Will 1060 be as bad for VR as previous pascal cards? (broken DP, no Asyn compute etc)

[piesquared]

Will be interesting to see if nv fixes the limitations in the rest of this generation's lineup. They obviously can't fix the architecture's lack of async compute support, however there is a slim chance they did fix the architecture with the smaller die, but would they be able to fix the broken DP Vive HMD support? It seems like it probably isn't fixable or else they wouldn't have released without fixing what would otherwise be a pretty glaring oversight.

 

[renderstate]

Troll thread. It needs to be closed ASAP.

 

[bystander36]

The Vive works, but atm, only through HDMI, which is not much of a limitation. So you use a different cord, big deal. And they already mentioned they have a fix on the way.

I'm not sure of async has any affect on VR. We'll see I guess.

 

[piesquared]

The Vive works, but atm, only through HDMI, which is not much of a limitation. So you use a different cord, big deal. And they already mentioned they have a fix on the way.

I'm not sure of async has any affect on VR. We'll see I guess.

I think I've read that async compute is a very nice feature for.developers making VR software, but yes we shall see (maybe the 1060 architecture is different from the 1070/80, so that it supports Async compute?). For the Vive though yes HDMI is possible, but just curious if nv can/will fix the broken DP implementation. I can't imagine how they could have missed it.(I know they've 'said' a fix is on the way, but they also 'said' Fermi will have dx12 support so who knows. If it was a simple fix, you'd think they would have fixed it before they released the card.

 

[Grubbernaught]

Perhaps you are thinking of async timewarp? Definitely a nice feature for VR, already supported by nvidia and oculus. Less so AMD and Vive.

Async compute doesn't benefit VR anymore than it does a standard game.

 

[3DVagabond]

We're going to have to wait for some VR reviews. FWIU AMD has the superior solution. That's yet to be proven though.

The DP bug with nVidia is real, though.

 

[bystander36]

I think I've read that async compute is a very nice feature for.developers making VR software, but yes we shall see (maybe the 1060 architecture is different from the 1070/80, so that it supports Async compute?). For the Vive though yes HDMI is possible, but just curious if nv can/will fix the broken DP implementation. I can't imagine how they could have missed it.(I know they've 'said' a fix is on the way, but they also 'said' Fermi will have dx12 support so who knows. If it was a simple fix, you'd think they would have fixed it before they released the card.

Async compute is nice for all games, but Nvidia's cards still do well despite their limited implementation. They did improve things with the latest series though.

The DP issue is a question mark. DP works fine in all other ways we know, and it's not even clear if it was Vive or Nvidia who is out of spec to cause the problem, but that doesn't affect your ability to use the Vive, as HDMI works. It's quite easy to have missed it, as Vive just released itself very recently.

As far as I'm concerned, I could care less if Nvidia made it work at this point. It's not needed, but it's always good to not have anything that isn't working.

How much async will affect VR is just a matter of waiting for reviews.

 

[antihelten]

Perhaps you are thinking of async timewarp? Definitely a nice feature for VR, already supported by nvidia and oculus. Less so AMD and Vive.

Async compute doesn't benefit VR anymore than it does a standard game.

AMD supports Async Time Warp (ATW) every bit as much as Nvidia, and quite possibly has a better implementation.

AMD utilizes async compute plus their quick response queue (QRQ) for the ATW shader.

AMD's QRQ appears (based on the little info available) to essentially combine preemption with async compute. In comparison Pascal performs ATW via preemption (something that Pascal should be much better at than Maxwell)

 

[littleg]

it's not even clear if it was Vive or Nvidia who is out of spec to cause the problem

It's pretty clear since it works fine on every other card but Pascal.

 

[bystander36]

It's pretty clear since it works fine on every other card but Pascal.

Maybe, maybe not. One thing that was mentioned with the DVI thing, is that Nvidia added some checks to make sure things run within spec. Their older GPU's didn't. This could simply be that Vive doesn't run completely within spec, and the GPU isn't allowing it to work out of spec. If so, this is likely a software/bios fix. It could be something else.

We don't know just yet.

 

[digitaldurandal]

There is a minor issue with 1070/1080 and Vive. Don't exaggerate it.

The 1070 and 1080 are supposed to be good VR cards as they have both Async Timewarp and Single Pass Stereo Rendering.

Somewhat OT.. I wonder if Single Pass Stereo Rendering affects 3D.

