2016 nMP

[Aldaris]

I've been lurking in this particular thread for months (this and the "a new Mac mini is almost certainly coming" have become my entertainment). I'm in the market for a Mac Pro primarily for running architecture rendering projects-right now I'm using a 2012 rMBP. Mainly running AutoCAD for Mac, Revit (Windows/boot camp) and pixelmator and quark for presentations. Overall my workflow is Mac based and have no desire to hackintosh.

I'm just an Arch Student and am not willing to drop 3k, on (practically) 3 year old tech with the launch day premium.

I know many arch firms have gone to running Windows/boot camp on iMacs just for the resale value of the machines-but I want the power to render without the iMac limits (which is basically what I'm doing with a rMBP) and new my Mac workflow.

I only bring this all up because I think there is still a reasonable market for a pro box-and I'd hate to see Apple drop a workstation class machine. Unfortunately I am feeling more and more uneasy with the direction Apple is heading, and the silence in the pro space is really causing a lot of folks in my position or similar to look elsewhere.

What good is the gateway drug/product of iOS when there is nothing to do any work on.

I'm looking for TB3 decent graphics offerings, I'd gladly take a current MP if there was a price drop-but it's maddening to wait to WWDC (without any tangible evidence that an update will happen).

I'm grateful for all the sleuthing going on looking at Intel ATI road maps to help guide a little.

Anyway sorry for the rant and detour.

Back to the Polaris/Fuji debate.

 

[poematik13]

I've been lurking in this particular thread for months (this and the "a new Mac mini is almost certainly coming" have become my entertainment). I'm in the market for a Mac Pro primarily for running architecture rendering projects-right now I'm using a 2012 rMBP. Mainly running AutoCAD for Mac, Revit (Windows/boot camp) and pixelmator and quark for presentations. Overall my workflow is Mac based and have no desire to hackintosh.

I'm just an Arch Student and am not willing to drop 3k, on (practically) 3 year old tech with the launch day premium.

I know many arch firms have gone to running Windows/boot camp on iMacs just for the resale value of the machines-but I want the power to render without the iMac limits (which is basically what I'm doing with a rMBP) and new my Mac workflow.

I only bring this all up because I think there is still a reasonable market for a pro box-and I'd hate to see Apple drop a workstation class machine. Unfortunately I am feeling more and more uneasy with the direction Apple is heading, and the silence in the pro space is really causing a lot of folks in my position or similar to look elsewhere.

What good is the gateway drug/product of iOS when there is nothing to do any work on.

I'm looking for TB3 decent graphics offerings, I'd gladly take a current MP if there was a price drop-but it's maddening to wait to WWDC (without any tangible evidence that an update will happen).

I'm grateful for all the sleuthing going on looking at Intel ATI road maps to help guide a little.

Anyway sorry for the rant and detour.

Back to the Polaris/Fuji debate.

If you don't want to drop 2799 or 3699 on a stock nMP you can always get an overly maxed out (12 core, 128GB ram, PCI-E ssd, 980 graphics, etc) 5,1 model on ebay for around 1700-2000, if of course you don't need TB. Also keep in mind the reason Apple doesn't really drop prices over time is because their products, unlike the PC competition, actually hold their resale value very well so the price you pay is actually worth it from a market standpoint. Disregard what the usual suspects on this forum will say about that.

The nMP is most likely fine for what you do, though. It's not like the specs have gotten WORSE in the past 3 years. They were great in late 2013 and they are still just as capable as they were back then, today. Has the average workflow really drastically changed in that past 2.5 years? No. Have there been significant advancements in Intel's Xeon chips since then? Definitely not lol. Aside from upcoming AMD GFX chips and TB3, there isn't much for Apple to upgrade to, so they are just waiting to overhaul everything at once, probably later this year.

IMO the yearly upgrade culture of iOS devices has really damaged the average consumer's mind and now everything is supposedly instantly obsolete the moment its released. It's really not the case. I edit 6k and 8k footage on my nMP 6c/d500/16Gb/1TB and it works amazingly. I might upgrade to the next model or put a 10core xeon and 128GB ram in my current one pending on whats new and offered.

