Announcement

Collapse
No announcement yet.

Unigine Superposition Is A Beautiful Way To Stress Your GPU In 2017, 17-Way Graphics Card Comparison

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

  • #21
    Originally posted by bug77 View Post
    For those expecting miracles from future drivers: the benchmark displays the GPU usage which in my case is pegged at 99-100%.
    That said, I saw that on Windows only, because on an up to date KDE Neon, the benchmark won't run. It disables my primary monitor and moves everything to the secondary instead. And when I try to launch it on the secondary display, it still won't start. It flashes an error message (way too fast to be read) and seems to not produce any log files. Running Nvidia 381 driver.
    GPU usage doesn't mean much if the driver is doing something inefficient behind the scenes.

    Comment


    • #22
      Not starting here on Debian...

      ./Superposition yields:
      Code:
      ** (gtk-launch:17411): WARNING **: Error retrieving accessibility bus address: org.freedesktop.DBus.Error.ServiceUnknown: The name org.a11y.Bus was not provided by any .service files
      gtk-launch: no such application Superposition
      [1]    17411 exit 2
      bin/launcher yields and stays like that doing nothing:
      Code:
      qt.network.ssl: QSslSocket: cannot resolve CRYPTO_num_locks
      qt.network.ssl: QSslSocket: cannot resolve CRYPTO_set_id_callback
      qt.network.ssl: QSslSocket: cannot resolve CRYPTO_set_locking_callback
      qt.network.ssl: QSslSocket: cannot resolve ERR_free_strings
      qt.network.ssl: QSslSocket: cannot resolve EVP_CIPHER_CTX_cleanup
      qt.network.ssl: QSslSocket: cannot resolve EVP_CIPHER_CTX_init
      qt.network.ssl: QSslSocket: cannot resolve sk_new_null
      qt.network.ssl: QSslSocket: cannot resolve sk_push
      qt.network.ssl: QSslSocket: cannot resolve sk_free
      qt.network.ssl: QSslSocket: cannot resolve sk_num
      qt.network.ssl: QSslSocket: cannot resolve sk_pop_free
      qt.network.ssl: QSslSocket: cannot resolve sk_value
      qt.network.ssl: QSslSocket: cannot resolve SSL_library_init
      qt.network.ssl: QSslSocket: cannot resolve SSL_load_error_strings
      qt.network.ssl: QSslSocket: cannot resolve SSL_get_ex_new_index
      qt.network.ssl: QSslSocket: cannot resolve SSLv2_client_method
      qt.network.ssl: QSslSocket: cannot resolve SSLv3_client_method
      qt.network.ssl: QSslSocket: cannot resolve SSLv23_client_method
      qt.network.ssl: QSslSocket: cannot resolve SSLv2_server_method
      qt.network.ssl: QSslSocket: cannot resolve SSLv3_server_method
      qt.network.ssl: QSslSocket: cannot resolve SSLv23_server_method
      qt.network.ssl: QSslSocket: cannot resolve X509_STORE_CTX_get_chain
      qt.network.ssl: QSslSocket: cannot resolve OPENSSL_add_all_algorithms_noconf
      qt.network.ssl: QSslSocket: cannot resolve OPENSSL_add_all_algorithms_conf
      qt.network.ssl: QSslSocket: cannot resolve SSLeay
      qt.network.ssl: QSslSocket: cannot resolve SSLeay_version
      qt.network.ssl: QSslSocket: cannot call unresolved function CRYPTO_num_locks
      qt.network.ssl: QSslSocket: cannot call unresolved function CRYPTO_set_id_callback
      qt.network.ssl: QSslSocket: cannot call unresolved function CRYPTO_set_locking_callback
      qt.network.ssl: QSslSocket: cannot call unresolved function SSL_library_init
      qt.network.ssl: QSslSocket: cannot call unresolved function SSLv23_client_method
      qt.network.ssl: QSslSocket: cannot call unresolved function sk_nu

      Comment


      • #23
        Originally posted by peppercats View Post

        GPU usage doesn't mean much if the driver is doing something inefficient behind the scenes.
        Only if the driver is doing redundant stuff (like rendering pixels you don't see). But both OpenGL and DX11 are too mature for that to be happening a lot.

        Comment


        • #24
          Originally posted by bug77 View Post

          Only if the driver is doing redundant stuff (like rendering pixels you don't see). But both OpenGL and DX11 are too mature for that to be happening a lot.
          There's a lot more reasons than that. it could have implemented the API in an inefficient way(e.g, excessive redundant state changes in OpenGL,) or misused the hardware, or just had a performance bug in general. Using RadeonSI as an example,

          Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

          Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

          Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite



          etc.,

          FWIW, a lot of the "performance bugs" fixed are not the driver itself but the application. One of the bigger issues with OpenGL is that nearly everything is just a suggestion and there's no guarantees. For example, there's nothing stopping you from telling OpenGL that a buffer is static and then updating it dynamically. AFAIK Nvidia ignores a lot of OpenGL hints because they're so misused unless something has changed with their driver recently.

          Comment


          • #25
            Originally posted by binstream View Post
            Here is the official page with download links: https://benchmark.unigine.com/superposition
            Hi
            Any chances for fixing this?
            While this game isn't quite the traditional RTs game, i was heavily enjoying managing unit swarms, building defense structures and taking out my enemies. While simplistic and lacking real depth, the game still has it's charm....until I hit Chapter 2 Mission 2. What in gods name is up with the nonstop spawning enemies despite having all of the Blue enemies structures. It makes no bloody sense other then to be supremely annoying.

            Comment


            • #26
              Two questions.
              1) Why is the linux version 2GB in size and the windows version 1GB? I would have thought the windows version should be bigger due to the extra dx11/dx12 renderer, and different compression.
              2) Im guessing this is due to thousands of people trying to download it at once, but I'm on a 1gigabit internet connection at my work facility and only downloading at 250kbps from the Unigine site. At this rate it will take 2 hours to complete. I did see a torrent link, so maybe I will see if I can get it faster over bittorrent, and maybe help to seed for a while.

              Comment


              • #27
                Originally posted by flubba86 View Post
                Two questions.
                1) Why is the linux version 2GB in size and the windows version 1GB? I would have thought the windows version should be bigger due to the extra dx11/dx12 renderer, and different compression.
                Neither is Windows version 2G nor is Linux 1 GB, but Windows version is 1276MB but Linux is 1564MB... offcial page is this:



                While Michael posted something else page with rough estemated and rounded sizes of these packages
                2) Im guessing this is due to thousands of people trying to download it at once, but I'm on a 1gigabit internet connection at my work facility and only downloading at 250kbps from the Unigine site. At this rate it will take 2 hours to complete. I did see a torrent link, so maybe I will see if I can get it faster over bittorrent, and maybe help to seed for a while.
                Yeah It was utter slow for me too So i just got it via torrent.
                Last edited by dungeon; 11 April 2017, 07:19 PM.

                Comment


                • #28
                  Originally posted by Licaon View Post
                  Not starting here on Debian...

                  ./Superposition yields:
                  Code:
                  ** (gtk-launch:17411): WARNING **: Error retrieving accessibility bus address: org.freedesktop.DBus.Error.ServiceUnknown: The name org.a11y.Bus was not provided by any .service files
                  gtk-launch: no such application Superposition
                  [1] 17411 exit 2
                  bin/launcher yields and stays like that doing nothing:
                  Code:
                  qt.network.ssl: QSslSocket: cannot resolve CRYPTO_num_locks
                  qt.network.ssl: QSslSocket: cannot resolve CRYPTO_set_id_callback
                  qt.network.ssl: QSslSocket: cannot resolve CRYPTO_set_locking_callback
                  qt.network.ssl: QSslSocket: cannot resolve ERR_free_strings
                  qt.network.ssl: QSslSocket: cannot resolve EVP_CIPHER_CTX_cleanup
                  qt.network.ssl: QSslSocket: cannot resolve EVP_CIPHER_CTX_init
                  qt.network.ssl: QSslSocket: cannot resolve sk_new_null
                  qt.network.ssl: QSslSocket: cannot resolve sk_push
                  qt.network.ssl: QSslSocket: cannot resolve sk_free
                  qt.network.ssl: QSslSocket: cannot resolve sk_num
                  qt.network.ssl: QSslSocket: cannot resolve sk_pop_free
                  qt.network.ssl: QSslSocket: cannot resolve sk_value
                  qt.network.ssl: QSslSocket: cannot resolve SSL_library_init
                  qt.network.ssl: QSslSocket: cannot resolve SSL_load_error_strings
                  qt.network.ssl: QSslSocket: cannot resolve SSL_get_ex_new_index
                  qt.network.ssl: QSslSocket: cannot resolve SSLv2_client_method
                  qt.network.ssl: QSslSocket: cannot resolve SSLv3_client_method
                  qt.network.ssl: QSslSocket: cannot resolve SSLv23_client_method
                  qt.network.ssl: QSslSocket: cannot resolve SSLv2_server_method
                  qt.network.ssl: QSslSocket: cannot resolve SSLv3_server_method
                  qt.network.ssl: QSslSocket: cannot resolve SSLv23_server_method
                  qt.network.ssl: QSslSocket: cannot resolve X509_STORE_CTX_get_chain
                  qt.network.ssl: QSslSocket: cannot resolve OPENSSL_add_all_algorithms_noconf
                  qt.network.ssl: QSslSocket: cannot resolve OPENSSL_add_all_algorithms_conf
                  qt.network.ssl: QSslSocket: cannot resolve SSLeay
                  qt.network.ssl: QSslSocket: cannot resolve SSLeay_version
                  qt.network.ssl: QSslSocket: cannot call unresolved function CRYPTO_num_locks
                  qt.network.ssl: QSslSocket: cannot call unresolved function CRYPTO_set_id_callback
                  qt.network.ssl: QSslSocket: cannot call unresolved function CRYPTO_set_locking_callback
                  qt.network.ssl: QSslSocket: cannot call unresolved function SSL_library_init
                  qt.network.ssl: QSslSocket: cannot call unresolved function SSLv23_client_method
                  qt.network.ssl: QSslSocket: cannot call unresolved function sk_nu
                  That is probably distro issue or due Debian's 9 ssl transition... it currently have 1.1 and 1.0 openssl.



                  Basically either link libssl.so and libcrypto.so to old 1.0.2 or better just copy/paste these to Unigine folder... whatever, just quick fix idea
                  Last edited by dungeon; 11 April 2017, 07:21 PM.

                  Comment


                  • #29
                    Originally posted by dungeon View Post
                    Basically either link libssl.so and libcrypto.so to old 1.0.2 or better just copy/paste these to Unigine folder... whatever, just quick fix idea
                    That fixed it (linking 1.0.2 of those 2).
                    Too bad QT keeps pushing this (now it's for 5.10 ~ Nov 2017?!?!): https://bugreports.qt.io/browse/QTBUG-52905
                    Last edited by Licaon; 11 April 2017, 08:16 PM.

                    Comment


                    • #30
                      No refraction or colour bleed, that I could see. Even the reflections looked like simple environment mapping, not ray-tracing. Still can’t do that in real time, huh?

                      Comment

                      Working...
                      X