool if this were faster on M2.

634b4ee6d40ad.png

>It will be approximately as fast as on a recent Intel Mac, and the Intel version of UF6 with Rosetta is definitely not much slower. I'm finding motion blur is much slower. This 38 hour render took 1.5 hours on Intel (3.1 GHz i5 dual-core, 8 GB RAM). The animation is a basic Mandelbrot zoom with motion blur at minimum setting. Pixels/sec on M2 dropped below 700 during motion blur but stayed above 4000 on Intel. My other tests show similar results. UF's motion blur looks really great, especially on 24 fps animations, which are what I make. It'd be super cool if this were faster on M2. ![634b4ee6d40ad.png](serve/attachment&path=634b4ee6d40ad.png)

http://www.youtube.com/fractalzooms

 
0
reply
376
views
4
replies
4
followers
live preview
Enter at least 10 characters.
WARNING: You mentioned %MENTIONS%, but they cannot see this message and will not be notified
Saving...
Saved
All posts under this topic will be deleted ?
Pending draft ... Click to resume editing
Discard draft