MP 1,1-5,1 MP5,1 - CPU upgrade beyond 3.33 GHz?

[quadra605]

Hi! My main machine is a cMP5,1 (mid 2012) & my back-up is a cMP4,1 (early 2009)--I purchased both new when they came out. I primarily work with Pro Tools & to a lesser degree FCP X and altho I've upgraded both cMP's, I was wondering if there is anything I'm overlooking on the 5,1 that I could change and get realistic improvements--especially with FCP X. The 5,1 is currently set up as follows:

1. Single 3.33 GHz 6-core Intel Xenon CPU
2. 32 GB 1333 MHz DDR3 RAM (four 8GB sticks)
3. Sapphire Pulse Radeon RX 580 GPU (8GB) in PCI Slot #2
4. Samsung 850 EVO 1TB SSD on Mercury Accelsior S PCIe 2.0 card (System Drive) in PCI slot #1
5. (2) WD Black 7200RPM (WDC WD4004FZWX) 4TB drives mounted internally, one each for AUDIO files & VIDEO files
6. OS X Sierra 10.12.6 (primarily because it works well with Pro Tools 2018)
7. Boot ROM Version: MP51.007F.B03
8. Drive Bay #3 has a 6TB HD for Time Machine, Drive Bay #4 is empty.

Is there a CPU upgrade I should be looking at? NVME drive? Faster HD for audio/video or faster HD connection?

Altho I don't have current plans to upgrade the 5,1 to Mojave, I'm planning on installing it on the 4,1 (Single 2.66 Quad-Core Intel Xeon CPU/32GB 1066 MHz DDR3 RAM/MSI RX560 4GB GPU) to test the latest FCP X upgrade & try it out with with Pro Tools 2018 & 2019. If all goes well, I'll eventually install Mojave on the 5,1.

Any advice/ideas/suggestions are appreciated!

 

[08380728]

You could put a Xeon W3690 in it.

You should question however the value of sinking more money into this machine.
It's very old and very SLOW, this thing is a dead end now.

Do an in page find W3690 on these two pages and see where this CPU is:

PassMark CPU Benchmarks - Single Thread Performance

Benchmarks of the single thread performance of CPUs. This chart comparing CPUs single thread performance is made using thousands of PerformanceTest benchmark results and is updated daily.

www.cpubenchmark.net

PassMark Intel vs AMD CPU Benchmarks - High End

PassMark Software - CPU Benchmarks - Over 1 million CPUs and 1,000 models benchmarked and compared in graph form, updated daily!

www.cpubenchmark.net

Intel Xeon W3690 Benchmark

Now look at the:
AMD Ryzen 9 PRO 3900
AMD Ryzen 9 3950X

3 times the performance, lower power:

AMD Ryzen 9 PRO 3900 Benchmark

AMD Ryzen 9 3950X Benchmark

That Xeon W3960 is seriously embarrassing right?

This lady shows you how simple it is to build a Hackintosh, she's done all the hard work, just copy her EFI folder onto your existing Mac Pro boot drive,
put it in the new machine and boom...

1st VANILLA! AMD Ryzen HACKINTOSH Build 2019 – It Cannot Be Easier! – MORGONAUT

morgonaut.cloud

 

[macguru9999]

You have a real mac, so its worth keeping a bit longer. A hackintosh will bring a different set of problems. For the 2012, i would consider an nvme drive in a pcie card with a heatsink. Such as a samsung 970 pro 1/2 or 1 TB to get a 1500mb/sec speed boot drive. Also, upgrade the firmware of course first to 144.0.0.0. Install mojave if you want and clone a system to the nvme. Its not worth upgrading the cpu to 3.46 unless you want more ram. You can have 48 but with an x5690 you can have 64g... if that matters. You could also raid a couple of 860 evos together for non spinning storage ...

 

[bsbeamer]

These processors are EOL’d by Intel and you should be able to find decent price deals. Also more used MP5,1’s and MP4,1>5,1’s being listed in next few weeks as MP7,1’s arrive.

