MP 1,1-5,1 Is It Possible To Make The cMP 5,1 and Final Cut Pro X Edit 4K Effortlessly?

[5comma1]

Hello fellow cMP pilots.

Thanks to all who contribute to the forum.
It’s really helpful and appreciated.

My primary objective is to make my Mid 2010 5,1 edit Final Cut Pro X 4K footage effortlessly, if that’s possible?

Final Cut Pro X (FCPX) frustration examples:
- adding a transition and wanting to view/play it immediately without stutter, and without waiting several seconds for it to render.
- scrubbing is not smooth on a 4K timeline.
- can’t play-back 4K on the timeline without dropped frames / without creating proxies.

Random FCPX notes:
I always transcode to ProRes on import.
I almost always export to H.264.
Most of my footage comes from the Sony A7III and Phantom 4 Pro.

I recently upgraded both CPU’s and GPU; noticed almost no real-world improvement when editing.
I recently learned the old-tech SATA 2 bus in the cMP is a known I/O bottleneck?
Am I on the right track thinking I/O is the culprit?

My boot drive is a 500GB Crucial SSD.
My remaining drives are all HD’s… listed at bottom of page.

FCPX lives on the SSD.
My FCPX Libraries live on one of the 6TB WDC 7200 HD’s.
Attached are two disk speed test results, in case they're helpful.

I’ve been doing a bunch of reading about I/O here, among other places:

PCIe SSDs - NVMe & AHCI

This is a general info thread for blade SSDs that can be used in the Mac Pro. This is a WikiPost so anyone with the proper credentials may edit it. NVMe SSDs can be used as a boot drive in the MP5,1 and MP6,1 with the latest firmware installed (beginning with version 140.0.0.0.0 for MP5,1 and...

forums.macrumors.com

These two NVMe PCIe Adaptors seem to have a good reputation:

HighPoint 4-Port M.2 SSD7101A PCIe Gen3 NVMe RAID Controller for Windows, Mac & Linux Systems.

Unlock the true performance of NVMe SSD's: the SSD7101A-1 is capable of delivering performance over 8X faster than onboard NVMe solutions, & over 40X faster than SATA SSD's. The unique architecture provides dedicated PCIe 3.0 X4 (32GB/s) bandwidth for each M.2 SSD, & unlocks the true performance ...

www.amazon.com

SYBA USA

SYBA USA-SYBA USA catalog..

www.sybausa.com

If you feel a PCIe Adaptor is the way I need to go, which NVMe drive blades have worked well in your experience, in terms of making a real difference speeding-up FCPX?

How might a backed-up RAID-0 factor into this whole thing, considering the drives I have on hand?

This is all a bit of a head-scratcher??
I don’t know what I don’t know.

Thanks very much in advance!


Specs
Dual 3.33 GHz 6-Core Intel Xeon (X5680’s)
Sapphire Radeon Pulse RX 580
48GB > 6 x 8gb > 1333 MHz DDR3
OS X > I0.14.6 Mojave
Firmware > 144.0.0.0.0
Final Cut Pro 10.4.7


Additional Drives
1TB Hitachi HD
2TB WDC HD
2TB WDC HD
4TB WDC HD
6TB WDC HD
6TB WDC HD

 

[krakman]

Your projects should live on a PCI SSD, there is a whole thread about NVME drives on this forum. Basically the cheap NVME drives start off fast but after sustained use they slow down, best suggestion is a Samsung 970 PRO NVME, if you have lots of money get a four SSD carrier board so you can have 4 NVME cards Raided together. (again there is a hug thread about these carrier boards)

Enable AMD hardware accel - there is a huge thread about this. This should prevent you having to transcode in the first place and speed up you final encode too!

I have and old SATA carrier board with 3 x 1TB ssd raided together it gives me 700MB/s transfer speed which is much faster than your Spinning HD can do. I edit Corporate/wedding videos. Some of the wedding videos have mulitcam sequences consisiting of 4 to 6 DSLR camera h264 streams plus 1 stream of Canon MXF video and one or two audio tracks, it all plays back smoothly.

My graphic card is the Vega 56, while it is faster at compute tasks than the 580 it has the same Hardware Accel chip. Unfortunately I don't have any 4k footage to test with.

