Hi All,
As per title of the thread: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
Currently I am ripping using two BD drives in parallel on my 6700k and also on my Synology via docker container and both take about the same time when two discs are ripped at the same time on the same device, be it 6700k or NAS.
Would RPi5 be able to handle the same task using 2 or single BD drives and achieve the same ripping times or should I expect a slower rip time and if so by how much if you have an experience of can guestimate?
Thanks
Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
The optical drives are the bounded resource (bottleneck). They are many, many times slower than your processor, et. al., even when running in a container. You should get pretty much the same speed.
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
I like what I hear but do you have any real life experience of it?
ARM CPU in RPi is not too great with single thread performance and that is what gets hammered with MakeMKV ripping. So to achieve max speed of BD drive you need certain processing power to handle IO etc.
ARM CPU in RPi is not too great with single thread performance and that is what gets hammered with MakeMKV ripping. So to achieve max speed of BD drive you need certain processing power to handle IO etc.
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
I have no experience with MakeMKV on a Pi, but... Considering that it is a separate copy for each device doing the ripping, it SHOULD be capable of running on separate cores on the Pi. The issue would be on the USB connection to the drives, and how it is dealing with that contention.
On non-Pi, I've run multiple rips at the same time as doing handbrake encodes, with no noticeable slow-down of either, so I don't think MakeMKV is putting much load on the processor.
On non-Pi, I've run multiple rips at the same time as doing handbrake encodes, with no noticeable slow-down of either, so I don't think MakeMKV is putting much load on the processor.
MakeMKV Frequently Asked Questions
FAQ about BETA and PERMANENT keys.
How to aid in finding the answer to your problem: Activating Debug Logging
FAQ about BETA and PERMANENT keys.
How to aid in finding the answer to your problem: Activating Debug Logging
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
I ordered some weeks ago CM5/16gb. Once it arrives I will let you guys know how it handles two copies of MakeMKV.
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
It can likely handle 128+ instances of makemkv as the bottleneck is USB. I've ripped 5 UHD discs simultaneously on a N100 CPU, but I did breakout the PCIe M.2 x4 to a USB 3.0 controller to connect 2 drives (also 2 with the onboard 3.0 and 1 with the onboard USB 2.0). Since that RPi has a M.2 that I believe is at least PCIe 2.0, it can do the same thing.
After looking at the RPI 5, I'm sure it can handle 5 drives but it's apparent that the connectivity is pretty poor. You're probably better off getting a small N100 as it's all around better and possibly cheaper with 16GB of RAM (although N100 can use up to 32GB).
The N100 device I used: https://www.aliexpress.us/item/3256805444933753.html
If you like big boards, there is a iTX motherboard that would work https://www.aliexpress.us/item/3256806939401375.html
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
Great info.flojo wrote: ↑Sun Jan 19, 2025 1:00 amIt can likely handle 128+ instances of makemkv as the bottleneck is USB. I've ripped 5 UHD discs simultaneously on a N100 CPU, but I did breakout the PCIe M.2 x4 to a USB 3.0 controller to connect 2 drives (also 2 with the onboard 3.0 and 1 with the onboard USB 2.0). Since that RPi has a M.2 that I believe is at least PCIe 2.0, it can do the same thing.You might be able to bifurcate it to 1 lane each and thus connect at least 11 drives to the RPI.The PCIe header is only single lane (x1) .
After looking at the RPI 5, I'm sure it can handle 5 drives but it's apparent that the connectivity is pretty poor. You're probably better off getting a small N100 as it's all around better and possibly cheaper with 16GB of RAM (although N100 can use up to 32GB).
The N100 device I used: https://www.aliexpress.us/item/3256805444933753.html
If you like big boards, there is a iTX motherboard that would work https://www.aliexpress.us/item/3256806939401375.html
My plan is to use 1 BD drive per existing 2 USB 3.0 ports on CM5 OE IO board or to even plug them both via single USB 3.0 port using USB 3.0 HUB. What do you think about such approach?
Why I am not so keen on miniPC is that I have about another 1500 discs to rip and after that is done MakeMKV will be only used for new purchases so few a month at best while CM5 has less power consumption and less heat to deal with than MiniPC and I plan to run it 24/7 as a media server. So if only I can get max ripping speed out of CM5 with two BD drives even if the setup is sub-optimal with how I will populate USB ports is fine with me as it will be only passing solution as in the end I can leave only of my two BD drives plugged in permanently there for this few a month discs that I will be adding.
But at the same time ripping 1500 discs is not a small task so would prefer to be able to do it at max speed possible.
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
It's worth noting to readers that what makes that statement true is 5 watts in power, however if you choose active cooling the same fans/coolers can be used with a N100 at the same effective cooling measurement.
For saving 5W with a RPI5 over a Nxxx CPU, you loose:
1. At least 4 lanes of PCIe 3.0, broken from M.2
2. Quicksync for encoding/transcoding
3. Ability to use more than 1 PCIe device natively
4. 32GB of SODIMM RAM (ddr4 or ddr5, but ddr4 does peg the N100)
5. 2.5gb ethernet
6. Using AMD64 software
7. AVX2
You decide if it's worth 5 more watts.
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
That is not the whole story in terms of power consumption differences:flojo wrote: ↑Sun Jan 19, 2025 1:43 pmIt's worth noting to readers that what makes that statement true is 5 watts in power, however if you choose active cooling the same fans/coolers can be used with a N100 at the same effective cooling measurement.
For saving 5W with a RPI5 over a Nxxx CPU, you loose:
1. At least 4 lanes of PCIe 3.0, broken from M.2
2. Quicksync for encoding/transcoding
3. Ability to use more than 1 PCIe device natively
4. 32GB of SODIMM RAM (ddr4 or ddr5, but ddr4 does peg the N100)
5. 2.5gb ethernet
6. Using AMD64 software
7. AVX2
You decide if it's worth 5 more watts.
So it can be up to 18W under load and in my use case probably 10-12W most of the time.
But either way these are not massive differences.
Re: Will running MakeMKV as docker container on RPi 5 slow down ripping speed / time using single or two BD drives?
Actually after doing calculations 12W more (estimated for my use case, where 18W more is the max) for N100 equates to about 32 euro per year more in electricity cost where I live.
That makes a difference as over few years CM5 cost of hardware goes to zero vs more power hungry N100 and as I don't really need extra power of N100 (which is anyway mostly in GPU area according to tests) then I think I will stick with CM5 idea and see how it goes.
That makes a difference as over few years CM5 cost of hardware goes to zero vs more power hungry N100 and as I don't really need extra power of N100 (which is anyway mostly in GPU area according to tests) then I think I will stick with CM5 idea and see how it goes.