swingmark also shows double improvement on the retina.
On 25.06.15 13:40, Andrew Brygin wrote:
Hello Sergey,
24/06/15 22:45, Sergey Bylokhov wrote:
Hi, Andrew.
Thanks for this report. As far as I understand it in case of retina
the lcd text is drawing faster after the fix than aa before the fix,
which means that we will not get a new regressions. So the fix looks
fine.
But on non retina our results still not so good, lcd text is slow:
485(was 16.4) vs 16508..... and the window for optimizations still
exists.
I agree that there is a room for further optimizations.
However, I do not think that it is possible to achieve the same
level of performance of the lcd rendering, as the aa rendering,
because of more complex nature of the lcd text rendering.
Thanks,
Andrew
global.dest=VolatileImg(Opaque),text.opts.font.fsize=6.0,text.opts.graphics.textaa=LCD_HRGB:
9-8087201-v00: 485.2052560 (var=0.57%) (2955.82%)
**|*********************************************************
**|*********************************************************
**|*********************************************************
global.dest=VolatileImg(Opaque),text.opts.font.fsize=6.0,text.opts.graphics.textaa=On:
9-8087201-v00: 16508.76580 (var=0.66%) (99.69%)
************************************************************|
************************************************************|
*********************************************************** |
On 19.06.15 15:54, Andrew Brygin wrote:
Hello Sergey,
the only part of the fix affects the performance of AA case: the
cache cell size.
In a case of retina, 13pt and 20pt glyphs do not fit the 16x16
cache cells,
so these benchmarks show better performance:
13pt: 40-80 times faster
20pt: 7-13 times faster
6pt shows the same results, because it fits the cache in any case.
Full benchmark results:
http://cr.openjdk.java.net/~bae/8087201/9/ogl-lcd-aa.res
Regarding the suggestion with creating a separate method for the fast
path possibility check: please note that we do this check and
calculate
the dstTextureID only once per whole glyph vector, but use the
dstTextureID
as an indicator for every glyph. So such change will affect
performance for
sure.
Probably we can masquerade the 'dstTextureID == 0' condition with
some
sort of a macro, like canReadDestinationDirectly() or something
like this.
Are you OK with this?
Thanks,
Andrew
19/06/15 13:57, Sergey Bylokhov wrote:
Hi, Andrew.
Can you additionally provide the bench data about aa(before/after
the fix) vs new lcd lcd?
Probably it well be more obvious if the code in OGLTextRenderer
1007 if (OGLC_IS_CAP_PRESENT(oglc, CAPS_EXT_TEXBARRIER) &&
1008 dstOps->textureTarget == GL_TEXTURE_2D)
Will be moved to the separate method and the check to the
possibility of fast blit will be clarified instead of:
if (dstTextureID == 0) {
Also your review request contains useful information like
fast/slow/read-after-write etc. I think this information can be
useful as a comments in the code.
On 18.06.15 17:39, Andrew Brygin wrote:
Hello,
could you please review a fix for 8087201?
The root of the problem is that we have to supply a content of
destination surface to lcd shader to compose the lcd glyph
correctly.
In order to do this, we have to copy a sub-image from destination
buffer to an intermediate texture using glCopyTexSubImage2D()
routine.
Unfortunately, this routine is quite slow on majority of systems,
and it
dramatically reduces the overall speed of lcd text rendering.
The main idea of the fix is to use a texture associated with the
destination
surface if it exists. In this case we have a chance to completely
abandon the
data copying. However, we have to avoid read-after-write in order
to get
correct results in this case. Fortunately, it can be achieved by
using the
GL_NV_texture_barrier extension:
https://www.opengl.org/registry/specs/NV/texture_barrier.txt
Beside this, suggested fix introduces following changes in OGL
text renderer:
* Separate accelerated caches for LCD and AA glyphs
We have a single cache which is initialized ether for LCD or
for AA glyphs.
If application mixes these types of font smoothing from some
reasons, we
have got a significant performance degradation.
For example, if we use J2DBench in GUI mode, then swing GUI
initializes the
accelerated cache for AA, and subsequent rendering of LCD text
always
uses 'no-cache' code path.
* Increase dimension of the glyph cache cell from 16x16 to 32x32.
This change gives significant performance boost on systems with
retina
(because of average size of rendered glyphs).
However, on systems where the fast path with destination
texture is not
possible for any reasons, this change may cause a performance
degradation
because of more extenceive usage of glCopyTexSubImage2D.
So, we probably may want to get a means to configure the cell
dimension
depending on system capabilities.
Performance results overview:
* MBP with Intel Iris (retina, texture barrier is available):
http://cr.openjdk.java.net/~bae/8087201/9/mbp-intel-iris.txt
* iMac with AMD HD6750M (no retina, texture barrier is available):
http://cr.openjdk.java.net/~bae/8087201/9/imac-amd-hd6750m.txt
* MBP with OSX10.8, NV GF9600M (no retina, no texture barrier):
http://cr.openjdk.java.net/~bae/8087201/9/mbp-10.8-NVGF9600M.txt
Please take a look.
Thanks,
Andrew
--
Best regards, Sergey.
--
Best regards, Sergey.