 

[Headfoot]

Perhaps you are thinking of async timewarp? Definitely a nice feature for VR, already supported by nvidia and oculus. Less so AMD and Vive.

Source? As far as I know this is entirely false.

 

[Yakk]

AMD supports Async Time Warp (ATW) every bit as much as Nvidia, and quite possibly has a better implementation.

AMD utilizes async compute plus their quick response queue (QRQ) for the ATW shader.

AMD's QRQ appears (based on the little info available) to essentially combine preemption with async compute. In comparison Pascal performs ATW via preemption (something that Pascal should be much better at than Maxwell)

In addition to the QRQ, AMD also will use their HMS hardware modules, which are similar to their ACE units, but with added ability to be more programmable.

 

[antihelten]

Somewhat OT.. I wonder if Single Pass Stereo Rendering affects 3D.

It should be possible to utilize it for 3D, but I doubt any of the traditional 3D solutions will implement it (all focus seems to be on VR as far as single pass stereo is concerned)

 

[imaheadcase]

I like how the OP thinks these issues are the end all be all of the card. As if VR is the selling point or the cards. lol

 

[antihelten]

In addition to the QRQ, AMD also will use their HMS hardware modules, which are similar to their ACE units, but with added ability to be more programmable.

This is not in addition to QRQ since QRQ is handled by said modules (in other words you couldn't do QRQ without them). They are called HWS (HardWare Scheduler) not HMS btw, and a single HWS module is basically equal to two ACE modules, plus a few extra features.

I like how the OP thinks these issues are the end all be all of the card. As if VR is the selling point or the cards. lol

It's definitely one of the selling points (Nvidia has spent a ton of time talking about the VR capabilities of Pascal), but yeah these issues are probably not going to have any major effect on the overall perception of the cards.

 

[96Firebird]

Source? As far as I know this is entirely false.

Vive doesn't support ATW...

 

[antihelten]

Vive doesn't support ATW...

It's probably more accurate to say that OpenVR (the SDK used by the Vive) doesn't support ATW. ATW is a software feature and if Valve decided to implement it in OpenVR someday, then the Vive should be perfectly capable of running it.

 

[bystander36]

In the end, all these marketing features are just different means to get to the same goal. We'll have to see reviews and benchmarks to see how it all turned out in practice.

 

[jj109]

Vive doesn't support ATW...

That's because ATW is overrated. It's a constant 15-20% performance penalty due to overdraw and causes weird artifacts when used constantly.

Unity and Unreal are getting SMP baked into the engine, so Pascal will probably leave all current cards in the dust when that happens. But VR's a small market and the portion that can use SMP performance is even smaller. For now it would only benefit people looking to supersample in VR.

 

[tviceman]

This thread is just as much a troll thread as if someone started one with the subject "will Vega be as bad for Polar ice caps as Hawaii was?"

Maybe Nvidia will release a driver fix for the DP issues with the Vive, giving users the option to check a compatibility box for a performance decrease.

 

[nurturedhate]

This thread is just as much a troll thread as if someone started one with the subject "will Vega be as bad for Polar ice caps as Hawaii was?"

Maybe Nvidia will release a driver fix for the DP issues with the Vive, giving users the option to check a compatibility box for a performance decrease.

About 80% of all threads on this forum are troll threads, from both sides.

 

[Sweepr]

This thread is just as much a troll thread as if someone started one with the subject "will Vega be as bad for Polar ice caps as Hawaii was?"

No doubt, should be locked.

 

[Headfoot]

It's probably more accurate to say that OpenVR (the SDK used by the Vive) doesn't support ATW. ATW is a software feature and if Valve decided to implement it in OpenVR someday, then the Vive should be perfectly capable of running it.

That was my impression. Pascal and GCN seem both perfectly capable of doing it...

 

[ioni]

The Vive works, but atm, only through HDMI, which is not much of a limitation. So you use a different cord, big deal. And they already mentioned they have a fix on the way.

I'm not sure of async has any affect on VR. We'll see I guess.

As a 1070 owner, it's a pretty big deal imo. Most people don't have DP monitors, but HDMI or DVI. If it's HDMI only, you need to get an adapter, which both adds to the cost of the card and the time before you can use it. I find it pretty laughable that people find a defective $400+ card acceptable.

I like how the OP thinks these issues are the end all be all of the card. As if VR is the selling point or the cards. lol

VR was one of the biggest selling points Nvidia marketed for Pascal with all that talk about rendering 2 images in 1 pass and SMP etc.