Gadgetory

everyone welcome to another episode of ask GN if you have questions for next week post them in the comment section below there's a good deal of them lately so it's just not possible to get to all of them but I will do my best to get to the most interesting ones that I think I can actually answer before getting to this one this is brought to you by our patreon backers you can go to patreon.com/scishow Feud like to help us that directly we have a new overwatch GPU optimization guide that is a preview of what our next patreon goal is because those graphics optimization guides while really fun and very insightful for how the game works are incredibly time intensive so check that out if you are interested we've also got a merch store that'll be up sometime the next week or so and I'll make a separate announcement about that and another ask Jen so first thing to point out I've got some really good questions this week I also wanted to bring some attention to our benchmarks in Andromeda I know that probably a lot of you are not remotely interested in Mass Effect Andromeda for various reasons even because it does not interest you or because you read the initial reviews or saw donkeys video either way there is some cool data there that's worth talking about so one of those things was we had some frame time plots we showed 30-second versus 122nd benchmarks and sure that they are more or less identical for the metrics that we required for the testing so the duration really didn't impact anything 30-second testing it was just fine there we also show the impact of various locations in Andromeda and how planetside versus Nexus versus tempest changed the performance which is thanks again to our discord patreon members because they provided a lot of savegame files to accelerate that process first question is from zeta who says and this was from the discord as well as ada said I'd like to let me let me preface this I have reworded his question because I believe it was about four words and then extrapolated the meaning based on the context so the question was basically I'd like to know if the heatpipe count or service area matters more for an cooller for example and assuming all else is equal is more are more heat pipes better of a smaller size than fewer heat pipes of a larger size so I sent this along to VSG of thermal bench it is a thermal specific website I've worked with VSG a good bit in the last few months or weeks on various benchmarks the guy knows his stuff I believe he's a post doctorate researcher and uses really high-end thermal equipment that 90% of people in the space have no access to so the guy knows his stuff I sent my question to him I'm gonna read his response here he said oh man this is an excellent question the diameter of a heat pipe does two things one influences the available service area of contacts with the fin stack for heat transfer and then two influences the volume bulk of heat transfer from the cold plate to the fin stack where the specific thermal conductivity is on a per volume basis so the number of heat pipes versus diameter of heat pipes plays into both hands and as you mentioned the third factor is air flow restriction this should not be as big of a factor relative to the air flow restriction from the fin stack so if you ignore it then it comes down to what CFD or computational fluid dynamics simulations for heat sources that say are the best way to go in practice there isn't any real difference between four six millimeter pipes and three eight millimeter pipes for low heat loads we see and that would be in computing and this is especially so on GPUs where more thinner heat pipes are adopted to increase contact surface area also the use of the exact type of heat pipe ends up being more of a retail availability and pricing thing than anything else but to answer the question I would say more thinner heat pipes where the fin stack is small would probably be the way to go so there's your answer that is from VSG of thermal vents Thank You VSP for helping out with that one a pretty interesting stuff great question and it looks like for a as he said a smaller fin stack which that would just be the actual you know aluminum block on top of the heat sink there's your fin stack for a smaller one of these he is suggesting that based on theory thinner heat pipes would be the way to go now a big note here for this question when I posed it to him I said you know we have to make a lot of assumptions so let's assume that the fin stacks are the same when we're comparing two units so we have two imaginary heat sinks in this scenario one has four six millimeter pipes and the other one has three eight millimeter pipes so that's our scenario for that scenario we made the assumptions that one the fin stack is the same it is the exact same and two the fan is the same three the cold plate is the same and also has the same design so some of them have those mirrored flat surfaces and some of them you can actually see the heat pipes and even the cracks between the heat pipes which is where you have the biggest potential for inefficiency in your thermal transfer because when you have those air gaps in there that's why we have thermal compound as as I've explained before that thermal compound and I can even see it on this unit which was used ages ago and cleaned it's still in there that's because the point of Tim is to fill the air gaps whether they are the microscopic imperfections or the bigger cracks between heat pipes because that's where your air warms up and creates hot pockets within ya know relation then that cold plate so very good question and again thanks to BSG for helping out with that next question is sir papa who is somewhat of a regular says question for the next ask GN or an idea for an entire episode which is the direction where you can add up going how does multitasking affect game performance if I have a few chrome tabs a stream Skype discord steam and battlenet etc open at the same time is my performance affected much maybe see how Rison can combat this with its many cores so we did definitely see this question we've thought about in the past so I brought this question up to Patrick Leith and when we were talking about the options for the show and he has been helping with all of the CP benchmarking lately he wrote the 1700 X review that's on the site so he's got hands on this stuff we of course in talking on this were immediately facing the the challenge of methodology and just execution in general the test because you look at this stuff and the methods used to get accuracy test to test when you're starting to introduce a whole bunch of things especially