My nMP Review (vs my old 4,1 Mac Pro)

[VirtualRain]

So I'm confused - are you saying that Aperture is basically not using the GPU at all? As a photographer I was going to upgrade the GPU but if there is no benefit that would change my purchase decision. Maybe aperture X will utilize the GPU more fully...

I wouldn't say that at all.

First, we don't really know what we are looking at in that graph. About all we can say with any certainty, is that the lines move when doing things in Aperture, which means Aperture IS using the GPU for OpenGL.

Second, the red line indicates how much the CPU is "waiting" on the GPU. So I'm not even sure seeing this go up is good or bad. This is how naive I am at these measurements.

So, I certainly would not alter your desired config on this rather terrible assessment

Given Apple's frantic focus on OpenCL I will not be surprised to see all of their Pro apps optimised towards it in their next versions. Perhaps that is why we are eagerly awaiting Aperture X for some time now...

Aperture basically never got updated to support any of the standards entirely, and looking at iMovie and Final Cut Pro X, there is no doubt this is where Apple is banking its mula. The other graphics developers are following suit, with Adobe also upgrading their CC apps. So, Aperture X here we go...

Yes, dual GPU support could be the very reason why Apple hasn't update Aperture to X yet. I'm waiting for it as well as I can't stand how Lightroom workflow works. The thing I'm wondering now is, do LogicProX makes use of the dual GPU

Totally agree. I certainly hope a version of Aperture that better utilizes the nMP is coming soon, however, even if not, I'm super impressed with how it's performing now.

Allthough I don't use Aperture, I hope it is upgraded, as that will force Adobe to tweak performance on Lightroom, and not just add new cameras.

Any chance you could install the LR demo and compare it to the performance of Aperture regarding the CPU and GPU usage?

Sorry man, I just have too much on the go to do any kind of LR testing for you... and I don't really know how to use the app which would either cause me to invest more time or produce test results that are in question... so it's probably best to wait until someone who uses LR regularly gets a nMP to do some benchmarking.

Thanks a TON for doing this! Very interesting and helpful. I assume Handbrake was maxing out all cores (hex core right)... If so then leveling off at 85degC really isn't that bad - though a part of me was wishing (just from an old computer builders point of view) with really just the CPU taking all the thermal mass it would stay closer to the 60degC range. But 85degC is way better than 100degC (like the imacs, minis and MBP's)...

Thanks again!

No worries... yeah, Handbrake was definitely maxing out all cores. I'd have to fire it up and test it again to be sure, but I seem to recall my oMP actually ran hotter under Handbrake than this machine (I vaguely recall seeing temps in the 90's with the 4,1).

 

[analog guy]

Thanks a TON for doing this! Very interesting and helpful. I assume Handbrake was maxing out all cores (hex core right)... If so then leveling off at 85degC really isn't that bad - though a part of me was wishing (just from an old computer builders point of view) with really just the CPU taking all the thermal mass it would stay closer to the 60degC range. But 85degC is way better than 100degC (like the imacs, minis and MBP's)...

Thanks again!

x2. thank you so much, VR!

i maxed out all the cores on a late-2013 iMac (i7 3.5) last night and was exceeding 95C with the fans on. granted, i haven't been able to do that in actual usage just yet, but it's a data point.

(have to make a keep/return decision on the iMac in a few days; unfortunately i won't have the nMP in time to make my own head-to-head comparisons...including LR.)

 

[analog guy]

No worries... yeah, Handbrake was definitely maxing out all cores. I'd have to fire it up and test it again to be sure, but I seem to recall my oMP actually ran hotter under Handbrake than this machine (I vaguely recall seeing temps in the 90's with the 4,1).

just tested my MP3,1 (2x2.8). intel power gadget doesn't work with multiple processors, but i was only able to get it to 57C with all 8 cores maxed out.

 

[slughead]

What he really said:

"...with a relatively heavy 4K render in Final Cut Pro. "

" I quit FurMark and kept the FCP render going, which brought system power down from ~380W to ~326W. It also slowly dropped fan speed from 1900 RPM down to 1500 RPM before smoothly alternating between 1200 RPM and 1450 RPM. "

He measured 35.2 to 38.6 dBA between 1193 and1440 RPM. And finally:

"The system never gets annoyingly loud, it just varies between silent and calmly reminding you that there’s a large fan inside. Even at full bore the Mac Pro’s fan noise is more pleasant than harsh."

Ah okay. It's been a while since I read it, so 35-38 on the FCP load.

That's still not all that loud as long as the tone isn't annoying. It is hardly the 20dBA Apple proclaimed on their website though.

 

[VirtualRain]

just tested my MP3,1 (2x2.8). intel power gadget doesn't work with multiple processors, but i was only able to get it to 57C with all 8 cores maxed out.

Hmm... ok, I must have been on drugs thinking 90's on the oMP then.

 

[analog guy]

Hmm... ok, I must have been on drugs thinking 90's on the oMP then.

hm, well i can't comment on any drugs....but maybe the 3,1s run a bit cooler than the 4/5,1s?

i only had the full load running for 10-15'. i was surprised temps were that low at full tilt. RAM on the other hand, gets hotter.

 

[LongSticks]

Thanks, msged him

Come on then, what did you find out? Been following your specviewperf requests for ages!!!!

 

[propower]

just tested my MP3,1 (2x2.8). intel power gadget doesn't work with multiple processors, but i was only able to get it to 57C with all 8 cores maxed out.

My oMP 4,1 2.93 Never got above 60degC even on Handbrake encodes. The heatsink on the CPU in those was serious stuff. Although I would have wished the nMP was similar ...

 

[ytoyoda]

Ah okay. It's been a while since I read it, so 35-38 on the FCP load.

That's still not all that loud as long as the tone isn't annoying. It is hardly the 20dBA Apple proclaimed on their website though.

I think, the noise level Apple publishes, is measured at 1 meter from the source and in an anechoic chamber.
I looked at AnandTeck' review again.
They measured it at 18", so less than 0.5 m, and when the fan is off, it was already 32.2 dBA, which was ambient noise.
So, you can't simply compare the value.

 

[wheelhot]

Come on then, what did you find out? Been following your specviewperf requests for ages!!!!

He hasn't replied yet

But if SolidWorks Real View is activated by default, that's indeed a good sign

 

[CaptainChunk]

I'm late to the party here, but thanks for your review, VirtualRain! It was very detailed and informative, and it's always nice to hear first-hand reviews from actual professionals that rely on their systems to get work done. Tech site reviews only do so much in that regard...


As a film/video editor, I think my biggest fear of ditching my trusty oMP (a 2008 model that's getting long in the tooth) in favor of an nMP has been losing the 4 internal drive bays. They've been pretty darned convenient in my case. Yes, they're SATA-II, but I've also never installed an SSD in that machine for anything other than boot. That's chiefly because their prices and capacities remain unrealistic for working with modern HD video codecs that chew up gobs of disk space.

On the other hand, elaborate RAID arrays have never really been my thing, either - for a couple of reasons. First, offline video editors such as myself don't really deal with uncompressed video much; that's what post houses are for. Secondly, I've always found that a proper backup strategy not too dissimilar from what you do is superior and less complicated, not to mention cheaper.


But before I can truly overcome the loss of those 4 internal SATA bays, I'd really like to see some manufacturers step up to the plate and offer up affordable 4-bay TB enclosures that are not overpriced turn-key solutions populated with proprietary drive modules. One would figure that by now, someone would have figured this out. So far, the closest things we have to this are LaCie's TB to eSATA adapter (I don't even know if that one supports port multiplication for enclosures) and a dual-bay dock from Highpoint that seems to accomplish a similar thing without needing an eSATA enclosure.

I have a LOT of bare SATA drives that I just slide in and out of my current oMP and external eSATA enclosures. It's much cheaper to do things this way, I get to choose the drive mechanisms I want to use and I don't have to deal with the mountain of power bricks that are a consequence to owning dozens of "turn-key" external hard drives.


It's sad because moving from an oMP to an nMP would yield so many good things for me...

1. Significantly faster CPU architecture
2. Memory upgrades that don't cost an arm and a leg (pricey FB-DIMMs have kept me stuck with 16GB of RAM for the last 4 years - it's really annoying)
3. Dual GPUs that wipe the floor with my current flashed 8800 GT in every conceivable measure
4. Standard (and wicked-fast) PCI-E based SSD boot volume
5. USB 3.0 ports that just work
6. No internal optical drive wasting space (people still use optical discs for data?)
7. Significantly smaller footprint; easier to move around

But again, forgoing internal drive expansion in favor of Thunderbolt 2.0 scares me at the moment. This would prove to be a good trade-off for me if and when I can have that BARE multi-bay TB enclosure that no one seems to want to make - this is really the only thing keeping me from pulling the trigger right now.

 

[VirtualRain]

But before I can truly overcome the loss of those 4 internal SATA bays, I'd really like to see some manufacturers step up to the plate and offer up affordable 4-bay TB enclosures that are not overpriced turn-key solutions populated with proprietary drive modules.

Don't let external storage be a hurdle... It's a solved problem, you just need to decide what route you want to go.

A lot of people here are opting for the Promise Pegasus2 which sells for $699 without drives. It's full TB2 and early comments are all very positive. It sounds like a great solution for those that want to migrate existing drives or bring their own. If you're in the market for new drives anyway, then I can vouch for the WD Thunderbolt Duo products (as I have one of those). They offer good value. Alternatively, if you're on a budget, maybe consider starting with a much cheaper USB3 enclosure to start with (Like the Oyen Digital Mobius). It's only $200 or so for 5 bays and receives high marks in reviews. You can always move to a Thunderbolt enclosure later when there are more affordable options.

As you point out there's so many advantages to the nMP... Get on it!

----------

I think, the noise level Apple publishes, is measured at 1 meter from the source and in an anechoic chamber.
I looked at AnandTeck' review again.
They measured it at 18", so less than 0.5 m, and when the fan is off, it was already 32.2 dBA, which was ambient noise.
So, you can't simply compare the value.

LOL... How did he get the fan up yo 1900RPM (Was that on the power virus torture test?)

In running Handbrake, my fan maxes out at 900RPM and as Anand's table says, it's not any louder than at idle (770RPM).

 

[CaptainChunk]

A lot of people here are opting for the Promise Pegasus2 which sells for $699 without drives. It's full TB2 and early comments are all very positive.

Wow. Somehow, I actually wasn't aware of this one's existence. Thanks for that! The original Pegasus was (stupidly) never offered as a barebones model. The price of the Pegasus 2 is a bit steep considering I'd likely never use its built-in hardware RAID, but I suppose it's reasonable all considering.

The Mobius actually looks wicked cool. That may be a good interim solution until TB stuff gets a little more affordable.

 

[Gav Mack]

The nMP is 45dBA with a reasonable FCP X load according to Anandtech, that's about what a lot of really powerful rigs do at load.

Having had one first hand with nothing to measure it apart from my ears, I would say the only way a powerful rig could make the same noise as the nMP it would have to be in an adjacent room with the doors shut!

This was on my desk next to my keyboard and mouse. At load my 3,1 under my desk and my hp 8000 across the room at idle made far more noise, and at nMP idle I could hear my modded ex490 micro server tucked out of visible sight more which until then was one of the quietest machines in the house. I have a pair of larger 4tb spinners to replace 3tb's to fit in the other quiet PC on my bench over the next few days, a modded Sony XL3 with a large heatpipe for CPU and a passive for GPU with slow spinning fan blowing across both that only really emanates noise you can tell from the PSU and spinners. My ears and memory will be able to tell how that rates against the little hex can I had in the same place barely over a week ago..

 

[slughead]

Having had one first hand with nothing to measure it apart from my ears, I would say the only way a powerful rig could make the same noise as the nMP it would have to be in an adjacent room with the doors shut!

That's just poor design and probably poor fans. Using larger, high quality fans running at lower RPM, cases like the Thor V2 run around 45dBA doing triple SLI GTX580's at full load.

That's not to mention the water-cooled solutions.

 

[LongSticks]

He hasn't replied yet

But if SolidWorks Real View is activated by default, that's indeed a good sign

That's a b****r!

Can you post if you do get a response, gonna be as interested and I'm a later adopter than you guys!

 

[Gav Mack]

That's just poor design and probably poor fans. Using larger, high quality fans running at lower RPM, cases like the Thor V2 run around 45dBA doing triple SLI GTX580's at full load.

That's not to mention the water-cooled solutions.

That's going into exotica territory, way away from out of the box workstation territory. I reckon you need to have a black can around you to truly appreciate it - I was sceptical and only totally won over by experiencing it's nothingness acoustically first hand.

All systems I have compared it to here are out of the box, and the XL3 and the ex490 which I have are only modded internally are certainly not poor designs as far as acoustics are concerned, the opposite. Both have only a single fan and are extremely quiet.

 

[slughead]

That's going into exotica territory, way away from out of the box workstation territory. I reckon you need to have a black can around you to truly appreciate it - I was sceptical and only totally won over by experiencing it's nothingness acoustically first hand.

All systems I have compared it to here are out of the box, and the XL3 and the ex490 which I have are only modded internally are certainly not poor designs as far as acoustics are concerned, the opposite. Both have only a single fan and are extremely quiet.

I'm not saying it's not quieter, I'm saying high power workstations don't need to run super loud. HP, Dell, and others just don't do a good job as it's not a priority. They'll have 130CFM 120mm fans and rev them up to max. No smart fan controlling.

Of course, I assume those boxes don't let the CPU get up to 90C...

 

[Infrared]

just tested my MP3,1 (2x2.8). intel power gadget doesn't work with multiple processors, but i was only able to get it to 57C with all 8 cores maxed out.

That is consistent with my experience of the 2008 MP, too. Those Xeons are more or less idling now here and their temperatures are 27C and 30C with an ambient temperature of 25C.

 

[Gav Mack]

I'm not saying it's not quieter, I'm saying high power workstations don't need to run super loud. HP, Dell, and others just don't do a good job as it's not a priority. They'll have 130CFM 120mm fans and rev them up to max. No smart fan controlling.

Of course, I assume those boxes don't let the CPU get up to 90C...

I agree, but anyone who is considering doing AV work on one of these versus a Windows workstation or even one of the older towers will be knocked out by the lack of noise, will not need their headphones on and can listen to their monitors instead on the desk!

Certainly my quiet machines don't let CPU temps go that high and neither do I on any of my Macs either, my fan control settings are far more conservative than Apples for sure

 

[wheelhot]

That's a b****r!

Can you post if you do get a response, gonna be as interested and I'm a later adopter than you guys!

Haha, no worries. I already have a thread looking for nMP owners to run a specviewperf benchmarking test. But it's taking longer then expected as not many users who run Windows + CAD, most only installed Windows to play games so it's taking longer then I liked. But no worries, I already placed an order for the base nMP, but expected shipment is in February, I'm hoping it's end of Jan or very early Feb

It looks like there are people looking forward into using the nMP with 3D CAD software as well (I'm not alone!)

FYI: I'm using SolidEdge and I wonder who would be running Siemens NX in OSX, I feel the only major client Siemens has who uses NX in OS X would be Apple

 

[slughead]

I think, the noise level Apple publishes, is measured at 1 meter from the source and in an anechoic chamber.
I looked at AnandTeck' review again.
They measured it at 18", so less than 0.5 m, and when the fan is off, it was already 32.2 dBA, which was ambient noise.
So, you can't simply compare the value.

Awesome. Considering the equation for sound dissipation (spherically, at least) is Power / (4πR^2) (inversely related to distance squared), and considering the sound of this thing is likely directed upward, all sorts of games can be played with the numbers here.

Personally, I'll compare the nMP to my gaming rig and I'm going to measure from outside my front door.

 

[Gav Mack]

Awesome. Considering the equation for sound dissipation (spherically, at least) is Power / (4πR^2) (inversely related to distance squared), and considering the sound of this thing is likely directed upward, all sorts of games can be played with the numbers here.

Personally, I'll compare the nMP to my gaming rig and I'm going to measure from outside my front door.

Lol. Aside from wheelhot's benchmark if someone hasn't already done it for him I have managed to successfully request my friends DVD of bf4 in advance and will try it out with crossfire in bootcamp. I think that will be rather more successful in getting the can truly cooking!

 

[AidenShaw]

Lol - as quiet as a gentle warm breeze too no doubt!

And they're also a bit larger than the Apple Tube.

However, some people value having 24 cores, 16 DIMMs with 512 GiB support, quad double-width Nvidia Quadros or Teslas over a quieter, smaller system with 64 GiB, 12 cores, and two desktop ATI GPUs.

Would you rather that your task takes 1 hour on a system that's barely loud enough to notice, or 4 hours on a system that you can't hear?

 

[slughead]

And they're also a bit larger than the Apple Tube.

However, some people value having 24 cores, 16 DIMMs with 512 GiB support, quad double-width Nvidia Quadros or Teslas over a quieter, smaller system with 64 GiB, 12 cores, and two desktop ATI GPUs.

Would you rather that your task takes 1 hour on a system that's barely loud enough to notice, or 4 hours on a system that you can't hear?

But... it's quiet!

Plus, everyone knows that real professionals use render-farms when they actually need that kind of power.