Friday, February 15, 2008

Parallel extensions for .NET framework



URL
XMLFILE

375042_ParallelFX_POV_Streaming.wmv d7a1e37d-9f86-4c89-a734-cfe8ce6efdb4 -4.0827494 welcome -1.42807705 jeff barnes 3.1667869 evangelist 6.4494649 parallelfx 8.8850851 quick agenda 11.45312135 part of series 15.420432 architects point of view 19.6069286 upcoming technologies 23.32517125 parallelfx extensions 26.4503005 leverage this 28.95114635 environment 31.79643315 do more with less 34.7357631 ubiquitous 37.8285386 multicore 40.20496035 couple of demos 42.2029479 plinq 45.3916633 using tpl 48.3632994 task parallel 51.0496316 library 52.98582205 download 54.73504655 ctp 58.0801633 point of view 61.52069165 architecting large projects 64.3629173 balancing 66.1119252 resources 69.3949683 adopting new technology 73.9563799 landscape 76.80207035 tearing apart 79.42715315 risks 81.8322903 mitigate them 86.9303776 impact on 90.2425533 future architecture 94.27495695 processing performance 97.02214175 gains 98.55528465 multicore 102.4018878 managed library 105.7413912 extension 109.05814175 past this type of 113.14995355 low level operations 118.4652699 specific libraries 121.68270865 intel and so on 124.21420075 multicore 127.0301418 lot easier 128.84037195 in code 132.5283232 see what amounts to 136.06302275 very very easy 139.06102955 data parallelism 141.24582205 plinq 142.71456015 parallel 145.935559 linq implementation 150.7829445 execute queries 154.78162135 first question 157.4398334 only one processor 161.1587075 gracefully degrade 165.56445695 sequential query 168.65665875 imperative 171.15919735 data parallelism 173.59547625 operation 175.6271803 for foreach 178.88034135 divide work in loop 185.2533118 imperative task parallelism 189.6602767 data driven 192.41057375 lightweight tasks 195.59876765 running in parallel 198.5381168 December ctp 202.4423175 since ctp 207.00627665 really just preview 211.1636361 ability to shape 214.88476085 Microsoft is planning 217.85320075 download 219.88410435 extension 222.60419395 feedback 224.0719604 questions 228.6984989 lets take couple example 232.29011115 parallel linq 236.5090239 simple c# linq 240.82264065 select 242.85492295 parallelism 245.5309479 all do to code 249.38785945 add .asparallel 253.66998755 one simple line edit of code 257.9200216 parallelism across 262.07962475 impact to existing code base 265.5754173 very minimal 268.82731865 task parallel library 272.4851463 incorporate 275.2037994 sequential 277.05008055 for next 279.45421545 standard 281.2050182 iterating 283.42344335 set of results 285.4856089 parallels 290.45429145 parallel. in front 294.86284925 executed in parallel 298.55224155 all cores of machine 302.01819055 using tpl loop over 305.89603405 ienumerable object 310.210186 top standard foreach 313.7413515 myclass 317.3333606 parallel. in front 321.42972115 foreach keyword 324.4926395 highlights 328.0564184 minimal impact 331.52492745 incorporate 334.1482801 classic 336.8672597 divide conquer 340.90226195 running through tree 345.46316555 left right 348.30989575 parallel. front of do 352.37271775 split up tasks 356.09058395 splitting things up 358.77608285 multiple cores 362.0595794 demonstration 366.06026985 count number of 369.2787268 valid isbn 372.8731452 three repetitions 376.1549887 time for processing 378.2169059 10000 379.7190749 100000 381.8438583 1000000 385.6578742 processing improvements 388.90592635 drop out powerpoint 391.50027445 code behind 394.4395057 this one of code samples 397.595881 download library 399.7829422 main routing 402.0675318 start up 405.97174495 three different 409.9076191 10000 up to million 413.2533867 sequential 415.1919082 stop watch 417.59923705 sample size 420.44409755 code above 422.50376195 implement 425.00516445 iterations 427.1630658 only change in code 430.6308334 adding as parallel 435.75652275 existing return 440.13356015 console application 442.54131415 run exe 444.38056245 live here 446.66491275 results come out 449.290169 three iterations 452.35134925 each sample size 455.5735057 interesting is 459.3528402 sequential 6ms 463.76198875 parallel 307ms 467.75947625 initial iteration 472.88579485 much higher startup time 479.01251595 parallelism kicks in 482.2629445 better performance 485.1699989 processing million 488.7032132 sequential 303 491.9539173 parallel 148 495.01397625 roughly half 497.0172381 dual core 499.29445585 processor laptop 502.86244565 results get new 506.54640025 framework extension 509.6698708 compiled 512.6726667 dual core 514.5155057 quad core 516.9874309 both machines 519.73655675 first iteration 523.26634015 higher startup time 528.1122268 2x performance gain 532.2703379 4x performance gain 534.4883107 example 537.0516395 results excel chart 540.80328575 orange sequential 544.58610775 light green parallel 549.11875065 example using quad core 554.86523475 4x performance gain 558.39834245 taking advantage\ 561.40119055 multicore 563.7418969 here another visual 567.1512234 demonstration 571.12034695 ray tracer demo 574.7472574 graphically intense 578.21399095 image on screen 580.9332166 divide conquer 584.5881996 drop out 587.43111225 fire up example 591.68638665 run this 594.02814065 clear out of screen 597.403627 first sequential mode 599.99782885 render 602.5932064 top to bottom 605.1212994 goes through 607.34239235 calculating 609.74805785 render this image 613.27883795 results down bottom 618.03236285 18 seconds 621.75054655 sequential processing 625.377627 process this parallel 628.0006089 render 629.3122631 couple 632.0327608 processing image 635.1592631 two places 637.75365655 simultaneously 641.31643545 finishes 645.19051025 starting middle 648.50515085 rendering top 655.2599887 results 13.46 seconds 660.911771 increase 663.4436996 performance 666.6967597 very visual perspective 669.507525 processing 673.5085341 how attacking problem 679.19594675 rough 27% 682.41596605 gain 685.447279 quad core about 53% gain 689.1036996 essential 691.102415 visually 692.8882359 watching 695.76058395 4 different places 699.63550915 parallel fx library 703.41983565 very visual feedback 708.2945397 graphically intensive 712.07376875 resources 717.26457715 huge impact design architectures 725.44811115 two really good 729.7030261 msdn articles 733.0769275 plinq 736.5170114 task parallel library 739.42074265 team blog 743.3301361 .net framework team 746.64464635 connect site 749.0496089 feedback 750.39058165 recap 752.11309985 adopting 754.20499895 early release 758.45928805 lot advantages 761.2998141 processing 763.4908368 multicore machines 766.04894105 process 768.73819055 large amounts of data 772.4593878 more prevalent 775.27144905 software 778.24154995 finally coming 780.95672115 speed 782.9602744 Moore’s law 786.83559525 software advances 791.7112268 take look parallelfx 795.6165579 .net extensions 798.77643085 less complexity 804.37177105 thank you 807.77509985 blog

No comments: