tag:support.primatelabs.com,2011-01-31:/discussions/geekbench/29801-linux-command-line-cuda-compute-fails-to-runPrimate Labs: Discussion 2018-09-26T15:03:38Ztag:support.primatelabs.com,2011-01-31:Comment/461134392018-09-20T15:15:39Z2018-09-20T15:15:39ZLinux command line CUDA compute fails to run<div><p>Hi Keith,</p>
<p>Thanks for reaching out. What output does <code>./geekbench4 --compute-list</code> give you?</p></div>Sarahtag:support.primatelabs.com,2011-01-31:Comment/461134392018-09-20T16:16:22Z2018-09-20T16:16:22ZLinux command line CUDA compute fails to run<div><p>keith@ASUS-Linux:~/Downloads/Geekbench-4.2.2-Linux$ ./geekbench4 --compute-list</p>
<p>CUDA0 0 GeForce GTX 1080 Ti0 1 GeForce GTX 10800 2 GeForce GTX 10800 3 GeForce GTX 1070 TiOpenCL0 0 GeForce GTX 1080 Ti0 1 GeForce GTX 10800 2 GeForce GTX 10800 3 GeForce GTX 1070 Tikeith@ASUS-Linux:~/Downloads/Geekbench-4.2.2-Linux$ </p>
<pre>
<code> From: Sarah <tender2+d737e3feee@tenderapp.com></code>
</pre>
<p>To: <a href="mailto:keith.myers@att.net">keith.myers@att.net</a> Sent: Thursday, September 20, 2018 8:15 AM Subject: Re: Linux command line CUDA compute fails to run [Geekbench #29801]</p>
<p>#yiv9939989771 pre {width:92%;margin:10px 2%;padding:5px 2%;background:#efefef;border:1px solid #d6d6d6;}#yiv9939989771 blockquote {margin-left:0;padding-left:1em;border-left:5px solid #ccc;}</p></div>Keith Myerstag:support.primatelabs.com,2011-01-31:Comment/461134392018-09-21T12:37:12Z2018-09-21T12:37:41ZLinux command line CUDA compute fails to run<div><p>What happens if you try the following command?:</p>
<p><code>./geekbench4 --compute CUDA --compute-platform-id 0 --compute-device-id 3</code></p>
<p>Alternatively, what happens if you replace the number after <code>--compute-device-id</code> with 0, 1, or 2?</p></div>Sarahtag:support.primatelabs.com,2011-01-31:Comment/461134392018-09-21T16:03:26Z2018-09-21T16:03:27ZLinux command line CUDA compute fails to run<div><p>keith@ASUS-Linux:~/Downloads/Geekbench-4.2.2-Linux$ ./geekbench4 --compute CUDA --compute-platform-id 0 --compute-device-id 3Error: Compute API 'CUDA' is not available in Tryout mode.Geekbench 4.2.2 Tryout : Geekbench 4 - Cross-Platform Benchmark</p>
<table>
<tr>
<td></td>
</tr>
</table>
<p>Geekbench 4 - Cross-Platform Benchmark<br>
Geekbench 4 measures your hardware's power and tells you whether your computer is ready to roar. How strong ... | |</p>
<p>|</p>
<pre>
<code> From: Sarah <tender2+d737e3feee@tenderapp.com></code>
</pre>
<p>To: <a href="mailto:keith.myers@att.net">keith.myers@att.net</a> Sent: Friday, September 21, 2018 5:37 AM Subject: Re: Linux command line CUDA compute fails to run [Geekbench #29801]</p>
<p>#yiv0005010536 pre {width:92%;margin:10px 2%;padding:5px 2%;background:#efefef;border:1px solid #d6d6d6;}#yiv0005010536 blockquote {margin-left:0;padding-left:1em;border-left:5px solid #ccc;}</p></div>Keith Myerstag:support.primatelabs.com,2011-01-31:Comment/461134392018-09-21T16:15:00Z2018-09-21T16:15:00ZLinux command line CUDA compute fails to run<div><p>Have you purchased a license yet? Only OpenCL is available in the Tryout version, so the CUDA Compute Benchmark will not run until you unlock either the regular or pro features. Here is a chart that compares the different versions: <a href="https://www.geekbench.com/versions/">https://www.geekbench.com/versions/</a></p></div>Sarahtag:support.primatelabs.com,2011-01-31:Comment/461134392018-09-21T16:34:31Z2018-09-21T16:34:31ZLinux command line CUDA compute fails to run<div><p>No, I was just running the test for comparison purposes to the Anantech Geekbench4 tests on RTX 2080Ti. I was unaware that the Tryout version was limited. I never found that chart you linked on the website till you just showed me.<br>
What tests are performed in CUDA mode? Are they the same tests as in OpenCL?</p>
<pre>
<code> From: Sarah <tender2+d737e3feee@tenderapp.com></code>
</pre>
<p>To: <a href="mailto:keith.myers@att.net">keith.myers@att.net</a> Sent: Friday, September 21, 2018 9:15 AM Subject: Re: Linux command line CUDA compute fails to run [Geekbench #29801]</p>
<p>#yiv9442757144 pre {width:92%;margin:10px 2%;padding:5px 2%;background:#efefef;border:1px solid #d6d6d6;}#yiv9442757144 blockquote {margin-left:0;padding-left:1em;border-left:5px solid #ccc;}</p></div>Keith Myerstag:support.primatelabs.com,2011-01-31:Comment/461134392018-09-26T15:03:35Z2018-09-26T15:03:35ZLinux command line CUDA compute fails to run<div><p>Yes, the workloads are the same for each Compute API. You can check out descriptions for each workload here: <a href="https://www.geekbench.com/doc/geekbench4-compute-workloads.pdf">https://www.geekbench.com/doc/geekbench4-compute-workloads.pdf</a></p>
<p>Here is some more general information on interpreting Geekbench 4 scores, which includes some more information about the Compute benchmark: <a href="http://support.primatelabs.com/kb/geekbench/interpreting-geekbench-4-scores">http://support.primatelabs.com/kb/geekbench/interpreting-geekbench-...</a></p>
<p>Let me know if there is anything else I can help with!</p></div>Sarah