Gadgetory

everyone welcome back to another ask Jian we shot three today one of them is on the patreon channel if you go to patreon.com/scishow Nexus is a bonus episode the other ones both will be on this main channel so you can check those out if you have questions for next week post them in the comment section below and I'll try to get to them that we do get quite a lot we have some really good questions this time one of them my favorite one I'm not going to be able to answer directly today but it prompted a really cool interview that we'll have coming up hopefully in the next week or so and that is going about the about Intel's ten nanometer struggle featuring hopefully David Cantor from real-world Tech I'll tell you more about that in a moment before that this video is brought to you by EVGA a 19th anniversary giveaway EVGA is giving away over $80,000 of components and it's retro themed that 19 year anniversary including two signed GN mod mats and also several full systems GPUs motherboards and more to participate click the link in the description below to start EVGA scavenger hunt screenshot competition or gaming events click the link below to learn more alright so before starting quick news on everything news update as always we restocked our blueprint logos if you were interested in them they sold out twice they're finally back hopefully they'll stay for a while now and we're hopefully not that your perspective and we also have the mod mats are in stock and shipping now so if you buy they are actually going out the door as orders come in now they're not backordered anymore final item before we start this posters brand new we really like the design of it it's a video card Anatomy poster so it's just kind of cool if you want something that shows like MOSFETs chokes and all that kind of stuff capacitors on it and explains where they are on the board we have them on the store store die Kara's next to sign up for all of those things alright let's get started properly first question I'm gonna tell you what it is and we're going to come back to it in a few days this question is from uncle curse burp who asked what do you think of Intel struggle at 10 nanometer and AMD moving to 7 nanometer with TSMC Intel CEO had to resign over the 10 nanometer struggle and they may truly be in trouble with the constant 10 nanometer delays in terms of AMD being able to deliver a more efficient product to the semi cost a market for reference here's a link and they link a semi accurate story not saying the story semi accurate it's that's the name of the website so David Cantor if you haven't seen him on our channel before we spoke with him about the sort of misnomer that is the phrase coup de coeur so we talked about what is really a coup de coeur isn't appropriate to call it a corps and we also have talked with him about CPUs talk to them about the SSD market so canter is a leading expert analyst he is a an actual technical analyst not just a market analyst he runs the website real world tech comm doesn't get as many posts but that's because each one is pretty in-depth previously wrote for microprocessor report and runs the real world Tech forums as well so canter is by all accounts an expert and can talk to this subject in a great way for our audience so I asked him do you want to contribute an answer for this episode we got to talking about it and basically decided you know what screw this let's just do a whole video on it because it would have been only two to three minutes whereas we can go fold that for the video so we'll have him on a just a Skype call pretty soon hopefully and quick quote from him when I was speaking on the phone with with Cantor about this idea he said ten nanometer is a little bit of a mystery to everyone what's the back-up plan so I think that's a that's a pretty good hook for the interview we'll do later on subscribe if you want to catch that basically I wish I had an answer this episode but I I really would love it's just dutiful like 15-20 minutes on it rather than three so next real first real question I guess does heat degrade CP is more than pure voltage and this was posted by Stefan Svensson who said I've heard different things from different sources but those heat degrade CP is more than pure voltage and put into the CPU freeze differently can one safely input higher voltages into the CPU as long as the temperature is low or will the high voltage degrade the CPU regardless of low temperatures I reached out to both Roman or Bauer and builds what about this Roman we'll probably be talking with him separately I don't know if it'll be a separate video or not or if I'll just revisit it next week but builds I'd had an answer he was immediately available there Bowers was I was busy with work so build so I'd said and I'll expand on this on my own but he said both are exponential as in the rate of degradation at like 150 degrees Celsius is really really fast compared to 100 degrees Celsius similarly 2 volts might kill a chip instantly when compared to 1.75 so to expand on builds roids answer the reason I reached out to these extreme overclockers exos years is because although I can speak to this from most practical standpoints these guys are bringing CPUs