starpu_sgemm_gemm.mirage 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. 0b0b0ce8 3686400 2.621440e+08 1.352609e+04 3.616534e+02 1.082087e+06 1.464687e+10 80
  39. 492beed5 33177600 7.077888e+09 3.550396e+05 8.949994e+03 2.840317e+07 1.009066e+13 80
  40. 24c84a50 11059200 1.769472e+09 8.875990e+04 2.237499e+03 7.100792e+06 6.306662e+11 80
  41. 4220e23d 14745600 2.097152e+09 1.078112e+05 1.983800e+03 8.624897e+06 9.301755e+11 80
  42. ####################
  43. # COMB_1
  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 (Comb1)
  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. 0b0b0ce8 3686400 2.621440e+08 6.589631e+02 8.406511e+00 6.787320e+04 4.473321e+07 103
  73. 492beed5 33177600 7.077888e+09 1.151398e+04 9.050114e+01 1.220482e+06 1.405348e+10 106
  74. 24c84a50 11059200 1.769472e+09 2.878495e+03 2.262529e+01 3.051205e+05 8.783425e+08 106
  75. 4220e23d 14745600 2.097152e+09 5.574713e+03 3.353004e+02 5.909196e+05 3.306125e+09 106
  76. ####################
  77. # COMB_0
  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 (Comb0)
  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. 0b0b0ce8 3686400 2.621440e+08 6.663664e+02 8.616537e+01 6.796937e+04 4.604980e+07 102
  107. 492beed5 33177600 7.077888e+09 1.150036e+04 8.404527e+01 1.207538e+06 1.388786e+10 105
  108. 24c84a50 11059200 1.769472e+09 2.875090e+03 2.101132e+01 3.018845e+05 8.679912e+08 105
  109. 4220e23d 14745600 2.097152e+09 5.579034e+03 3.672012e+02 5.857985e+05 3.282348e+09 105
  110. ####################
  111. # COMB_2
  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 (Comb2)
  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. 0b0b0ce8 3686400 2.621440e+08 6.181769e+02 5.174143e+01 6.181769e+04 3.848198e+07 100
  141. 492beed5 33177600 7.077888e+09 1.148096e+04 7.289415e+01 1.205501e+06 1.384086e+10 105
  142. 24c84a50 11059200 1.769472e+09 2.870240e+03 1.822354e+01 3.013752e+05 8.650538e+08 105
  143. 4220e23d 14745600 2.097152e+09 5.580581e+03 3.970717e+02 5.859610e+05 3.286558e+09 105