iMac Pro user experiences?

Get answers to your questions about color grading, editing and finishing with DaVinci Resolve.
  • Author
  • Message
Offline

Ed Rudolph

  • Posts: 131
  • Joined: Thu Apr 18, 2013 12:43 am

iMac Pro user experiences?

PostThu Dec 28, 2017 10:01 pm

I'm curious to hear stories from anyone beginning to user the new iMac Pro.
Offline

Michael Tiemann

  • Posts: 284
  • Joined: Fri Jul 05, 2013 11:22 am

Re: iMac Pro user experiences?

PostFri Dec 29, 2017 2:18 am

As a RED user I can say it's been really great so far. I have 10 core/64GB RAM/Vega 64 GPU/4 TB SSD. I also have a Red Rocket-X connected via TB3 in a Sonnet box. It cuts through 8K files like a knife through hot butter. Today is literally the first time in years that I felt like my computer was helping me work faster rather than holding be back.

There are a few times I can remember such a large step-up in performance. Moving from floppy disks to a 10MB hard drive in 1976 (Cromemco Z2-D); moving from Sun 2 to Sun 3 workstation (1986); moving from SPARC to MIPS R10000 (1996); moving from Pentium to dual-Xeon (Mac Pro 2008); and now. It's a big deal--and not a moment too soon. I've been using a 2012 Mac Pro and a 2013 new Mac Pro, and both have long fallen behind what my cameras have been demanding. Now I feel like I'm back to being able to do creative things without constantly bumping into performance limitations of one sort or another. Which feels great!
MacOS Sierra Version 10.13.5
iMac Pro (2017)
3 GHz Intel Xeon W
64GB 2666 MHz DDR4
Radeon Pro Vega 64 16 GB
Red Rocket-X in Sonnet III-D via Thunderbolt 3
Offline

Uli Plank

  • Posts: 2856
  • Joined: Fri Feb 08, 2013 2:48 am

Re: iMac Pro user experiences?

PostFri Dec 29, 2017 3:10 am

May I ask about heat and noise?
Plus, regarding the Red bug under current High Sierra, I suppose you run it under Metal?
Resolve Studio and Fusion Studio
iMac 2017 Radeon Pro 580 8 GB VRAM and 32 GB RAM
Offline

Michael Tiemann

  • Posts: 284
  • Joined: Fri Jul 05, 2013 11:22 am

Re: iMac Pro user experiences?

PostFri Dec 29, 2017 3:42 am

I'm using the default, reported to be OpenCL. It just works. Noise of iMac pro much less than spinning disks and Sonnet expansion box. Headphones help.
MacOS Sierra Version 10.13.5
iMac Pro (2017)
3 GHz Intel Xeon W
64GB 2666 MHz DDR4
Radeon Pro Vega 64 16 GB
Red Rocket-X in Sonnet III-D via Thunderbolt 3
Offline

Charles Lyons

  • Posts: 26
  • Joined: Mon Dec 04, 2017 8:18 pm

Re: iMac Pro user experiences?

PostFri Dec 29, 2017 5:11 am

Thanks for the report. How does it run without the Red Rocket (which, at $6500, costs almost as much as the entire computer and is I imagine a large part of the performance step up)?
Resolve 14.3.0.014
iMac Pro 10-core 3.0 GHz Xeon
64GB RAM / Vega 64 GPU w 16GB VRAM / 2TB SSD
High Sierra 10.13.4 (17E199)
Offline
User avatar

rick.lang

  • Posts: 8207
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

iMac Pro user experiences?

PostFri Dec 29, 2017 5:17 am

I saw results from a Geekbench 4 Test in which Metal performed very slightly worse than OpenCL. Didn’t expect that.

One peculiar aspect I’ve seen reported (AppleInsider and confirmed by other early adopters): the iMac Pro attempts to keep the processors cool by throttling the clock so your CPU may be running at 3.9 GHz but will slow to 3.1 GHz until the processor is cooler and then it ramps up to 3.9 GHz and the cycle continues. Eventually after awhile the system will ramp up the fan to help keep things cooler.

That dynamic variable clock may mean this isn’t a gamers’ machine, but that’s not likely a concern for most of us.


Sent from my iPhone using Tapatalk
Rick Lang
Offline
User avatar

rick.lang

  • Posts: 8207
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

iMac Pro user experiences?

PostFri Dec 29, 2017 5:20 am

Overall though I echo Michael’s comment about editing video on the 10 core machine: like cutting warm butter.

Michael, can you do me a favour and run the Blackmagic Speed Test on the 4 TB internal flash with 128 KB block size? Love to know how well it’s going to handle BMD video with the maximum internal SSD. Posting a sample screen capture if the he Speed Test results would be interesting to see.


Sent from my iPhone using Tapatalk
Rick Lang
Offline

Michael Tiemann

  • Posts: 284
  • Joined: Fri Jul 05, 2013 11:22 am

Re: iMac Pro user experiences?

PostFri Dec 29, 2017 12:25 pm

Here's the screen capture. Dunno how to set block size in the speed test app, but used 5GB stress as the marker:

4TB SSD DiskSpeedTest.png
4TB SSD DiskSpeedTest.png (583.61 KiB) Viewed 3608 times
MacOS Sierra Version 10.13.5
iMac Pro (2017)
3 GHz Intel Xeon W
64GB 2666 MHz DDR4
Radeon Pro Vega 64 16 GB
Red Rocket-X in Sonnet III-D via Thunderbolt 3
Offline

Glenn Venghaus

  • Posts: 685
  • Joined: Wed Jan 01, 2014 9:56 pm
  • Location: Amsterdam , The Netherlands

Re: iMac Pro user experiences?

PostFri Dec 29, 2017 4:35 pm

Saw in i-fixit teardown that the drives are 2 in raided formation .
So drive speed wise it seem there is little to wish for. Not that you would store all your multi TB data on there but its a snappy cache ;-)
So far so good (if you ignore the initial price tag scare)

Can you try a long render and see what the temp does ? That would be "the" number one biggest question mark we all have after the trashcan debacle.
Beatstep / APC-40 Resolve Edition Custom Controllers posttools.tachyon-consulting.com
HACK OSX10.12.4 | 4790K | 2xGTX980ti | 32GB | DRS 15.x
V: CUDA387.99 | 378.05.05.05f02 | DTV10.8.6 | MiniMon-SDI >HDlink > Eizo
A: 5.1RME-FireFace
Offline
User avatar

rick.lang

  • Posts: 8207
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: iMac Pro user experiences?

PostFri Dec 29, 2017 9:21 pm

Michael, very sweet!


Sent from my iPhone using Tapatalk
Rick Lang
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostSat Dec 30, 2017 12:56 pm

Michael Tiemann wrote:As a RED user I can say it's been really great so far. I have 10 core/64GB RAM/Vega 64 GPU/4 TB SSD. I also have a Red Rocket-X connected via TB3 in a Sonnet box. It cuts through 8K files like a knife through hot butter. Today is literally the first time in years that I felt like my computer was helping me work faster rather than holding be back.


It just shows how much behind Apple was/(still is) in terms of workstations. Well- they don't have one, simple as this. Even new Mac Pro is not a real workstation replacement.

What about noise during heavy working (without headphones on ears)? I would also like to see CPU/GPU throttling graph on long terms renders etc.
Offline
User avatar

rick.lang

  • Posts: 8207
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: iMac Pro user experiences?

PostSat Dec 30, 2017 6:31 pm

To compare to Michael’s screenshot with the 4 TB internal flash storage, another person tested with a 2 TB 10-core and had 2,996 MB/sec Write and 2,450 MB/sec Read.

I know every time you run these tests you’ll get some variations with no parameters changing. So it looks like there isn’t a significant performance difference between 2 TB and 4 TB SSD options.


Sent from my iPhone using Tapatalk
Rick Lang
Offline

Ed Rudolph

  • Posts: 131
  • Joined: Thu Apr 18, 2013 12:43 am

Re: iMac Pro user experiences?

PostSat Dec 30, 2017 6:38 pm

Per Marc Wielage's question about the Resolve UI on this display, any problems?
Offline

Dermot Shane

  • Posts: 1270
  • Joined: Tue Nov 11, 2014 6:48 pm

Re: iMac Pro user experiences?

PostSat Dec 30, 2017 7:22 pm

i have a cleint asking me about this machine, has anyone done a 72hr stress test, full load in normal -to- cramped office space conditions?

heat managemt is a concern, i'd have to tell them to avoid untill someone does this, not in a well ventlated lab, but in a real world office /desk crap around/ walls close to back of machine, typical edit suite type of enviroment - 72hr straight under 100% load

any takers? any one doing this as a normal part of bedding in a new machine? i do this with all my workstations, but if they failed a stress test, and only once one did, HP came on site and replaced it no questions asked within a few hours

the concern is that the Apple store would ignore, bs, mis-state , blame, dodge, avoid, wiggle, anything but sort overheating issues, like the first few years of the trashcan
Offline
User avatar

Francois Dompierre

  • Posts: 164
  • Joined: Thu Jun 23, 2016 7:42 pm
  • Location: Montréal, Canada

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 6:02 am

Dermot Shane wrote:i have a cleint asking me about this machine, has anyone done a 72hr stress test, full load in normal -to- cramped office space conditions?

heat managemt is a concern, i'd have to tell them to avoid untill someone does this, not in a well ventlated lab, but in a real world office /desk crap around/ walls close to back of machine, typical edit suite type of enviroment - 72hr straight under 100% load

any takers? any one doing this as a normal part of bedding in a new machine? i do this with all my workstations, but if they failed a stress test, and only once one did, HP came on site and replaced it no questions asked within a few hours

the concern is that the Apple store would ignore, bs, mis-state , blame, dodge, avoid, wiggle, anything but sort overheating issues, like the first few years of the trashcan


I have a 23-hour render session under way on the 10-core model (55 hours of C300mk2 footage transcoding to ProRes). I'll keep you posted! It's been a few hours and I don't even hear the fan, it's much more quiet than the maxed out regular 2017 iMac, surprisingly. There's plenty of hot air coming out of the back vent, but the fans are bigger/quieter. The machine is barely warm after hours of all 10 cores running full blast, I'm impressed.

I did a few tests (exported the same project, a 42mins tv show) on my 3 machines (2012 12-core MP with Titan X, 2017 iMac maxed out and 2017 10-core iMac Pro with Vega 64) and the iMac Pro is much, much faster, more so than anticipated. Now I kinda want to buy the 18-core. I'll report the details of my tests when I have a moment.
Last edited by Francois Dompierre on Sun Dec 31, 2017 2:43 pm, edited 1 time in total.
François Dompierre
Senior Colorist / Finishing Editor
Montréal, Canada

2012 Mac Pro 12-core 3.33GHz | 32GB RAM | Dual Titan Xp 12GB
DeckLink Mini Monitor 4K | Teranex Mini Converter (as a LUT box)
macOS 10.13.4 | Resolve Studio 15 beta 2
Offline

Dan Sherman

  • Posts: 364
  • Joined: Fri Jul 01, 2016 11:07 pm

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 6:17 am

rick.lang wrote:I know every time you run these tests you’ll get some variations with no parameters changing. So it looks like there isn’t a significant performance difference between 2 TB and 4 TB SSD options.


You need to be careful when doing tests like this, because depending on who wrote the testing software, the OS settings, and the amount of ram free, you can get highly inaccurate results.

Both OS X and Windows will use RAM as a cache. Thus to ensure accurate results use the largest possible sample sizes.
Offline

Dermot Shane

  • Posts: 1270
  • Joined: Tue Nov 11, 2014 6:48 pm

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 5:27 pm

Merci Francois!

have a 23-hour render session under way on the 10-core model (55 hours of C300mk2 footage transcoding to ProRes). I'll keep you posted! It's been a few hours and I don't even hear the fan, it's much more quiet than the maxed out regular 2017 iMac, surprisingly. There's plenty of hot air coming out of the back vent, but the fans are bigger/quieter. The machine is barely warm after hours of all 10 cores running full blast,


this is excatly the workflow my cleint is targeting, getting dalies turned around quickly to ProRez, currently useing a 5.1, so the imacpro should be a significant improvement in terms of speed

one thing their 5.1 has going for it is reliablity
Offline

Dermot Shane

  • Posts: 1270
  • Joined: Tue Nov 11, 2014 6:48 pm

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 5:42 pm

Merci Francois!

have a 23-hour render session under way on the 10-core model (55 hours of C300mk2 footage transcoding to ProRes). I'll keep you posted! It's been a few hours and I don't even hear the fan, it's much more quiet than the maxed out regular 2017 iMac, surprisingly. There's plenty of hot air coming out of the back vent, but the fans are bigger/quieter. The machine is barely warm after hours of all 10 cores running full blast,


this is excatly the workflow my cleint is targeting, getting dalies turned around quickly to ProRez, currently useing a 5.1, so the imacpro should be a significant improvement in terms of speed

one thing their 5.1 has going for it is reliablity
Offline
User avatar

Francois Dompierre

  • Posts: 164
  • Joined: Thu Jun 23, 2016 7:42 pm
  • Location: Montréal, Canada

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 5:45 pm

Dermot Shane wrote:Merci Francois!

have a 23-hour render session under way on the 10-core model (55 hours of C300mk2 footage transcoding to ProRes). I'll keep you posted! It's been a few hours and I don't even hear the fan, it's much more quiet than the maxed out regular 2017 iMac, surprisingly. There's plenty of hot air coming out of the back vent, but the fans are bigger/quieter. The machine is barely warm after hours of all 10 cores running full blast,


this is excatly the workflow my cleint is targeting, getting dalies turned around quickly to ProRez, currently useing a 5.1, so the imacpro should be a significant improvement in terms of speed

one thing their 5.1 has going for it is reliablity


The render finished without problems, stayed cool and quiet all the way. Apple designed quite the cooling system. You can tell the classic iMac's cooling system is getting near its max design capacity with the 2017 maxed out iteration (you often hear the fans, and the iMac does get hot). But the iMac Pro's cooling design is plenty sufficient for the current iteration of the iMac Pro (and has room to accommodate more for the next couple of years no doubt).

