NVIDIA shows signs ... [2008 - 2017]

Status
Not open for further replies.
I actually saw DX11 software on DX11 hardware for AMD. Instead of the same CUDA/DXCompute Presentation from March, not even a mention of OpenCL or DX5 Compute shaders.

Publicly demo'ing unreleased hardware isn't usual IHV behavior so I don't really know what new stuff you expected to see. There wasn't a peep at Computex 2006 about DirectX 10 and then we got G80. AMD simply seems to be applying the CPU habit of building hype around products months before launch via demos, silicon porn etc....

Yeah, they have shown other stuff too; some soldier guy with cybernetic hand and apparently some computeshader demos based on the old froblin demo, and then couple tesselation demos based on the old froblins demo and ruby demo (mountains)


Nope, missed those. Thanks.
 
the soldier, as explained in the thread here is from the dx11 sdk and, who said it ran at minutes per frame on the software renderer? versus 14/30 fps on Evergreen.

I wish everyone would follow intel and AMD and show those embryo's when they had the opportunity.
 
even a single triangle can take several minutes on the reference rasterizer, hardly a way to perform comparisons with real hardware.
 
even a single triangle can take several minutes on the reference rasterizer, hardly a way to perform comparisons with real hardware.

Yea, but we now have WARP, which is considerably faster.
Then again, it could STILL take several minutes to render some stuff. It's not as fast as my X3100, and that can take its sweet time rendering stuff aswell, such as 3DMark Vantage :)

By the way, what SDK is that then? I checked the D3D11 samples from the March 2009 SDK and couldn't find the soldier?
 
By the way, what SDK is that then? I checked the D3D11 samples from the March 2009 SDK and couldn't find the soldier?

Herr Spille mentioned running them as well:

http://www.pcgameshardware.com/aid,686395/AMD-shows-DirectX-11-Tessellation-in-action/News/

PC Games hardware has used this demo to explain the effect of hardware based Tessellation in our German print magazine. But since we don't have access to DirectX 11 hardware yet, the demo was run on the CPU via Microsoft's Reference Rasterizer. Although we used a overclocked Core i7 processor we didn't reach interactivity at the highest Tessellation level 32 but sometimes had to wait for several minutes before the video information was updated.
 
It's the wobbly face, that's in the SDK, not the soldier. Should've been more clear. Sorry :(
 
I was assuming that people already realised that the Ion brand had been repurposed for some time.

Yep, looks like MCP79 is now "Ion" regardless of the platform. From a marketing perspective it's a no-brainer. It seems like Tegra is the real cash cow though - it has the potential to completely replace chipset revenue and then some. Assuming it goes anywhere.
 
Since I mentioned I quit in this thread earlier. I thought I'd provide an update. Whether anyone cares is another matter. But :p

Yesterday me and Several guys at Nvidia had a phone conference call about me and the user group/SLIZONE. Went over various issues. And worked out some of the disagreements. After the discussions most of my concerns have been alleviated and actually have some things I am looking forward too.

So basically. Grievances worked out, I'm back to working with Nvidia. Got alot to look forward too and more stuff to talk about soon.

Chris
 
Since I mentioned I quit in this thread earlier. I thought I'd provide an update. Whether anyone cares is another matter. But :p

Yesterday me and Several guys at Nvidia had a phone conference call about me and the user group/SLIZONE. Went over various issues. And worked out some of the disagreements. After the discussions most of my concerns have been alleviated and actually have some things I am looking forward too.

So basically. Grievances worked out, I'm back to working with Nvidia. Got alot to look forward too and more stuff to talk about soon.

Chris

Good for you Chris! Happy you got back to doing.. what you probably like doing most. (cleaning other people's mess and work a bit with computers)
 
Since I mentioned I quit in this thread earlier. I thought I'd provide an update. Whether anyone cares is another matter. But

Yesterday me and Several guys at Nvidia had a phone conference call about me and the user group/SLIZONE. Went over various issues. And worked out some of the disagreements. After the discussions most of my concerns have been alleviated and actually have some things I am looking forward too.

So basically. Grievances worked out, I'm back to working with Nvidia. Got alot to look forward too and more stuff to talk about soon.

Chris

That's cool to hear, Chris; glad things worked out for you.
 
Since I mentioned I quit in this thread earlier. I thought I'd provide an update. Whether anyone cares is another matter. But :p

Yesterday me and Several guys at Nvidia had a phone conference call about me and the user group/SLIZONE. Went over various issues. And worked out some of the disagreements. After the discussions most of my concerns have been alleviated and actually have some things I am looking forward too.

So basically. Grievances worked out, I'm back to working with Nvidia. Got alot to look forward too and more stuff to talk about soon.
Good to hear, you being in it gives me a bit more faith in their focus group. :)
 
Patent trolls are such pains for everyone. I am glad in this case someone stood up to Rambus and I hope that this ruling stands regardless of how much money the lawyers need to make.
 
Patent trolls are such pains for everyone. I am glad in this case someone stood up to Rambus and I hope that this ruling stands regardless of how much money the lawyers need to make.

Yes. I hope this opens up a negative trend against rambus regarding lawsuits. All it really takes is one loss.


Thanks guys :)
 
Rambus may appear to be like a patent troll, but I think the reality is unclear.

If you've actually followed the whole history...Rambus isn't innocent, but neither is anyone in JEDEC, especially DRAM vendors.

Rambus has made some huge contributions in terms of moving forward high speed interfaces. If you look at the real story, the DRAM vendors were trying to screw Rambus out of any money for their contributions and original work, while Rambus was trying to screw the DRAM guys for using some things which rambus had pioneered and patented. Nobody is innocent.

Like it or not, unlike most patent trolls, Rambus actually does make IP that other people use. They design memory subsystems (e.g. PS3), they have designed DDR2 and DDR3 memory controllers as well (at the behest of 3rd parties that had difficulty getting high performance memory controllers).

Anyway, if you look at what Rambus is doing (http://www.realworldtech.com/page.cfm?ArticleID=RWT120307033606&p=2), it's pretty interesting.

They are moving away from a parallel command bus towards something that looks like CSI - where command/addressing and data are all treated similarly and running at the same frequency.

Intel's aged front-side bus runs the data bus substantially faster than the command bus. The biggest the difference between the two, the lower the efficiency. GDDR1-5 works the same way.

I suspect GDDR6 or 7 will have to move to a differential, point to point command link in the future.

DK
 
Status
Not open for further replies.
Back
Top