 

[tommy chen]

first you should update to FCPx 10.4.8
here the metal support is very much improved.

HDDs are much too slow to handle 4k material without problems.

a single PCIe blade with a good NVMe gives you about 1500 MB/s
compared to the 100 to 200 MB/s of an HDD.

if you work with multiple layers, of course the
HP7101a-1 the dream. here you can reach
about 2800 MB/s on a single NVMe or
5000 MB/s from two NVMe as raid 0

 

[XNorth]

‘‘Effortless’’ 4K editing comes only with recent Thunderbolt Macs. The MP 5,1 will never have smooth 4K scrubbing.

Fast storage can make the 5,1 usable. Three recommendations I have used...

Fastest/expensive: NVMe drives, RAID 0 with a PCIe card ≈ 3000MB/s, capacity limited to 8 TB, 12 TB on a 6-blade card.

Fast/affordable: SATA SSD drives, RAID 0 ≈ 1000 MB/s. No PCIe card necessary, there are inexpensive enclosures that fit in the optical drive bay and connect directly to the SATA connector.

Large capacity/expensive: 7200 RPM HDD, RAID 0 NAS with 10G Ethernet card ≈ 500 MB/s. Slower but only solution if massive capacity is needed.

no matter what the set up, the 5,1 slows down quickly the longer the timeline and more effects/complex the project.

 

[Snow Tiger]

Hello fellow cMP pilots.

Thanks to all who contribute to the forum.
It’s really helpful and appreciated.

My primary objective is to make my Mid 2010 5,1 edit Final Cut Pro X 4K footage effortlessly, if that’s possible?

Final Cut Pro X (FCPX) frustration examples:
- adding a transition and wanting to view/play it immediately without stutter, and without waiting several seconds for it to render.
- scrubbing is not smooth on a 4K timeline.
- can’t play-back 4K on the timeline without dropped frames / without creating proxies.

Random FCPX notes:
I always transcode to ProRes on import.
I almost always export to H.264.
Most of my footage comes from the Sony A7III and Phantom 4 Pro.

I recently upgraded both CPU’s and GPU; noticed almost no real-world improvement when editing.
I recently learned the old-tech SATA 2 bus in the cMP is a known I/O bottleneck?
Am I on the right track thinking I/O is the culprit?

My boot drive is a 500GB Crucial SSD.
My remaining drives are all HD’s… listed at bottom of page.

FCPX lives on the SSD.
My FCPX Libraries live on one of the 6TB WDC 7200 HD’s.
Attached are two disk speed test results, in case they're helpful.

I’ve been doing a bunch of reading about I/O here, among other places:

PCIe SSDs - NVMe & AHCI

This is a general info thread for blade SSDs that can be used in the Mac Pro. This is a WikiPost so anyone with the proper credentials may edit it. NVMe SSDs can be used as a boot drive in the MP5,1 and MP6,1 with the latest firmware installed (beginning with version 140.0.0.0.0 for MP5,1 and...

forums.macrumors.com

These two NVMe PCIe Adaptors seem to have a good reputation:

HighPoint 4-Port M.2 SSD7101A PCIe Gen3 NVMe RAID Controller for Windows, Mac & Linux Systems.

Unlock the true performance of NVMe SSD's: the SSD7101A-1 is capable of delivering performance over 8X faster than onboard NVMe solutions, & over 40X faster than SATA SSD's. The unique architecture provides dedicated PCIe 3.0 X4 (32GB/s) bandwidth for each M.2 SSD, & unlocks the true performance ...

www.amazon.com

SYBA USA

SYBA USA-SYBA USA catalog..

www.sybausa.com

If you feel a PCIe Adaptor is the way I need to go, which NVMe drive blades have worked well in your experience, in terms of making a real difference speeding-up FCPX?

How might a backed-up RAID-0 factor into this whole thing, considering the drives I have on hand?

This is all a bit of a head-scratcher??
I don’t know what I don’t know.

Thanks very much in advance!


