[OpenJDK 2D-Dev] [8] request fro review: 8005530: [lcms] Improve performance of ColorConverOp for default destinations
Andrew Brygin
andrew.brygin at oracle.com
Fri Feb 1 15:41:41 UTC 2013
Hello Phil,
please take yet another look at updated webrev:
http://cr.openjdk.java.net/~bae/8005530/webrev.02/
Summary of changes:
- LCMS.c: unused native method getTagSize() was removed.
- mapfiles for lcms library are updated accordingly to changes in
native methods of LCMS
(affects new and old build systems, is visible on solaris)
Thanks,
Andrew
On 1/29/2013 1:52 AM, Phil Race wrote:
> This is all fine by me.
>
> -phil.
>
> On 1/28/2013 1:40 AM, Andrew Brygin wrote:
>> Hello Phil,
>>
>> I have updated the fix according to your comment:
>>
>> http://cr.openjdk.java.net/~bae/8005530/webrev.01/
>>
>> Beside this, I have included some improvements related to
>> access to profile header (which is quite important in case
>> of decoding jpeg images with embedded color profile),
>> and to raster processing.
>> New benchmark to estimate the performance of image
>> decoding was also added to the fix.
>>
>> Please take a look to a comparison of performance level
>> below.
>>
>> Regarding to the change in build files: I have verified that
>> it works for both old and new build systems. I am pretty
>> sure that the change is safe (the HIGHEST level of optimization
>> is already used for other libraries in jdk).
>>
>> Would you like to see the change in build systems as a
>> separate fix?
>>
>> Thanks,
>> Andrew
>>
>> ======================================================================
>> Options common across all tests:
>> cmm.colorconv.ccop.ccopOptions.content=photo
>> cmm.colorconv.ccop.ccopOptions.srcType=BYTE_3BYTE_BGR
>> cmm.colorconv.embed.embedOptions.Images=MEDIUM
>> cmm.opts.profiles=1001
>>
>> cmm.colorconv.ccop.op_draw,cmm.colorconv.ccop.ccopOptions.dstType=BYTE_3BYTE_BGR,cmm.colorconv.ccop.ccopOptions.size=250:
>>
>> OpenJDK Baseline: 0.008438818 (var=18.29%) (100.0%)
>> *************************************** |
>> ****************************************|
>> ****************************************|
>> OpenJDK Fix: 0.012336448 (var=7.36%) (146.19%)
>> ****************************************|******************
>> ****************************************|*****************
>> ****************************************|****************
>> cmm.colorconv.ccop.op_draw,cmm.colorconv.ccop.ccopOptions.dstType=BYTE_3BYTE_BGR,cmm.colorconv.ccop.ccopOptions.size=4000:
>>
>> OpenJDK Baseline: 0.900596556 (var=2.93%) (100.0%)
>> ******************************************************|
>> ***************************************************** |
>> ******************************************************|
>> OpenJDK Fix: 0.193663771 (var=7.52%) (110.1%)
>> ******************************************************|*****
>> ******************************************************|**
>> ******************************************************|*****
>> cmm.colorconv.ccop.op_draw,cmm.colorconv.ccop.ccopOptions.dstType=COMPATIBLE_DST,cmm.colorconv.ccop.ccopOptions.size=250:
>>
>> OpenJDK Baseline: 0.009548438 (var=16.26%) (100.0%)
>> ***********************************************|
>> ***********************************************|
>> ********************************************* |
>> OpenJDK Fix: 0.011871784 (var=7.76%) (124.33%)
>> ***********************************************|************
>> ***********************************************|***********
>> ***********************************************|********
>> cmm.colorconv.ccop.op_draw,cmm.colorconv.ccop.ccopOptions.dstType=COMPATIBLE_DST,cmm.colorconv.ccop.ccopOptions.size=4000:
>>
>> OpenJDK Baseline: 0.927800441 (var=7.74%) (100.0%)
>> ****************************************************|
>> ****************************************************|
>> ****************************************************|
>> OpenJDK Fix: 0.247631935 (var=12.63%) (110.92%)
>> ****************************************************|*****
>> ****************************************************|*****
>> ****************************************************|*****
>> cmm.colorconv.ccop.op_img,cmm.colorconv.ccop.ccopOptions.dstType=BYTE_3BYTE_BGR,cmm.colorconv.ccop.ccopOptions.size=250:
>>
>> OpenJDK Baseline: 0.171872303 (var=13.62%) (100.0%)
>> ******************************************|
>> ******************************************|
>> ******************************************|
>> OpenJDK Fix: 0.238933601 (var=3.17%) (139.02%)
>> ******************************************|****************
>> ******************************************|****************
>> ******************************************|****************
>> cmm.colorconv.ccop.op_img,cmm.colorconv.ccop.ccopOptions.dstType=BYTE_3BYTE_BGR,cmm.colorconv.ccop.ccopOptions.size=4000:
>>
>> OpenJDK Baseline: 0.615214994 (var=7.92%) (100.0%)
>> ******************************|
>> ******************************|
>> ******************************|
>> OpenJDK Fix: 0.001296456 (var=2.15%) (195.98%)
>> ******************************|*****************************