 

[Aldaris]

I totally agree with the yearly upgrade cycle of iOS devices (probably more iPhone than anything else). I imagine that the current offerings would meet my needs, the issue I have is that it is near the end of its product cycle (based of buyers guide and looking at time between updates). Knowing that feature jumps/bumps are coming (TB3 primarily in my case, especially changing the connector). I'm more apt to waiting even if it drives me to the asylum, than jumping in the current model. With the history, between the quicksilver G4, mdd G4, and even the early G5 machines, the Pro models saw either a regular spec bump be it graphics/memory/storage/cpu or price drop (even with apple standards, they were very modest).

poematik13 Do you find 16Gb fine for your nMP? Looking back would you recommend more?

 

[poematik13]

TB3 is all but confirmed to be in the next nMP. There is a device code in the newer 10.11 betas that hints at a desktop with 10 USB-C connectors. Right now the nMP has 6 TB2 and 4 USB- so 10 TB3 (in USB-C connector shape) is logical as a next step to combine all ports into one standard.

16GB works for me right now, I rarely use more than 12. If I'm gonna upgrade next I'll just go all in with 64 or 128. 128 on the nMP means I have to step down to 1333mhz though, but I doubt I'll see a difference.

 

[AidenShaw]

TB3 is all but confirmed to be in the next nMP.

On the other hand, there's a very significant chance that Apple will simply declare the MP6,1 to be the "death of the pickup truck", and drop the line - citing poor sales for a system that nobody asked for.

The presence of code supporting a machine that will never be sold is irrelevant. Engineers prepare for systems on the drawing boards.

Bean counters decide whether to move a system from the drawing board to production.

Apple is run by bean counters, not engineers.
[doublepost=1457579314][/doublepost]

16GB works for me right now, I rarely use more than 12. If I'm gonna upgrade next I'll just go all in with 64 or 128. 128 on the nMP means I have to step down to 1333mhz though, but I doubt I'll see a difference.

That's odd - I'm getting 1866 on my E5-2600 v2 systems with 16 DIMMs. Odd that Apple would drop to 1333 with a measly 4 DIMMs.

 

[p.l]

On the other hand, there's a very significant chance that Apple will simply declare the MP6,1 to be the "death of the pickup truck", and drop the line - citing poor sales for a system that nobody asked for.

The presence of code supporting a machine that will never be sold is irrelevant. Engineers prepare for systems on the drawing boards.

Bean counters decide whether to move a system from the drawing board to production.

Apple is run by bean counters, not engineers.
[doublepost=1457579314][/doublepost]

That's odd - I'm getting 1866 on my E5-2600 v2 systems with 16 DIMMs. Odd that Apple would drop to 1333 with a measly 4 DIMMs.

View attachment 620493

Fine Print on 32gb Modules in a Mac Pro.

Special 32GB module use information: Due to processor design and memory architectures, 32GB module utilization occurs at a bus speed of 1066MHz with Apple Mac Pro 2013 E5-16XXv2 4-Core, 6-Core, and 8-Core processor equipped models, 800MHz with factory 12-Core or after-market E5-26XX equipped models.

 

[flat five]

I'm in the market for a Mac Pro primarily for running architecture rendering projects-right now I'm using a 2012 rMBP. Mainly running AutoCAD for Mac, Revit

so Autodesk hasn't added this button to AutoCad yet?
; )

(screenshot from fusion360)

 

[Zarniwoop]

Latest AMD drivers for Windows added AMD XConnect technology to connect eGPU via TB3. Lets see will Apple ever adopt this tech. Maybe they could start to sell a secondary cylinder with Fiji GPU's? (Maybe not)

http://support.amd.com/en-us/kb-articles/Pages/AMD_Radeon_Software_Crimson_Edition_16.3.aspx

 

[Zarniwoop]

If MP v2 comes along with new Intel E5-EP 1600 v4 series within a month, AMD has only Pitcairn (D300) and Amethyst (Tonga XT, M395X) for 125W TDP class, suitable for the MP thermal core. Nvidia has GM206 (GTX 960) or GM204 (980M) - unless Apple goes back to an one GPU solution or reinforce MP cooling and PSU. Or they sack the whole cylinder concept and come up with something new.