under it's under liquid nitrogen they're pushing high voltages so they've got the really extreme version of the answer for you so both are exponential has builds Lloyd said to expand on his answer though so it does heat the great CP is more than just pure voltage it depends on what part of the cpt are talking on if you're talking about the CPU core that's one thing let's talk first about another component though the IMC or the SOC on if you're talking AMD so the integrated memory controller has its own voltage and on Intel the IMC if whenever you're working with let's say memory overclocking you're having trouble stabilizing a memory Yoseob you might need to push your system agent voltage on Intel or your SOC voltage on AMD same idea over all different names slightly different well actually different applications as well but same idea so if you have to push the SI or the SOC voltages up this is an instance where although your temperatures could be perfectly fine and well well with inspect pushing the voltage on either one of these two specific areas SOC or the IMC can cause serious degradation on the IMC over time so if you start pushing let's say 1.4 volt system agent on Intel on modern Intel like skylake architecture and onward then absolutely voltage will cause degradation on the CPU not the cores but the memory controller so pushing 1.4 si over time to stabilize your memory overclock eventually and this could be a couple months it could be six months probably not longer than a year though almost certainly under a year if you're at one point 4 volts probably within six months but at some point there will be degradation and that degradation manifests in a primary way either well mostly requiring more voltage now to sustain the same clock or you're just not able to ever hit that clock again and you have to drop clocks but keep the same voltage or you drop really hard and you lower the bowl to an appropriate safety level so that is one area where just pure voltage can cause serious degradation without necessarily having the heat component at least not presented to you through core temperatures something else that's hot but adjust the voltage plus the heat on that component where you can't get an IMC temperature and any sauce for that I'm aware of that is the that's one component of your question the other component is yes actually four cores four V cords but different you can push pretty high voltages and typically under most normal user scenarios I'll define a moment typically you will hit a thermal limitation before you hit a deadly voltage typically not always so when I say normal user scenario I mean it's not deleted you didn't put liquid metal on it it's out of the box thermals as you might say and pulling that CPU out of the box put it into a socket and I'm putting a cooler on it that's maybe like 240 millimeters max for a radiator or an air cooler under those conditions it's pretty likely that you are going to hit a thermal throttle scenario before you start hitting really seriously deadly V core and I mean in most cases we've tested with like coffee late with a six core or something you might hit thermal throttle at one point for two volts which is around where you kind of cut off for safety anyway maybe one point four or five which isn't all that much worse but start pushing any deadly amount of voltage is is going to require more exotic cooling or D LEDs and things like that but there's still the other part of your question okay so I'm not running a deadly voltage but my temperature is really high well for the most part especially on Intel and AMD CPUs are a bit different Intel CPUs have currently basically zero response to temperature except for a my throttling or not it's very its binary it's like it's a bit just basically a boolean question of am i a hundred degrees or on some in my 105 degrees if no then let's just keep running the clock as high as it can if yes we'll drop it back until we sustain t.j.maxx or temperature junction max and in those cases of Intel CPUs you can run like 90 degrees pretty much without problems for a very long time the degradation there is it's very difficult to plot we don't have a real answer it would take years and years of testing and dozens of CPUs at that sustained temperature to really know the answer but 90 degrees is not that bad now it's it's not a number people want to see generally yeah even it was just an OCD thing you might want to see like 80 or something like that but 90 is reasonable it's just that I mean it's not it's not a pleasing number it's just it's not killing your CPU literally though and the voltage in theory if you're under like 1.42 is gonna be fine as well for V cores just watch out for SA and SOC and these got a bit of a different temperature scale can't remember it off the top of my head because they have a few different swings so like AMD is TDP suggests it's calculated partially on temperature as you would expect and because they are directly related TDP for Andy is assumes an optimal temperature of about sixty point eight degrees Celsius for the cores and beyond that point you start getting some D rating sort of TDP worsens that's pretty normal and the also their CPUs do have more issues