As for turning around ProRes dailies, compared to the 5,1 Mac Pro, here's one of the quick tests I ran:

42mins 1080p ProRes4444 clip transcoded to 422HQ:

17min 39sec on our Mac Pro 2012 12-core 3.46GHz (note the faster, upgraded CPU's here)
9min 40sec on our iMac Pro 2017 10-core 3.00GHz

With all cores working full blast on both machines, the iMac Pro is roughly twice as fast! Both machines have 32GB of RAM, and the export was going from one SDD drive to another in both cases, so drives do not account for the difference. Given the performance and cooling efficiency of the 10-core iMac Pro, I think the 18-core model is well worth it. Both Xeon chips are 140W, so the cooling performance should be identical. But 18 cores @ 2.2 vs 10 cores @ 3,0 -- roughly 25% faster I guess, if we extrapolate total GHz? That's an expensive upgrade for just 25% more. Maybe the 10-core is the sweet spot if budget is an issue. But if budget is no issue, I think the 18-core is going to be just as cool and quiet as the 10-core.
Last edited by Francois Dompierre on Wed Jan 03, 2018 12:12 pm, edited 3 times in total.
François Dompierre
Senior Colorist / Finishing Editor
Montréal, Canada

2012 Mac Pro 12-core 3.33GHz | 32GB RAM | Dual Titan Xp 12GB
DeckLink Mini Monitor 4K | Teranex Mini Converter (as a LUT box)
macOS 10.13.4 | Resolve Studio 15 beta 2
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 6:47 pm

Now it would be nice to see clock speed for CPU and GPU during such a render.
Question is- how much was clock reduced during this render? If it's just slightly or not at all then great. If there was a big reduction then your 10 cores CPU is not used very well at all. It's like many laptops which have powerful CPUs, but never use 100% of their power for more than 5min.
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 6:50 pm

Francois Dompierre wrote:Given the performance and cooling efficiency of the 10-core iMac Pro, I think the 18-core model is well worth it. Both Xeon chips are 140W, so the cooling performance should be identical. When all cores are working, the 18-core chip runs at 2.2GHz vs 3.0GHZ for the 10-core, and that's how it stays at 140W. But 18 cores @ 2.2 is still going to be faster than 10 cores @ 3,0. The question is by how much. Roughly 25% faster I guess, if we extrapolate total GHz? That's an expensive upgrade for just 25% more. Maybe the 10-core is the sweet spot if budget is an issue. But if budget no issue, I think the 18-core is going to be just as cool and quiet as the 10-core.


18*2.2=39.6
3*10=30
so in theory 18 cores is about 30% faster, but in reality this is going to be probably less, e.g. 20% and in some tasks it may end up that 10 core version is actually faster (due to much higher clock).
Offline
User avatar

Francois Dompierre

  • Posts: 164
  • Joined: Thu Jun 23, 2016 7:42 pm
  • Location: Montréal, Canada

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 6:57 pm

Andrew Kolakowski wrote:Now it would be nice to see clock speed for CPU and GPU during such a render.
Question is- how much was clock reduced during this render? If it's just slightly or not at all then great. If there was a big reduction then your 10 cores CPU is not used very well at all. It's like many laptops which have powerful CPUs, but never use 100% of their power for more than 5min.


With all cores running, the Xeon chips do not throttle below their rated speed. You can expect a minimum of 3.0GHz with all cores running, and something between 3.0 and 4.3 when less cores are in use. Same for any Xeon chip, in PC or Mac. Difference is, in a bigger box (PC or Mac) with better cooling, they will on average stay at a higher clock (closer to 4.3 than 3.0). But if you look at the Xeon chips available in the latest HP Z-class workstations, they are CRAZY expensive. The more affordable options do not go all the way to 4.2, so it would appear that the iMac Pro is not that expensive after all. Man, Intel makes a ton of money on those Xeon's!! It will be interesting to see if Threadripper forces them into lowering their prices.

http://store.hp.com/us/en/pdp/hp-z6-g4- ... /z6-buynow
François Dompierre
Senior Colorist / Finishing Editor
Montréal, Canada

2012 Mac Pro 12-core 3.33GHz | 32GB RAM | Dual Titan Xp 12GB
DeckLink Mini Monitor 4K | Teranex Mini Converter (as a LUT box)
macOS 10.13.4 | Resolve Studio 15 beta 2
Offline

Dermot Shane

  • Posts: 1270
  • Joined: Tue Nov 11, 2014 6:48 pm

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 7:32 pm

if you look at the Xeon chips available in the latest HP Z-class workstations, they are CRAZY expensive. The more affordable options do not go all the way to 4.2, so it would appear that the iMac Pro is not that expensive after all.


