Playback on Linux swamps 4 cores?

Good luck with that [grin]

It could be that the HDDs are the bottleneck in your PC. If it it supports SATA3 you could see a remarkable speed-up with SSDs. I put an SSD in my wife’s PC and the throughput was better but with SATA2 it did not achieve full capability. SATA2 = 300 MB/s, SATA3 = 600 MB/s, m.2 = 16 Mb/s (!)

I can’t remember the guy’s name - he invented the first “lugable” PC running CP/M - his motto was something like - It only has to be good enough. That used to tick me off because I knew there was often a lack of craftsmanship in modern manufacturing. But, as I grow older (70 soon) I am much less of a perfectionist. As I said few people appreciate great craftsmanship anyway. Most people are happy enough if something just “works.” Best to you -=Ken=-

Osborne-1 by Adam Osborne.

I used Google to verify my recollection: Osborne hit the …um… bit bucket at 64, in 2003.

BTW I’m older than you and still… see above in this thread. ROFL

YES! That’s the guy. I just read Wikipedia. Forgot he checked out so early. Mensa member…

Instead of Google use Startpage, searx.me, Quant or even Yandex (Chinese). Google is NOT your friend. Google now does EVIL.

So you’re older than me, huh? When I started trying to do video editing on the computer it was considered a luxury to have two 40 GB HDDs. You would have to archive off your project in order to start another. The 2 MB cache in the HDDs was a real limitation. When they came out with 8 MB cache then it was so much better. What we have now days was dream stuff. -=Ken=-

Adm. Grace Murray Hopper’s nano-second wires (never got one) and the first “bug”, 8" floppies and the nibbler to make them two-sided, 6502’s (KIM-1 w/ 1K SS RAM equal to a UNIVAC I w/ mercury delay lines), 6800’s, 8080’s, z80’s (woohoo!), Apple I literally on a bread board, DEC PDP 11/20 with RT11, PDP 11/45 running RSTS (OS written in BASIC!!) with core memory, Eckert & Mauchly (taught his granddaughter in a CompSci101), BASIC put together by some guy named Bill Gates for some company called Microsoft, C/PM by Cdr. Gary Kildahl, Dr. Dobb’s Journal - running light with over-byte… BTDT - t-shirt long since gone

ADDED: Hmmm that could probably be sung to Billy Joel’s We Didn’t Start the Fire…

Sounds like you got there before I did. I started out building a Heathkit H-89 running HDOS and later cp/m. Jumped from that right into an IBM PC clone by an outfit named PCs Unlimited run by a kid named Michael Dell out of his dorm room. Later, when I first saw DOS 2.0 I was elated - because we could actually segregate out dozens of files into “subdirectories.” My great aunt used to say she lived in the good old days - except, she couldn’t remember what was so great about them. Some things are so much better now. -=Ken=-

I think the KIM-1 was the best example of now and “good?” old days. The UNIVAC I needed a huge room, cooling that would otherwise support whole buildings, The CPU (sic) was literally large enough to walk around in (there was a glass door). The tapes were steel tapes that meant operators had to wear safety shoes. By comparison, the KIM-1 board was about 8 1/2 x 11, was faster than the UNIVAC, and had the same amount of memory in some chips. The UNIVAC pumped a wave into a column of mercury, picked up the pulses at the other end, and recycled them, unless they were changed (memory write). The KIM-1 (along with its contemporaries) used a standard audio cassette player. Cooling was just ambient temperature in the room. The only thing missing was the UNIVAC’s wicked cool console. At least the z80, S100 bus, Cromemco case had really neat switches and (IIRC) LED’s like the PDP-11’s. Woohoo!

Anyway, it’s worth remembering how far we’ve come in an amazingly short amount of time. From 8080’s that creaked along on BASIC or (gasp!) assembler to whining about saturating four cores with data on a 5 Tb HDD.

ADDED: Just for laughs, I looked up the Cromemco Z1 info. Delivered price for this marvelous microcomputer: $2500 1970 dollars. Dial in inflation and… ah, you don’t wanna know what amounts to in 2017 dollars.