 

[shaunp]

I've been lurking in this particular thread for months (this and the "a new Mac mini is almost certainly coming" have become my entertainment). I'm in the market for a Mac Pro primarily for running architecture rendering projects-right now I'm using a 2012 rMBP. Mainly running AutoCAD for Mac, Revit (Windows/boot camp) and pixelmator and quark for presentations. Overall my workflow is Mac based and have no desire to hackintosh.

I'm just an Arch Student and am not willing to drop 3k, on (practically) 3 year old tech with the launch day premium.

I know many arch firms have gone to running Windows/boot camp on iMacs just for the resale value of the machines-but I want the power to render without the iMac limits (which is basically what I'm doing with a rMBP) and new my Mac workflow.

I only bring this all up because I think there is still a reasonable market for a pro box-and I'd hate to see Apple drop a workstation class machine. Unfortunately I am feeling more and more uneasy with the direction Apple is heading, and the silence in the pro space is really causing a lot of folks in my position or similar to look elsewhere.

What good is the gateway drug/product of iOS when there is nothing to do any work on.

I'm looking for TB3 decent graphics offerings, I'd gladly take a current MP if there was a price drop-but it's maddening to wait to WWDC (without any tangible evidence that an update will happen).

I'm grateful for all the sleuthing going on looking at Intel ATI road maps to help guide a little.

Anyway sorry for the rant and detour.

Back to the Polaris/Fuji debate.

It's annoying - lack of roadmap is getting to be unacceptable. I have a nMP and it's a nice, but not great machine. Like you I also have the 2012 rMBP that I desperately want to upgrade because of the memory limitations. I run a lot of VMware and 16 really isn't enough. All the other vendors have laptops with 64GB and mobile Xeon now. I know mobile Xeon is really an i7 with a different badge but it enables 64GB RAM. Apple have announced nothing on this. If I can't get a decent laptop out of them, then I'll look at Dell as their products are shipping and are getting great reviews, and at that point I'll ditch my nMP and go back to my old PC and upgrade it as and when I need to.

Personally I think the writing is on the wall and Apple only want to sell iMacs, iPhones, iPads and low-end MacBooks. Doesn't anyone in Apple remember where they came from? Enthusiasts and Professionals don't seem to matter any longer.

 

[koyoot]

http://venturebeat.com/2016/03/09/o...-the-best-graphics-software-across-platforms/

Pretty interesting news.

A primary goal for creating this technology is to bring CUDA applications such as Octane to Apple’s Metal GPGPU API on OSX and iOS, where support for OpenCL 2.1, Vulkan, and OpenGL ES compute is noticeably absent, Urbach said.

Funniest part is that it looks like they took what AMD gave in OpenGPU initiative(Botlzman, and CUDA Compiler) and made it better.

Benefits of Open software.

P.S.

It’s pretty ready, and it answers questions about whether this could be done,” Urbach said. “It runs on the other cards at the same speed as it runs on Nvidia cards.

So, yeah, 4TFLOPs AMD GPU will execute it at the same speed as 4 TFLOPs GPU from Nvidia.

 

[tuxon86]

"Jules Urbach, chief executive of Los Angeles-based Otoy, said in an exclusive interview with GamesBeat that Nvidia’s CUDA language is superior and enables much richer graphics software. Hence, OpenCL hasn’t provided a true market alternative to CUDA. That’s why building the CUDA “cross compiler” was an important task. And Urbach said the Otoy research and development team was able to do it in 9 weeks."