i'm seing a match for the imacpro 3*10 = 30/32g priceing in a z6 @ $2895

i'm counting turbo as my z820's run at full turbo for days and days with dropping speed, i'm assumeing the cooling in the z6 is every bit as good

now it's also true that the full wack z8 is crazy expensive, but then compare the maxed out imacpro 18*2.2 = 39.6 -vs a maxed out z8 28*3.8*2 = 212.8.... yea, apples to some giant awesome fruit that's 5.5 times larger.... and costs 10x more ;-)

or a z6 with comparable spec to the 10core imacpro @ 4.5k -vs- 10k USD...

i'd be buying the z6 myself, but based on your very valuable feedback, i'd advise my clients to go ahead with the 10 core imacpro and keep the 5.1 handy for atleast six months as it took that long for the overheating issues with the trashcan to start appearing
Offline

Christopher Irion

  • Posts: 44
  • Joined: Wed Jun 25, 2014 8:40 pm
  • Location: San Francisco/Putney, VT

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 9:10 pm

Hi Ed-

Can't report good or bad so far as the migration from Time Machine backup produces error permissions messages saying Resolve does not have permission for cache clip/.gallery/Capture paths and I can't update them in project settings. Also, very strangely cannot give read/write permissions to any user, administrator or myself or add additional users.

Apple techs did some root level processes through the terminal that they were very surprised to find din't fix the problem. Of course, with a brand new machine they have been very attentive but were unable to find a fix. Supposedly they are spending their holiday trying to figure it out with a phone conference on Tuesday morning, but I am beginning to suspect that it may be a resolve 14.0.1/OS 10.13.2/PostgreSQL issue.

I'm planning to post elsewhere in more detail to see if I can get some response to potential upgrade issues unrelated to iMac Pro.

Frustrating to not have any sense of what this thing can really do!
Resolve 15.0.0B.065

iMac Pro (2017
10.13.2 (17C212)
3.2 GHz Intel Xeon W (10 Core)
64 GB 2666 MHz DDR4
Radeon Pro Vega 56 8 GB
Offline
User avatar

rick.lang

  • Posts: 8207
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 9:50 pm

As I stated previously, the 10-core under load is throttled to keep the temperature moderate and the fans fairly constant. The 10-core seems to run mostly at 3.9GHz and dips briefly to just below 3.2GHz until the temperature drops and then quickly resumes using 3.9GHz. So most of the time with multiple cores active, you get much more than Apple’s published clock speed.


Sent from my iPhone using Tapatalk
Rick Lang
Offline
User avatar

Jack Fairley

  • Posts: 652
  • Joined: Mon Oct 03, 2016 7:58 pm

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 10:17 pm

I'm more interested in the thermals for the 18c model. Glad to see Apple actually innovated with the cooling in the Pro, but such a packed chip could still be a nuclear inferno on long jobs.
Offline

Dermot Shane

  • Posts: 1270
  • Joined: Tue Nov 11, 2014 6:48 pm

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 10:43 pm

had not looked at the hp store before, for the same price as the 10c imacpro i can config a z6 with dual 18c's / 64g/dual 1080Ti's

z6; 18*3.7*2 = 133.2Ghz
imac; 18*2.2*1 = 39.6Ghz
for the same price

not sure the imacpro is very cost effective, if it was me, id likely get a moderately spec'd z6 for 5-6k, and a licence of Nucoda look for 3k to make Prorez dalies before i 'd spend 10k on an imacpro and .3k on a Resolve licence

but i am telling my clents to go ahead with caution, they are kinda afraid of anything without an apple logo on it ;-)
Offline
User avatar

Francois Dompierre

  • Posts: 164
  • Joined: Thu Jun 23, 2016 7:42 pm
  • Location: Montréal, Canada

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 11:38 pm

Dermot Shane wrote:had not looked at the hp store before, for the same price as the 10c imacpro i can config a z6 with dual 18c's / 64g/dual 1080Ti's

z6; 18*3.7*2 = 133.2Ghz
imac; 18*2.2*1 = 39.6Ghz
for the same price



Yeah but the Z6 is 30% off today only (year-end sale) and it doesn't include a 5K screen, and you're comparing Max Turbo boost speeds to minimum base guaranteed speeds in the procs, so not exactly fair ;) But I guess thats's the million dollar question, what's the actual speed of those chips when under full load on all cores for a long time? Regardless, Mac or PC, I've worked with both over the years and in the end there are a lot of good options and what matters most is what we do creatively with those beautiful machines! Happy New Year guys!
Last edited by Francois Dompierre on Sun Dec 31, 2017 11:59 pm, edited 2 times in total.
François Dompierre
Senior Colorist / Finishing Editor
Montréal, Canada

2012 Mac Pro 12-core 3.33GHz | 32GB RAM | Dual Titan Xp 12GB
DeckLink Mini Monitor 4K | Teranex Mini Converter (as a LUT box)
macOS 10.13.4 | Resolve Studio 15 beta 2
Offline
User avatar

Francois Dompierre

  • Posts: 164
  • Joined: Thu Jun 23, 2016 7:42 pm
  • Location: Montréal, Canada

Re: iMac Pro user experiences?