>> ******************************|****************************
>> ******************************|*****************************
>> cmm.colorconv.ccop.op_img,cmm.colorconv.ccop.ccopOptions.dstType=COMPATIBLE_DST,cmm.colorconv.ccop.ccopOptions.size=250:
>>
>> OpenJDK Baseline: 0.065811677 (var=6.25%) (100.0%)
>> ****************|
>> ****************|
>> ****************|
>> OpenJDK Fix: 0.236734693 (var=7.65%) (359.72%)
>> ****************|*****************************************
>> ****************|******************************************
>> ****************|******************************************
>> cmm.colorconv.ccop.op_img,cmm.colorconv.ccop.ccopOptions.dstType=COMPATIBLE_DST,cmm.colorconv.ccop.ccopOptions.size=4000:
>>
>> OpenJDK Baseline: 0.763357432 (var=1.53%) (100.0%)
>> ********|
>> ********|
>> ********|
>> OpenJDK Fix: 0.001278227 (var=3.47%) (724.88%)
>> ********|**************************************************
>> ********|*************************************************
>> ********|**************************************************
>> cmm.colorconv.ccop.op_rst,cmm.colorconv.ccop.ccopOptions.dstType=BYTE_3BYTE_BGR,cmm.colorconv.ccop.ccopOptions.size=250:
>>
>> OpenJDK Baseline: 0.110134436 (var=1.92%) (100.0%)
>> ***************************|
>> ***************************|
>> ***************************|
>> OpenJDK Fix: 0.234074823 (var=4.62%) (212.54%)
>> ***************************|******************************
>> ***************************|*******************************
>> ***************************|******************************
>> cmm.colorconv.ccop.op_rst,cmm.colorconv.ccop.ccopOptions.dstType=BYTE_3BYTE_BGR,cmm.colorconv.ccop.ccopOptions.size=4000:
>>
>> OpenJDK Baseline: 0.391746749 (var=16.2%) (100.0%)
>> ****************|
>> ****************|
>> *************** |
>> OpenJDK Fix: 0.001300390 (var=2.29%) (383.4%)
>> ****************|*******************************************
>> ****************|*******************************************
>> ****************|******************************************
>> cmm.colorconv.ccop.op_rst,cmm.colorconv.ccop.ccopOptions.dstType=COMPATIBLE_DST,cmm.colorconv.ccop.ccopOptions.size=250:
>>
>> OpenJDK Baseline: 0.110379746 (var=1.38%) (100.0%)
>> ***************************|
>> ***************************|
>> ***************************|
>> OpenJDK Fix: 0.238619309 (var=3.32%) (216.18%)
>> ***************************|******************************
>> ***************************|******************************
>> ***************************|*******************************
>> cmm.colorconv.ccop.op_rst,cmm.colorconv.ccop.ccopOptions.dstType=COMPATIBLE_DST,cmm.colorconv.ccop.ccopOptions.size=4000:
>>
>> OpenJDK Baseline: 0.846569883 (var=10.33%) (100.0%)
>> *************|
>> *************|
>> *************|
>> OpenJDK Fix: 0.001290600 (var=2.88%) (453.39%)
>> *************|*********************************************
>> *************|**********************************************
>> *************|*********************************************
>> cmm.colorconv.embed.embd_img_read:
>> OpenJDK Baseline: 0.299874002 (var=7.04%) (100.0%)
>> ************|
>> ************|
>> ************|
>> OpenJDK Fix: 0.003114456 (var=9.07%) (494.37%)
>> ************|*********************************************
>> ************|**********************************************
>> ************|**********************************************
>> cmm.profiles.getHeader:
>> OpenJDK Baseline: 136.4378577 (var=4.34%) (100.0%)
>> *|
>> *|
>> *|
>> OpenJDK Fix: 6858.467330 (var=1.36%) (5026.81%)
>> *|**********************************************************
>> *|**********************************************************
>> *|**********************************************************
>> cmm.profiles.getNumComponents:
>> OpenJDK Baseline: 135.4737465 (var=2.49%) (100.0%)
>> *|
>> *|
>> *|
>> OpenJDK Fix: 6903.416838 (var=4.1%) (5095.76%)
>> *|*********************************************************
>> *|********************************************************
>> *|********************************************************
>>
>> Summary:
>> OpenJDK Baseline:
>> Number of tests: 15
>> Overall average: 18.159329310271605
>> Best spread: 1.38% variance
>> Worst spread: 18.29% variance
>> (Basis for results comparison)
>>
>> OpenJDK Fix:
>> Number of tests: 15
>> Overall average: 917.5243389674637
>> Best spread: 1.36% variance
>> Worst spread: 12.63% variance
>> Comparison to basis:
>> Best result: 5095.76% of basis
>> Worst result: 110.1% of basis
>> Number of wins: 15
>> Number of ties: 0
>> Number of losses: 0
>>
>>
>> On 1/17/2013 1:54 AM, Phil Race wrote:
>>> The variables
>>> 154 ShortComponentRaster shortRaster;
>>> 155 IntegerComponentRaster intRaster;
>>> 156 ByteComponentRaster byteRaster;
>>>
>>> are each used only in a localised few lines of code,
>>> can we move them to the block where they are used so
>>> that their scope is limited to that ? I don't know for a fact
>>> but I can imagine that the VM can then skip allocating stack for them
>>> until you enter that block which actually needs them.
