Hi Vaughan, Your choral score is beautiful, Most impressive. Well done. Another fantastic use of lilypond.
It seems your score has become a standard benchmark for lilypond performance tests now! A couple of points. It is very often the case that when running Linux on Virtualbox or similar under Windows or Mac OS X that the performance, both disk IO and CPU, even though on the same machine, is better. While I have never looked into this deeply, it is readily observed. I am not convinced that it is due to the design of UNIX because Mac is essentially UNIX under the hood as well. On my Windows 10 machine your score compiles in about 55s and on Linux in a VM on the same machine about 35s. Others here have similar figures. Also, you mentioned you use MIDI for score checking. You could separate the midi generation and the PDF generation. Run the midi when you need it, and generate the PDF score normally. This would speed up your workflow a lot. You could even have a small script that kicks off both compiles in parallel at the same time. I am sure it does not matter of the midi finishes later [I think it is the midi that takes the time?]. Just an idea. Andrew On 21 November 2016 at 21:57, Knut Petersen <knut_peter...@t-online.de> wrote: > Am 17.11.2016 um 16:19 schrieb Knut Petersen: > >> Hi Vaughan! >> >> I’ve included a large project if anyone with a newish computer would like >>> to test their compilation time. The main file is MDSM.ly. It takes my >>> computer between 4 and 5 minutes to compile. >>> >> >> > I tried my old computer ... Pentium-M Dothan 1.86 GHz on an AOpen > i915GMm-hfs mobo with 2 GB RAM. > One hardware update: OpenSuSE linux system installed on SSD, all the other > parts are older than 11 years. > That old machine only needs between 138 and 140 seconds! > > The old machine takes about 5 times longer to compile your score than my > fast PC, and more than 50% of that > is related to the increased system clock. > > But: If I do a full build of lilypond, the old machine takes about 23 > times as long as the modern computer ;-) > > Some further comments: With some slight modifications you could compile > Agnus, Credo, etc as individual scores > and combine the pdf using a program like pdftk. That reduced the time to > build the pdf to a bit less than 8 seconds > on the modern hardware. > > cu, > Knut > > My system: >> ========= >> mainboard: asus h97 pro gamer >> cpu: i7-4790K, 4.00 Ghz (turbo speed: up to 4.4 GHZ,) >> SSD: Samsung SSD 850 PRO 512GB >> os: Linux, based on openSuSE Tumbleweed >> For the test I used lilypond 2.18.2. >> >> Compile times: >> ============ >> 1st run: 27,123 seconds >> 2nd run: 26,697 seconds >> 3rd run: 26,700 seconds >> 4th run: 26,901 seconds >> > > > _______________________________________________ > lilypond-user mailing list > lilypond-user@gnu.org > https://lists.gnu.org/mailman/listinfo/lilypond-user >
_______________________________________________ lilypond-user mailing list lilypond-user@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-user
