March 31, 2014

Want a Desktop Video Supercomputer? New NVIDIA & AMD GPUs May Make It a Reality

GTX_Titan_Z_nvidia graphics card processing unit gpu processor video card high definition hd video 4kOver the last year or so, major software such as Adobe Premiere Pro and REDCINE-X have joined DaVinci Resolve in leveraging GPU power for improved performance. GPU acceleration coupled with greater native RAW format support and forward strides in GPUs themselves means good things for the future of independent media production. And concerning that last part about GPUs getting better, well, both NVIDIA and AMD had some pretty serious things to announce over the past week -- NVIDIA with its $3000 dual-GPU GeForce GTX Titan Z, and AMD with its FirePro W9100 featuring 16GB of RAM. Check below for more details.

NVIDIA GeForce GTX Titan Z

GTX_Titan_Z_nvidia graphics card processing unit gpu processor video card high definition hd video 4k

Apparently featuring 2,880 cores and 6GB of VRAM per GPU (for a total of 5,760 processing cores and 12GB VRAM between both GPUs), the Titan Z is, according to AnandTech, "NVIDIA's obligatory entry into the dual-GPU/single-card market, finally bringing NVIDIA’s flagship GK110 GPU into a dual-GPU desktop/workstation product." As of right now, core clock rate and power consumption for the card have not been announced, but AnandTech had this to say based on other available information:

The GPU clockspeed [should be] around 700MHz, nearly 200MHz below GTX Titan Black’s base clock (to say nothing of boost clocks). NVIDIA’s consumer arm has also released the memory clockspeed specifications, telling us that the card won’t be making any compromises there, operating at the same 7GHz memory clockspeed of the GTX Titan Black.

Tom's Hardware has the following to add (citing the video below but restated here for emphasis):

Huang compared the new Titan Z to Google Brain, which features 1,000 serverspacking 2,000 CPUs (16,000 cores), 600 kWatts and a hefty price tag of $5,000,000 USD. A solution using Titan Z would only need three GPU-accelerated servers with 12 Nvidia GPUs total.

The following video was featured on the NVIDIA blog.

I'm pretty sure we all hope each card is assembled like a Transformer, too. As mentioned, the launch price for the Titan Z is slated at $3000. Pricing for the GPU teased by AMD last week, on the other hand, has not yet been announced. Speaking of which...

AMD FirePro W9100

amd firepro graphics card processing unit gpu processor video card high definition hd video 4k post production digital cinema workflow_beauty

Though not a dual-GPU card, the FirePro W9100 isn't playing around either -- featuring an impressive 16GB of VRAM with 2816 processing cores, AMD's announcement specifically targets high-end production (among other things) as a field which stands to benefit from the card. Once again AnandTech has the following to say: "AMD is banking on their 16GB of VRAM giving them a performance advantage [in 4K video encoding and image processing] due to the larger working sets such a large memory configuration can hold." Case in point:

A Faster & More Optimized Future

In yet another interesting announcement in the graphics world last week, NVIDIA outlined a revised product roadmap for its GPU lines. Apparently, the GPU architecture the company plans to unleash in 2016, affectionately dubbed Pascal, will feature several previously announced features. The feature of Pascal not unveiled earlier is perhaps its most interesting one: it will connect via a totally new interface dubbed NVLink. According to AnandTech NVIDIA "is looking at what it would take to allow compute workloads to better scale across multiple GPUs." And, apparently the 16GB/s (at 8 gigatransfers per second, or 8GT/s) of PCIe 3.0 -- or even the proposed 16GT/s of the upcoming PCIe 4.0 -- are not and will not be sufficient for Pascal to communicate among members of its family optimally.

Considering each card's ability to access memory at rates which may exceed 250GB/sec, it's not terribly difficult to see why NVIDIA would want a wider pipe. In terms of bandwidth NVLink will feature 8 lanes shoveling 20Gb/s apiece, meaning 20GB/s overall (according to AnandTech this also equates to 20GT/s). NVIDIA won't be abandoning PCIe by any means, as some proposed implementations envision multiple GPUs interfacing to each other via NVLink, but to CPU via PCIe; others see one or several GPUs talking to the CPU directly via NVLink. One version of the former is plotted out below:

Back to the Titan Z and FirePro W9100: it's unclear how these two workstation graphics units will stack up again each other, in terms of both performance and price. Neither card's clock rate is presently available, and the AMD unit's cost is also a question mark. To give you some kind of idea, the FirePro W9100's predecessor, the W9000, is going for $3400 on Newegg right now. These are pieces of hardware clearly not geared toward the casual user, and $3K is no amount to be brushed off. This is not to mention that we're living in an age of global shutter, interchangeable lens, 12-stop Super35 sensor camera systems shooting 4K directly to post-ready codecs that a filmmaker could purchase for the same price. Well, at least one.