>>>
>>> As you note,this takes into account old and new builds in the
>>> makefile changes
>>> Have you tested both ?
>>> And unless you are 100% confident in this [minimal] change
>>> you should get sign off from the build group as they've recently
>>> requested.
>>>
>>> You should add "noreg-perf" as a label in the JIRA
>>>
>>> -phil.
>>>
>>> On 1/9/2013 4:25 AM, Andrew Brygin wrote:
>>>> Hello Jennifer and Phil,
>>>>
>>>> could you please review a fix for CR 8005530?
>>>>
>>>> CR: http://bugs.sun.com/view_bug.do?bug_id=8005530
>>>> Webrev:http://cr.openjdk.java.net/~bae/8005530/webrev.00/
>>>>
>>>> This fix improves performance of ColorConvertOp.filter()
>>>> operation in case of lcms. The fix can be divided into
>>>> three separate changes:
>>>>
>>>> * provide support for custom component images in
>>>> LCMSImageLayout.
>>>> This change affects the case of conversion to default
>>>> destination.
>>>>
>>>> * provide a way to process whole image, instead of
>>>> scan-by-scan processing, if both source and
>>>> destination images do not contain padding samples
>>>> (i.e. next scan starts immediately after previous).
>>>>
>>>> * increase optimization level for lcms library from
>>>> LOW to HIGHEST. This change affects both new and
>>>> standard build systems.
>>>>
>>>> A benchmark comparison below illustrates the increase
>>>> of performance:
>>>>
>>>> Options common across all tests:
>>>> testname=cmm.colorconv.ccop.op_img
>>>> cmm.colorconv.ccop.ccopOptions.srcType=BYTE_3BYTE_BGR
>>>> cmm.colorconv.ccop.ccopOptions.content=photo
>>>> cmm.opts.profiles=1001
>>>>
>>>> cmm.colorconv.ccop.ccopOptions.dstType=BYTE_3BYTE_BGR,cmm.colorconv.ccop.ccopOptions.size=250:
>>>>
>>>> OpenJDK Baseline: 0.152008134 (var=1.02%) (100.0%)
>>>> *************************************|
>>>> *************************************|
>>>> *************************************|
>>>> OpenJDK Fix: 0.245116358 (var=2.0%) (161.25%)
>>>> *************************************|**********************
>>>> *************************************|**********************
>>>> *************************************|*********************
>>>> cmm.colorconv.ccop.ccopOptions.dstType=BYTE_3BYTE_BGR,cmm.colorconv.ccop.ccopOptions.size=4000:
>>>>
>>>> OpenJDK Baseline: 0.914826498 (var=0.41%) (100.0%)
>>>> ***************************|
>>>> ***************************|
>>>> ***************************|
>>>> OpenJDK Fix: 0.001310043 (var=2.24%) (221.48%)
>>>> ***************************|********************************
>>>> ***************************|********************************
>>>> ***************************|********************************
>>>> cmm.colorconv.ccop.ccopOptions.dstType=COMPATIBLE_DST,cmm.colorconv.ccop.ccopOptions.size=250:
>>>>
>>>> OpenJDK Baseline: 0.060737151 (var=1.57%) (100.0%)
>>>> ***************|
>>>> ***************|
>>>> ***************|
>>>> OpenJDK Fix: 0.242736486 (var=1.53%) (399.65%)
>>>> ***************|*******************************************
>>>> ***************|********************************************
>>>> ***************|********************************************
>>>> cmm.colorconv.ccop.ccopOptions.dstType=COMPATIBLE_DST,cmm.colorconv.ccop.ccopOptions.size=4000:
>>>>
>>>> OpenJDK Baseline: 0.559251559 (var=0.7%) (100.0%)
>>>> *******|
>>>> *******|
>>>> *******|
>>>> OpenJDK Fix: 0.001306904 (var=1.05%) (838.16%)
>>>> *******|***************************************************
>>>> *******|****************************************************
>>>> *******|****************************************************
>>>>
>>>> Summary:
>>>> OpenJDK Baseline:
>>>> Number of tests: 4
>>>> Overall average: 0.053373173444166644
>>>> Best spread: 0.41% variance
>>>> Worst spread: 1.57% variance
>>>> (Basis for results comparison)
>>>>
>>>> OpenJDK Fix:
>>>> Number of tests: 4
>>>> Overall average: 0.12261744826137347
>>>> Best spread: 1.05% variance
>>>> Worst spread: 2.24% variance
>>>> Comparison to basis:
>>>> Best result: 838.16% of basis
>>>> Worst result: 161.25% of basis
>>>> Number of wins: 4
>>>> Number of ties: 0
>>>> Number of losses: 0
>>>>
>>>> Thanks,
>>>> Andrew
>>>>
>>>
>>
>
More information about the 2d-dev
mailing list