Specs
Dual 3.33 GHz 6-Core Intel Xeon (X5680’s)
Sapphire Radeon Pulse RX 580
48GB > 6 x 8gb > 1333 MHz DDR3
OS X > I0.14.6 Mojave
Firmware > 144.0.0.0.0
Final Cut Pro 10.4.7


Additional Drives
1TB Hitachi HD
2TB WDC HD
2TB WDC HD
4TB WDC HD
6TB WDC HD
6TB WDC HD

Install a pair of RX 480 8GB blower type cooling graphics cards in your cMP , in PCIe Slots 1 and 4 . Modify the I/O bracket of the card in slot 4 so it fits .

Programs like Blender , Resolve and FCPX are optimized for dual matching GPUs . Usually , programs would take advantage of a GPGPU array simply to boost compute operations . But with FCPX , one GPU is dedicated for compute and the other is dedicated for UI . There's a reason why the Mac Pro 6,1 ( 2013 ) has a pair of matching GPUs .

Rendering is a compute operation .

Scrubbing is an UI operation .

Enable AMD hardware acceleration . You will also get H.264 format support .

Activate AMD hardware acceleration

[For existing OpenCore users, you can always download my latest OpenCore package at post #1314 for update] Full HWAccel achieved now! Everything is covered in the OpenCore thread. The step by step tutorial to enable AMD hardware acceleration is now at here, but I strongly recommend you go...

forums.macrumors.com

Install a HighPoint SSD7101A-1 four slot NVMe M.2 PCIe Card in PCIe Slot 2 . Use Intel 660p or HP EX 920 or 950 drives . Dedicate one drive for your boot device . We can not boot from RAID . Only use your boot drive for the OS and apps . Do not place data there . The other three matching drives you can RAID 0 together to form your working drive ( where you write projects to live ) and maybe other data if it needs to be read fast .

Install a 10 Gbps USB Card or a 10 GbE ethernet card in PCIe Slot 3 , for data import and export .

Use latest version of Mojave . I would not use Catalina in a production cMP .

The two RX 480 you want looks like this and can be powered by your Mac without any hacks . Each one only requires a single standard six pin to mini six pin PCIe power booster cable .

 

[tommy chen]

‘‘Effortless’’ 4K editing comes only with recent Thunderbolt Macs. The MP 5,1 will never have smooth 4K scrubbing.

.......

no matter what the set up, the 5,1 slows down quickly the longer the timeline and more effects/complex the project.

I'm afraid I must disagree with you there
on my main 5.1 i can even cut UHD H265 and scrubbing in FCPx and davinci resolve.

ok, multicam with UHD H265 is not possible, but in prores 444
(i usually convert everything or get prores from the arri alexa anyway)

and with my setup there is no slowdown - exept noise reduction oder neural engine slowmotion in resiolve

 

[5comma1]

Your projects should live on a PCI SSD
Enable AMD hardware accel

Thanks much for your thoughts, krakman.

I'm going to look further into PCI SSD's, for sure.

Thanks for the hardware acceleration suggestion, too.

I might attempt hardware acceleration on the RX580 if I were comfortable.
The task seems above my pay-grade... and somewhat risky.
So many terms and processes I'm completely unfamiliar with.
Too many little got'cha's, too, like needing to re-do the process after a security update,
and somewhat diminished video quality.
I like to keep my tech as close to stock as possible so it doesn't become buggy/unreliable.

I must admit, though... I sure would appreciate that speed bump!

When doing research about which GPU to purchase, I was under the impression the RX580 would provide a noticeable speed bump in FCPX, without any sort of hacking of the card. Nope.

 

[5comma1]

first you should update to FCPx 10.4.8

if you work with multiple layers, of course the
HP7101a-1 the dream.

Updated FCPX to 10.4.8, IndioX... thanks for the suggestion. Haven't yet tried it out.

What was meant by working with "multiple layers?"
Do you mean layers of media stacked above the primary story line?

 

[5comma1]

‘‘Effortless’’ 4K editing comes only with recent Thunderbolt Macs. The MP 5,1 will never have smooth 4K scrubbing.

Fast storage can make the 5,1 usable. Three recommendations I have used...

Fastest/expensive: NVMe drives, RAID 0 with a PCIe card ≈ 3000MB/s, capacity limited to 8 TB, 12 TB on a 6-blade card.

