Gadgetory

everyone we're back another episode of ask GN coming out of the AMD risin event from earlier this week and we've got another one coming up so a lot of travel in about a six day period and we've got benchmarking of course for rising going on in between check back for the review once that's live don't jump up there so the risin question that I've seen a few places is about PCIe lanes people keep asking how many PCIe gen3 does lanes does I have for the GPU we've answered this several times now but I guess just to do it one more time and maybe we'll put this in the title this time so it's more it's easier for people to find we've got a block diagram that we made earlier this month or actually last one that just after CES that diagram shows everything put that on the screen I suppose so you can see in the diagram again that we made its sixteen a PCIe at 3.0 lanes off of the CPU going to the D GPU and then those can be MUX t' which is not special there's just some boards have markers on them or pax chips through px and we've seen that with amasai boards at CES and a couple of the other am4 motherboards but it's 16 lanes out of the CPU for the GPU then there are another 4 PCIe lanes going to the chipset that's X 370 P 350 etc and then that splits into your PCIe gen 2 general-purpose or a USB 3 lanes and there are also PCIe gen3 lanes that act as i/o and those go into other devices which again we've got in this block diagram you can see the three choices it's a pic 1 scenario so a B or C in pic those that is the PCIe lanes set up on Rison it's nothing crazy as far as strictly GPU again 16 lanes out of the CPU itself which is sort of an SOC at this point and then the rest is just the usual IO Lane assignments if you want to run em about 2 devices via PCIe rather than SATA so that hopefully answers the question it's very simple if you wanted to run one x16 card you could if you wanted to run 2 by 8 you could or you could start muxing and go down as as desired depending on the motherboard and the support for the board next question is from Sir Papa who says question for next Sen our delta temperature is affected by room temperature more than just subtraction for example the air is 20 degrees on the cpu's at 50 the deltas at 30 as usual but if the air is 30 with the CPU then run at 60 or would it be hotter this is a really good question this is basically asking is temperature when you're using delta T going to be a linear scale and the answer is it depends now we use delta T for a few reasons one of them is because the fluctuation is normally so small that a nonlinear scaling is kind of irrelevant it tends to be within a degree or so with a lot of our tests if that depending on kind of which room they were running and what coolers were used in things like that but as for the answering the question I actually I ended up reaching out to VSG from thermal bench and VSG I've worked with for a while now on kind of validating externally some of our thoughts and our test methods he runs thermal bench he as I understand it is some form of post doctorate researcher does a ton of work with thermal equipment that plenty of folks don't have access to so the guy is definitely credentialed to answer so I asked and the answer was more about read-through here if the thermal resistivity of the die or package is a linear function also then sure it would be linear with older CPUs and he says the even up to 22 nanometers it was mostly fine meaning mostly a linear scaling between ambient and the IHS or whatever might be measured VSG then said I did a test of my 4770k with no heat load in my hot box and just buried the hot box ambient temperature from 25 to 80 Celsius in increments of 2 Celsius at a time he then it mapped the average CPU core temperature and the increase in the average core temperature was nearly the same as the increase in the ambient temperature after steady-state was reached which admittedly took a few hours at times and then VSG also says I had a thermistor in contact with the IHS and the increase in IHS was not the same even with study state accounted for the slope of increase was higher for the IHS than the CPU cores but the IHS slope of increase was but the IHS varied between 1.4 Celsius increase to 2.2 or 2.4 Celsius increase due to the lower temperature increase of previous cycles so he's saying that the results was slightly off one point four to two point something Celsius between ambient which was controlled in a hot box and the IHS and says that was a result of the thermal resistivity of the IHS alloy having a quote hyperbolic function with temperature so it started off slow and then picked up as such once it picked up with heat transfer to the CPU cores also changed so it's basically for the particular chip you test in the 4770k and potentially onward it was a nonlinear function increase from the ambient temperature of the hot box and the IHS temperature or the core temperature as it may be so they forty on one did not equal 40 on the other they were offset by one point four to two point four cell Z's in this particular example he later told me that with the more densely packed 40 nanometer chips this might be exacerbated so that's something to certainly keep in keep in mind and we have a lot of challenges to look into with rise and thermal benchmarking and previously FX had some similar challenges so this is something we're researching for Rison most of the benchmarks I've done with CPU temperature thus far meaning non cooler temperature but the CPU reviews those thermal benchmarks I tend to do within generation or within chip so just 7700 K benchmarked at different voltages things like that so then you're looking at just a delta one bench to the next rather than looking across architecture because for a lot of reasons that's not always easy especially if you're relying on software measurements for the temperatures because again as previously Andy at least with FX and we'll talk about rise and shortly does not use the same temperature scale as Intel so they're not linearly comparable but back to the ambient stuff VSU kind of concluded by saying what started off as a nice trend of CPU core temps increasing linearly with ambient temps suddenly went all over the place I imagined