PostSun Dec 31, 2017 11:39 pm

rick.lang wrote:As I stated previously, the 10-core under load is throttled to keep the temperature moderate and the fans fairly constant. The 10-core seems to run mostly at 3.9GHz and dips briefly to just below 3.2GHz until the temperature drops and then quickly resumes using 3.9GHz. So most of the time with multiple cores active, you get much more than Apple’s published clock speed.


Sent from my iPhone using Tapatalk



What, that's good news! What utility are you using to monitor the CPU clock speed under macOS ?
François Dompierre
Senior Colorist / Finishing Editor
Montréal, Canada

2012 Mac Pro 12-core 3.33GHz | 32GB RAM | Dual Titan Xp 12GB
DeckLink Mini Monitor 4K | Teranex Mini Converter (as a LUT box)
macOS 10.13.4 | Resolve Studio 15 beta 2
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 1:42 pm

rick.lang wrote:As I stated previously, the 10-core under load is throttled to keep the temperature moderate and the fans fairly constant. The 10-core seems to run mostly at 3.9GHz and dips briefly to just below 3.2GHz until the temperature drops and then quickly resumes using 3.9GHz. So most of the time with multiple cores active, you get much more than Apple’s published clock speed.


Sent from my iPhone using Tapatalk


If 2 or 4 cores are only active then that's normal that clock is raised to turbo speed even for long time.

My question was about 100% load for e.g. 6H. This will show how cooling is performing. Fact that Mac can stay cool during heavy load over long time is one "good" thing. If during this time clock is going to be lowered to e.g. 2.5GHz this means that there is a compromise compared to PC workstation based on the same CPU.
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 1:46 pm

Dermot Shane wrote:had not looked at the hp store before, for the same price as the 10c imacpro i can config a z6 with dual 18c's / 64g/dual 1080Ti's

z6; 18*3.7*2 = 133.2Ghz
imac; 18*2.2*1 = 39.6Ghz
for the same price

not sure the imacpro is very cost effective, if it was me, id likely get a moderately spec'd z6 for 5-6k, and a licence of Nucoda look for 3k to make Prorez dalies before i 'd spend 10k on an imacpro and .3k on a Resolve licence

but i am telling my clents to go ahead with caution, they are kinda afraid of anything without an apple logo on it ;-)


HP won't run forever at turbo speed at 100% load, so this is not fair comparison. It still doesn't change fact that HP can have dual Xeons. At the end these are the same CPUs. My question is how much Apple slows CPU (also GPU) during long processing at 100% load.

http://appleinsider.com/articles/17/12/ ... under-load

So it takes only 30 seconds for 8 core CPU to hit max allowed temp when GPU is also utilised quite heavily.
All of their tests were rather short e.g. (12min), but they are promising. The only worry is that even 8 cores CPU is very quickly operating at 94C. This is very high, but the good point is that iMac wasn't noisy according to article, so maybe with manual fan control it can be cooled better (for e.g. overnight render).
Offline

Dermot Shane

  • Posts: 1270
  • Joined: Tue Nov 11, 2014 6:48 pm

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 3:26 pm

HP won't run forever at turbo speed at 100% load, so this is not fair comparison


but it will it seems

the longest render i have done was a 95 min feature at DCi4k x 6 timelines to JPEG2000 XYZ, not disk bound, not GPU bound, CPU's ran at 100% for the duration and the fans did not need to spool up

in a rack with good ventlation, and the room is kept cool

my concern is haveing visted my cleints "facility" the room is small, crowded, has poor ventlation, and the machine crunching dalies was stuffed into a corner with disk houseings, speakers, paperwork etc everywhere

i'd be pretty sure an 840 in the same space would run very hot as it needs 24" clear air front and back
Offline

Dan Sherman

  • Posts: 364
  • Joined: Fri Jul 01, 2016 11:07 pm

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 5:53 pm

Dermot Shane wrote:
HP won't run forever at turbo speed at 100% load, so this is not fair comparison


but it will it seems

the longest render i have done was a 95 min feature at DCi4k x 6 timelines to JPEG2000 XYZ, not disk bound, not GPU bound, CPU's ran at 100% for the duration and the fans did not need to spool up


That's not how the turbo boost works.

