And the guide to hardware accelerated transcoding:
This is another interesting forum post that isn’t TOO old about the differences / limitations of various generations of hardware encoding options:
I had a 9th gen dual core Pentium gold transcoding more streams than I could open on my desktop (around 10-12) - I only recently replaced it with a 9700k because I needed additional horsepower for deinterlacing and burning in subtitles on older 480i DVD rips.
Definitely check out the links that @NateTheBrewer posted.
The two NVENC chips on your M6000 will blow through H.264 and H.265 (8-bit only) transcodes at 30X the speed of the CPU. There’s also a dedicated NVDEC decoder section on the main chip’s die that offloads that work to the GPU as well.
There are also CUDA-enabled filters for stuff like deinterlace and noise reduction that make quick work of things that make a CPU weep. I don’t know if Plex implements those at all but they’re certainly available for offline transcoding.
I use an open-source command line interface to the Nvidia API called NVEnc (AVAILABLE AT GITHUB). There are also a couple of GUI’s to this package like Staxrip but I personally find them more confusing than they’re worth.
Just don’t try to use Nvidia GPU’s with Handbrake. The Handbrake guys apparently got pissed with Nvidia for not supporting them as well as Intel does so Handbrake’s NVENC implementation is horrible and they refuse to support NVDEC or CUDA.
Also take head of my earlier note about quality using the NVENC chips on the Maxwell card you have. People say that it’s noticeably worse than what you get from the newer cards. I can’t personally vouch for that. All I know is that the NVENC output from my P4000 is indistinguishable from a software x.265 encode and goes 30X faster.
As @NateTheBrewer shows in his last post, adding even a weak Nvidia card like the Quadro P400 gives a Plex server more transcode power than it will ever need. Highly recommended if for no other reason than it keeps those Xeons running cool.
@MonsterMaxx - Yeah, I’ve been thinking about pulling the trigger with OnShape but I’m too caught up in the sunk-cost fallacy of all those years of SolidWorks support to do it. Maybe this year . . . .
(FYI to those who complain about the cost of a Plex Pass: Unless you’re a monster company with more than 100 developers a SolidWorks license will set you back somewhere between $4K and $20K per seat plus between $1K and $5K per seat, per year in support costs depending on the options you need. You then need about $5K of high-end, certified hardware if you want to run it without crashing every twenty minutes. My eyes roll whenever people complain about spending $120 on a lifetime Plex Pass. They have never experienced the pain of having a SolidWorks distributor sucking $5k out your left teat year after year after year.)
Here’s the response I sent to my VAR when they tried to get me to re-up
I need simulation too. It’s like $10k for what I’d need to get back current on SW.
I probably shouldn’t say this…
I’ll be honest with you, SW lost me some time ago over this kind of thing. $5k here, $10k there, And the bugs are enough to drive me mad. I’ve been on SW since ’95, I’ve seen it all.
Onshape is looking very appealing to me.
Files. I hate SW and files. SW single threaded PDM is the only thing worse than File structures. I spend an hour the other day un-screwing a file mess.
Bugs in SW, updates, patches, reinstalls, staying current, upgrading files, CRASHES.
Sharing, omg, Onshape is so easy to share my work. A click and send a link. So easy.
Robust modeling and fea (via altair)
Managed releases and ECOs.
Extreme hardware requirements for SW, where onshape runs in the cloud and on anything with a decent gpu. It’ll run on my laptop which is 5yrs old. FEA is near realtime.
Their prices are quite reasonable too.
My next ‘upgrade’ will be Onshape. Sorry.
And PTC is throwing a lot of weight behind it, it’s only going to get better.
Here’s the downside to Onshape.
When you translate a SW file, you lose all your parametric features.
Yes, you still have a body you can operate on, but it’s dumb.
I’m going to keep SW around to support legacy stuff, but I’m gone as soon as I start a whole new project.
Those who would complain too strenuously about Plex development should take a lesson from @MonsterMaxx’s last posts.
We’re worried in this thread because Plex won’t automatically scrape metadata for our music library. Imagine spending three years designing a new jumbo jet only to have your $20K software package “lose” 10,000 4mm fasteners because it forgot where it put the file for that part.
Hell hath no fury like an engineer talking to their SolidWorks rep.
I got another friend in the hospital due to Covid and hooked him up as well as my buddy 4000mi away. Oh, and I bought him a plexpass.
Also finally got the drives to expand the entertainment array on the server, getting the data off my workstation. The restoration has been running all night and still has 4hrs to go. Lot of data.
But then plex and all it’s data will be on the one server which just finished getting a ton of upgrades. It’s not new, but every system was touched and upgraded. This should last another 5yrs until it’s time for a new server.
As to the Solidworks v. Onshape discussion. I’m deep into it with both companies sales departments. Both of them think (and rightly so) that I’m buying a new CAD package when I do my next project. SW’s team is now talking 3Dx which can be SW, but doesn’t have to be. There’s a new program called DesignX which is supposed to take SW’s place in cloudspace. SW can run in the cloud, but not well, it’s layers on layers and still file based. Onshape is also working hard to get my business. I have live demos from both of them coming up in the next few days. Should be interesting. I still think I’m going to Onshape, but I’ve done business with the SW guys a loooonnnnnnggggg time and am willing to let them show me what they have.