Dorico performance benchmarks?

Interesting! It looks like our chips are fairly comparable, but your times are substantially faster. I wonder if that’s due to the crap I have running in the background, RAM speed, Mobo, or something else? My Mobo is an ASRock Z390 Taichi, and Dorico is running on an NVMe m.2 drive.

Ah. Thanks. I’ll have to make some tests. I’m also curious as to the difference between the performance of the MP and the MBP.

I get 4702 and 796 here. i9 eight-core 2019 MacBook Pro with 32GB RAM, driving two monitors.

If you’re starting from the Modern Orchestral template, won’t it automatically default to some larger paper size?

Ah, yes: A3. Scratch that idea.

Can’t wait to try it when I finally get my hands on the new Ryzen 5900. Single-core performance for that chip is unreal…

On a laptop, Intel Xeon E3-1545M at 2.9 GHz, 32 GB RAM, condensing test took 8388 ms.

Interestingly, I then opened InDesign, Excel, and a bunch of browser tabs. Same test took only 8400 ms. I guess because it’s a RAM thing. But @FredGUnn it makes me think having other processes open probably doesn’t make a huge difference, unless they’re actually taxing the CPU.

Ok, I’m now down to 5090 ms on the 1000 bar condensing test with the HSSE (SE) playback template, 4909 with Silence selected. I made a bunch of BIOS tweaks, mostly disabling Intel SpeedStep and C-state controls. I had already made some required Windows Thunderbolt tweaks, but I guess never made it down to the BIOS tweaks recommended at the end of this UAD guide, which seemed to speed things up a bit.

You’ve got a laptop with Xeons? How big is it?

It’s a Dell Precision. It won’t win any awards for portability, but it’s not bad at all. Maybe 6 lbs? 15-inch screen.

I’ve just run the condensing test on an M1 MacBook Pro (with 16GB RAM).

4031ms and 608ms, in spite of the Rosetta 2 translation. That’s with the HSSE+HSO (Pro) playback template.

With Silence template: 3862ms and 592ms.

That’s not fair!

3 Likes

Oh my God! I should have waited one more year to change my mpb…

2 Likes

Me too! I’d be lost without the larger screen when on the move, and updated as I was due to be away from home for six months of this year. That’s another story…

Yes Dan, great idea! A few weeks ago I upgraded some stuff and did a few tests of my own. I’ll throw out some ideas in point form.

  • I thought a good “realistic” benchmarking file would be the Dorico Prelude or any of the other example files. Not unrealistically long or dense, and everyone can access it.
  • I timed exporting files to the desktop. XML, audio, PDF etc. There were huge differences in XML export time after upgrading.
  • I haven’t tried this yet but if you’re on a Ryzen CPU I think you can disable cores via the Ryzen Master app. Could be useful to see if tasks favour core count vs clock.
  • On a similar note, you could enable/disable XMP to see if RAM speed has any effect on things. Probs not.
  • Maybe this would ruffle feathers but there are lots of tasks that you could compare to Sib/Fin/MS if you have access to them. I have Sib on my mbp and it’s blazing fast compared to Dorico on the same machine. I can upload video at some point.
  • other misc. tests: duplicating flows, duplicating players, time it takes to “activate” a project when switching between multiple projects

So far, things are snappier after moving from i7 6700 to 3800xt. But the experiment is a bit spoiled because it’s a newer mobo and it’s now running off a pcie 4.0 nvme… I’ll run some tests and post results. I’ll be following this thread!

5 Likes

Okidoke, so running @FredGUnn’s 3 flute, 1000 bar condensing test five times I averaged:

Shiny new PC:
5338 ms enabling condensing
4616 ms disabling condensing
For 100 bars:
850 ms enabling condensing
764 ms disabling condensing

Dank old MBP:
8020 ms enabling condensing
6846 ms disabling condensing
For 100 bars:
1147 ms enabling condensing
1018 ms disabling condensing

I’m assuming it’d take longer with more notes to think about…

If anyone wants to try it, take the 100-bar condensed file and export uncompressed XML with “Export layouts as separate files” to your desktop. I also ticked “Create folder for exported files” because of the potential mess. The PC did it in 75625 ms, the MBP in 107108 ms. Big difference imo.

btw current PC setup is:
Gigabyte B550 Aorus Master
Ryzen 7 3800XT (8 core 16 thread, 3.9GHz base, 4.7GHz boost)
64GB 3200MHz
2080ti …let’s be honest it’s for games

Dank old MacBook Pro is:
MBP retina mid 2012
i7 something (4 core 8 threads, 2.6 GHz base, 3.6GHz boost)
8Gb 1600MHz RAM
Mojave

Did you forget the attachment?

I think he means to use the modern orchestra 100-bar example we’ve been using.

Ah! Understood.

Yeah I assume the confusion came from “take the 100 bar condensed file”. I just didn’t want anyone to wait around for 10 minutes to export 1000 bars of xml.

2019 Macbook Pro i9 8-core 2.4GHz, 32GB RAM

condensing test

1000 bars:
enable 4,656 ms
disable 4,085 ms

100 bars:
enable 729 ms
disable 688 ms

MusicXML export 77,633 ms

This M1 MacBook Pro does the export (101 bar version) in 57757ms.