Does This Kill The Mac Pro?

[John-michael Seng-wheeler]

And how do you expect it to get that information?

[Jeremy Garchow]

[John-Michael Seng-Wheeler] “And how do you expect it to get that information?”

Through the motherboard like everything else.

[John-michael Seng-wheeler]

[Jeremy Garchow] ”
Through the motherboard like everything else.”

And there we have the problem.

No current PCIe GPU sends display signals anywhere but out the back of the computer. (The current PCIe bus doesn’t allow for it. They can send data, but not a display port signal.)

Apple will need to make a GPU that’s capable of this, however they do it. (I outlined the options as numbers 2, 3a and 3b)

[Jeremy Garchow]

[John-Michael Seng-Wheeler] “No current PCIe GPU sends display signals anywhere but out the back of the computer. (The current PCIe bus doesn’t allow for it. They can send data, but not a display port signal.) “

I’m confused, dude.

Yes, the ports are on the card to go out to a monitor, but the physical data can be sent anywhere. That’s how the GPU is used to process in applications like PPro or Shake or Color. The data is available anywhere as long as it’s plugged in. The data is processed on the GPU and sent to other devices (via PCIe) such as a Kona card. I don’t have to take the displayport out and send it to the Kona, for example. It’s just data.

[John-michael Seng-wheeler]

It’s a different kind of data. CUDA is meant to be used that way. Sending the whole data stream back doesn’t work that way.

Ever taken a screen shot and found a video or something was just a black box? Current GPU cards aren’t designed to send the video stream back through the PCIe connection, they’re designed to send it out of the computer.

[Jeremy Garchow]

[John-Michael Seng-Wheeler] “Ever taken a screen shot and found a video or something was just a black box?”

Yeah, that’s called copy protection.

Alright, at this point I am going to need proof John-Michael. I think I fully understand where you are coming from, and I don’t think the system operates in the way you are describing.

Color does not have or need Cuda, and I still need to get GPU processed images out to the Kona, and those images are not coming out of my DVI port into my Kona to get there, which is how you seem to be describing how Thunderbolt works.

Jeremy

[John-michael Seng-wheeler]

Yeah, I think you’re right about it being copy protection.

I have no idea how well a GPU would or would not work used this way. There’s still the problem that the GPU needs to send the video data to the Thunderbolt controller and that means going through the north Bridge. (the signal destined for the Thunderbolt controller would need to be separated from the signal going back to the processor, But that’s not unsurmountable, it would just require a North Bridge Specially designed for the process. Sending the data all the way back to the processor would be a bit silly, but without a special north bridge it wouldn’t be possible without going through the processor.)

I agree it is posible, but it still seems like a bad way to do it.

A built in GPU or a custom one seems far more likely.

[Walter Soyka]

[Jeremy Garchow] “Alright, at this point I am going to need proof John-Michael. I think I fully understand where you are coming from, and I don’t think the system operates in the way you are describing.”

I think you’re both right.

The PCIe bus was not intended to push data to the graphics card and receive fully rendered data back, but it is flexible enough to allow this to be done. I’d suggest that a 16x card may be important here.

I have heard that this was one of the big challenges that Assimilate ran into when porting SCRATCH to the Mac. On Windows, they use a NVIDIA daughtercard with SDI out, so the processed image data doesn’t have to come back through the PCIe bus. On the Mac, they didn’t have this option, so they have to push rendered frames back from the card to the computer. Guaranteeing realtime performance was apparently quite an engineering challenge.

Even if it’s possible to render graphics on one card, pipe the rasterized frames over the PCIe bus in real time, and output them over another card, this would be lousy system design. You’d be building a large bottleneck into the system and soaking up several PCIe lanes for no reason at all.

Jeremy, check out the Thunderbolt process diagram PDF [link], page 3. It shows two ways that DisplayPort may be passed directly into the Thunderbolt controller (which I think is more or less what John-Michael is suggesting is necessary), preserving PCIe bandwidth for data instead of display.

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

[Jeremy Garchow]

Thank you. I have seen the diagram, but I have not read that paper:

“For some power users, optimal workflows can be had with workstation performance and expandability while using a thin and light laptop. Thunderbolt technology enables using the thinnest and lightest laptops, connected, with “in the box” performance over a single external cable, to high-performance external media drives, HD displays, HD media capture and editing systems, as well as legacy I/O hubs and devices, for the utmost in performance, simplicity and flexibility.”

I keep flip flopping. Good thing I am not running for public office.

Now I’m back in the camp that Thunderbolt makes sense for laptops/iMac types and not desktops, and this paper pretty much spells it out, but doesn’t say the word. In the fine print, they are comparing thunderbolt to eSata, fw, USB, not PCIe 2 or 3. Laptops/non full length PCIe computers will get performance that they have never seen before. Once Thunderbolt catches up, then it will make sense on desktops.

Jean-Michael, with this information, you are right. If Thunderbolt really is the traffic cop to the whole data structure, then Thunderbolt limits a desktop’s capability.

Now, it’s back to if Apple will release a MacPro without Thunderbolt. I think they will now that I have changed my mind sixteen times. Otherwise, it’s a 4x MacPro which doesn’t make sense.

Sorry for my incorrect ramblings, I apologize to you J-M S-W. This is miniaturization in the flesh.

Thanks, Walter, for pointing out the document.

I don’t think Apple will kill the MacPro as they can’t quite yet. Of course, I could be wrong, I’ve certainly proved that today.

Jeremy

[Kevin Patrick]

I haven’t read this thread enough times to get to the point where I can say “Ah, I see … ”

But, I was looking at some recent info on Intel’s latest high end processors, the Core i7 Sandy Bridge E. My take on the review this site did was that it appears to focus on increased multi-processor, multi-threaded performance. Which I assume applications like FCP X and After Effects. (referencing the recent benchmark data from Barefeats) The article notes that the first models will be 6 core, but the die clearly shows a layout for 8 cores. It also has 4 channels of memory, as opposed 4 channels in the current i7, providing almost double the bandwidth.

But it appears Intel removed the integrated graphics core. Here’s where I was wondering how this relates to what you guys have been talking about.

How does the removal of on-die GPU affect all of this?

What does this mean to systems that are PCIe based and systems that have Thunderbolt?

Or does it even matter?

Would this be the CPU that people have speculated Apple has been waiting for? Assuming of course they are still planning on updating the Mac Pro.