The 3.33>3.46 single processor upgrade is minimal performance gain. Have seen reports at less than 6% boost.

 

[solaris8x86]

1. Single 3.33 GHz 6-core Intel Xenon CPU
2. 32 GB 1333 MHz DDR3 RAM (four 8GB sticks)
3. Sapphire Pulse Radeon RX 580 GPU (8GB) in PCI Slot #2
4. Samsung 850 EVO 1TB SSD on Mercury Accelsior S PCIe 2.0 card (System Drive) in PCI slot #1
5. (2) WD Black 7200RPM (WDC WD4004FZWX) 4TB drives mounted internally, one each for AUDIO files & VIDEO files
6. OS X Sierra 10.12.6 (primarily because it works well with Pro Tools 2018)
7. Boot ROM Version: MP51.007F.B03
8. Drive Bay #3 has a 6TB HD for Time Machine, Drive Bay #4 is empty.

I'm an intensive-FCPX user with the similar hardware you have. To invest cheapest budget and to obtain biggest improvement is to enable h264 hardware acceleration to your existing RX580 graphic card and give it a try first. It should boost a lot when you are editing the timeline components. See in another thread. https://forums.macrumors.com/threads/activate-amd-hardware-acceleration.2180095/

If that doesn't suit your need. I have some advises:

I have Dual Xeon 3.46Ghz CPU x 2 on my Mac Pro 2010 with Vega 56 hardware acceleration enabled to FCPX. The result is the FCPX doesn't use much CPU power anymore excepts when on editing HEVC clips.

When editing a h264 or h265/HEVC clip on the timeline. The processing workload ratio in between the GPU and CPU power is about 98% vs 2%.

Therefore, upgrading CPU could be just a waste when working on FCPX to your existing situation if you have a hardware enabled GPU graphic card such as Vega 56 (work out directly and no need any Mod) or Vega 64 (need a power cable Mod to PSU). Notice that in such case. You must upgrade to OSX 10.14.6 in order to use these GPU cards in optimal performance (driver level).

To FCPX. Any storage (SSD or NVMe) which can achieve a [4K random read at 30MB/s] is more than enough in timeline editing. Those NVMe performance figures about the sequential read/write at 1500MB/s...2700MB/s etc are technically useless to FCPX. Because almost all applications on Mac rely on random I/O to improve performance (except disk-to-disk backup or file transfer). Not by the sequential I/O. So NVMe is nice to have but not really that noticeably useful to FCPX and you won't find much different even after upgrading to NVMe. A SSD (500MB/s read/write that kind) runs at native SATA3 6Gbps speed will do its job well on FCPX if your GPU has h264/h265 hardware acceleration enabled in Mojave. If you can afford, the second priority is to upgrade to NVMe storage, and the third priority is the CPU.

Upgrading CPU or to have more CPUs is only useful if you want to do software-based h264 encoding such as on Handbrake application to produce a better compression quality to the movie than GPU hardware h264/h265 encoding can provide. In such case, you need more CPU power, not more GPU power. But on FCPX, you need GPU power, not CPU power. For example, if you would like to export a Blu-ray Disc movie to a MKV file and want to achieve the best compression quality to the movie file. You must use software-based h264 encoding by Xeon CPU. The GPU hardware encoding is fast but it cannot achieve quality (you would find rough and blocks in the pixel).

Telling from my experience on FCPX.

 

[08380728]

solaris8x86 > I'm assuming here your experiences were with NVMe SSD's were all flash based, yeah?

Your experiences may have been somewhat different had you had an Intel Optane SSD PCIe NVMe in the mix because practically all Flash based SSD are on the same level in terms of random read performance and the user isn't going to notice much difference here. Where as the Optane is going to perform about 5x in this area.

