------- Comment From mranw...@us.ibm.com 2018-09-26 13:11 EDT------- I had to rebuild the utilities in scripts since those were the problem, they were x86 binaries.
The problem is fixed with this kernel: user@deb3qwsp1:~$ cat /proc/version Linux version 4.15.0-34-generic (jsalisbury@kathleen) (gcc version 7.3.0 (Ubuntu 7.3.0-16ubuntu3)) #38~lp1792102 SMP Wed Sep 12 19:55:58 UTC 2018 user@deb3qwsp1:~$ nvidia-smi |grep -A3 Memory | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 Tesla V100-SXM2... On | 00000004:04:00.0 Off | 0 | | N/A 31C P0 34W / 300W | 0MiB / 16128MiB | 0% Default | -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux in Ubuntu. https://bugs.launchpad.net/bugs/1792102 Title: Ubuntu18.04: GPU total memory is reduced Status in The Ubuntu-power-systems project: In Progress Status in linux package in Ubuntu: In Progress Status in linux source package in Bionic: In Progress Bug description: == Comment: #0 - Michael Ranweiler <mranw...@us.ibm.com> - 2018-09-10 19:26:14 == Due to a recent change for powernv, now the total GPU memory is no longer available. This impacts performance for any application/benchmark has a large GPU memory utilization. Previous amount of memory : 16128MiB Current amount of available memory : 15360MiB From Anton, describing the recent change.: powerpc/powernv: Increase memory block size to 1GB on radix Memory hot unplug on PowerNV radix hosts is broken. Our memory block size is 256MB but since we map the linear region with very large pages, each pte we tear down maps 1GB. A hot unplug of one 256MB memory block results in 768MB of memory getting unintentionally unmapped. At this point we are likely to oops. Fix this by increasing our memory block size to 1GB on PowerNV radix hosts. Fixes: 4b5d62ca17a1 ("powerpc/mm: add radix__remove_section_mapping()") This is fixed with: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=7acf50e4efa6 To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu-power-systems/+bug/1792102/+subscriptions -- Mailing list: https://launchpad.net/~kernel-packages Post to : kernel-packages@lists.launchpad.net Unsubscribe : https://launchpad.net/~kernel-packages More help : https://help.launchpad.net/ListHelp