That's what we've been saying to you for the past months... Nice of you to see the light at least.
[doublepost=1457618868][/doublepost]In other news:

"NVIDIA has been named VMware’s "Global Technical Partner of the Year," as well as its "European Regional Technical Partner of the Year." NVIDIA and VMware have been close partners in delivering the next generation in end-user computing. NVIDIA GRID and VMware Horizon dramatically transform how companies can deliver amazing graphics from the cloud or data center in order to support collaboration among global design and engineering teams and their suppliers."

 

[koyoot]

That is only his opinion. CUDA has much better documentation, and support, than OpenCL, that is no question. But from compute CAPABILITIES perspective, they are the same. There is nothing that CUDA allows that OpenCL could not allow. Unless you believe in Nvidia magic. But the implementation and running software on it is easier.

P.S. Great to know that VMware praises Nvidia for their marketing efforts.

 

[tuxon86]

That is only his opinion. CUDA has much better documentation, and support, than OpenCL, that is no question. But from compute CAPABILITIES perspective, they are the same. There is nothing that CUDA allows that OpenCL could not allow. Unless you believe in Nvidia magic. But the implementation and running software on it is easier.

P.S. Great to know that VMware praises Nvidia for their marketing efforts.

And yet you were the one trying to pass this as a win for your side. How fast can you switch gear when presented with reality.

And I guess you're the expert around here about marketing effort with all the amd press release that you keep foisting on us.

 

[koyoot]

Can you stop saying things what you think about what I was writing, and am writing and stick to the truth(what I actually was writing, and am writing)?

About the other part I think it is not even worth responding.

P.S. To neutrals: I think the fact that everyone is abandoning OpenCL, including the creators of it makes it obvious why: CUDA compilers for Applications. Metal for sure will have compiler as at least extension for it.

P.S.2 Im genuinely curious how will CUDA apps perform on AMD GPUs, knowing they have more compute horsepower than Nvidia ones.

 

[tuxon86]

Can you stop saying things what you think about what I was writing, and am writing and stick to the truth(what I actually was writing, and am writing)?

About the other part I think it is not even worth responding.

P.S. To neutrals: I think the fact that everyone is abandoning OpenCL, including the creators of it makes it obvious why: CUDA compilers for Applications. Metal for sure will have compiler as at least extension for it.

P.S.2 Im genuinely curious how will CUDA apps perform on AMD GPUs, knowing they have more compute horsepower than Nvidia ones.

Maybe you should calm down. Your first phrase is painful to read.

As for your P.S.2: Too bad those horsepower are coupled with a ****** transmission because all that extra power seems to go to waste in actual use beside mining for bitcoins.

 

[koyoot]

Tuxon, Compute is compute, you will not run from that. Unless - you believe Nvidia Magic.

 

[Aldaris]

so Autodesk hasn't added this button to AutoCad yet?
; )

(screenshot from fusion360)

View attachment 620510

Cloud rendering is there in revit, and I've used it-there are times where I don't want to rely on cloud rendering though. You can get great results, but what happens if you loose a net connection?

 

[Hank Carter]

Cloud rendering is there in revit, and I've used it-there are times where I don't want to rely on cloud rendering though. You can get great results, but what happens if you loose a net connection?

Or if you are working with confidential information that can not leave your network.
Or if you are working with files that are gigabytes in size and would take too long to upload.
Cloud computing is also not free. You need to pay for those CPU cycles.

Cloud computing is no magic bullet.

 

[AidenShaw]

Tuxon, Compute is compute, you will not run from that. Unless - you believe Nvidia Magic.

¿¿ FP16 == FP32 == FP64 == char == short == int == long == longlong ??

 

[tuxon86]

Tuxon, Compute is compute, you will not run from that. Unless - you believe Nvidia Magic.

And yet people use NVidia and Cuda for that... Funny...

 

[Machines]

128 on the nMP means I have to step down to 1333mhz though, but I doubt I'll see a difference.