Fast/affordable: SATA SSD drives, RAID 0 ≈ 1000 MB/s. No PCIe card necessary, there are inexpensive enclosures that fit in the optical drive bay and connect directly to the SATA connector.

Large capacity/expensive: 7200 RPM HDD, RAID 0 NAS with 10G Ethernet card ≈ 500 MB/s. Slower but only solution if massive capacity is needed.

no matter what the set up, the 5,1 slows down quickly the longer the timeline and more effects/complex the project.

Really appreciate your three recommendations, XNorth.
NVMe seems like the way I'm headed.
I'm not swimming in the green stuff, so with the 7,1 Mac Pro being out of budget... NVMe looks like the cats meow.

 

[krakman]

When doing research about which GPU to purchase, I was under the impression the RX580 would provide a noticeable speed bump in FCPX, without any sort of hacking of the card. Nope.

To be fair every time Apple release a new version of FCPX there are claims of improved performace with Metal supported GPU, so we may see more improvements yet. Or as someone else pointed out you may see an improvement with 2 Graphics cards installed but the quickest step up in speed will be to install an NVME SSD and keep your projects on it.

 

[5comma1]

Install a pair of RX 480 8GB blower type cooling graphics cards
Enable AMD hardware acceleration:
Install a HighPoint SSD7101A-1 four slot NVMe M.2 PCIe Card

Thank you, Snow Tiger— for the suggestions and RX 480 pic.

Regarding a second GPU
Would I need to use an identical pair of RX 480's, or could I keep my existing RX 580 and purchase one RX 480?

Regarding the HighPoint SSD7101A-1 four slot NVMe M.2 PCIe Card
Forgetting about the addition of a second GPU for a moment... if I were to go with the Highpoint card and NVMe drives... do you have any sense of what kind of a FCPX speed enhancement I might experience?
In other words... with read/write speeds greatly enhanced... would operations of the FCPX UI improve? Do you think timeline scrubbing and playback would improve? Would I no longer need to create proxy files for 4K?

Thanks again!
[automerge]1580155825[/automerge]

To be fair every time Apple release a new version of FCPX there are claims of improved performace with Metal supported GPU, so we may see more improvements yet. Or as someone else pointed out you may see an improvement with 2 Graphics cards installed but the quickest step up in speed will be to install an NVME SSD and keep your projects on it.

Got'cha, krakman.
Understood and appreciated.
Thanks.

 

[Snow Tiger]

Thank you, Snow Tiger— for the suggestions and RX 480 pic.

Regarding a second GPU
Would I need to use an identical pair of RX 480's, or could I keep my existing RX 580 and purchase one RX 480?

Regarding the HighPoint SSD7101A-1 four slot NVMe M.2 PCIe Card
Forgetting about the addition of a second GPU for a moment... if I were to go with the Highpoint card and NVMe drives... do you have any sense of what kind of a FCPX speed enhancement I might experience?
In other words... with read/write speeds greatly enhanced... would operations of the FCPX UI improve? Do you think timeline scrubbing and playback would improve? Would I no longer need to create proxy files for 4K?

Thanks again!

Yes , a matching pair of the blower type RX 480s . The RX 580 is similar but you don't want to risk rendering data being sent to the wrong card and causing a kernel panic . It's a conservative build practice .

So , remove the RX 580 and deploy elsewhere .

I'm a system builder , not an editor . But I do get a lot of feedback from editing clients on what works .

The highpoint card with three decent SSDs in a RAID 0 array will have around 5,000 MB/s throughput . Editing 4K RAW at the lowest settings requires around 1,300 MB/s on the working drive . This is incidentally TB2 performance . So , this array should be able to handle formats that are less demanding due to compression methods .

Here's the performance results of a 6TB RAID 0 array of three 2TB Intel 660p NVMe SSDs on the Highpoint in a cMP . This was built for a classic music composer's scoring machine . He's overjoyed and even his local tech was impressed :

I'm not swimming in the green stuff, so with the 7,1 Mac Pro being out of budget... NVMe looks like the cats meow.

Oh gods , now you're speaking like me 🤣 .

