[webkit-qt] QtTestBrowser slowness when launched from visual studio

Mon Dec 13 18:29:05 PST 2010

Some more strange stuff that I noticed. It seems if I visit yahoo.com first in the QtTestBrowser, the test results improve noticeably. In particular, the string subset of test suites get a major boost. I have repeated following like 10 times so I am sure it is not a random thing. Following are my results.

Test result directly landing at sunspider page when launching application in visual studio with debugging enabled -

http://www2.webkit.org/perf/sunspider-0.9.1/sunspider-0.9.1/results.html?{"v": "sunspider-0.9.1", "3d-cube":[75,79,65,71,80,65,76,68,70,62],"3d-morph":[15,14,14,14,14,17,15,15,15,14],"3d-raytrace":[31,32,31,30,31,31,31,31,31,31],"access-binary-trees":[6,6,6,7,5,6,6,6,5,6],"access-fannkuch":[18,18,18,19,18,18,18,18,18,19],"access-nbody":[12,13,13,13,13,13,13,13,13,13],"access-nsieve":[5,6,6,5,5,6,5,5,5,5],"bitops-3bit-bits-in-byte":[4,4,4,5,4,4,4,4,4,4],"bitops-bits-in-byte":[7,6,7,7,7,7,6,7,7,6],"bitops-bitwise-and":[9,9,9,9,9,9,9,9,9,9],"bitops-nsieve-bits":[10,9,9,9,9,9,9,9,10,10],"controlflow-recursive":[4,4,4,4,4,4,4,3,4,4],"crypto-aes":[31,29,29,28,29,29,29,29,28,30],"crypto-md5":[5,5,5,5,6,5,5,6,5,6],"crypto-sha1":[5,5,4,4,4,4,5,4,4,4],"date-format-tofte":[38,39,38,39,38,40,39,39,39,39],"date-format-xparb":[27,28,27,27,27,27,27,27,27,27],"math-cordic":[9,9,9,9,9,9,9,9,9,9],"math-partial-sums":[14,15,14,15,14,15,15,14,15,14],"math-spectral-norm":[7,7,8,7,7,8,8,7,7,7],"regexp-dna":[14,14,14,14,14,14,14,14,15,14],"string-base64":[42,42,40,42,40,42,40,39,38,43],"string-fasta":[38,38,38,39,38,38,38,38,37,38],"string-tagcloud":[234,123,89,190,114,122,87,135,88,98],"string-unpack-code":[144,155,135,162,126,125,153,147,128,140],"string-validate-input":[62,58,69,65,76,61,73,61,62,70]}

string:               414.0ms +/- 9.3%
    base64:              40.8ms +/- 2.8%
    fasta:                  38.0ms +/- 0.9%
    tagcloud:           128.0ms +/- 27.0% (Note the 27% deviation here)
    unpack-code:        141.5ms +/- 6.5%
    validate-input:      65.7ms +/- 6.5%

Test result after hitting yahoo.com first and then launching the application in visual studio with debugging enabled (no such gain experienced when visiting google.com first) -

http://www2.webkit.org/perf/sunspider-0.9.1/sunspider-0.9.1/results.html?{"v": "sunspider-0.9.1", "3d-cube":[39,39,38,38,39,40,41,40,39,38],"3d-morph":[14,14,14,14,13,14,14,15,13,14],"3d-raytrace":[28,28,28,28,28,27,28,29,28,28],"access-binary-trees":[6,6,5,7,5,5,6,6,6,5],"access-fannkuch":[18,19,18,19,19,18,18,0,18,19],"access-nbody":[13,12,13,13,12,12,12,13,13,13],"access-nsieve":[5,5,5,5,5,6,5,5,5,5],"bitops-3bit-bits-in-byte":[4,4,4,5,5,5,5,5,5,4],"bitops-bits-in-byte":[6,7,7,7,7,7,7,7,7,7],"bitops-bitwise-and":[5,4,4,6,5,5,5,5,6,5],"bitops-nsieve-bits":[10,10,9,9,10,10,10,9,9,10],"controlflow-recursive":[3,4,4,4,4,4,4,4,4,4],"crypto-aes":[25,25,25,25,25,25,25,25,25,24],"crypto-md5":[5,6,5,5,6,5,5,5,5,6],"crypto-sha1":[5,4,5,4,4,4,4,4,4,4],"date-format-tofte":[39,37,37,38,37,37,37,38,37,37],"date-format-xparb":[27,26,26,27,27,27,26,26,26,27],"math-cordic":[9,9,9,9,9,8,9,9,9,9],"math-partial-sums":[14,14,15,14,14,14,15,14,14,15],"math-spectral-norm":[7,8,8,7,7,8,7,7,7,7],"regexp-dna":[14,15,14,15,14,15,14,15,14,14],"string-base64":[24,25,24,25,24,25,24,24,24,24],"string-fasta":[35,35,35,35,35,35,35,36,35,35],"string-tagcloud":[39,39,39,39,39,39,39,40,39,39],"string-unpack-code":[85,87,84,84,84,87,87,90,86,88],"string-validate-input":[39,39,39,39,39,39,39,39,41,39]}