sustaining overclocks once you get into the 80s especially once you're into the 90s so that doesn't mean it's bad doesn't mean that Andy's worse or better than Intel it means they approached differently so this particular aspect is not a better or worse it's a these things work differently but to answer your question well I think I read it but build Joy's answer covers the extreme and with Ellen 2 and extreme high voltages where as he said 2 volts might just straight kill a chip but you're never going to hit 2 volts docking at one point seven five volts heat there it's very unlikely I'll even do 1.5 volts if it's not deleted with liquid metal and you're not eating some extreme cooler like a giant water cooler or chiller or something like that it's for the most part you will probably thermal throttle before you kill something with voltage that's not always true though see the entire rest of this answer for why hopefully that gives you a pretty good outline I feel like that's fairly complete there's dummy more there but hopefully that helps next question and our thanks to builds wide from actually hardcore overclocking by the way for helping with the answer you can check him out actually hardcore overclocking on youtube not to be confused with actually running stock which was his other side channel our mining GP is safe to buy use those from guy in Vera who asked now that mining rigs are being chopped up and sold off are those GPUs safe to buy or should they be avoided but like the plague well assuming it's not one of the ones that was hit by the flood in China let's just do that was it's ignore those first the idea of mining GPUs so I the concern here I would assume is hey this thing's been running a hundred percent load for its whole life does that mean it's degraded in any serious way and the answer kind of depends whenever you're talking used obviously there are million variables we don't know what the owner did I have no idea how well they took care of it whatever so any answer I give you is not going to be a blanket yes or no as far as would I buy a mining GPU used let's see that if the price is significantly reduced and I feel like I have recourse as a buyer then yes I would consider I use mine in GPU however I I've got kind of my own any like personal idiomatic ways of or idiosyncratic ways I should say of dealing with this stuff so my idiosyncrasies for this kind of scenario I just always feel a little uneasy buying used but that's not nicely for a logical reason it's just I mean it kind of is based on logic but I like the idea of knowing that there's a warranty there the component could fail for any reason it might not even be the miners fault it might just have been eventually waiting to fail because there's something bad on it so I don't personally like buying used parts if I can avoid it but there's absolutely totally an argument for why you should do it we buy you stuff for business like tables chairs all that stuff all the time so I totally understand it I don't want to like say that you should and I'm just saying that keep my perspective in mind which is that I feel uneasy doing it myself which means that I feel really uneasy recommending anyone buy use because if something happens I don't want to be responsible for it so with that disclaimer out of the way you take on the risk all that stuff let's talk about the real technical aspect of this technically a mining GPU has been under maybe a hundred percent load for its whole life when a miner is presumably working in optimal scenarios they should have the GPU under power target so typically miners we run 7080 percent power target something like that I don't know exactly but it's it's below 100 percent it's certainly not over clocks like on or twenty percent power because you instantly lose all of the the profit margin there on power so they're running below power target in theory the fans are also running kind of fast so hopefully if there is a dead component soon it's going to be a fan not like a cap or something so in theory the fans are running fast power targets low vrm shouldn't be too abused the memory vrm might be a bit abused if they overclock the memory and memory via arms weaker that's certainly true in a lot of cars the core vrm gets way more focused so it's possible that the memory vrm gotta be used as possible that it ran a bit hot but MOSFETs can take like a hundred twenty-five degrees Celsius before there's any real Assyria's problems and doctors same thing so then what components are we concerned about well MOSFETs can pop even though they can run high temperatures but that might just be a bad batch so it's really hard to distinct to draw a line distinguishing between mining abuse and what is going to fail anyway capacitors are where your real concern is especially anything that's kind of located in a hot spot on the card where there might not be air flow because the cooler designs bad so this comes down to a lot of assumptions like we're assuming that the cooler is pretty good to begin with we're assuming that everything's getting decent air flow we're assuming the miner was running it at maybe below the power target and thus below the temperatures this is what you'll hear a lot