 

[naerct]

What a great response to the OP's original inquiry. I have brought a couple of 4,1's up to date, primarily to have bootable NVMe, which really helps with my Adobe Photo apps. I don't really have anything new to offer , but I may have some suggestions that don't go to the extremes, yet may help the OP. I tried a single 960 at first, before the blades were bootable in the cMP. When the 144.0.0.0 ROM flash happened, I decided on a PCI card that used 2 blades and had a fan and a large heat sync. But the best part was that it nearly doubled the speed by converting the 16X PCI 2 slot to a PCI 3 slot. It's made by IO Crestline and also sold as Syba ($200) I get 2800MB/s reads with the 970 EVO. I don't use Raids on the NVMe devices. Mojave doesn't seem to like RAIDs for booting, it worked fine for a long time, but since I only had 2 blades, I wanted to keep my scratch files for the OS and PS, etc separated, using only one controller for the boot and one for the PS and Lightroom stuff. The HighPoint cards are even better, but not in my price range. If you have a cMP full of old spindle drives, I have a good solution. Most spindle drives don't get better than 150MB/s. Using SoftRaid Raid 5, I used all four drive bays plus one of the optical bays for a 5-drive (4TB). This makes those same drives work at 450MB/s reads. This gives me 16TB of space and redundancy for the failure of one disk. When that happens, it's still fully functional so you can order a new drive and go back to work. SoftRaid 5 is about $175 and doesn't work with APFS which is required with Mojave. I just use APFS for the SSDs and HFS+ for the SoftRaid data disks. I've been using it in a eSATA 4-bay external as well and I set up a Thunderbay unit for an iMac as well. All have been working for about 4 years, and SR also has excellent drive information and diagnostics that DU omits.

 

[5comma1]

Yes , a matching pair of the blower type RX 480s . The RX 580 is similar but you don't want to risk rendering data being sent to the wrong card and causing a kernel panic . It's a conservative build practice .

So , remove the RX 580 and deploy elsewhere .

I'm a system builder , not an editor . But I do get a lot of feedback from editing clients on what works .

The highpoint card with three decent SSDs in a RAID 0 array will have around 5,000 MB/s throughput . Editing 4K RAW at the lowest settings requires around 1,300 MB/s on the working drive . This is incidentally TB2 performance . So , this array should be able to handle formats that are less demanding due to compression methods .

Here's the performance results of a 6TB RAID 0 array of three 2TB Intel 660p NVMe SSDs on the Highpoint in a cMP . This was built for a classic music composer's scoring machine . He's overjoyed and even his local tech was impressed :

View attachment 890918



Oh gods , now you're speaking like me 🤣 .

Thanks for the speed numbers, Snow Tiger. Those ARE impressive!
How those numbers translate into real-world FCPX experience is lost on me, but I trust that kind of an improvement will make a difference; it would have to. I'm glad your client is happy with his setup.

It may be contained in your previous post and I'm misunderstanding... but... how about if I were to purchase a second RX 580 instead of the two RX 480's? Is that an option? Is my RX 580 not a blower type?

Thank you once again.
[automerge]1580161194[/automerge]

What a great response to the OP's original inquiry. I have brought a couple of 4,1's up to date, primarily to have bootable NVMe, which really helps with my Adobe Photo apps. I don't really have anything new to offer , but I may have some suggestions that don't go to the extremes, yet may help the OP. I tried a single 960 at first, before the blades were bootable in the cMP. When the 144.0.0.0 ROM flash happened, I decided on a PCI card that used 2 blades and had a fan and a large heat sync. But the best part was that it nearly doubled the speed by converting the 16X PCI 2 slot to a PCI 3 slot. It's made by IO Crestline and also sold as Syba ($200) I get 2800MB/s reads with the 970 EVO. I don't use Raids on the NVMe devices. Mojave doesn't seem to like RAIDs for booting, it worked fine for a long time, but since I only had 2 blades, I wanted to keep my scratch files for the OS and PS, etc separated, using only one controller for the boot and one for the PS and Lightroom stuff. The HighPoint cards are even better, but not in my price range. If you have a cMP full of old spindle drives, I have a good solution. Most spindle drives don't get better than 150MB/s. Using SoftRaid Raid 5, I used all four drive bays plus one of the optical bays for a 5-drive (4TB). This makes those same drives work at 450MB/s reads. This gives me 16TB of space and redundancy for the failure of one disk. When that happens, it's still fully functional so you can order a new drive and go back to work. SoftRaid 5 is about $175 and doesn't work with APFS which is required with Mojave. I just use APFS for the SSDs and HFS+ for the SoftRaid data disks. I've been using it in a eSATA 4-bay external as well and I set up a Thunderbay unit for an iMac as well. All have been working for about 4 years, and SR also has excellent drive information and diagnostics that DU omits.