webpages it's it's not going to be easy to make sure everything's consistent one run to the next so yes we see your question and we have some ideas for it we've brainstormed a bit one thing something like even chrome doesn't really hog CPU resources when it's idling so if you're running an exclusive fullscreen you're generally not gonna see a whole lot of CPU resource consumption by Chrome the one outlier there would be something like twitch or really any streaming service and in that scenario if you're still streaming stuff other than maybe music while you're playing a game and exclusive fullscreen on a monitor it's probably because you have two monitors so when talking about Patrick I think the way to properly do this and do it justice would be a multi-monitor setup unless we wanted to do some kind of like windowed mode gaming which I don't really like that idea so we would probably have to go multi monitor because otherwise Windows is just gonna stifle all of those tasks in the background that aren't being used when you're exclusive full screen to another application in theory there's some trick places where that's not true particularly with some webpages but that's generally how it should work so multitasking isn't really multitasking if you're exclusive full screen and battlefield one for instance and you've got Excel and Word open doing nothing in the background so to do this justice will do multi monitor the next thing is what what tasks do we use the test so because we're mostly doing this or exploring doing this as a user request it was a popular one I would much rather hear from you all what you would like us to investigate for use in this benchmarking as the other tasks aside from gaming so leave a comment below I will have a top-level comment that's pinned to the top of the comments thread reply to that one and tell me what kind of things you want open in the background while we're doing tests and use the assumption that we're doing a multi-monitor setup so you're gonna you can have the game here and the other stuff here the one thing we're looking into immediately would be something like a stream now to do that you have some complications you have to make sure that the amount of data being received is the same in the processing being done by the browser is the same and by the CPU of course as a result we have a solution for that already brainstorm that one up earlier so that's not gonna be a problem streaming is one thing excel word I don't know if really those would be engaged so let me know what you think that is more of a user request at the end of the day but yes we absolutely see your question and I've thought about it so it's a possibility next question Omar Baja says when a company is signing the same exact GPU with different factory overclocks for a price premium do they have the same overclocking Headroom parenthetically at neglecting silicon lottery or are they bend I'm asking about the normal I'm I'm asking about the normal not like me known to be bend kingpin cards in my case I'm specifically looking for the X of X 480 GTR 1 version runs 12 88 the other 13 38 for a $20 price premium so there's another one where I wanted to pitch this to manufacturers because I thought this you know this is a good question we've kind of addressed it in the past but in the past when I have addressed it it's been just based on information I've collected and I wanted official responses this time so we reached out to EVGA in Asus because they were most immediately available to us for before this video so I asked EVGA inside the same thing to them if you sell all these different SKUs let's ignore things like kingpin are they Bend how are they Bend how do you do that process EVGA said that more accurately they do pre testing they didn't want to use the word bending and I actually agree after following through the process so EVGA does a series of tests to make sure that bass and boost reach a certain minimum and their tests for bass are particularly abusive for something like the FTW series and so they'll they they will pass a high percentage of the cards that might be FTW targeted but the ones that do not pass the minimum bass and boost test will not be given that higher ski brand name instead something on the same PCB would end up with the DT Series if you're familiar with a DT series that's basically what that is it means that it can still overclock pretty well and they were telling me a lot of those cards still the same Headroom as the FTW cards but they won't have the same high bass or boost pre-configured clock because they were failing and again the abuse of synthetic tests for for validation to make sure the product should be wherever it is ultimately going so they call it pre testing in that regard and they also said a high percentage of them will pass and become the target card like FTW evj will mount the GPU to the board so you have to do that you buy the GPU from Nvidia they ship you a whole bunch of chips you build or buy the boards in the case of these cards it's build some of them might be reference boards but these aren't and then you take the chip you take the board and you put them together you assemble the card effectively and test it so they do that that determines the testing determines which skew within the boards available skews it will become FTW DT or otherwise we asked if they ever tried to test the GPU and find that it's so good they want to pull it and put it on a different card like kingpin and they said that they have done that in the past but it's a huge amount of work and it's kind of risky because when you're dealing with BGA mount things desoldering it effectively and putting it on something else is not a trivial process so this is a quote from Asus they said it's not really bending per se as you have to test the chip to understand the differences not only in clocking capabilities but also temp ranges at those clock speeds so pre testing and this was in response to my question is a good way to say it that said you will still have the chip lottery in some cases as even after pre testing for a certain clock or temperature level there's still some headroom that occurs which is why you had a GTX 1080 that maybe did 50 megahertz over the pretest Oh see that another one might get you 100 megahertz over for example those very top chips we do a separate pass on for our ROG cards like matrix and Poseidon and in talking with them further about this they're basically saying that there was our og cards that matrix Poseidon and otherwise go through a whole lot more validation and so because they go through more validation they get a higher skill