tighter packed eyes and with end 14 editor process on Andy and Intel will be worse in this regard so basically he was seeing something like a 40 Celsius for ambient and an IHS temperature of maybe thirty five point four or forty two thousand ambient and 37.7 IHS something like that so no it's not perfectly linear we had a pretty long conversation about it but there's plenty more but that's the basics check out thermal bench for more of that type of stuff next question is from ramos studios who says hey Steve I was wondering if I were to remove the fan from an air cooler such as the 212 Evo or dark rock pro 3 could the cooler still provides a sufficient cooling to the chip relying just on the airflow generated by the case fans ask because some of the newer air coolers are big and chunky and look like they can dissipate a lot of heat but the fans are loud or ruin the aesthetic of the cooler the 212 Evo I absolutely would not remove the fan from that is not an impressive cooler by any stretch it's okay and there's a reason it's like 30 bucks or less so now I would not remove the fan from the 212 Evo depending on your CPU of course that is worse or better because of the lower TDP chips it's not going to be such a problem when you're running something like maybe 40 watt chip under the the cooler but I still wouldn't take the fan off because you could always just run it out of lower rpm it doesn't have to be allowed that doesn't fix your aesthetics problem but you could run a lower rpm and at least just get some airflow through there because the the fins will sink the heat and they'll sit there waiting to be dissipated so if it's not dissipating the heat you kind of run out of room to to soak the heat from the CPU there are passive coolers though and they are massive Zalman made one years ago is it some Zalman cube cooler it looked like a Borg cube and that I believe had no fan it was a passive cooler and I don't know if it was even anywhere remotely within spec for the socket because it was a heavy thing so possible yes but you generally don't use them with overclocking ksq type CPUs don't use them with higher TDP CPUs and they need to be huge if they're gonna be passive so I would not recommend taking the fan off and just rely on the case airflow because it's not gonna be as directed and the air pressure by the time that air gets to the CPU cooler the air pressure is not gonna be that high so you it's off it might as well be passive really next question Laura's passage says hey Steve got a question while we all eagerly await the launch of bikers Nexus I've heard a surprising number of people say that there are inherent compatibility issues when mixing and matching AMD Intel and NVIDIA components when building a PC for example some say that you should only go with an AMD GPU if you're going with an AMD CPU as well are there compatibility problems between and the GPUs and Intel CPUs do NVIDIA GPUs have compatibility issues with Andy CPUs it goes on to ask more about that is it safe to mix and match components I know everyone pretty much knows the answer to this I think on our channel but I wanted to point out just as a more basic question no it doesn't really matter there's not gonna be performance compatible there should be compatibility issues between those platforms listed what would matter is if you are trying to go with something like the dual GPU setup that AMD can do between some of their ap use and some of their lower end graphics cards that would require of course and I am the APU and an AMD GPU to work you can't use that otherwise and that's because there's some memory unification there and whenever you do any kind of unifying of memory you want direct access to the memory controller on whatever CPU or APU it is and if you're going across architecture it's not it's there's normally some latency because you can't get direct access among other things but that would be the only instance where it really matters there are other things in sort of enterprise environments where and videos got there special interface for their high end tesla class accelerators but that's not something will ever encounter when we're using PCIe so now does not really there's no no inherent compatibility Intel and AMD can't work together if you wanted to name the GPU on an Intel motherboard or something next question Justin vishton says any chance you guys will test the heat pipes on motherboard v arms actually how cool anything or are they just there for show yeah I'm interested in it so to sort of reiterate what I've said in the past a lot of these better VRMs that have enough surface area I mean enough phases are just good enough components in the vrm don't necessarily need that heatsink to begin with because VRMs again can sustain 125 150 Celsius something like that so a lot of the time the heatsink is for show it's hopefully still helped sink the heat if it's got fins and contact and it's aluminum or copper or something you're still soaking heat and hopefully still spreading it anticipating it and heat spread is always a good thing well for the most part but that doesn't mean that they need to be cooled by passive heat sinks it just means that it's not a bad idea now you could take them off in a lot of instances if you wanted to again so these overclockers like I was talking about with build Zoids comment previously some of these overclockers just ripped the heat sinks off anyway and they don't need them on there but as for the heat pipes and things like that you'd assume for the most part that they work unless it's there are a couple examples where on motherboards I've seen bad heat sink designs like the MSI gain Pro carbon which big surprise the vrm heatsink there is just a giant block of aluminum and solid all the way through and there's no fins so it just like if you put a a 2 inch cubed block of copper or aluminum on a CPU it would be a lot worse in that state cooling or a sinking heat then if you finned it so if you dice it into fins and you remove some of the mass so it's it's actually less copper or aluminum ash but it has more surface area because you've diced it and there that will cool a whole lot better than just a solid block of aluminum on top of VRM a bunch of MOSFETs so there are definitely bad