With that said, this is also a file-based, data-centric age of filmmaking in which computers are ubiquitous and essential in basically every stage of production. The justification for spending $3000 on a single piece of computing hardware isn't so unlike that of spending the same amount for a solidly-spec'd camera as part of a gear kit. If it enables, and furthermore, facilitates, the creation or manipulation of creative media (and hopefully pays for itself in the process), it may be worth it for you. On the other hand, if all that wild business about gaming on 4K displays at maxed-out settings and frame rates is a necessity for you, one of these cards may be just as valuable.

For all the technical details, check the links below. Many thanks to AnandTech for the information and images.

Links:

Your Comment

38 Comments

5,760 processing cores installed in my computer is like putting 500 quarter mile dragster engines in a MiniCooper.

For only $3000.00 it's pretty tempting to do!!

March 31, 2014 at 2:03AM, Edited September 4, 11:56AM

0
Reply
Enjoying some G...

Sounds almost like an April fools joke?

March 31, 2014 at 2:51AM, Edited September 4, 11:56AM

0
Reply
JPS

I only have a quad core.... so I need this. At least that's the excuse I want to use to justify the $3000.00. Like Tim Allen says, "More power!"

But $3000.00 is a steal for this. Google Brain costs so much more, $5,000,00.00. There's my excuse! It's like the man that was trying to save money so he took the bus all the time. One day he came home panting and said "Honey, you'd be proud of me. I just saved $1.75. I ran home behind the bus." She said, "Ohhhh, why didn't you run home behind a taxi! You could have saved $9.00!"

Seriously though, I know this thing is going to be a godsend to lots of people. Technology is becoming astronomically more powerful and intriguing and cost is going down so fast!

March 31, 2014 at 2:54AM, Edited September 4, 11:56AM

2
Reply
Gene

Just keep in mind that you will not be able to fully utilize the power of this card with just a quad-core processor. You need top-end processing power to be able to get this card working hard.

March 31, 2014 at 4:11PM, Edited September 4, 11:56AM

3
Reply
Brian

Thanks Brian. It's a nice piece of advice.

I won't be getting this just yet as I really don't need this much horse power. But per chance I do want to have a lot of computing power someday what computer would you recommend for this?

March 31, 2014 at 7:14PM, Edited September 4, 11:56AM

0
Reply
Gene

Brian can you explain more, it says thousands of cores, this makes no sense compared to one's computer which is usually 4 cores, so I have no idea how much time will actually be saved in rendering time using one of these cards at all. I use sony vegas. If anyone has a site benchmarking these high end cards compared to intel iris pro 4000 built in to the board graphics I would like to see it to really know what difference these cards make for the large amount of money they cost.

Thanks

April 4, 2014 at 1:05PM, Edited September 4, 11:56AM

5
Reply
Dylan

To be clear Dave, the GTX Titan Z from nVidia is not a workstation graphics card. Those would be the Quadro series. The GTX line is still for the most part aimed at the gaming scene. Not to say you can't use some of those cards for work application as I am using a GTX 780 for Premiere with tremendous results.

March 31, 2014 at 3:07AM, Edited September 4, 11:56AM

2
Reply

Thanks for the GTX 780 info.

March 31, 2014 at 10:40AM, Edited September 4, 11:56AM

0
Reply
Gene

Last time I checked you could trick the software to recognize gtxs as Quadro card with little firmware diddling.

April 1, 2014 at 1:41PM, Edited September 4, 11:56AM

1
Reply
Natt

I run DaVinci Resolve. Been looking at PCI expansion to stack up cards, but what they're developing is pretty exciting. https://www.youtube.com/watch?v=prBKjJnUtSA

March 31, 2014 at 3:27AM, Edited September 4, 11:56AM

0
Reply

Not to rain on anyones parade but until software can even use the current cards properly (Speedgrade doesn't even boot with 2 gpus and hasn't done that for four months after the last update) it's kind of a moot point to release better ones.

