VR ready laptop or amplifier with Pascal GPU. . — Oculus
New to the forums? Click here to read the "How To" Guide.

Developer? Click here to go to the Developer Forums.

VR ready laptop or amplifier with Pascal GPU. .

NacaryusNacaryus Posts: 206
Nexus 6
Do you guys think i should change my Alienware 17 R2 with an msi ready laptop..or wait sometime and buy a Pascal GPU to connect through graphics amplifier.. i guess Pascal with amplifier wins but i still wanted to ask fellow VR gamers. ..

Comments

  • MradrMradr Posts: 3,348 Valuable Player
    edited April 2016
    Laptops are ok, but only supported laptops. I would recommend getting a real PC vs a laptop for VR, but I understand the needs for them. Most laptops can't be used because they have a "m" class gpu inside. The supported laptops have desktop cards instead. The issue is with Optimus and there isn't anything that can be done at this point leaving almost 90% of todays laptops in the dust because - well - they were design for a different idea of gaming.

    I would go with the single laptop if it supports it. There will be VR ready laptops power by Pascals as well in the future. Amplifiers have to run over a bus that may or maybe not able to support VR alone or have some sort of limits we haven't seen yet. It's the only way that might allow todays laptops to work.

    https://www.reddit.com/r/Alienware/comments/3l0790/can_the_2015_alienware_17_r2_work_with_oculus_rift/

    Also, I recommend looking the problem up in the MANY TONS OF THONSUNDS OF THREADS over the forum and readit about it. It's a common enough question that there is a ton of reading media over it.
  • r00xr00x Posts: 389
    Nexus 6
    @Mradr ;

    It's great that you're trying to help but I don't feel this information is not (entirely) correct.

    That said I will start by saying I agree with you - only certain laptops are officially recognised by Oculus as working (those with desktop GTX 980 chips) and if the option is there to get a desktop, it's the better way to go.

    HOWEVER - the problem is not Optimus and it's not the fact that they use "m" class GPUs.

    The problem seems to be in detecting the headset - whether the GPU has a direct connection to the headset via the HDMI port, or not. This is because these ports are more often than not wired to the CPU instead (to use its low power integrated graphics solution).

    Optimus, meanwhile, is a technology for handling seamless graphics switching between a low-power and a high-power GPU, along with some cool PCIE compression stuff. Hence it is almost always present on these systems, because it's there to facilitate switching between the integrated GPU in the processor and the bigger Nvidia chip.

    So Optimus happens, only by coincidence, to also exist on these systems. Therefore it's true that Optimus is a good indicator that a laptop might not work, but it doesn't mean it won't work for sure.

    The true issues, in short then, are:

    1) Most vendors wire the HDMI port to the CPU's integrated graphics chipset, breaking compatibility (nothing to do with Optimus, but Optimus will be present on these systems)

    2) The current crop of "m" variant GPUs, while perfectly capable of driving the headset (as long as issue #1 doesn't get in the way), officially are too slow and do not meet the minimum specs (how much of a problem this will be depends on what you're doing in VR). You may find it works fine, or it's juddery and horrible, depending on the GPU and the game/software you're using.
  • r00xr00x Posts: 389
    Nexus 6
    Afterthought - to expand on my above post - it's perfectly possible for a system to have Optimus extant and running and still drive the Rift, so long as the headset is connected directly to the Nvidia GPU. Some laptops are indeed wired this way (generally laptops which use, or can use dual Nvidia mobile GPUs in SLI since I believe they cannot route video through the iGPU).

    Some users have also reported certain ports not working but others working okay. For instance the HDMI port didn't work (wired to CPU) but the displayport connector did (ostensibly, was likely connected to the discreet GPU instead). Obviously this is terribly unreliable and anecdotal information because there can be variance even within a laptop vendor's model families. It really depends massively on which laptop is being used.
  • NacaryusNacaryus Posts: 206
    Nexus 6
    Well let me ask this way..i could buy a 3k $ VR ready notebook with a 980, provided that i sell my old one and add money on it..

    Or i could keep The laptop and buy a superior Pascal GPU to upgrade with a graphics amplifier.. believe me if i had the budget, i would keep the laptop and buy a solid desktop for VR :smile:
    I understand form The threads i read, amplifier works
  • MradrMradr Posts: 3,348 Valuable Player
    edited April 2016
    r00x said:
    Some users have also reported certain ports not working but others working okay. For instance the HDMI port didn't work (wired to CPU) but the displayport connector did (ostensibly, was likely connected to the discreet GPU instead). Obviously this is terribly unreliable and anecdotal information because there can be variance even within a laptop vendor's model families. It really depends massively on which laptop is being used.
    Aka, why I just keep it simple ;) Too much information can be a problem as well.

    Anywhoo, we don't know the answer OP. Amplifier are full on new and we haven't seen them really in the wild much.

    Other things to consider as well:
    1) Does your laptop even support the bus connection required to make it work?
    2) What are the time delay running over this bus. How much latency does it add?

    Over all ~ it's going to be recommend getting a VR ready laptop than trying to force one to work really.
  • r00xr00x Posts: 389
    Nexus 6
    @Mradr ;

    Oculus have the same approach - better/simpler to not support most laptops in general!

    To play devil's advocate though, there may be people who could have perfectly well run the Rift who've gone out and dumped a ton of money on a desktop they might not have needed. Very likely WOULD benefit from, yes, but not necessarily need. Giving all the details would make those people happy, I'm sure! :smiley: 

    Where graphics amplifiers are concerned, I wouldn't worry so much about the bus. At least not for latency- the bus in question will essentially be PCIE (carried via bespoke interconnection in the case of things like Alienware, or Thunderbolt in others) so we'd do as much good wondering about the latency between the processor and the PCIE slots in our desktops.