they're sold for more and theoretically they at least will be sold at a higher reference or a higher base and boost speed if not actually boast a higher overclock potential there with Pascal as we've seen that's not too exciting so that's how EVGA and ace used to it I asked a whole lot more questions and they're waiting to get on calls with the Taiwan teams at some point this week or next and then they'll get back to me so we might have more of that next question is from Sam the man who loaned us his Titan XP several months ago Sam says how does it rise and chip binning at work I can barely see a difference when overclocked in all three of the r7 lineup so what's different apart from the price tag is this to do with ASIC quality you know that's an interesting question when we did our r7 1700 review which was largely positive especially coming off the 1800 ex review a big reason that review was was seemingly so much more positive than the 1800 ex review is because our r7 1700 effectively invalidated the existence of our 1800 ex and that's because it can clock the same or even higher in some cases but that's just the chip lottery as Asus was saying so the fact that you could achieve these higher clocks if you are an overclocking type did kind of raise some questions and with the 1700 X it was even more questions because it's it's dead in the middle of a weird range where the 1700 can do what the 1700 X can do with all the stuff that's out in the first production pretty reliably there's a good chance you can at least hit 1700 ex stock clocks and therefore achieve 1700 X performance because there's really no other difference of note so what are the differences then well first of all the X the X denotes a wider XFR range for well XF are I guess redundant extended frequency range for the 1700 X and the 18 or X processors as far as binning I don't have a hard answer for you and the thing I'm most curious about is what happens after this first production run so to phrase that question basically if our 1700 today can do 4.0 gigahertz and achieve the same performance as a 1700 X or an 1800 X even when they are overclocked what's going on in the future where these things might get more binning based on demand in the market so a 1700 today the ones provided to us we bought one and we had one provided to us by Andy those today could be bend down chips where they were potentially dropped into a lower skew to fill demand where they could have actually passed as something like a 70 under tax or an 18 ordered X in terms of validation and this kind of blends in with a previous question as well because these are all the same architecture same r7 everything it's just the skew is different the clocks are different so then your separation comes from two things market demand and meeting it and validation in other words did the 1700 that we and the are selling for cheaper fail validation to be an 1800 X chip and if so we we mark it down we saw it at a lower clock so I don't know how that works in the will work in the future I don't know how they're doing it now and I'm not sure if anyone does if you've seen a store it's great please supposed to blow like an official actual source but if a concern there is the 1700 we can recommend pretty easily today but in six months I don't know if those things are actually going to be lower performance overhead for overclock and then they are depending on how they've scaled those initial shipments so the answer unfortunately to your question is I don't know how the rise and bending works there's been some discussion of it online but I don't have anything official for you and so I'm not going to try and guess as for the differences again their ends is between the three if you're overclocking it's not a whole lot other than basically silica mater and apparently how your temperatures are read which will have old thing on that separately so yeah hopefully that answers that basically our current suggestion as of today is if you are buying rise and probably get a 1700 and overclock it and be happy and save $200 or maybe a 1700 X 1800 X doesn't seem worth it unless maybe your IT department won't let you overclock or something like that and wait which case that's exiting core audience for us so next question is from Ethan Lachlan who says will EVGA make an AMD video card no they will not the only reason I'm bringing this back up is because I'm still seeing it in a lot of video comment sections just because you can rearrange the letters EVGA into V eg a test not mean EVGA is making an ante video card in fact we spoke with EVGA CEO not too long ago probably back in February just after CES during the icx launch and he told all the press in the room very proudly that he has never stepped foot into a meeting with AMD that was something he was very proud of so I think we can firmly say that they're not making a vega card next question is a short one Orion agape says where are the am 4 X 300 motherboards we talked about this in our 1800 X review the delay on those mini ITX boards with X 300 and similar chipsets is because they're complicated to make so 4 X 300 specifically in speaking with some manufacturers who make the boards those have an extra layer of complexity literally in that you have to have an extra PCB layer to pull things off at least for this particular vendor we spoke to in a way that was satisfactory again to that particular vendor so that means that there's more lead time to actually end during this thing and that I would guess is why there's a delay next question the last question out of Acts says can you do benchmarks with old overclocked 2600 K 3770k and compare them to your stock sixty seven hundred and seventy seven hundred K and gaming benchmarks to show how feasible it is to overclock your old CPU to compete with the new ones I know my 4.8 gigahertz 2600 K outperforms a stock 700 K and CB Z's benchmark I would like to see how they can compare and real gaming benchmarks stay tuned we already did those tests Patrick's articles written I'm soon turning it into a video and adding some stuff to it so that's going to come up soon the 3770k is not in there but the 2600 K is in there and we did overclock it I think to 4.7 gigahertz so it's pretty close to yours and we have a stock 6700 K so you're in luck exactly what you asked for has already been done we'll probably be online next week sometimes thank you for watching as always patreon.com slash gamers next to the up side directly leave questions in the comments below or if you're already a discord member from patreon leave them in the special ask GN channel thank you for watching I'll see you all next time