from miners it's well it was it was below the load that's on gaming it would have it and it's not thermal cycling so it doesn't have the abuse of thermal expansion and contraction constantly through its life like it would with game and it was just a hundred percent always-on and there's absolutely some truth to that but it's it's also not quite that simple so if you have a capacitor on the board that's not getting good airflow and it was in Iraq in China where there's not like necessarily it's higher temperature just ambient temperatures higher maybe they had a whole bunch of GPUs in a line whatever if it's in kind of poor conditions it's possible that a capacitor could be running at 105 degrees Celsius it's not too common capacitors time to get these and cooling from the fans but if that's the case if it's a bad pool or whatever then absolutely you could buy a car that has a cap nearing end of life 105 Celsius maybe some caps are gonna be 5,000 hours something like that so if you're running 24/7 105 C it's gonna die inside of like a year but 105 sees a whole lot and it's not that common that caps get that hot we typically see them in the 60s to 80s in like a hundred percent load 100% power target gaming scenario or like a prime crunching type or a power buyer scenario so the answer to your question is this is really hard to answer and there are a lot of things to consider and we're making a lot of assumptions one of them is that the cooler doesn't suck if it's a card that has a bad cooler and it was being used 100 percent of the time even with reduced power consumption from miners who set up properly it could still be a risky buy because if the coolers bad it doesn't matter what the miners did at some point you've got stuff that's getting hot even though they're not pulling as much power as normally it's still in a higher temperature than then your average room ambient temperature in most serious cases so I don't know I would say yes it is worth considering buying a used mining GPU there's a good chance that it was under less strain that a lot of used gaming GPUs in some instances but in those instances where it was under more strain than the average gaming GPU it was a hell of a lot more strain so you're more likely to see a failure it's just by somewhere where you've got some buyer protection I know there are going to be people who take issue with this answer because they're gonna be people who are miners who feel or people who bought use mining GPUs and say my anecdotal evidence suggests it's completely ok and I agree with you it's definitely completely ok if you get lucky but it's also easy to get screwed over and you just don't know what you're buying it doesn't have like a CARFAX report or something like that so until we know we have a way to know what you're buying I feel uneasy suggesting it but I would say that there's a good chance you're okay it's just when it's not okay it will be a catastrophic failure and that's how GPUs failed by the way it's not like performance degradation it's not going to be you buy a 1080 as performing like a 1070 it's gonna be you buy a 1080 and either works great and you never notice that was use remaining or it doesn't work that's probably gonna be the two modes you get GBS don't really degrade and performance like in gaming they just have they work fine until they don't and that's because when they don't it was a catastrophic failure capacitor died MOSFET blew up something like that but anyway in the very least PI just replaced their own place so hopefully I covered all my basis there and addressed most of the definitely going to be posted criticism of the answer I'm sorry if you don't agree I did my best to cover all possible angles of it so hopefully I got the angle that you cared about but anyway that's that's you asked my opinion that's how I feel about it it's not a hard fact necessarily that you should or shouldn't buy one next one prime95 vs. ADA so Devin Dykstra says when I was overclocking my CPU I noticed an odd phenomenon when I overclocked it to high 8 of 64 and prime both crashed my system but when I ran stress tests at the same time it was stable I would think that running both would put even more stress on the CPU so what's going on here they actually won't you should only use one real stress testing program if it's going to pull 100% of resources for the component of testing so you should use either only prime or only ADA I don't particularly like ADA's stress testing tests so I would suggest most other things prime95 smaller 50s I would suggest blender is a good one too that's a realistic one if you don't like prime because it's not super realistic blender uses a VX it's still really abusive it's not quite as abusive as small F of T's and prime so using both you're gonna run into an issue with thread scheduling where Windows is bouncing between the two applications and windows basically sees two applications calling for a hundred percent resources and Windows is sitting there freaking out trying to figure out how I can keep both of these applications running simultaneously and so the solution is to do thread scheduling and the switch resources back and forth between them so you're toggling stuff on and off