Announcement

Collapse
No announcement yet.

Suggested article: RoCM vs normal OpenCL hash rates cryptocurrencies benchmarks

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

  • Suggested article: RoCM vs normal OpenCL hash rates cryptocurrencies benchmarks

    I said in the title I would like to see RocM vs OpenCL benchmarks of Hash rates on cryptocurrencies like ethereum or so.

  • #2
    I do have Ethereum / ethminer available via PTS. Do other cryptocurrencies have automated benchmark modes available? If so, please provide links/references and could then possibly make it a reality.
    Michael Larabel
    http://www.michaellarabel.com/

    Comment


    • #3
      I couldn't make claymore etherminer work on RocM, it just doesn't work it doesnt detect the opencl interface. But you could do a benchmark of just ethminer compared to the most common ethereum graphics card 470/480/460 and 560/570/580 and a performance per watt/dollar. And maybe when Vega is coming it will help because Vega 56 is said to have a 70megahash rate, and then you could compare this results with Vega. MIning is very important for AMD , that leverage their sales and there was a shortage of supplies, now you can get a rx480 cheaper than a Rx470 because it has the same hash rate and the rx480 is more power hungry. In the upcoming days of Vega releases if it has the hash rate they say there will be a shortage of supplies. IMHO. And most miners are running linux it has a performance advantage over windows. Again IMHO the mining community is huge, linux friendly and eager to buy new hardware.
      Last edited by Kayote; 08-08-2017, 10:22 PM.

      Comment


      • #4
        https://www.eteknix.com/amd-radeon-r...tes-calculated
        calculation on vega hash rate:
        Applying this to Vega 56, it converts to 409.6GB/s divided by 7.8GB giving a hash rate of 52.5 MH/s. For the RX Vega 64, it is 483.8GB/s divided by 7.8GB which results in approximately 63.4 MH/s.
        so for Vega 56 hash rate is double of the most used ethereum card (rx 470) and for Vega 64 is 2.5 times.

        Comment


        • #5
          Most miners are using Pentium dual Core or just Celerons on old motherboards. Rocm puts a lots of thing into the table SenseMI and other technologies only found in i7 and most miners use pci express 2 and rocm is pci express 3, in the long run RoCM should outperform normal amdgpu pro performance so for miners, is one thing to keep an eye even if the increment of performance is minimal, it does take to account because miners run 24/7 365 days so in the end of the year/week or whatever that minimal increment makes a huge difference.

          Comment


          • #6
            Sorry I read your reply very quickly I thought you meant other mining applications. I'm only into ethereum but I will investigate.

            Comment


            • #7
              Maybe you should compare a Pentium Dual Core with a Pci Express 2.0 vs a Ryzen or a i7 pci express 3.0 to see if it's worthless for miners to upgrade their cpu and motherboards, and to see if rocm takes advantage of the advertised features (leverage of pci express 3.0, SenseMI, heterogenous memory, etc). It's an uneven comparison but most miners don't care much about the cpu because it makes no difference.
              Last edited by Kayote; 08-08-2017, 11:18 PM.

              Comment


              • #8
                Suggested articles: With what drivers and GPUs is it even possible to run the ethminer benchmark, and then not hang the GPU doing it? https://bugs.freedesktop.org/show_bug.cgi?id=101584

                Anyway, for clover you want
                Code:
                diff --git a/libethash-cl/CLMiner_kernel.cl b/libethash-cl/CLMiner_kernel.cl
                index 7661e7425..b37d6c4d6 100644
                --- a/libethash-cl/CLMiner_kernel.cl
                +++ b/libethash-cl/CLMiner_kernel.cl
                @@ -1,3 +1,4 @@
                +#pragma OPENCL EXTENSION cl_clang_storage_class_specifiers : enable
                 #define OPENCL_PLATFORM_UNKNOWN 0
                 #define OPENCL_PLATFORM_NVIDIA  1
                 #define OPENCL_PLATFORM_AMD
                Until they have OpenCL 1.2 with "static".

                Comment

                Working...
                X