designs but it should still sync keep better than nothing and you still probably don't need it overall but no I haven't tested any specifically it is something we have the capability and the tools to do and I'm interested in it but we've got plenty to deal with them to shield and all the other thermal stuff with Rison and everything else right now next question Josh Orenburg has been a supporter for a long time asks question for the next stn what's the difference between the advertised endurance as in the warranty and raw NAND endurance for an SSD should the SSD last significantly longer as indicated by Ron and endurance or should it only last as much as indicated in the warranty my samsung eep 850 Evo is seventy five terabytes on the warranty where it's 500 Ron and endurance for this one this is one I'm only bringing up not because I know the best answer off the top of my head but because I know the best place to get the answer and it's it's I would prefer to point use a it's a tech report for this one so tech report back when Scott was still there running the show and when his I forget the guys name one of his staffers at the time I think the guy has since left was doing a bunch of SSD nan tests and so they set up test benches with SSDs running non-stop for a really long time I think it was at least a year or at least approaching a year so they had several disks on there I think they had Intel they had Samsung absolutely this one may have been included and they just burned them tons of reads and writes so they're going through the p/e cycles non-stop for this whole period to see which SSDs died when do they die and who hangs on the longest almost like a survivor reality show of SSDs as that was actually a really cool feature to follow they did several updates on it throughout the test period and eventually wrote the conclusion of course which you could check out but I would recommend going through the whole series because I actually kind of fun to read and see which of these SSDs dies off and how to say degrade and performance how do they start showing flaws and faults when do the testers think that it's kind of nearing an end of life for a particular SSD I think they did a couple petabytes of write and read read and write cycles on one of them but I should say write any race cycles programming race so that would be where I'd go for that question I I don't have the best answer but they have a really cool content piece that I wanted to highlight because there's there's no point trying to compete with that it's probably the best one out there for endurance testing that was done by a third-party media outlet next question bad dragon 5 says hey Steve I've always wondered how a rain is rendered in modern games like GTA 5 or battlefield 1 is it just green space like outside of the players viewpoint it's only raining in your guy's face or is it world space like it's raining in the world if the former has any game ever tried the latter and what kind of impact on the game might we see between the two if you do something like world space it would be a bigger impact this is something that we want to look into for a separate series actually so thank you for the idea we'll be working on rain and weather effects and games coming up but the short of it is that no it's not gonna be world space it's probably rendered kind of in the frustum that your your whatever your viewpoint is but I don't have off top my head examples of games where they've tried to do world space specifically but we will look around for the upcoming series so subscribe for that as always or the last two questions here ml North says you're passing comment about radiators being mounted with tubes at the bottom made my one week old H 100 IV to pump absolutely silent whereas before on performance setting I could hear noise why is this information not more widely known or accepted I don't know if you look around at a lot of YouTube videos you'll see that very frequently the youtubers mount with the tubes at the top maybe doing a short build whatever I guess doesn't matter a whole lot but for best silence and performance and endurance mostly silence though yeah mount mount tubes at the bottom I'm glad it helped you hopefully some other folks will pick that up as well does matter a little bit based on manufacturer but in general that's kind of the rule tubes on the bottom because you want the pressure to be forcing the water down not with a big air gap above the tank last question greasy pommel says steve is the wooden wall behind you real or is it a green screen it is real we built that wall maybe over a year ago now maybe two years ago might be two years old at this point built it out of adhesive aged wood so this is wood is real and very expensive unfortunately it's like sticker wood so we stuck it on and built a folding wooden panel box so this is actually two halves I think the split the splits may be right here or it's right here actually so if we were to pull this off it just mounts onto the wall there's an angled piece of wood at the top that this sockets into so the wall itself is not screwed into the wall its screwed into sort of a mounting mechanism and then that's got an angle in it so if a cat claws on the thing that's not gonna fall down and then the cool thing is we can pull it off the wall and fold it so it can fit through the door frames and be moved if we move studios or something so now it's actually real we have a video on making it you can search the channel I think for video set or GN set another two videos on it and it's progressed a lot since then so thanks for all the questions as always if you have more for next week post them in the comments below and as kind of an aside a side note I was on the awesome hardware show with Paul and tile a couple days ago they had some audio issues at the beginning let's get the first three minutes but after that it was pretty good we had a good time that was at the am the event which I can say now because obviously we were all there for the Rison thing and we'll have more news on rising shortly so subscribe as always comments below for the questions for next week patreon.com slash gamers nexus if you'd like to help us out directly thanks for watching I'll see you all next time