fightingfi
2[H]4U
- Joined
- Oct 9, 2008
- Messages
- 3,231
Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
It's going to be great for people with Intel and AMD CPUs that have an iGPU in them. It might put you over the hump so that you can enable a few more graphics settings.
Yea that's cool as hell! Should have made the title of the thread more exciting.
If this takes off, I wonder if it would be possible/practical to make a dual gpu card, with a gpu from each team. Ignoring cost, and likely licensing/marketing issues with a cross-branded card, it might not be the worst idea considering it could offer fall-back single gpu compatibility modes for those titles that either don't work well with mutli-gpu dx12, or have a strong brand preferance between nvidia/amd.
Likely, cost and other compromises would make the card impractical, but it would be interesting to see.
Will all DX12 games be able to leverage this by default?
has to be implemented by the dev I think
These Oxide guys are on the ball and on the bleeding edge of graphics developement.
Not to sure but was this not already possible under OpenCL ?
I wonder if the GPU companies want you to slap 2 - 3 of your old GPUs together and get better perf than their flagship without the problems of SLI/CF.
Pretty sure they dont want that.
Pretty sure this will never be a reailty then, devs will have no incentive
I hope it does though, in the next 5 years maybe.
This could change the upgrade scene forever, in the long term.
You wouldn't be able to tell from the gameplay footage, which seems more like an indie supreme commander knockoff with graphics from 2005
We'll see. This is simple AFR, and the scaling is unexciting. But it does work, so at least we have that one confirmed
You can use OpenCL to spread tasks across multiple GPUs, but it's not well suited for rendering games like D3D 12 is. It could be used for other tasks like physics, but there hasn't been much interest in doing that so far.Not to sure but was this not already possible under OpenCL ?
does anyone recall lucid hydra? I wonder how this would compare to that solution ?
Now this would be a dilemma, imagine if PRIME was gifted a high end pc with a 980ti and a fury x working in tandem for dx12 games. Do you think he'd take the increased performance or purge his system and cut out the amd card?
I'm thinking the latter.
no
The message you have entered is too short. Please lengthen your message to at least 3 characters.
You can use OpenCL to spread tasks across multiple GPUs, but it's not well suited for rendering games like D3D 12 is. It could be used for other tasks like physics, but there hasn't been much interest in doing that so far.
amvidia r980furyti the ultimate gpu, the fanboy killer edition
Well, I imagine PhysX would still be disabled due to Nvidia's Vendor ID check. And I certainly wouldn't be surprised if Nvidia decides to sabotage this explicit multi-GPU function in a future driver update because SLI mode wouldn't be "certified" when used in conjunction with an AMD card.
You can do anything you want in OpenCL, including writing a complete reali-time 3D rendering engine with multi-GPU support. However it doesn't have any built-in framework for such a thing, so you would have to write it all from scratch, which would be much harder than using D3D. And it's designed for compute, not graphics, so you would miss out on a lot of harware acceleration.I thought that the "current" driver would allow this under OpenCL spread tasks to whichever compute form was available rather then stuck to a single gpu by default..
You can do anything you want in OpenCL, including writing a complete reali-time 3D rendering engine with multi-GPU support. However it doesn't have any built-in framework for such a thing, so you would have to write it all from scratch, which would be much harder than using D3D. And it's designed for compute, not graphics, so you would miss out on a lot of harware acceleration.
Yes, you can send tasks to any combination of GPUs regardless of vendor, as long as they support OpenCL. You can even distribute a workload over GPUs and CPUs at the same time.I was not talking about the DX12 feature I'm talking about driver support for accessing multiple GPU from different vendors in one machine with OpenCL.
Sure they will piss all over Microsoft just because that is very good way of doing business ...
Are the DX12 functions used in mixed GPU setups core parts of DX12 functionality? Meaning is access to these functions required for DX12 certification?