Speaking of great responses to the OP's original inquiry, naerct, your's is fantastic, too.
I'm working on digesting it now.
Thank you very much!

 

[Snow Tiger]

Mojave doesn't seem to like RAIDs for booting, it worked fine for a long time,

The latest macOS versions do not allow booting from RAID . I think High Sierra was the last OS that supported that .

SoftRaid 5 is about $175 and doesn't work with APFS which is required with Mojave. I just use APFS for the SSDs and HFS+ for the SoftRaid data disks. I've been using it in a eSATA 4-bay external as well and I set up a Thunderbay unit for an iMac as well. All have been working for about 4 years, and SR also has excellent drive information and diagnostics that DU omits.

RAID for data drive arrays is permitted and it's best to use HFS+ ( AKA Mac OS Ext Journaled ) formatting with cMPs . Avoid APFS whenever permitted with cMPs . We have no choice with the boot devices , but with data drives and drive arrays it's important to use HFS+ for performance reasons .

 

[Snow Tiger]

It may be contained in your previous post and I'm misunderstanding... but... how about if I were to purchase a second RX 580 instead of the two RX 480's? Is that an option? Is my RX 580 not a blower type?

The Apple recommended RX 580 card is actually a bit wider than two single wide PCIe slots . Your cMP only has 5 slot positions ( one double wide slot and three single wide slots ) . Therefore , while it's possible to physically install two official RX 580 , they'll hog all the slots . You might find room for an additional thin single wide card , but you'll risk damaging something .

And then we need to talk about power . The official RX 580 card has a single 8 pin PCIe booster power connector . The way to feed this one card is from the logic board's two six pin mini PCIe connectors , using an adapter cable ( 2 x 6 pin mini PCIe to 1 x 8 pin standard PCIe ) . In order to install two of these cards , you will need to find additional sources of booster power . The way to do that is with the Pixlas mod or to gain access to an external PSU . My personal cMP has four holes drilled into its side panel access door for exactly this reason . If you were really determined , you could install a pair of Vegas in your cMP .

Now you know why the RX 480s are so nice . Each one only takes up two single wide slots and only needs a single 6 pin mini PCIe power connector from the logic board . They are also only around 15 percent slower than the 580 .

 

[XNorth]

I'm afraid I must disagree with you there
on my main 5.1 i can even cut UHD H265 and scrubbing in FCPx and davinci resolve.

ok, multicam with UHD H265 is not possible, but in prores 444
(i usually convert everything or get prores from the arri alexa anyway)

and with my setup there is no slowdown - exept noise reduction oder neural engine slowmotion in resiolve

Yes, you can cut and scrub with the MP 5,1.

But I am of course comparing my 5,1 with my MPB 16", 4K native H264. There is no comparison. Everything is more effortless and smoother on the MBP 16.

 

[5comma1]

Now you know why the RX 480s are so nice . Each one only takes up two single wide slots and only needs a single 6 pin mini PCIe power connector from the logic board . They are also only around 15 percent slower than the 580 .

Understood, Snow Tiger.
Thanks for the thorough explanation about why dual RX 480's are preferred.
Much appreciated.

 

[5comma1]

SoftRaid 5 is about $175 and doesn't work with APFS which is required with Mojave. I just use APFS for the SSDs and HFS+ for the SoftRaid data disks. and SR also has excellent drive information and diagnostics that DU omits.

naerct... some questions about your 5 spindle drive 16TB RAID, please:

1) do you backup this RAID, or do you completely trust the 'if one fails' design of the SoftRaid RAID 5?
2) I assume you purchased and are using SoftRaid to create your RAID 5 because it's not possible to create this same RAID 5 using OS X? Is this the case?
3) What does the acronym 'DU' mean?

