F5F Stay Refreshed Hardware Desktop Warning about Epyc Naples

Warning about Epyc Naples

Warning about Epyc Naples

Pages (3): Previous 1 2 3
T
TebrynArbandi
Junior Member
46
03-14-2017, 04:05 PM
#21
I plan to run several virtual machines and Docker containers, but my main requirement for processing power comes from Plex. I’d like to eliminate the need for separate libraries for 4K HDR and 1080p SDR when duplicates exist. The $105 purchase of the 7401p seems far less expensive than future storage costs. With Plex HDR to SDR tone mapping, it should handle this efficiently. Do you know how well this will work given that CPU cores won’t be divided in the same way as with VMs?
T
TebrynArbandi
03-14-2017, 04:05 PM #21

I plan to run several virtual machines and Docker containers, but my main requirement for processing power comes from Plex. I’d like to eliminate the need for separate libraries for 4K HDR and 1080p SDR when duplicates exist. The $105 purchase of the 7401p seems far less expensive than future storage costs. With Plex HDR to SDR tone mapping, it should handle this efficiently. Do you know how well this will work given that CPU cores won’t be divided in the same way as with VMs?

A
AK_Sanda
Member
106
03-16-2017, 10:03 AM
#22
I’m not very familiar with PLEX systems, but I hope it would gain from quicker processing units rather than extra threads unless you’re handling many streams at once. I’d love to know how it performs for you and what core utilization looks like per stream. I don’t anticipate such high numbers. EPYC tends to excel with parallel tasks instead of single jobs like PLEX. I often use BOINC as a case in point. I currently run an EPYC 7601 with PROXMOX, while BOINC operates within a LXC container. Native OS versus container Linux doesn’t really matter much for my workloads. The VM overhead is noticeable, so consider it or pin CPU cores to the same NUMA node if possible. You might also look into Huge Pages—they can boost VM performance noticeably.
A
AK_Sanda
03-16-2017, 10:03 AM #22

I’m not very familiar with PLEX systems, but I hope it would gain from quicker processing units rather than extra threads unless you’re handling many streams at once. I’d love to know how it performs for you and what core utilization looks like per stream. I don’t anticipate such high numbers. EPYC tends to excel with parallel tasks instead of single jobs like PLEX. I often use BOINC as a case in point. I currently run an EPYC 7601 with PROXMOX, while BOINC operates within a LXC container. Native OS versus container Linux doesn’t really matter much for my workloads. The VM overhead is noticeable, so consider it or pin CPU cores to the same NUMA node if possible. You might also look into Huge Pages—they can boost VM performance noticeably.

R
ruzzie
Member
65
03-17-2017, 09:37 AM
#23
According to recent information, Plex transcoding operates with multiple threads, though VC1 video is an exception. Most users experience around eight to ten concurrent streams smoothly, which should suit your needs.
R
ruzzie
03-17-2017, 09:37 AM #23

According to recent information, Plex transcoding operates with multiple threads, though VC1 video is an exception. Most users experience around eight to ten concurrent streams smoothly, which should suit your needs.

Pages (3): Previous 1 2 3