-
Theoretical MacPro Replacement
-
[Phil Brockett] “HP Pavilion w/ i7 4 core 3.8 GHz;”
Base model MacPro is 6 core Xeon. You really should be looking at HP Z400 or Z800 IMHO.
-
“Base model MacPro is 6 core Xeon. You really should be looking at HP Z400 or Z800 IMHO.”
You are correct. The Z800 is certainly a better workstation and surely has a much, much longer mean-time-between-failure rate than the Pavilion. It also is a better comparison to the MacPro in quality. But really, you can run the same software on the Pavilion (or iMac for that matter) as you can with the MacPro. And the software hasn’t really caught up to the hardware yet and may not for a long time. Refer to 64bit versions from 32bit (which most products came first on the Windows platforms first). Point being is that if the software can’t keep up, what’s the point of bigger, better machines?
A apologize for being a little flip in my prior post. My point is that, from a users perspective, the hardware is more interchangeable than the software (e.g., all the posts about FCP7 being replaced by FCPX and trying to find a software package that would have the shortest learning curve to replace 7, which took them years and years to master).
Assuming most people rely on some type of 3D package as well as compositing software and something to tie everything together (FCP7, X, Avid, etc.), it is the 3D that really requires the computing horsepower. Given a budget of $X, and the requirment for those three key software packages, I’d probably (theoretically) lean toward getting the least expensive hardware that will make those three systems go and also getting less expensive (ie slower) computer(s) to act as a render farm for the 3D thereby no need for the background rendering.
-
Which Apple pro apps were you going to run on this non existent computer?
How much more than a windows 7 machine were you going to pay?
Craig, go sell a video, and start producing something. You have too much time on your hands, or perhaps Apple is paying you to shill for them.
DBK
-
About the macmini cluster + G4 cube…
[IMG]https://i20.photobucket.com/albums/b236/livingfortoday2/minicube.jpg%5B/IMG%5D
You could fit like 4 MacMini 2011 in there..chain them up with thunderbolt and you have a 16-core CPU beast,with 65 GB of RAM. And 8TB of HDD. Just saying it could be done, prettily today.hehehe
Quobetah
New=Better -
While I’d like to see a new MacTower, my growing conviction is that it’s over. Less than 30% of all CPU sales are desktop. Jobs said you should be afraid to obsolete yourself and the shrinking revenue from MacPro’s all but guarantees that there will not be anything like it.
https://www.macobserver.com/tmo/article/why_the_mac_pro_wont_last/
I hope I’m wrong but I think your idea while interesting is pure fantasy in the minds of Apple engineers.
I wonder if you can use thunderbolt to share CPUs in which case taking 2-3 mini’s and putting them in a 1U rack might be a more likely.
-
[Phil Brockett] “Assuming most people rely on some type of 3D package as well as compositing software and something to tie everything together (FCP7, X, Avid, etc.), it is the 3D that really requires the computing horsepower. Given a budget of $X, and the requirment for those three key software packages, I’d probably (theoretically) lean toward getting the least expensive hardware that will make those three systems go and also getting less expensive (ie slower) computer(s) to act as a render farm for the 3D thereby no need for the background rendering.”
I’d still go the other way — as much power as possible, given the budget, on the artist’s desk. Just getting the cheapest machine possible and backing it with a render farm will cost you money every day the system is in use.
With both 3D and compositing, once the artist gets something roughed in, refining it requires iteration. Make a change, do a test render (or partial render), evaluate the result, repeat as necessary. Network renders add a lot of overhead (gather assets, submit render, wait, receive finished render) and may require the artist to step outside of their main app dozens of times a day.
A fast machine on the artist’s desk will allow him or her to iterate more rapidly, getting good results faster, or getting better results in comparable time.
Talent is expensive. Computers (even fast ones) are cheap in comparison. I don’t think it pays to save a couple thousand dollars on a computer when it will cost you far more in artist time over the life of the machine.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
[Dennis Radeke] “I wonder if you can use thunderbolt to share CPUs in which case taking 2-3 mini’s and putting them in a 1U rack might be a more likely.”
The problem with your thinking is that doesn’t take into account the GPU. The GPU is more responsible for tasks these days then ever before and this is quite specific to Apple’s products whether it be FCPX and especially Motion. So even a commodity box will have to have a GPU and FCPX is all about selling hardware so Apple is pushing GPU performance harder in FCPX and Motion. Apple needs a box with a good GPU sans monitor (they’d like you to buy their $1000 Thunderbolt monitor). Neither the Mini nor the iMac fit the bill. The mini is stackable but sans GPU. The iMac has a decent GPU but isn’t stackable. Result is a stackable box with a good GPU and that’ll have to be bigger than the Mini by necessity.
-
If you need serious GPU power (same argument for MacPros) then you can add it via an expansion chassis.
I also think some people put to much emphasis on the GPU in an editing system. If you rely solely on GPU to solve editorial and compositing tasks, you’ve simply moved the problem from the CPU to the GPU. The key is to use both in balance.
Finally, who’s to say you can’t gang a bunch of ‘okay’ GPU’s together as well? Adobe has been asked for multiple GPU enablement for CUDA since CS5 was released. With PremierePro 5.52 we’re taking a first tiny step with Maximus (Quadro + Tesla) for more GPU power.
Again, it’s the idea that the market size might not be large enough for a company like Apple to support with a traditional tower. Besides, Steve always said not to be afraid to obsolesce yourself. Reading the book like everyone else and find it fascinating. I think the multiple mini concept would make sense for Apple and for a larger addressable market including some editorial and rendering type jobs.
-
[Dennis Radeke] “If you need serious GPU power (same argument for MacPros) then you can add it via an expansion chassis. “
Thunderbolt doesn’t currently support a 16 lane GPU or do you have something else in mind.
[Dennis Radeke] “who’s to say you can’t gang a bunch of ‘okay’ GPU’s together as well?”
Are you then saying that ganged 4x is fine? If that’s so that might be interesting. Apple, though, has been using the GPU of late to motivate Mac sales as per FCPX/Motion requirements. Apple wants you to buy more boxes (the nature of commodity) so I can’t help but think an in built GPU would be best for their economics.
-
[Dennis Radeke] “I wonder if you can use thunderbolt to share CPUs in which case taking 2-3 mini’s and putting them in a 1U rack might be a more likely.”
I don’t think you could with today’s technology. You can’t even use multiple i7 CPUs on the same motherboard.
Beyond that, I think Thunderbolt at 10 Gb/s is too slow for sharing CPUs and RAM. DDR3-1333 peaks at 10,667 MB/s.
[Dennis Radeke] “I think the multiple mini concept would make sense for Apple and for a larger addressable market including some editorial and rendering type jobs.”
I’d guess the market for scalable server clusters for enterprise applications, web, storage, and databases far outweighs the market for video-oriented clusters, and Apple chose to exit the enterprise market last year.
If Apple were heading back in the cluster direction, I’d think they’re be a hint of cluster support in Grand Central Dispatch. Even if they introduce a new cluster technology tomorrow, it will take a long time before this technology could be integrated in apps across the industry.
All that said, I’d love to be wrong on this. I do agree with you both that it would be really cool to connect a few computers with simple cables and instantly multiply your compute power.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events
Reply to this Discussion! Login or Sign Up