Gadgetory

everyone welcome to another episode of ask GN where we answer questions as many as we can anyway from the comments so as always leave a question below I've seen a couple repeats from the past few episodes so I am trying to pull those up to actually talk about them but we won't get to them all but do leave a question below and I'll try to get to it for next episode before we get into the content today it is brought to you by AMD so we've got a couple Andy devices on the table the RX 480 of course is at one of the highest end cards that while the highest end card that they make in the current polaris generation and that's available around $250 gigabyte models and then the free sync devices that we've been talking about for the last few episodes so we got one of those on the table because the RX 40 is can drive 1440p displays now so that's pretty cool but the only other thing to note I guess is that this look looks fake looks like spent photoshopped in but I assure you we have painstakingly arranged it so that it is not reflecting too much light from the lights that you all likely take for granted because there's about 2000 LEDs in my face so the first question is from comenzó rollit AKA Sam who loaned us the Titan XP previously Sam says is there a need for a hyper threading on ten core CPUs at like the 69 50 X if you're going to use them for gaming or can you disable it to gain more stable overclock if needed so it's a good question you can definitely disable the hyper threading I know Sam knows this but for those who don't you can go into BIOS and actually turn it off and it's a pretty interesting way to just test is there actually gain from hyper threading and in some games there is so you'll see this in our battlefield one CPU benchmark it probably went live yesterday we're filming this before it went live but the CBO benchmark shows that a 6700 K with hyper threading actually performs reasonably different from the 6700 K without hyper threading and it was I think it was about maybe a 10% difference I think was about a 10% difference so in that particular scenario the FPS was so high anyway because we're using a 10-8 ftw at 1080p but there was a real difference that you could see with hyper-threading and that's also true primarily for low values so when you're looking at 1% in point 1 percent low is in our metrics or the longer frame time is the worst frame times in a benchmark that's where we see the biggest difference with hyper-threading it's true for Metro last light and it's true for battlefield one as well so that's dated yes there's a difference from hyper-threading we've also seen a difference with the 59 30k which is not a 10 core CP like this one Sam's talking about but I don't have the 10 core CPUs that are out right now so I don't know how much impact hyper-threading has with those I can say we've seen an impact in FPS again about 10% and that would help you get to 144 Hertz if you're trying to push a 144 Hertz display or something like that but it does depend on the game quite a lot depends on how they've optimized their game now as for other use cases I know with Adobe software for example a lot of those kind of cap out at utilizing threads maybe six cores efficiently so there's limits in production software some production software use that record that you throw at it or every thread that you throw at it so there's going to be a more of a benefit there than with games but not always because the Dobies Creative Suite is actually pretty limited and the threads you can utilize which is a shame but they're getting better with it as for gaming there like I said depends on the game if I had in your case if I had a 10 core CPU you're basically looking at do I want to potentially increase my overclock a little bit and get the clock rate game gain in games that are clock sensitive or do I want to keep hyper-threading and get the hyper threading gain and games that our thread sensitive and I would say that in most of our current testing and I need to catch up on the more recent launches other than battlefield most of our current tests we we do see more of a gain from the core increases the clock rate increases then from the thread count in creases but that's changing as we roll forward so I'll let you know how all that goes as we continue to have to do next question is from dirt zoo who says there's a quick there's a question for the am the expert cat who we just kicked out of the room I built a red machine for battlefield 1fx 9370 gigabyte motherboard to XFX for 80s and crossfire DirectX 12 does not seem to support crossfire in this game week a battlefield one he says DirectX does not seem to support crossfire tried the x11 but it is unplayable tearing and just bad things happen in crossfire do you think the x12 will support cross right in the future so what are we doing here oh I'm to 144 Hertz monitor okay so yes that also depends on the game I guess the rest of the question here is actually also kind of important saying I'm getting 90 ish fps now and on ultra with one monitor but I used if I use two it stutters bad lower settings a resolution does not seem to help so I don't think it's a power issue addressing the last part of the question 850 watts is plenty for two for 80s and well yeah 1993 70 you're pushing it a bit but not that much you're still you should still have enough power uh-huh so I don't think that's gonna be your issue unless your PSU is not outputting what it should be but that seems that's uncommon and unlikely especially for an 80 plus gold P SES but as for the rest of this crossfire is again going to be game dependent so in DirectX 12 versus DirectX 11 the only game that really really does multigp well right now is ashes and that is more of a benchmark than a game but ashes does what's called MDA multi display adapter this is different from link display adapter or the more common implementation and that's something we've talked about in the for in in the past with the ashes benchmarks with 1080s with the crossfire 480 testing we did and the sli 1070 testing we did which needs to be updated but MDA is what you're gonna see with the x12 that kind of bypasses the need for a bridge and goes through the PCIe bus instead and then it also reduces some of your latency concerns but it's not really prevalent right now so