Premiere uses about 5% the power of the GTX690. AE only uses the GPU for very specialized tasks (raytracing but who uses that when there's element 3d?) Resolve uses cards pretty well but after you've gotten realtime playback there's really not much more to gain?

March 31, 2014 at 4:36AM, Edited September 4, 11:56AM

0
Reply
Teuvo Pakkanen

I thought most of the rendering falls on the CPU, gpu is for playback (normal video editing) and the more cuda cores the easier it is play multiple layers as well as high bit rate footage. If I was doing mostly heavy VFX, cad stuff I would invest in a real workstation with duel CPUs, a couple quadros etc. With my experience using basic layers, 2k raw as well as 4k play fine on my i7 machine with 1 gtx780 ti. A better investment would be drive space for most of us doing 4k, as long as you have at least a gtx 670 I think we're fine getting to the next round of high power stuff. I read next year is the big year for gpu performance, I guess we'll see.

March 31, 2014 at 11:46AM, Edited September 4, 11:56AM

0
Reply
Anthony Marino

True dat. I wish they'd re-engineer some speed into SpeedGrade. It's so laggy!

Hopefully Adobe stays ahead of the curve (well, they've already fallen behind) as CUDA processing seems like it's going to get left behind for things like OpenCL.

March 31, 2014 at 4:57PM, Edited September 4, 11:56AM

0
Reply
Ty

Premiere heavily uses the GPU when doing scaling (GPU scaled 4k -> 1080p looks gorgeous), Neat Video takes advantage of the GPU which can make a huge difference if you'd prefer to run it on a lot of footage but can't due to CPU limitations...plus there's a big rise in GPU powered unbiased renderers in the CGI world.

April 1, 2014 at 1:59AM, Edited September 4, 11:56AM

0
Reply
Gabe

What is the best price/performance graphics card for the old MacPros?

March 31, 2014 at 4:40AM, Edited September 4, 11:56AM

1
Reply
Tulio

Depends which model Mac Pro you have. I have a 4,1 and have tried everything up to a GTX 780 in it. The problem is power - the Mac Pro doesn't have the power options that a PC has - so you have to jerry-rig power from the CD drive or other source to power cards like to GTX770 and higher. I'm currently running the PNY GTX 760 OC as it can use the normal power from the graphics slot and dual cables from the mobo and it works well with Premiere and FCPX. It's an off-the-shelf card that runs around $250 and I like it better than the Mac version GTX 680. The ply issue with off-the-shelf cards that aren't "Mac" is you have a blank screen until it fully boots to the desktop - but I'm using SSD so it only takes about 18 sec. to boot so I don't miss the Apple logo.

March 31, 2014 at 1:01PM, Edited September 4, 11:56AM

0
Reply
Lance Bachelder

thanks. it is a bit of a pity not being able to use those machines with FCPX.

March 31, 2014 at 4:13PM, Edited September 4, 11:56AM

0
Reply
Tulio

This is great news for 3D users, especially with Maya 2015's native viewport 2.0 and Vray tugging so hard on that GPU.

March 31, 2014 at 5:08AM, Edited September 4, 11:56AM

6
Reply
nate

I was thinking the same about GPU rendering, especially regarding Octane Render by Otoy.
http://render.otoy.com/
But I keep hearing that GPU rendering is not "production quality" compared to a CPU renderer such as say, Arnold. Do you, or anyone else, have any views on this, especially with organic subjects ?

March 31, 2014 at 8:12AM, Edited September 4, 11:56AM

2
Reply
Saied

I haven't ever heard of anyone saying that GPU rendering isn't "production quality", and to me it doesn't make much sense seeing that it's just a means of calculating. Perhaps you've heard this because some studios may choose to render on a farm that's CPU-centric?

