<div>I also tested for text, and the result was unaccountable.</div>
<div> </div>
<div>I couldn't find the method for scaling text in skia canvas, </div>
<div>so I compared between cairo and skia with differentiating fontsize and text length,</div>
<div>then the result was the skia is about 10 times faster than cairo in drawing text.</div>
<div> </div>
<div>I think the test was wrong, but I can't find the wrong point.</div>
<div>I compared the execution time between SkCanvas::drawText and cairo_show_text.</div>
<div> </div>
<div>If cairo_show_text is really so slower than similar function in skia, what can I do for getting a faster execution time in cairo? </div>
<div>Using the cairo_show_glyphs can be better?</div>
<div><br> </div>
<div class="gmail_quote">On Thu, Feb 10, 2011 at 10:16 AM, Siarhei Siamashka <span dir="ltr"><<a href="mailto:siarhei.siamashka@gmail.com">siarhei.siamashka@gmail.com</a>></span> wrote:<br>
<blockquote style="BORDER-LEFT: #ccc 1px solid; MARGIN: 0px 0px 0px 0.8ex; PADDING-LEFT: 1ex" class="gmail_quote">
<div class="im">On Wednesday 09 February 2011 08:23:51 Siarhei Siamashka wrote:<br>> On Wednesday 09 February 2011 05:28:46 cooolheater wrote:<br>> > Thank you for your kind explanation.<br>> > I used pixman-0.21.4 for testing.<br>
> > As you guessed, we are using SIMD and are finding method for NEON<br>> > acceleration.<br>> > Could you let me know the bilinear scaling interfaces in pixman and<br>> > where the SIMD optimization will be applied?<br>
><br>> You can look here for the start:<br>> <a href="http://cgit.freedesktop.org/pixman/tree/pixman/pixman-bits-image.c?id=pixma" target="_blank">http://cgit.freedesktop.org/pixman/tree/pixman/pixman-bits-image.c?id=pixma</a><br>
> n-0.21.4#n189<br>><br>> But applying optimizations locally just for this small function is not<br>> going to provide the best performance, it's kind of like swinging a<br>> large polearm in a narrow passage is not so effective.<br>
<br></div>And here is an example of such patch attached. Performance improvement is not<br>impressive at all. Who cares if it's now let's say ~15x slower than nearest<br>scaling instead of ~30x?<br><br>Obviously we need a better solution.<br>
<br>--<br>Best regards,<br><font color="#888888">Siarhei Siamashka<br></font></blockquote></div><br>