[
https://issues.apache.org/jira/browse/PDFBOX-3988?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16240143#comment-16240143
]
Maciej Matecki commented on PDFBOX-3988:
----------------------------------------
Thanks. That allows me to save a few hundred ms. There's still big difference
between first and the second page:
{code}
Page 0 rendered in 1204 ms
Page 1 rendered in 149 ms
{code}
Above that's the pdf file where two pages are the same. So let's say that
something is cached and the second page is much faster.
However, if I test the pdf where each page is different results look like that:
{code}
Page 0 rendered in 1172 ms
Page 1 rendered in 763 ms
Page 2 rendered in 208 ms
Page 3 rendered in 56 ms
Page 4 rendered in 78 ms
Page 5 rendered in 161 ms
{code}
The interesting thing in the last commit to my small benchmark project there're
two pdfs rendered:
{code}
try (PDDocument document = getPDFDocument("/test-pdf-18.pdf")) {
renderPages(document, dpi);
System.out.println("---------------");
} catch (Exception e) {
System.out.println("Ups. " + e.getMessage());
}
try (PDDocument document = getPDFDocument("/test-2p.pdf")) {
renderPages(document, dpi);
System.out.println("---------------");
} catch (Exception e) {
System.out.println("Ups. " + e.getMessage());
}
{code}
And the results looks like that:
{code}
Page 0 rendered in 1083 ms
Page 1 rendered in 718 ms
Page 2 rendered in 185 ms
Page 3 rendered in 64 ms
Page 4 rendered in 82 ms
Page 5 rendered in 164 ms
Page 6 rendered in 1216 ms
Page 7 rendered in 184 ms
Page 8 rendered in 104 ms
Page 9 rendered in 99 ms
Page 10 rendered in 112 ms
---------------
Page 0 rendered in 123 ms
Page 1 rendered in 113 ms
---------------
{code}
So the first page of the second PDF is rendered really quickly.
Is there something more to initialize earlier so later results are better?
> Performance issue when rendering first page of PDF
> --------------------------------------------------
>
> Key: PDFBOX-3988
> URL: https://issues.apache.org/jira/browse/PDFBOX-3988
> Project: PDFBox
> Issue Type: Improvement
> Components: Rendering
> Affects Versions: 2.0.7
> Reporter: Maciej Matecki
> Attachments: calls.png
>
>
> Let say that you want to generate PNG for all pages in PDF.
> The generation of the first page is really slow, the second one is quite fast.
> For example test PDF contains two same PDF pages.
> First page renders in: ~2000ms
> Second one: ~220 ms
> It looks like for the first page (inv.0) there's 359 ms overhead just for
> creating the font.
> [^calls.png]
> Tried to use other library to perform the same operation with the same pdf
> file and I was able to retrieve BufferedImage of the first (slower) page in
> ~750 ms.
> For the second page the result was almost the same like for the PDFBox.
> It looks like there's the place to improve performance when starting
> rendering the PDF page.
> Or do you have any advices how to improve performance in the example project?
> Test project: https://github.com/mmatecki/pdfboxperformance
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
