starpu_sgemm_gemm.attila 4.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150
  1. ##################
  2. # Performance Model Version
  3. 45
  4. ####################
  5. # COMBs
  6. # number of combinations
  7. 4
  8. ####################
  9. # COMB_3
  10. # number of types devices
  11. 1
  12. ####################
  13. # DEV_0
  14. # device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3)
  15. 0
  16. ####################
  17. # DEV_0
  18. # device id
  19. 0
  20. ####################
  21. # DEV_0
  22. # number of cores
  23. 1
  24. ##########
  25. # number of implementations
  26. 1
  27. #####
  28. # Model for cpu0_impl0 (Comb3)
  29. # number of entries
  30. 4
  31. # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx
  32. 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0
  33. # a b c
  34. nan nan nan
  35. # not multiple-regression-base
  36. 0
  37. # hash size flops mean (us) dev (us) sum sum2 n
  38. 492beed5 33177600 7.077888e+09 3.328725e+05 1.185902e+04 2.563119e+07 8.542747e+12 77
  39. 24c84a50 11059200 1.769472e+09 8.321812e+04 2.964755e+03 6.407798e+06 5.339217e+11 77
  40. 0b0b0ce8 3686400 2.621440e+08 1.421718e+04 3.409134e+02 9.098993e+05 1.294364e+10 64
  41. 4220e23d 14745600 2.097152e+09 1.008105e+05 2.361630e+03 8.064841e+06 8.134670e+11 80
  42. ####################
  43. # COMB_0
  44. # number of types devices
  45. 1
  46. ####################
  47. # DEV_0
  48. # device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3)
  49. 1
  50. ####################
  51. # DEV_0
  52. # device id
  53. 0
  54. ####################
  55. # DEV_0
  56. # number of cores
  57. 1
  58. ##########
  59. # number of implementations
  60. 1
  61. #####
  62. # Model for cuda0_impl0 (Comb0)
  63. # number of entries
  64. 4
  65. # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx
  66. 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0
  67. # a b c
  68. nan nan nan
  69. # not multiple-regression-base
  70. 0
  71. # hash size flops mean (us) dev (us) sum sum2 n
  72. 492beed5 33177600 7.077888e+09 1.123499e+04 6.785566e+01 1.190909e+06 1.338033e+10 106
  73. 24c84a50 11059200 1.769472e+09 2.808747e+03 1.696392e+01 2.977272e+05 8.362706e+08 106
  74. 0b0b0ce8 3686400 2.621440e+08 6.738679e+02 4.393713e+01 6.873452e+04 4.651489e+07 102
  75. 4220e23d 14745600 2.097152e+09 5.557425e+03 3.241733e+02 5.835297e+05 3.253957e+09 105
  76. ####################
  77. # COMB_2
  78. # number of types devices
  79. 1
  80. ####################
  81. # DEV_0
  82. # device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3)
  83. 1
  84. ####################
  85. # DEV_0
  86. # device id
  87. 1
  88. ####################
  89. # DEV_0
  90. # number of cores
  91. 1
  92. ##########
  93. # number of implementations
  94. 1
  95. #####
  96. # Model for cuda1_impl0 (Comb2)
  97. # number of entries
  98. 4
  99. # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx
  100. 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0
  101. # a b c
  102. nan nan nan
  103. # not multiple-regression-base
  104. 0
  105. # hash size flops mean (us) dev (us) sum sum2 n
  106. 492beed5 33177600 7.077888e+09 1.123077e+04 9.504466e+01 1.179231e+06 1.324463e+10 105
  107. 24c84a50 11059200 1.769472e+09 2.807693e+03 2.376116e+01 2.948078e+05 8.277894e+08 105
  108. 0b0b0ce8 3686400 2.621440e+08 6.672056e+02 3.376608e+01 6.805497e+04 4.552295e+07 102
  109. 4220e23d 14745600 2.097152e+09 5.553764e+03 3.500896e+02 5.831453e+05 3.251521e+09 105
  110. ####################
  111. # COMB_1
  112. # number of types devices
  113. 1
  114. ####################
  115. # DEV_0
  116. # device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3)
  117. 1
  118. ####################
  119. # DEV_0
  120. # device id
  121. 2
  122. ####################
  123. # DEV_0
  124. # number of cores
  125. 1
  126. ##########
  127. # number of implementations
  128. 1
  129. #####
  130. # Model for cuda2_impl0 (Comb1)
  131. # number of entries
  132. 4
  133. # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx
  134. 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0
  135. # a b c
  136. nan nan nan
  137. # not multiple-regression-base
  138. 0
  139. # hash size flops mean (us) dev (us) sum sum2 n
  140. 492beed5 33177600 7.077888e+09 1.124174e+04 2.629960e+01 1.180383e+06 1.326963e+10 105
  141. 24c84a50 11059200 1.769472e+09 2.810435e+03 6.574900e+00 2.950958e+05 8.293519e+08 105
  142. 0b0b0ce8 3686400 2.621440e+08 6.002221e+02 2.259043e+01 6.242310e+04 3.752080e+07 104
  143. 4220e23d 14745600 2.097152e+09 5.577722e+03 1.615194e+02 5.912385e+05 3.300529e+09 106