Thank you!

 

[naerct]

Thanks for the speed numbers, Snow Tiger. Those ARE impressive!
How those numbers translate into real-world FCPX experience is lost on me, but I trust that kind of an improvement will make a difference; it would have to. I'm glad your client is happy with his setup.

It may be contained in your previous post and I'm misunderstanding... but... how about if I were to purchase a second RX 580 instead of the two RX 480's? Is that an option? Is my RX 580 not a blower type?

Thank you once again.
[automerge]1580161194[/automerge]

Speaking of great responses to the OP's original inquiry, naerct, your's is fantastic, too.
I'm working on digesting it now.
Thank you very much!

Thanks 5

 

[naerct]

naerct... some questions about your 5 spindle drive 16TB RAID, please:

1) do you backup this RAID, or do you completely trust the 'if one fails' design of the SoftRaid RAID 5?
2) I assume you purchased and are using SoftRaid to create your RAID 5 because it's not possible to create this same RAID 5 using OS X? Is this the case?
3) What does the acronym 'DU' mean?

Thank you!

I backup the internal Raid5 with the 4-bay external and a USB3, 8TB HD Elements. I also have most of that stuff on the original 1TB and 3TB disks, stored at a different location, so YES. No, I don't rely on the redundancy for safety, mostly for not having to stop working, and then having it completely rebuild itself with the new drive.
Yes, I did a lot of research before my purchase. I found out the bad part of using a hardware raid as an external backup. In this case if the hardware raid fails, you need to go out and buy that exact same unit in order to get your data back. So you end up replacing a bad situation with a replacement bad situation. So I went for the SoftRAID which is also more versatile, RAIDing drives anywhere in your system, and have Raids inside of Raids. The Apple Raid only has 0 or 1, striped or mirror. Obviously RAID0 gives you the best performance and only requires 2 disks, but if either drive(s) fail, all your data is lost. First you have to order a new drive before restoring your RAID. I find that acceptable, but not ideal. If you add a third drive and use RAID5, your speed should be comparable to two drives in Raid0, but you useable storage in only equal to two drives. The benefits of SR are numerous, but they can't deal with APFS...yet.
DU is Disk Utility, the Mac disk manager.

 

[Melbourne Park]

Yes , a matching pair of the blower type RX 480s . The RX 580 is similar but you don't want to risk rendering data being sent to the wrong card and causing a kernel panic . It's a conservative build practice .

Now that Radeon RX 480 GPUs are unobtainable, what pair of GPUs do you now use that has enough room to fit and meet the rest of the cMP's criteria? Most GPUs that Apple approve of seem to be 40mm wide, and thereby take up two slots.

 

[edgerider]

my two cent on this :
I’ve now done something like 10 maxed out cMP 4.1 or 5.1... and have used them extensively over the past 5 years in a true broadcast environment.
all of wich was only possible because of the help of good people here!

regarding video editing in 4k all of what is mentioned here is quite true but I feel that it is more of hobbyists / passionated one man band/ solo entrepreneur feedback.

no jugement or pun intended here.

I know own a 16 core 7.1 with 384 gb of ram and two rx5700, an anfeltec 6ssd card and for 90% of the task it is a tad faster than my maxed out old cMP but it is not « night and day » as long as I dont use new format such as hvec or heif.

the neglected key to make a 5.1 fly is workflow and understanding video format.

I sold one of my former cMP 5.1 to a small production company I was working with and I ended up having to take it back because « their 15 inch macbook pro were faster » so I went there to see what their workflow was and their filmer/editor was filming in Avhcd on a sony a7r, offloading the rushs on a qnap nas, and start editing with the macbook.

I do not work like this everything I shoot is raw or high bandwidth codec, so I have always ingested my footage then convert everything to prores, then worked on it.