string:               223.9ms +/- 0.8%
    base64:              24.3ms +/- 1.4%
    fasta:               35.1ms +/- 0.6%
    tagcloud:            39.1ms +/- 0.6%
    unpack-code:         86.2ms +/- 1.6%
    validate-input:      39.2ms +/- 1.2%

Test results when launching the application without  debugging enabled, the fastest case (loading test suite straight up) -

http://www2.webkit.org/perf/sunspider-0.9.1/sunspider-0.9.1/results.html?{"v": "sunspider-0.9.1", "3d-cube":[21,20,19,19,22,19,19,19,19,19],"3d-morph":[14,14,14,14,14,13,14,14,14,14],"3d-raytrace":[20,21,20,20,21,21,22,21,20,21],"access-binary-trees":[5,5,4,4,4,5,5,6,5,6],"access-fannkuch":[18,18,18,18,18,18,19,18,20,18],"access-nbody":[13,12,12,12,12,12,12,12,12,12],"access-nsieve":[5,5,5,5,5,5,4,5,5,5],"bitops-3bit-bits-in-byte":[4,4,5,5,4,4,4,4,4,4],"bitops-bits-in-byte":[7,6,6,7,7,7,7,6,7,6],"bitops-bitwise-and":[4,5,4,5,5,5,5,5,5,4],"bitops-nsieve-bits":[9,9,10,9,10,10,9,9,9,9],"controlflow-recursive":[4,3,4,4,4,4,4,4,4,4],"crypto-aes":[13,15,14,14,14,14,14,13,14,14],"crypto-md5":[5,5,4,4,4,5,5,5,4,4],"crypto-sha1":[4,4,4,3,4,3,3,4,3,4],"date-format-tofte":[24,24,24,23,24,23,24,24,24,24],"date-format-xparb":[20,20,21,20,20,20,20,20,20,20],"math-cordic":[9,9,8,9,9,8,8,8,8,9],"math-partial-sums":[15,15,14,15,14,15,14,14,15,15],"math-spectral-norm":[7,7,7,7,7,7,8,7,7,7],"regexp-dna":[14,14,14,14,14,14,14,14,14,14],"string-base64":[15,15,16,16,16,16,15,16,15,15],"string-fasta":[17,23,23,24,24,24,23,24,23,23],"string-tagcloud":[26,26,26,28,27,26,27,26,26,28],"string-unpack-code":[47,48,46,45,47,48,49,48,46,45],"string-validate-input":[23,24,22,22,22,23,24,23,23,21]}

string:               134.5ms +/- 1.6%
    base64:              15.5ms +/- 2.4%
    fasta:               22.8ms +/- 6.6%
    tagcloud:            26.6ms +/- 2.3%
    unpack-code:         46.9ms +/- 2.1%
    validate-input:      22.7ms +/- 3.0%

From: webkit-qt-bounces at lists.webkit.org [mailto:webkit-qt-bounces at lists.webkit.org] On Behalf Of Baldeva, Arpit
Sent: Monday, December 13, 2010 5:39 PM
To: webkit-qt at lists.webkit.org
Subject: [webkit-qt] QtTestBrowser slowness when launched from visual studio

Hi,

I have QtTestBrowser building inside visual studio with the help provided earlier on this forum. I am building the "release" binary. Today, I noticed that the sunspider JavaScript benchmark performance results are twice as slow when I launch the executable from inside visual studio(Start Debugging - F5 command). The test suite finishes in about 800 ms. Note that when launched from inside visual studio without debugging(Start without debugging - Ctrl+F5), the performance is quite nice (The test suite finishes in about 380 ms). My binary is same in both the cases(release). While I expect some performance hit when launching with debugging, I don't expect the performance to drop in half. Is this is a known thing?

So far I have tried following.

*         Disabled exceptions.

*         Added _NO_DEBUG_HEAP=1 in all the projects and rebuilt. I have already defined QT_NO_DEBUG and NDEBUG(That came by default in generated solution).

Any help is appreciated.

Thanks.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-qt/attachments/20101213/d65dfe71/attachment-0001.html>