    Bandwidth, however, is a very real concern. Especially if (as is likely the case) the laptop's internal LCD is going to be used to display the "preview" window which Oculus won't let us turn off (why not? who knows). Even with Optimus running (a good thing in this case due to the PCIE compression), the preview window gobbles a reasonable bit of bandwidth on a narrow PCIE connection. This isn't an issue for desktops because the monitor is usually connected to the same card as the headset (therefore never needs go on the PCIE bus) and the bus is the maximum x16 width anyway.

    I've used "DIY" eGPU implementations and they tend to struggle with the preview window, typically they top out at PCIE 2.0 x1 which is like 500MB/s in each direction. By comparison Alienware's graphics amplifier uses PCIE 3.0 x4 - that's 4GB/s each way, or 8 times as much bandwidth!!! The new Thunderbolt 3 eGPU docks similarly can offer this much bandwidth, which should be more than enough to handle the workload.

    So, other than mysterious, unpredictable compatibility issues we can't forsee (things of an esoteric nature, like the weird issues people have with some USB 3.0 controllers and the Rift) I don't anticipate any problems using a graphics amplifier as a VR solution. It would be more a question of whether OP selected a good enough GPU (GTX 970 or above, as we know) and whether the processor in their laptop was up to the task.

    That said, @Nacaryus it's undeniably true that the "pure" VR laptops, those with the desktop GTX 980 (NOT 980m) chips, are going to be the best laptop you could get for VR in terms of peace of mind, if not also performance*, and it's also undeniably true that all things being equal, unless there is some really compelling reason you need a laptop, it would be better to get a desktop for VR at this point in time.

    * With one exception: eGPU implementations result in much cooler running laptops because the heat-intensive rendering is happening off-board in another enclosure. This can result in laptops that run cooler, quieter and sometimes last longer than their all-in-one counterparts!
  • DreamwriterDreamwriter Posts: 1,101
    3Jane
    edited April 2016
    r00x - are you sure? I thought the whole point of Optimus was that the high powered GPU's HDMI was connected to the onboard GPU so that it could choose to output the low-power onboard GPU instead of the high powered one when high-powered graphics weren't needed. Without that connection, Optimus isn't possible, and without Optimus, that connection isn't needed. That internal connection *is* Optimus. Thus, the problem is most definitely Optimus, and Optimus is specifically what you need to look for.

    As for the OP, I definitely don't recommend buying a laptop for VR - if you already have one then fine, but paying a lot of money for something that may or may not work seems a mistake to me (a brand calling itself VR-ready may be mistaken, may just be calling themselves that because it's high powered and not because they've done testing on Oculus Rift). Especially when Oculus doesn't support it - what if you have problems with your Rift, and something isn't working? The moment Oculus support realizes you have a laptop they'll say that must be the source of your problems and you won't get any more help.
  • LuscaLusca Posts: 171
    Art3mis
    Not sure if you have looked but there have been a number of people confirming that the Alienware graphics amplifier works well with CV1. The bus bandwidth may hit you for 5% performance, although it depends on the card and whether you have an external monitor to mirror your preview screen on, as that may be responsible for some of the bandwidth suckage (pushing the preview back to the laptop display). From what I've read the bus is not a limitation, except if you are pushing data back to the laptop display. You don't have to worry about the hdmi connection anymore with this scenario. But.. you might want to wait until reports come in that show how Pascal and Polaris do in the amplifier.
  • r00xr00x Posts: 389
    Nexus 6
    @Dreamwriter ;

    Yep, quite sure! Optimus is primarily a software technology as far as I know and that was one of its selling points to laptop vendors - no complicated bespoke motherboard designs.

    The hardware it relies on lies in the GPU itself, generally - this is why I can take a laptop that has no Optimus support (a Lenovo X230 for instance, which has no NVIDIA GPU and something like Intel HD4000 graphics) and bolt on an appropriate eGPU and suddenly I can take advantage of the exact same graphics switching/passthrough that a laptop with Optimus from the factory can do.

    Optimus pretty much relies on the PCIE bus to do everything. There is no re-routing of HDMI or any other display port - it's much lower-level than that. It's not that "HDMI" is being routed through the integrated GPU that's causing the problem - rather, there is no "HDMI" connection to the big GPU at all. Framebuffer data is passed to the iGPU over the PCIE bus and then the iGPU pushes this out via HDMI (or to the internal LCD, of course).

    The little iGPU being wired to the HDMI port is the problem - this means there is no direct connection to the big GPU.

    In an eGPU/graphics amplifier solution, the big GPU is sat on the desk beside the laptop and directly accessible (we can plug stuff straight into it). Therefore the Rift is detected just fine. The most tangible concern here is whether there is sufficient bandwidth over the PCIE link to not only feed the big GPU, but also to carry its framebuffer data back so the laptop's LCD can display the Oculus preview window (which, again, for some reason they force us to use and we can't disable it). This has proven a very real concern for "DIY" eGPU implementations because the bandwidth there is tiny (usually PCIE 2.0 x1) but on proper Amplifiers the link is about 8 times faster and probably OK (seems so, as @Lusca mentioned, from what people with amplifiers say, it works fine and the bus is not a limitation).
  • LuscaLusca Posts: 171
    Art3mis
    @r00x thanks for an excellent explanation of the issue
  • Felixm477Felixm477 Posts: 131
    Art3mis
    eGPU's are are thing now and non mac laptops are coming with TP3. we should have a sticky thread for this kinda topic as im currently also shopping around for a laptop that can do this. 
Sign In or Register to comment.