I have posted about speed issues with Dorico before, but I’ll admit that working with Steve’s Rite of Spring files, I have no speed issues at all. As others have mentioned, processor speed is probably your main issue here.
As some have suggested, you might make it a practice to work in galley mode, and don’t do any fine work until you have the entire piece keyed in. There is a temptation to want to make everything look right early in the process, but it’s usually better not to fight the program. Most of the scores that I have worked with have been translated from Sibelius or Finale files and then reconstructed - although I find that if you can do as much of that reconstruction as possible in the legacy program, the better. Untangling manually-condensed Sibelius scores is more easily done in Sibelius.
I understand that your scores aren’t condensed yet, but it is best to leave that as late as possible in the process, as well as any fine adjustments to positioning. The more you can leave in the default settings/positions as possible, the less slow down you will be prone to.
For reference, I’m using an i9-14900K (3.2 GHz) and 128GB RAM (8 core/32 thead). If you are comparing Geekbench scores, that is more than twice the speed as yours on both single and multiprocesser performance. You might also take a look at your video card. Switching from an Nvidia GTX730 to a GTX1060 made a huge difference in the speed of my screen redraws, and that isn’t even one of the newer breeds of graphics cards. If you are using onboard video, it is competing with your processing for bandwidth.
I have noticed that the speed at which Dorico switches from Engrave to Galley view is still slow. (7 seconds in an 88 page trumpet concerto, fully condensed - 4 flows.) My big beef is that Dorico has a lot of unnecessary mouse clicks and over-crowded menus and dialog boxes, which if made more efficient, would speed up your input as well, even if you still had to wait for screen redraws.
I just got bit by this problem. I was revisiting an old score. 30 staves, 500 bars (full orchestra). There is a significant lag during note entry. The delay between pressing a key on my keyboard and the ghost note registering is about 2 seconds. The little green MIDI light in the lower right, however, registers the MIDI activity immediately.
While my experiments have shown that the size of the score is definitely a factor, I can greatly speed things up by changing the playback engine to anything BUT NotePerformer. This is odd because Dorico isn’t playing notes while I enter them, so it’s not clear why the active playback engine should have any effect on the speed of note entry.
It could be the complexity of the expression map: NotePerformer is one of the more larger and more complex expression maps, and as such, when Dorico refreshes playback data (which it does after most note input and editing operations) this could have an impact in a larger project, even though Dorico tries to minimise the amount of recalculation it’s carrying out.
So using a lighter playback template, e.g. plain old HALion Sonic Selection, might help.
Yes, switching to HALion Sonic does results in much faster response. While I understand why one might need to refresh the playback data after note entry, it’s not clear why it needs to be done during note entry. Particularly if “Play notes during note input…” and “Play changes of shadow note pitch…” are off. I don’t mind waiting a second or two to enter or exit note entry, but waiting two seconds between each press of the MIDI keyboard is frustrating.
Yes, it would be nice if Dorico could update the playback pipeline only when the application is idle (similar to how it now handles e.g. proofreading) but unfortunately that’s not how things are set up at the moment.
Well, makes sense, userbenchmark.com indicates around 25-30% difference with my machine. SSD also for project storage?
Just checked CpuMonkey as well, Multi-core difference is even higher, obviously.
My conclusion would be if you’re handling massive scores (bigger even than Stravinsky…) plus the eventual VST overhead (more RAM), you might need to look at a faster machine…
Updated GPU to dedicated pcie (NVIDIA GeForce GTX 1070)
Open from HUB: 41s (no change)
Uncondensing: 13s (no change
Condensing: 16s (no change)
To Concert pitch: 46s (2s change)
To transposed: 46s (3s change)
Specs:
Operating System
Windows 11 Home 64-bit
CPU
AMD Ryzen 5 5600G 42 °C
Cezanne 7nm Technology
RAM
32,0GB Dual-Channel DDR4 @ 1596MHz (16-20-20-38)
Motherboard
Micro-Star International Co., Ltd. MPG B550 GAMING PLUS (MS-7C56) (AM4) 34 °C
Graphics
PHL 275E1 (2560x1440@59Hz)
4095MB NVIDIA GeForce GTX 1070 (ASUStek Computer Inc) 36 °C
ATI AMD Radeon Graphics (ATI)
SLI Disabled
CrossFire Disabled
Storage
931GB Seagate ST1000DM003-1CH162 (SATA ) 26 °C
931GB Seagate ST1000DM003-1ER162 (SATA ) 22 °C
223GB KINGSTON SA400S37240G (SATA-2 (SSD)) 37 °C
465GB Seagate ST500DM0 02-1BD142 USB Device (USB )
Optical Drives
Optiarc DVD RW AD-7260S
Audio
Focusrite USB Audio
Alright.
I’ll have to work on the fence a little before it starts raining again, but then I’ll run the flute benchmark and the Stravinsky one on Noah’s new machine for everybody’s continuing enjoyment. Stay tuned…
Perhaps a little off-topic, but I’m still surprised by these timings on fairly long projects.
I don’t use Dorico for long projects, but this makes me think:
Dorico is now integrated into Cubase. Can anyone give some feedback on the impact on the slowness in Cubase since Dorico’s integration?
@MarcLarcher, does it matter if the condensing is done in page mode or engrave?
And the timings come from an exported diagnostics file, right?
And it’s supposed to be using the silent PTempl.?
B.