ashes is the only game that's really doing it for the most part I don't think battlefield one does but maybe someone does know and can count I haven't actually researched that yet DirectX 11 I it's so basically you have an issue of latency with multi GPUs always going to be a concern because especially with AFR or something like that where the GPUs have to communicate and then build frames together intersperse the frames one after the next so there's always going to be a worse frame time output with multi-gpu so if you have two rx for ATS you might actually see that stuttering you're talking about where I think the word you used was tearing but normally it's more of like a micro stutter type thing where the latency between frames is disparate enough that the gap exceeds sort of an eight millisecond window so once you have more than eight milliseconds one frame to the next it starts to become more visible to a human player that's where you see the stuttering and that often happens with the multi-gpu just because of the latency involved with to display devices computing the frames so as for your specific question I guess I would have to know more is this happening with other games or not if the way you could troubleshoot this you can buy first of all power is ever a concern is for anyone out there you could buy what's called a kilowatt power meter for the wall they're not expensive they're maybe 20 or 30 bucks not the most accurate tool on the planet but pretty damn good for testing how much power your computer draws so that's just kind of general advice if you have a system you've built you're trying to figure out do I have Headroom here to add another card you can buy one of those for a couple bucks 20 bucks or so and see how much power you're drawing at the wall so that would answer that question I don't think it's a power issue but I think it's basically a game issue so the 9370 will definitely bottleneck either so the the VF ones benchmark we just did the CPU benchmark should be live on the channel did not test the 9370 we tested the 83-70 and we're seeing some pretty decent bottle knockin on the GTX 980 ltw hybrid so you can check out that content if you're curious more about about that bottleneck next question is from John Smith he says coming from one of your CP review videos I have a question could you add gaming and recording as one of your CD and GPU benchmarks so we've actually tested that first of all recording software nowadays is mostly moving to encoders on the GPU so these are physical components on the GPU that handle encoding or capture of live video or streaming of live video and they are a lot more efficient than the way that it used to be done which is to the point of this question on the CPU so these fraps that's all CPU so you're recording with fraps pretty much uncompressed lossless AV is that hammer the framerate down about 50% lower than what it was when you're not recording fraps is doing that on the CPU so that back when that was really one of the only ways to do this that would be a great benchmark because it was the only way to do it so now though this software like shadow play and like GVR or whatever it's called these days it does it all on the GPU encoder and OBS can do it on the encoder on the GPU as well whether it's aim the or Nvidia and video has special codecs that are supported and the does it as well in different ways but for the most part GPU encoding is supported with all the major software suite I know of for video capture as for the impact we have an old old video now shadow play versus gbr versus fraps you can still find it on the channel and an old article as well same title and that shows that with shadow play from memory I think it was about a 3% hit to performance overall the actual frame rate performance in a game and that's pretty good GPR was very close to that it was a couple percent but it did depend a little bit I think on may have depended on the software settings but GBR was pretty damn close and that was AMD a solution at the time they've rebranded it a few times since so I'm not sure what it's called right now OBS does take more of a hit so OBS is harder to configure there's a lot more options it's more built for streaming that would be one that might be worth adding to a benchmark but for the most part I would say we probably won't be benchmarking with recording scenarios because for CPUs because so much of it is just done on the GPU now and the hit again is like three percent depending on the south resolution if you're using something accelerated by Nvidia or AMD specifically the hit is so small that it's it's negligible but OBS would be more of a concern for sort of live streaming and that's still it's getting optimized on the hardware so it's kind of fading away as an issue but it's still worth discussing so hey we'll look at that in the features and benchmarks or something like that next question is from sir papa who says for the next ask GN why is it that AMD has a larger performance gain on dx12 than nvidia this is for the new and old cards will this be the same going forward Vega vs. Volta Vega versus Walt I don't know I don't know anything about the architecture of either of those and that is really what matters the first part of this question why is the performance gained larger for AMD a lot of that is dx11 optimization NVIDIA invested a ton of resources into their drivers or dx11 optimization so if you watch our interview with Rajat Kaduri you can kind of learn in there that what these driver teams do for Nvidia and AMD both they'll go into the game basically indirectly through the driver and they will rewrite how the game is drawing different effects so if md says you know our hardware handles this better if we use these shaders or if we use this arrangement and the sort of the the hardware stack what components specifically within the GPU are being used to handle what processing is demanded by the game or the engine nvidia has invested a lot of resources in dx11 optimization I would say that's probably a lot of it because you know there's this discussion where the architectures are different and people keep hearing a ce4 and E or compute preemption for NVIDIA and I think the common internet place to jump is well am these scales better because of a CES but that's not all there is to