I know more and more studios are turning toward their GPU's as the hardware becomes more affordable. One example is PIXAR utilizing Katana for a quick "render preview" (which turned some heads at SIGGRAPH 2013): https://www.youtube.com/watch?v=LACmRpMYOak&feature=share

March 31, 2014 at 6:59PM, Edited September 4, 11:56AM

0
Reply
nate

April 1, 2014 at 12:13AM, Edited September 4, 11:56AM

1
Reply
nate

Nate, thanks for those links. This article has a section CPU vs GPU which is quite interesting:-
http://www.fxguide.com/featured/the-art-of-rendering/
By the way, OctaneRender Version 2.0 looks amazing.
http://render.otoy.com/forum/viewtopic.php?f=7&t=38935

April 1, 2014 at 11:08AM, Edited September 4, 11:56AM

1
Reply
Saied

Would this help, as an example, in converting Raw into a proprietary format such as XAVC, AVC-Ultra, HEVC, VP9 and so forth?

March 31, 2014 at 9:26AM, Edited September 4, 11:56AM

2
Reply
DLD

Why not just get 2 gtx 780 ti's? $1400 vs $3000, how much speed would you really gain? I'm using one 780 ti and it's great, I think next year the big update is coming, new architecture, processors etc.

March 31, 2014 at 10:48AM, Edited September 4, 11:56AM

0
Reply
Anthony Marino

One big difference is in the ram...2x6gb vs 2x3gb. Can make or break it depending on what you're using it for. Super important for CGI rendering like Octane.

April 1, 2014 at 2:03AM, Edited September 4, 11:56AM

1
Reply
Gabe

Yours is not a bad card and has about 2300 cuda cores, which very impressive in itself. But this Titan is dual GPU and has 5,760 cores combined. I'd love to stick 2 of them in 2 PCI Express slots and leverage over 11,000 cuda cores. 4K realtime playback with many nodes on resolve would be a breeze.

April 1, 2014 at 6:49AM, Edited September 4, 11:56AM

0
Reply
Benjamin Forrest

All this cool tech always on the horizon. This is my main issue with switching to a new Mac Pro - you're stuck with whatever GPU you order - not that it's underpowered at all. But with a tower it's so nice to swap out GPU's as they come available and always have an up-to-date machine.

March 31, 2014 at 1:04PM, Edited September 4, 11:56AM

7
Reply
Lance Bachelder

I still remember rocking my Voodoo2. Upgraded from an ATI Rage, I believe.

March 31, 2014 at 1:22PM, Edited September 4, 11:56AM

0
Reply
Swissted

I was thinking of going for duo titan for my soon to buy duo-8core hp Z820 workstation...
yea i think I'll still go for then,this aint out yet...plus its 3k...while 2 titans its 2k...pc all the time for mi handles 3d modelling n animation very well...n also video editing

March 31, 2014 at 2:57PM, Edited September 4, 11:56AM

0
Reply
mathenge

Still cheaper than a Mac

March 31, 2014 at 5:32PM, Edited September 4, 11:56AM

5
Reply
Ryan Maelhorn

Yeah yeah whatever. How do I connect this to my PS1?

March 31, 2014 at 5:53PM, Edited September 4, 11:56AM

2
Reply
Edgar

something to consider after seeing some tests versus the red rocket. if it can replace the red rocket i'd definitely go for it

March 31, 2014 at 7:49PM, Edited September 4, 11:56AM

0
Reply

I can't wait to read about the performance this thing brings to cybercurrency miners...

April 1, 2014 at 1:10AM, Edited September 4, 11:56AM

2
Reply
trackofalljades

April 1, 2014 at 1:31AM, Edited September 4, 11:56AM

0
Reply
DLD

but but... will it play Crysis on full?

April 1, 2014 at 12:13PM, Edited September 4, 11:56AM

0
Reply
mike

At the current rate of GPU card development, I wonder how long it will take for the internal GPU's in the mac pros to be outdated(as far as performance goes. not usability).

April 3, 2014 at 10:44PM, Edited September 4, 11:56AM

6
Reply

Any thoughts on this conversation from a hackintosh perspective? I'm thinking about building one. I'd love to get straw poll of thumbs up or down.

April 28, 2014 at 3:56PM, Edited September 4, 11:56AM

2
Reply
Earnest Reply

Having read this I thought it was very informative. I appreciate you taking the time and
effort to put this article together. I once again find myself spending way too much time
both reading and commenting. But so what, it was still worth it!

June 16, 2014 at 6:47AM, Edited September 4, 11:56AM

0
Reply