A 128GB memory configuration in the nMP will operate at 800 MHz , with the factory 12 Core models . This is also true with aftermarket E5-26XX processor upgraded machines . This is year 2008 performance .

 

[flat five]

You can get great results, but what happens if you loose a net connection?

then it renders locally.
idk, i was just curious if Autodesk was going to be integrating cloud rendering in all of their software or just a select few.
(so, i take it, it's not an option in AutoCad right now?)

Cloud computing is no magic bullet.

right.. just another option that many people can in fact take advantage of.

Or if you are working with confidential information that can not leave your network.

well, yeah.. but in those cases (say, Boeing working on some secretive govt project or whatever).. those companies more likely than not will have in house render farms etc..

but i'm not really sure why you should use these kinds of examples as negatives unless you're actually in the position for it to be a negative aspect.. if it affects someone somewhere then cool.. if it doesn't affect you (which it likely doesn't) then that's cool too.

i mean, corresponding a design via email is also 'cloud computing'.. it's decently secure these days.. definitely secure enough for most projects.

Or if you are working with files that are gigabytes in size and would take too long to upload.

compare uploading a couple GB to render on 128 cpu cores (or whatever) VS rendering locally on 8 cores.. i think you'll find it's worth the upload time.. as in - way worth it.

Cloud computing is also not free. You need to pay for those CPU cycles.

again.. compare that price to the price you'll pay for local CPU cores.. it's way (way!) cheaper to outsource the cycles.

 

[Aldaris]

flat five- As far as I know cloud rendering isn't on AutoCAD for Mac, but it is on Revit. The strange thing with AutoDesk, is how each product is fairly independent of each other, I know different teams steal ideas like cloud rendering buttons, and will implement them at some point down the road, they just aren't as cohesive (quick and seamless integration) as smaller outfits. Which is completely understandable, and no way a slam to them, it's just that the bigger guys move slower, kinda like this discussion on a nMP, be it the 'fault' of intel, ATI, or other big entity.

As for cost, last time I rendered a Revit Project in the AutoDesk Cloud, I believe it was free to students (or a certain of cloud credits given for a period of time), along with free educational licenses for most if not all of their products. Very good points on end price comparison, and finding what works best for your workflow. Not a direct comparison, but some folks I know will put out some cash for a decent wide format plotter, others will send a PDF (digitally or USB) to a local printer. My preference would-be to have one locally, but for now the local guys will do.

With all that said, I think thats the general feeling of many people who look into the Mac Pro, smaller/independent outfits/firms/studios/offices, regardless of industry, that appreciate the option to have enough muscle (when they need it) to meet there needs, and not have to rely on others as much. And of course the Mac Pro isn't their only tool. Opposed to the big boys like Pixar and others that have render farms and the like.

 

[flat five]

With all that said, I think thats the general feeling of many people who look into the Mac Pro, smaller/independent outfits/firms/studios/offices, regardless of industry, that appreciate the option to have enough muscle (when they need it) to meet there needs, and not have to rely on others as much. And of course the Mac Pro isn't their only tool. Opposed to the big boys like Pixar and others that have render farms and the like.

for clarity- i'm certainly not trying to say something like "don't worry about capabilities of the local computers.. do it all in the cloud" ; )

i do most of my renders on my local network at the moment (that being an imac master and mbp as a node).. rarely do i go to the cloud unless i'm under a time crunch.. i'm interested in the next gen rendering software which is reliant on gpgpu as opposed to cpu based raytracing as i don't think i'll ever be able to justify doing something like a 24+ cpu core system since it's far too expensive for the comparatively slight increase in performance.

also, i'd rather have a 'pro' system for local rendering simply because every time i render now (with the imac & mbp fans going all night etc), i always have a worry in the back of the mind as to whether or not i'll have functional computers in the morning or just a pile of melted glass/metal/aluminum

(i haven't had any heat issues with the current setup since i've switched over from a cmp for rendering but at the same time, my confidence level in the hardware isn't the same as it was with the mac pro.)