You have a base clock, an all core boost clock, and a max single core boost clock. The marketers only talk about the core clock, and the single core max boost clock.

for example, if memory serves this is the 10 core chip in the imac pro.

https://ark.intel.com/products/125042/I ... e-3_30-GHz

core clock: 3.3 GHz
single core boost clock: 4.5 GHz

All core boost is usually 100 or 200 Mhz over the core clock, so lets say 3.5 GHz.


The only way you are running all cores at maximum frequency is with an unlocked and custom cooled chip, and you can't do that with any Xeons.



Based on your post I assume you where monitoring the OS performance metrics, and to be frank those are crap, regardless of what OS you are using. To see what your machine is really doing you need third party monitoring software that isn't trying to dumb down the information for the lemmings of the world.
Offline
User avatar

rick.lang

  • Posts: 8207
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

iMac Pro user experiences?

PostMon Jan 01, 2018 6:48 pm

François, see AppleInsider reviews and tests of the 10-core iMac where they describe the near constant fan and the cycling of the CPU clock.

Here’s a graphic of that behaviour sent to me by a friend with the 10-core with Vega 64 and 1 TB SSD and 64 GB memory:
Image

Sent from my iPhone using Tapatalk
Rick Lang
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 7:40 pm

This also confirms that iMac's Pro CPU runs at high temp overall. It's fine somehow, but shows that cooling is stressed a lot.
80C at 34% CPU load- quite high temp again. Many heavily overclocked PCs (when done well) never go above eg. 75C.
I've noticed the same on my Macbook Pro that any stress for CPU means very high temp.
Offline
User avatar

Francois Dompierre

  • Posts: 164
  • Joined: Thu Jun 23, 2016 7:42 pm
  • Location: Montréal, Canada

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 7:52 pm

rick.lang wrote:François, see AppleInsider reviews and tests of the 10-core iMac where they describe the near constant fan and the cycling of the CPU clock.

Here’s a graphic of that behaviour sent to me by a friend with the 10-core with Vega 64 and 1 TB SSD and 64 GB memory:
Image

Sent from my iPhone using Tapatalk


Thanks Rick! I looked at the article, very interesting info, and it looks like the iMac Pro CPU is performing very well indeed. I was expecting the chip to stay at the base clock of 3.0 when all cores were in use, but we get way more than that. Can't wait to see how the 18-core performs in that regard (throttling, base clock, all cores working, etc).
Last edited by Francois Dompierre on Mon Jan 01, 2018 8:04 pm, edited 1 time in total.
François Dompierre
Senior Colorist / Finishing Editor
Montréal, Canada

2012 Mac Pro 12-core 3.33GHz | 32GB RAM | Dual Titan Xp 12GB
DeckLink Mini Monitor 4K | Teranex Mini Converter (as a LUT box)
macOS 10.13.4 | Resolve Studio 15 beta 2
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 7:57 pm

What is minimum throttling, 10%?
Is CPU temp at around 95C?

Does Mac Fan Control app not working yet with iMac Pro?
Offline
User avatar

Francois Dompierre

  • Posts: 164
  • Joined: Thu Jun 23, 2016 7:42 pm
  • Location: Montréal, Canada

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 8:13 pm

Here's what my 10-core looks like after 30mins of all 10 cores rendering in Resolve (50% utilization = 100% of real cores and 0% of hyper threading cores, which Resolve doesn't seem to be using for this particular export). The computer is barely warm and the fans are very quiet... The CPU stays at around 90C, which is within spec (I checked on Intel's website). I barley get any throttling and I have sustained 3.8GHz on all 10 cores, not bad at all. I'll report back if it changes after a couple of hours... but I don't see why it would change, half hour of 100% rendering is as hot as it's gonna get... if the room temp stays the same... So basically this CPU has plenty of room to clock down when in really hot environments (which Apple has to account for when designing computers). I get 3.8 in my 20C/70F office, but someone in a hot apartment in Dubai with no AC might get 3.0..?

Now I really want to see how the 18-core performs, and what average GHz it stays at during an 18-core render session (in a properly temp controlled environment).

IMG_3506 copy.jpg
IMG_3506 copy.jpg (271.62 KiB) Viewed 2987 times
François Dompierre
Senior Colorist / Finishing Editor
Montréal, Canada

2012 Mac Pro 12-core 3.33GHz | 32GB RAM | Dual Titan Xp 12GB
DeckLink Mini Monitor 4K | Teranex Mini Converter (as a LUT box)
macOS 10.13.4 | Resolve Studio 15 beta 2
Offline

Dan Sherman

  • Posts: 364
  • Joined: Fri Jul 01, 2016 11:07 pm

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 9:04 pm

Francois Dompierre wrote:The CPU stays at around 90C, which is within spec (I checked on Intel's website). I barley get any throttling and I have 3.8GHz on all 10 cores, not bad at all! I'll report back if it changes after a couple of hours...