it and these scaling looks as good as it does because they're dx11 performance just wasn't as good as Nvidia's was and part of that is GCN architecture so what you end up with is scaling that either looks potentially slightly negative on Nvidia or equal at best and scaling that looks really good for AMD depending on the implementation doom with Vulcan would be an example especially with their actual implemented async compute and ashes is another good example but an Ashes you look at the dx11 performance with an D it's really bad and the frame times are really bad too so DX 12 there is actually the implementation of it a DAC is actually really useful and that's again because of the optimization done on the x11 drivers by Nvidia and because of the optimization done by Andy for DX 12 they've shifted their focus but there's not as much driver interfacing with these new api so with this is what I again talked about with Raja kanuri going forward a lot of this onus will be on the developers to build their software so it actually leverages the APS properly the way you have games like Battlefield Wan which put out really really a spotty performance at the x12 best-case scenario you're about equal maybe one FPS better with the x12 but the frame times are all over the place and that's because the developers and AMD and NVIDIA can't do a lot to fix that because this is the game talking more directly to the hardware the driver is not really interfacing as much in the middle it's not interfering with what's being drawn as much as it used to with the x11 or OpenGL so the short answers the question is what why is it that AMD has a larger performance Gandy x12 if you're looking at percentages AMD will look infinitely better that's because of the x11 performance is not great and a lot of games tested and then there is something to be said for the asynchronous hardware support stuff like that and video does async compute as well through compute preemption for the most part and we'll talk about that more in the future once Vega and Volta actually have architecture information out there but right now we don't we don't have any information so I can't predict what it'll look like in the future but for now yeah and these seemed decent scaling on games that actually support the extra properly if the developers have optimized for dx12 and for Andy hardware but a lot of game still zero scaling so it's it's a mess right now ap eyes are a big mess you can watch our battlefield 1 GPU benchmark to learn more about it and why that is the case next question - Kulu says do you use for mark sometimes it's a very very simple question I guess I use fur mark to validate Tom's Hardware its findings on the EVGA 1080 FTW so that's what they use to generate the numbers that generated so we use it to validate their numbers and see if there's actually concern I don't use it in everyday testing we use other tools normally next question is Peter custom who says you mentioned it as possible in the previous previous video to move workload from the CPU to the GPU what possibilities are there out there to do that I have an FX 8350 which is really bottleneck in my bf1 performance so I want to use this shift as an interim solution until Zen comes out so unfortunately what I was talking about in that video is not really something you can just do as a user this is when I was talking about shifting the workload from the CPU to the GPU I was talking about api's this is related to the sir powers question about the developers actually using an API to pull draw calls off of the CPU so the GPU can focus on them without needing to communicate with the CPU for every single piece of primitive or geometry drawn and that's that's not something you can control as a user that's done by the developers so the best thing you can do and battlefield one will definitely battle a bottleneck on an 8350 we show that again in our cpu benchmark for the game that should be out by now the best thing you can probably do short of buying a new CPU is to overclock it so if you have a decent cooler on there or if you buy these in cooler should be able to overclock that thing pretty decently in battle one does respond to clock rate increases we saw that in our testing so I would recommend that as the immediate solution or the interim solution until you get Zen because it looks like you're waiting for Zen based on this question last question is apple juice apple juice balm says hey gamers Naxos what is the theoretical minimum size that a GPU die set could be manufactured to before electrical resistance halts the shrinking process that's a pretty hard question then then the follow up to that also by the same user also what does the naughtiest thing the cat has done while you were trying to film the first part of the question the one about the GP diet I can't really answer that I'm not an expert in silicon engineering so that's something that we will certainly be asking experts in the future as we talk to Intel and AMD more maybe at CES I don't have an answer off the top my head several years ago I spoke to a former Intel employee who had moved around the industry a bit talk to him about silicon manufacturing process things like that he was saying that around 8 nanometers there starts to be a big wall and the the actual tools used to manufacture the product would change but I don't know much more than that so I don't have a good answer for you I'm sure there are people who have researched this I'm sure and on champey has talked about it but I'll look at that in the future as for the second part the question the naughtiest thin the quote unquote that the cat has done while trying to film was probably knocking over video cards I think in one of our comparisons she we made it we even included a clip where she just hopped up on the table over on that side of the table just jumped up while we were filming and then proceeded to knock a video card over but it didn't fall off the table so it's all good so that is it for this episode as always patreon like the post love it it helps out directly links in the description below subscribe for more I'll put a link to one of these things down there if you're interested in them I put a link to the free sink monitors and that'll be an affiliate link so thank you for watching I'll see you all next time you