You should mention exactly how. Some tools show threads as processes. If you ps axww | grep [b]in/shotcut then you should only see one. (| grep [s]hotcut shows additional for the launch script.) Either you are looking at threads or some processes that failed to exit from a previous run.

Again, the tool. Some tools show #cores x 100 %; others 100% regardless of number of cores. More than likely you see 4 threads for 4 cores with the workload spread between them. If they were maxed-out, I expect your tool to be showing total 400% CPU.

I used htop, and its internal filter set to shotcut. It’s effectively the same as ps… etc. only it’s recurring.

I just launched shotcut with no leftover processes (htop and shotcut in filter). At the moment there are two command line processes and nine processes with only bin/shotcut. All of them are sleeping because I don’t have anything for shotcut to do. Hmmm… let’s try Open Other… OK, using the Count and playing it back… bin/shotcut processes jumped up to 15!!! (the two shell processes are still there).

Well! That’s interesting. I tried to move the count to the timeline and shotcut and all of the htop entries vanished in a serious hurry. That can’t be good…

Richard, I have been trying to track down the excellent System Monitor app in my Linux Mint. It identifies as “System Monitor” but the real name is “gnome-system-monitor.” Excellent, recommended. -=Ken=-

Oops - I was repeating what the label on the icon is. BTW, my installation doesn’t launch. It spins its wheels briefly and… poof! gone! I tried once to uninstall and reinstall it - no luck. Anyway, look into htop (either in the synaptic tool or the software store tool). There’s also top - 3 guess where htop comes from.

By default, htop shows threads whereas the ps options I gave does not. In htop Setup > Display options, see Hide userland process threads. Don’t try to analyze the thread count and usage unless you are the programmer. It’s very complicated, even some parts of it when you get into HTML effects and the web runtime I do not understand.

I got the right bin/shotcut PID from ps and I’m watching it with htop (the other shotcut items are, of course, gone after hiding userland threads). At the moment, playing back a video clip, I’m seeing the process’ CPU utilization at somewhere around 280-300%. Not quite what I expected.

Is Shotcut saying your clip is interlaced?
Does your video mode in settings cause it to upscale?

Are you using the nvidia binary driver to get properly accelerated OpenGL? You can run glxinfo | grep OpenGL. Also, Shotcut’s application log shows some info, for example on Windows where I am now it says:
[Debug ] Mlt::GLWidget::initializeGL begin
[Info ] Mlt::GLWidget::initializeGL OpenGL vendor “NVIDIA Corporation”
[Info ] Mlt::GLWidget::initializeGL OpenGL renderer “GeForce GTX 980 Ti/PCIe/SSE2”
[Info ] Mlt::GLWidget::initializeGL OpenGL threaded? true
[Info ] Mlt::GLWidget::initializeGL OpenGL ES? false
[Debug ] Mlt::GLWidget::initializeGL end

The clip is marked as Progressive. The video mode is HD 1080p 59.94 fps - this matches the source video’s specs.

glxinfo shows the Nvidia driver 384.98 is being used. The GPU is a GTX-1050 Ti - 4 Gb.

[Debug ] begin [Info ] OpenGL vendor "NVIDIA Corporation" [Info ] OpenGL renderer "GeForce GTX 1050 Ti/PCIe/SSE2" [Info ] OpenGL threaded? true [Info ] OpenGL ES? false [Debug ] end

Have I missed anything?

Sorry to interject into the conversation here but regarding your mention of Eckert and Mauchly, the HTPC case that I designed/built last year was “branded” in honor of their historic creation, the ENIAC. As I don’t have a photo of it handy, here’s a rendering of the CAD model itself:

It’s a 17 Watt TDP machine running Kodi that’s based on the Intel 1037U CPU. I’m in the process of building a second one for my shop (although based on a different, 7 Watt TDP motherboard). ENIAC lives! Hey, it’s cool that you taught one of their granddaughters! :slight_smile: BTW, I learned to program on the Processor Technology Sol-20:

Sure doesn’t look like the room-sized ENIAC I’ve seen in pictures.

I’d forgotten about MITS. I was subscribed to PE when the 8008 cover article came out. Little did anyone know… LOL