lroundf_gen.S 3.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138
  1. /*
  2. * Math library
  3. *
  4. * Copyright (C) 2016 Intel Corporation. All rights reserved.
  5. *
  6. * Redistribution and use in source and binary forms, with or without
  7. * modification, are permitted provided that the following conditions
  8. * are met:
  9. *
  10. * * Redistributions of source code must retain the above copyright
  11. * notice, this list of conditions and the following disclaimer.
  12. * * Redistributions in binary form must reproduce the above copyright
  13. * notice, this list of conditions and the following disclaimer in
  14. * the documentation and/or other materials provided with the
  15. * distribution.
  16. * * Neither the name of Intel Corporation nor the names of its
  17. * contributors may be used to endorse or promote products derived
  18. * from this software without specific prior written permission.
  19. *
  20. * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  21. * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  22. * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
  23. * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
  24. * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  25. * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  26. * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  27. * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  28. * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  29. * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  30. * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  31. *
  32. *
  33. * Author Name <jingwei.zhang@intel.com>
  34. * History:
  35. * 03-14-2016 Initial version. numerics svn rev. 12864
  36. */
  37. .file "lroundf_gen.c"
  38. .text
  39. ..TXTST0:
  40. # -- Begin lroundf
  41. .text
  42. .align 16,0x90
  43. .globl lroundf
  44. lroundf:
  45. # parameter 1: %xmm0
  46. ..B1.1:
  47. .cfi_startproc
  48. ..___tag_value_lroundf.1:
  49. ..L2:
  50. subq $24, %rsp
  51. .cfi_def_cfa_offset 32
  52. movss %xmm0, 16(%rsp)
  53. movq $0, (%rsp)
  54. movl $0, 8(%rsp)
  55. ..B1.2:
  56. stmxcsr 8(%rsp)
  57. movsd 16(%rsp), %xmm1
  58. xorpd %xmm2, %xmm2
  59. andl $-2, 8(%rsp)
  60. ldmxcsr 8(%rsp)
  61. cvtss2si %xmm1, %rax
  62. addss %xmm1, %xmm2
  63. movq %rax, (%rsp)
  64. stmxcsr 8(%rsp)
  65. ..B1.3:
  66. testb $1, 8(%rsp)
  67. jne ..B1.6
  68. ..B1.4:
  69. pxor %xmm2, %xmm2
  70. movq (%rsp), %rax
  71. cvtsi2ssq %rax, %xmm2
  72. movss 16(%rsp), %xmm1
  73. movaps %xmm1, %xmm0
  74. subss %xmm2, %xmm0
  75. andps .L_2il0floatpacket.1(%rip), %xmm0
  76. ucomiss .L_2il0floatpacket.0(%rip), %xmm0
  77. jp ..B1.5
  78. je ..B1.7
  79. ..B1.5:
  80. addq $24, %rsp
  81. .cfi_def_cfa_offset 8
  82. ret
  83. .cfi_def_cfa_offset 32
  84. ..B1.6:
  85. movq (%rsp), %rax
  86. addq $24, %rsp
  87. .cfi_def_cfa_offset 8
  88. ret
  89. .cfi_def_cfa_offset 32
  90. ..B1.7:
  91. pxor %xmm0, %xmm0
  92. comiss %xmm0, %xmm1
  93. jbe ..B1.10
  94. ..B1.8:
  95. comiss %xmm2, %xmm1
  96. jbe ..B1.5
  97. ..B1.9:
  98. incq %rax
  99. movq %rax, (%rsp)
  100. addq $24, %rsp
  101. .cfi_def_cfa_offset 8
  102. ret
  103. .cfi_def_cfa_offset 32
  104. ..B1.10:
  105. comiss %xmm1, %xmm2
  106. jbe ..B1.5
  107. ..B1.11:
  108. decq %rax
  109. movq %rax, (%rsp)
  110. addq $24, %rsp
  111. .cfi_def_cfa_offset 8
  112. ret
  113. .align 16,0x90
  114. .cfi_endproc
  115. .type lroundf,@function
  116. .size lroundf,.-lroundf
  117. .data
  118. # -- End lroundf
  119. .section .rodata, "a"
  120. .align 16
  121. .align 16
  122. .L_2il0floatpacket.1:
  123. .long 0x7fffffff,0x00000000,0x00000000,0x00000000
  124. .type .L_2il0floatpacket.1,@object
  125. .size .L_2il0floatpacket.1,16
  126. .align 4
  127. .L_2il0floatpacket.0:
  128. .long 0x3f000000
  129. .type .L_2il0floatpacket.0,@object
  130. .size .L_2il0floatpacket.0,4
  131. .data
  132. .section .note.GNU-stack, ""
  133. // -- Begin DWARF2 SEGMENT .eh_frame
  134. .section .eh_frame,"a",@progbits
  135. .eh_frame_seg:
  136. .align 1
  137. # End