projects / platform / upstream / ne10.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b79e33b) | patch
Tuning floating point RFFT for GCC 4.9.0
authorPhil.Wang <phil.wang@arm.com>
Fri, 13 Mar 2015 10:10:26 +0000 (18:10 +0800)
committerPhil.Wang <phil.wang@arm.com>
Wed, 18 Mar 2015 10:23:18 +0000 (18:23 +0800)
commit97dd73d886ad75a11f5121bf72ccc3dcd28d1e2a
tree45ce540748bbce0d7bc90ca58eff04467c94627ftree | snapshot
parentb79e33bb5bba2d02a7b0cfdc1b163a71db7239eccommit | diff
Tuning floating point RFFT for GCC 4.9.0

    Cortex-A53 (1.69GHz)
    GCC 4.9.0, with -O2
    Android-L, AArch64

    |   R2C FFT Time Cost in ms|
    |size|Ne10|pffft|pffft/Ne10|
    |  32| 118|  254|      215%|
    |  64| 126|  198|      157%|
    | 128| 109|  177|      162%|
    | 256| 126|  154|      122%|
    | 512| 122|  165|      135%|
    |1024| 143|  162|      113%|
    |2048| 153|  188|      123%|

    The larger the last column is, the faster Ne10 is.

Change-Id: I8921fc83afb8c7307ffd0fcb2a4bb1a88b349339
modules/dsp/NE10_rfft_float32.neonintrinsic.c diff | blob | history
Domain: Machine Learning / ML Framework;