Aside from that, the most important critical point you make is upgrading the video card, as you say this is going to have the most dramatic affect on rendering, but is there enough power in the cMP to run these things, are they not 300W? Radeon Vega Frontier Edition Air Retail I mean.

In that case, more valid is to build a hackintosh for cheap. He'd have a whole new smoking fast system end to end night and day faster than any cMac Pro. No power PCIe power limitations imposed like in the cMac Pro etc

 

[solaris8x86]

solaris8x86 > I'm assuming here your experiences were with NVMe SSD's were all flash based, yeah?

Your experiences may have been somewhat different had you had an Intel Optane SSD PCIe NVMe in the mix because practically all Flash based SSD are on the same level in terms of random read performance and the user isn't going to notice much difference here. Where as the Optane is going to perform about 5x in this area.

I have a (1) NVMe drive at 1500MB/s, (2) a SAS RAID-0 drive at 1500MB/s and (3) a single SSD at SATA3 500MB/s installed to my Mac Pro 2010. When working on FCPX timeline on these 3 different type drives. The smoothness performance is almost identical when GPU hardware acceleration is enabled. So I can confirm to FCPX case. A general SSD runs at native speed 6Gbps will do its job fine on FCPX timeline. Since this thread is all about FCPX. So I can conclude this result within the scope.

Of course, to other applications. I can't say.

 

[08380728]

I really do not understand people's motivation or their logic for using RAID on SSD let alone NVMe SSD. There's not the same performance gains compared that seen on mechanical disks.
I know it is often advised but thats' likely more habitual past experience with HDD rather than examining the real world application variables on SSDs.

ReadWrite rates on SSD Raid is only marginally improved.

Seek and access is significantly slower on SSD Raid.

Wear levelling of cells are equal which consequently results int two drives failing at the same time once MTBF is reached.

SMART is not supported in RAID.

The sum of these disadvantages should be considered more carefully before deciding if a few MB/s sequential R/W is worth it.
[automerge]1576945205[/automerge]
Anyway, if I was this guy and have a bit of budget, it's be sinking it into a new build to gain significant performance in every area on the system rather than singularly pin pointing 1 metric. These old machines as said are just too slow for day to day use let alone pushing video around.

 

[solaris8x86]

I really do not understand people's motivation or their logic for using RAID on SSD let alone NVMe SSD. There's not the same performance gains compared that seen on mechanical disks.
I know it is often advised but thats' likely more habitual past experience with HDD rather than examining the real world application variables on SSDs.

FCPX users only concern the result. And It's a done result.
Pseudo theory without practical achievement should be avoided and necessary.

 

[tommy chen]

You could put a Xeon W3690 in it.

You should question however the value of sinking more money into this machine.
It's very old and very SLOW, this thing is a dead end now.

Do an in page find W3690 on these two pages and see where this CPU is:

PassMark CPU Benchmarks - Single Thread Performance

Benchmarks of the single thread performance of CPUs. This chart comparing CPUs single thread performance is made using thousands of PerformanceTest benchmark results and is updated daily.

www.cpubenchmark.net

PassMark Intel vs AMD CPU Benchmarks - High End

PassMark Software - CPU Benchmarks - Over 1 million CPUs and 1,000 models benchmarked and compared in graph form, updated daily!

www.cpubenchmark.net

Intel Xeon W3690 Benchmark

Now look at the:
AMD Ryzen 9 PRO 3900
AMD Ryzen 9 3950X

3 times the performance, lower power:

AMD Ryzen 9 PRO 3900 Benchmark

AMD Ryzen 9 3950X Benchmark

That Xeon W3960 is seriously embarrassing right?

This lady shows you how simple it is to build a Hackintosh, she's done all the hard work, just copy her EFI folder onto your existing Mac Pro boot drive,
put it in the new machine and boom...

1st VANILLA! AMD Ryzen HACKINTOSH Build 2019 – It Cannot Be Easier! – MORGONAUT

morgonaut.cloud

what a load of rubbish