And the sad truth here, is that to work with this type of workflow, the key ingredient is a true professional storage based on SAS when you work with very high bandwidth rushes such as 4kp60 Prores4444HQ. a cMP 5.1 with a rx5700 and a promise Vtrack or a Netapp ds4246 storage , twin x5690,128gb of ram, with one sm951 for boot and 1 samsung 970 pro for cache and an areca 1882 in slot 2 will make circle around the most upgraded 6000$ macbook pro cutting prores 4k.
the macbook would choke with two stream of prores4k where I could have 3 4kprores stream in my timeline.
because in prores the processors dont do much...

BUT in 4k mjpeg or avhcd or hvec, the macbook would deal with it like a champ while the cMP would be sluggish as hell even with hardware gpu acceleration on...

so the final point I want to make is :

if you work mainly on « prosumer codec » with short clips, and fast paced workflow for youtube very new and very compressed, go for a newer machine such as a imac 5k or a 16 » with a egpu and some kind of thunderbolt storage that is above 1000mb/s.

If you mainly work with raw and pro codec such as avid Dnx /prores/cineDng need to do accurate grading and edit very long features, go with a cMP and a true 12/16/24 drive SAS direct attached storage and a very high end raid card that takes 8gb of cache. use only SM951 for boot and 970pro for cache.

at the end I was rocking my cMP with a 5slot 24bay pcie expension enclosure 16x with one extra rx580 and an areca 1880 and this thing was flying.

The 7.1 is just capable of both and I run it with a promise Vtrak 16bay enclosures hoked up to a areca 1882, and it is rock solid.

I kept one cMP and some netapp bays that are in my basement for archival, hooked up with a atto fastframe nt12 to have 192 Tb of « cold storage » that run at 800mb/s.

netapp or vtrack sas chassis can be had for as little as 500$ with 2tb drive, and when hooked up with a 300$ areca 1880 raid card they deliver steady 1500/1800 mb/s R/W when most fast ssd will drop down to 40mb/s when you R/W more than 200gb of files...

a 5 minute video with dslr or mirorless footage will never exceed 200gb of data, so a newer machine will always be faster... But once you hit the « longer than 15 minutes, 3 camera, 4k60 and 300/500 mb/second bandwith with uncompressed data in read/write the cMP with pro grade sas and a good Gpu such as the radeon7 or a rx5700 will litterally smoke anything else than the new macpro or a very high end Imac pro with a promise R8 thunderbolt 3 storage...

last thing : if you work professionally, you can not rely on « forum fixes » this is exactly why 6-7 years ago I went from hackintosh to real cMP when they went below 2000$ used.

and this is why now, I went from cMP to a 7.1.

my top 5.1 config with the pcie expender was still plenty fast enough for what I was doing, but once I started to have to instal opencore to enable gpu acceleration, I flet I was back to hackintosh and this was just not an option for me, as it is just a tool.

by the end of the day, unless you get a fully upgraded machine used from someone for less than 1500$ and spend the additional 1000$ to get a pro grade SAS storage to match it, a 2000$ used high end imac 5k . or a used high end 2500$ macbook pro with a egpu will be way smoother.

I hope it will help you.

 

[e1me5]

Nothing really to add, except of my experience.
I've replaced the GPU of my 5,1 with an RX580 with hardware acceleration enabled and added a USB 3.1 PCI card for faster I/O. 4K video of Sony Fs7 and A7iii cut and play just fine from an external 7200 rpm RAID 0 Caldigit VR2 drive. But the Phantom 4 Pro files are kneeling the mac when they are on the FCP timeline. I'm starting to believe the DJI codec is just a mess and the only way to work with it is by transcoding it to PR.

 

[edgerider]

Nothing really to add, except of my experience.
I've replaced the GPU of my 5,1 with an RX580 with hardware acceleration enabled and added a USB 3.1 PCI card for faster I/O. 4K video of Sony Fs7 and A7iii cut and play just fine from an external 7200 rpm RAID 0 Caldigit VR2 drive. But the Phantom 4 Pro files are kneeling the mac when they are on the FCP timeline. I'm starting to believe the DJI codec is just a mess and the only way to work with it is by transcoding it to PR.

no it just rely on hardware coprocessor of newer consumer machines witch are not present on xeon processor : same with go-pro... my GF macbook air reads go-pro footage better than my 7.1....🤦‍♂️