Gadgetory

hey everyone I'm Steve from gamers access tonight and we're back with another episode of ask GN where we answer your questions so if you have questions for next episode post them below we'll get to a couple of them and see what we can learn for this week we're gonna start off with some CPU stuff and this question is from uh - I think Anthony is that it asks I was always wondering is hyper-threading software and does that mean every CP user to have it as it stands now I don't like hyper threading it's an elite CPU thin and that's why it's not supported because it's a niche thin how can we get more CPU CPU cores to organize threads better and be a bit across all CP ran so first of all a couple things here hyper threading is part of something called simultaneous multi-threading or SMT that word hyper threading is specifically intel's implementation of SMT and Andy can do it as well do they just don't call it hyper threading with that stated does every CPU have it should at all should every CPU have it not necessarily hyper threading is not software as was sort of asked here it is on the hardware level it's sort of a switch that in Intel's case they can throw the switch but there's additional validation that needs to be performed so that does add to cost and in terms of being the so-called elite CPU thing I guess that depends on your perspective but there's definitely uses for hyper threading and places where you wouldn't want to use it and in some gaming scenarios there's really no gain at all and actually we've seen maybe one FPS advantage non hyper threaded equivalent CPU is if their clocks the same and then in other cases like video production you see actually pretty big gains from having extra thread so it depends less on the CPU this is answer the last part of the question saying how do we get CPU cores to organize the threads better that's not really where the problem is the problem is on the software organizing the threads and assigning threads or tasks to different threads so with hyper threading say you have a fork or CPU with eight threads Windows just sees eight threads it doesn't really care how many cores they're all are all it knows is that there's eight threads and maybe CryEngine says let's use six of those for a game logic game I game rendering physics and two other a couple other things if that happens it'll just task out those items to the threads as appropriate as they're defined within the software the game engine windows etc and that's not really on the CPU side the CPU doesn't go seeking for things to assign itself it needs to be told what to do at some level by the host even though the CPU does control everything ultimately at the end of the day so that hopefully answers part of that question as for hyper-threading not being supported because it's a niche thing that same thing there hyper-threading isn't really something that is explicitly supported or not supported it's just the usage of threads and how many threads can be utilized or spawned by the game engines and the software not really so much on the hardware side more entirely on the software side so that stated a couple things to sort of wrap this question hyper-threading is part of SMT or simultaneous multi-threading and thee does this thing where they they're modules for their older architectures and some of the current architectures they'll run sort of a wone module with maybe one FP or floating-point processing unit and then a couple of integer units or a L use and that's sort of their version of SMT depending on Intel and a.m. these definitions of threads or cores they define them a little bit differently between their architectures so that's important to note as well it's not really a linear comparison core to core for these two different architectures between Intel and AMD because they define them differently right now next question is from Andika who asks right now both AMD and Intel are integrating better and better graphics hardware on their CPUs definitely true do you think this trend will continue in the future how far are we from VR capable eye GPUs or integrated graphics processors so starting with the VR question I'm not really sure how far away we are it depends on how VR evolves and what you define as VR capable because of course there can be lower requirement VR applications that aren't as high render quality but I was who are definitely a little ways off right now they're not a GPS aren't that powerful right now it'll be a little while they might be able to run some VR applications or VR desktops but running sort of VR games we're a bit away from that now in terms of integrating better graphics hardware this is pretty interesting and maybe not where the question was intended to go but yes the trend will continue and the interesting part of that is because GPUs are almost sort of replacing CPUs in a lot of ways CPU will stick around but the importance of the GPU has grown tremendously over the last few years and a lot of that is the api's that are being used and the ability to sort of push processes or tasks to the GPU rather than the CPU GPU being parallel so it can handle things better when there's a lot going on at once and as AMD is a good example here as these different processing architectures grow and Intel and AMD build to enable their IGP is to handle some of the basic processing tasks within any post environment the importance of GPUs will continue to grow CPUs will continue to assign more die space to their GPUs and that's I think where we're gonna see a lot of the advanced and in the immediate future for CPU architectures next question is from Ronaldo who says what are your thoughts on Nix gaming so I think we address this in one of the first ask GN episodes the thing with Linux gaming is it is very interesting and I was really hoping that steams OS would sort of build this Linux gaming community and it has it's definitely built up an ecosystem where now developers have some level of support from a major publisher valve where they can get either financial or programming support to support Linux so that's definitely grown and that's a good thing but steam OS did sort of fall off a cliff it's not really as visible as it was intending to be I think steam machines lot of height for them they've kind of disappeared now so I'm not really sure what they're doing with that or what their plans are for the immediate future but it does t-mike it's kind of hibernated in terms of linux gaming in general of course you don't need steamos to do life game it's it's very important I think especially as Windows sort of has historically alienated some of its user base with different moves that Microsoft's making but you've got all these driver issues you have software support issues so it's it's not really gonna take off until more people start adopting Linux and developers can internally justify their expense to build on their development cost and port or support Linux and of course the drivers are also an issue so my thoughts on it purely from sort of an industry standpoint Linux is incredibly important we build our servers on it it's important for competition in the market it's important for offering a free solution if you don't need Windows or don't want it but it's not entirely there yet for gaming of course to get it there though it needs users so sort of a catch-22 in that regard last question is from Ron Mo's who is asking about dx12 again we have more dx12 content every week and Ron says what are your thoughts about dx12 in Nvidia they seem to do bad lately and dx12 tests due to their proprietary middleware game works I believe so first of all game works is not the reason that they have some poor performance and dx12 benchmarks that's that's sort of just completely unrelated the reason that the performance is the way they looks is because the architecture is on a hardware level so and vidya doesn't do the sort of same processing architecture that AMD does where Andy has asynchronous compute engines and those are used to asynchronously queue up tasks we've talked about this a lot in the past video so leave those for more discussion on that but asynchronously queues tasks which was built sort of for mantle and mantle is really very similar to dx12 at a top level so Andy's good at that type of processing where things can get queued up without building and waiting on other tasks in the pipe and video doesn't do it the same way and that's something we'll talk about more as Pascal begins to continue weekend the information I'll start talking about GPU architectures and how the build or the pipeline for GPUs and the rendering pipeline impacts performance in dx12 and other api's but it's really not to do with game works it's just to do with the hardware architecture so that is all for this week as always comments below if you have more questions hit the link the patreon link - patreon the postal video if you want help us out directly thank you for watching I'll see you all next time