Where did you see 90C was ok?

I find that temp highly suspect, as 95 to 105C is usually within thermal shutdown range.

It's usually advised to keep temps in the 70 to 75C range or lower for Intel chips, with momentary jumps (a few seconds) over 80C ok. Sustained periods over 80C usually leads to a substantially shortened life span do to thermal degradation.
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostMon Jan 01, 2018 9:21 pm

Those temps really high. Apple does allow for CPU to go to very end if its ability, which is good, but in the same time bit worrying.
I hope at some point there will be an app to control fan speed and force it to higher RPM to get bit lower temps.
There may be some logic to slow down CPU after it operates to long at such a high temps.
Offline

Dan Sherman

  • Posts: 364
  • Joined: Fri Jul 01, 2016 11:07 pm

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 4:26 pm

Found an article, showing some pretty severe throttling.
http://appleinsider.com/articles/17/12/ ... under-load
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 6:15 pm

It's the same article which I posted.
There is no throttling there. Are you talking about turbo speeds?
Offline

Dan Sherman

  • Posts: 364
  • Joined: Fri Jul 01, 2016 11:07 pm

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 7:39 pm

Andrew Kolakowski wrote:It's the same article which I posted.


Sorry, I missed that.

Andrew Kolakowski wrote:There is no throttling there. Are you talking about turbo speeds?


I'd call this throttling! At one point the Frequency drops to just above 2.0 GHz, and CPU usage dropping to almost zero several times.

Image
Offline

Simon Brown

  • Posts: 88
  • Joined: Wed Sep 02, 2015 12:06 pm

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 8:11 pm

That looks like an illustration of multiple consecutive runs, as discussed in the article. The 0% CPU utilisation areas are the areas where the tests aren't running. And as there's no utilisation, there's no need to keep the processor running at max clock, so it doesn't. I'm fairly certain that if the benchmarks had caused the CPU clock to drop that low, the article would mention it.
Offline

Dan Sherman

  • Posts: 364
  • Joined: Fri Jul 01, 2016 11:07 pm

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 8:40 pm

Simon Brown wrote:That looks like an illustration of multiple consecutive runs, as discussed in the article. The 0% CPU utilisation areas are the areas where the tests aren't running. And as there's no utilisation, there's no need to keep the processor running at max clock, so it doesn't. I'm fairly certain that if the benchmarks had caused the CPU clock to drop that low, the article would mention it.



Check the video at about 1:05, to me Max makes it sounds like it was going on during the individual test runs.
Offline

Simon Brown

  • Posts: 88
  • Joined: Wed Sep 02, 2015 12:06 pm

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 8:43 pm

The video says exactly the same thing the article says.

After the second test, each additional run would cause the iMac Pro to slow down the CPU when the temperature reached roughly 94C, which caused the clock speed to drop from 3.9ghz to about 3.6ghz for a second or two. This allowed the CPU to drop below 92C, and the clock speed to rise back to the maximum turbo boost of 3.9GHz.


Nothing about the clock speed dropping to 2GHz.
Offline

Dan Sherman

  • Posts: 364
  • Joined: Fri Jul 01, 2016 11:07 pm

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 8:57 pm

Then something isn't right with the graph, or his explanation, because at several points the frequency graph drops down below 3.6Ghz and it's still showing ~100% cpu usage.
Offline

Dan Sherman

  • Posts: 364
  • Joined: Fri Jul 01, 2016 11:07 pm

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 9:04 pm

Simon Brown wrote:Nothing about the clock speed dropping to 2GHz.


Look at the graph!
Offline

Andrew Kolakowski

  • Posts: 3780
  • Joined: Tue Sep 11, 2012 10:20 am

Re: iMac Pro user experiences?

PostTue Jan 02, 2018 10:58 pm

Those dips last only few seconds. There is no time scale on these graphs which can be misleading.
It needs proper test for few hours etc, not few minutes. Then you can really tell how much CPU is throttled (and if at all).
Offline

Simon Brown

  • Posts: 88
  • Joined: Wed Sep 02, 2015 12:06 pm

Re: iMac Pro user experiences?

PostWed Jan 03, 2018 12:26 am

Dan Sherman wrote:
Simon Brown wrote:Nothing about the clock speed dropping to 2GHz.


Look at the graph!

What I am pointing out is, neither the article nor the video mentions those, and as I said earlier, as they coincide with CPU usage dropping to zero, they're likely in between runs of the benchmark. I don't consider clocking the CPU down when it has nothing to do to be a problem.
Next

Return to DaVinci Resolve

Who is online

Users browsing this forum: Jonathan Happ, Liam O'Brien and 22 guests