Announcement

Collapse
No announcement yet.

22-Way AMD/NVIDIA OpenCL Linux Benchmarks To Start Off 2015

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Here's somthing interesting: Here's my benchmark of Pyrit on an ASUS G750JM with a Maxwell-based NVIDIA GeForce GTX 860M 2GB and Intel HD Graphics 4600 (optimus enabled) on Arch Linux.


    Pyrit 0.4.0 (C) 2008-2011 Lukas Lueg http://pyrit.googlecode.com
    This code is distributed under the GNU General Public License v3+

    Calibrating...

    Running benchmark (43241.5 PMKs/s)... -

    Computed 43241.47 PMKs/s total.
    #1: 'OpenCL-Device 'Intel(R) HD Graphics Haswell GT2 Mobile'': 3786.0 PMKs/s (RTT 3.1)
    #2: 'OpenCL-Device 'GeForce GTX 860M'': 40141.5 PMKs/s (RTT 2.9)
    #3: 'CPU-Core (SSE2)': 592.3 PMKs/s (RTT 2.6)
    #4: 'CPU-Core (SSE2)': 478.5 PMKs/s (RTT 3.1)
    #5: 'CPU-Core (SSE2)': 456.3 PMKs/s (RTT 3.3)
    #6: 'CPU-Core (SSE2)': 476.8 PMKs/s (RTT 3.3)
    #7: 'CPU-Core (SSE2)': 457.5 PMKs/s (RTT 3.3)
    #8: 'CPU-Core (SSE2)': 471.7 PMKs/s (RTT 3.1)

    These results are done with the OpenCL version, CUDA 6.5 installed. Intel's Beignet was also installed.

    With CUDA, these are the results:

    Pyrit 0.4.0 (C) 2008-2011 Lukas Lueg http://pyrit.googlecode.com
    This code is distributed under the GNU General Public License v3+

    Calibrating...
    Running benchmark (29808.8 PMKs/s)... \

    Computed 29808.82 PMKs/s total.
    #1: 'CUDA-Device #1 'GeForce GTX 860M'': 28350.3 PMKs/s (RTT 2.8)
    #2: 'CPU-Core (SSE2)': 521.1 PMKs/s (RTT 3.1)
    #3: 'CPU-Core (SSE2)': 524.2 PMKs/s (RTT 3.0)
    #4: 'CPU-Core (SSE2)': 509.9 PMKs/s (RTT 3.0)
    #5: 'CPU-Core (SSE2)': 516.2 PMKs/s (RTT 3.1)
    #6: 'CPU-Core (SSE2)': 516.6 PMKs/s (RTT 3.0)
    #7: 'CPU-Core (SSE2)': 516.3 PMKs/s (RTT 3.0)
    #8: 'CPU-Core (SSE2)': 508.0 PMKs/s (RTT 3.1)

    Here, I had disabled Intel's Beignet (after upgrading the Linux kernel to 4.0.4-2-ARCH) due to Haswell related errors with Beignet, see this and this , more details on the same here

    Below are my results after upgrading to CUDA 7.0 with the latest NVIDIA driver (as of today):

    Pyrit 0.4.0 (C) 2008-2011 Lukas Lueg http://pyrit.googlecode.com
    This code is distributed under the GNU General Public License v3+

    Calibrating...
    Running benchmark (41515.0 PMKs/s)... -

    Computed 41515.03 PMKs/s total.
    #1: 'OpenCL-Device 'GeForce GTX 860M'': 41442.0 PMKs/s (RTT 2.9)
    #2: 'CPU-Core (SSE2/AES)': 509.9 PMKs/s (RTT 3.1)
    #3: 'CPU-Core (SSE2/AES)': 519.2 PMKs/s (RTT 3.1)
    #4: 'CPU-Core (SSE2/AES)': 522.6 PMKs/s (RTT 3.0)
    #5: 'CPU-Core (SSE2/AES)': 515.8 PMKs/s (RTT 3.1)
    #6: 'CPU-Core (SSE2/AES)': 495.6 PMKs/s (RTT 3.1)
    #7: 'CPU-Core (SSE2/AES)': 501.0 PMKs/s (RTT 3.5)
    #8: 'CPU-Core (SSE2/AES)': 511.8 PMKs/s (RTT 3.0)

    Conclusion: Pyrit is MUCH faster with OpenCL on Nvidia Maxwell Gen 1 (GM107 ) than with CUDA.

    Regards,

    Brainiarc7.


    Comment

    Working...
    X