you're putting stuff in memory you're pulling stuff out of memory and all of that means that you have a lot of CPU downtime while Windows is trying to keep both applications running rather than just saying you know what I can't handle this I'm gonna give all my attention to prime in which case you would have the first scenario you described so what it's doing instead is it's trying to keep both running you're getting the prioritization and the threads switching between the applications memory is being used and dumped and that means that you're going to end up with a sort of like false sense of stability so just use one application next one is from a fourth tuna who says can GN possibly review gaming mice it's known in cyber sport enthusiast community that even among top tier game you know my sensors button switches do differ in effect performance players collect data based on reviews and personal experience there's a lack of expertise about sensors their technology and parameters we've done a couple of videos with people like Chris pate from Logitech who's a great engineer NPM about sensors the trouble with reviewing mice we used to do a lot of peripheral reviews but we got out of it and it's because they're so damn subjective it's just it's really hard as someone who these days tries to lean on data and data collection it's hard to review something subjective and feel confident in my conclusion because ultimately the thing with subjectivity is that people won't agree with you and that's fine but it's different when it's something like that personal like a mouse where like a game review if you read game reviews and you think I don't agree with that critic at all that's fine you don't have to like call them names for it it's just that in the future you know that you shouldn't be reading that critic because they might not like the same kinds of games as you do so it's important for subjective reviews to find someone who has the same style as you and typically agrees with you as opposed to what most the internet thinks which is that you need to be constantly at odds and adversarial with everybody so my street views fall under that category it's really hard to do right there are ways to do technical testing you can test the latency you can test input response time we've done it but we worked with Logitech to do it and we don't have a permanent set up for that so I've thought about it I would love to review mice and a technical fashion that's like repeatable way to collect data but it's just not something we can do right now like I mean what are you gonna do test button endurance and get a pneumatic arm to sit there and push the button 50 million times and then you have one sample so it's really hard to do and we got out of it but I'd love to look at my skin it's just I don't have any present plans that said if you have ideas I'm not a mouse enthusiast anymore I don't have the time to be anywhere so if you kind of spend a lot of times on forum time on forums and you look at these user reviews of it let me know what kind of data the people collect is it objective if it's objective data that I can maybe collect myself let me know and we can look into it for next time if it's subjective let me know what that is - but just I'm probably not going to test it if it is to more good full tower case is George Lambros says host IVA hope you're doing well I am thank you very much I know you mostly cover mid tower cases but do you plan on reviewing some full tower cases as well I made the mistake six years ago to buy the 800 D and now educated me I ripped the drive cage apart for to 140 fans in the front glued Silverstone air filters on the front that's cool now I have much better temperatures but that means all my seven drives are in the air held by cables or sitting on the bottom of the case thing it's time to upgrade so which one would you recommend except the half while the half X is kind of old anyway so I don't really feel great about recommending it anymore but okay so I was reading the rest of this a few things here I also have drives on the bottom of my case and I shouldn't but I do you you've got a couple options if you want to keep the full Tower approach which I completely get I actually personally prefer full towers and use them myself then there are a few options the dark base pro 900 s out there as you mentioned that's one of the better ones we've looked at lately the view 71 is a fairly big case I guess it's not technically a full tower and I don't remember how many drive cages it has sorry but look into the view 70 wanted performed way better than we thought it would and it is full tempered glass case but it has enough gaps that actually perform pretty well so that's also a big-ish dark base pro 900 for sure the biggest we've looked at lately see 700 p we actually decently liked the cosmos see 700 P I know a couple of other reviewers really didn't like it and it is kind of clumsy to work with if you want to invert it I would just suggest not doing that but it's a large case does have some not great air flow but you can you can work around it with that one that's the upside those are really that's that's what I got for you case labs if you have a ton of money like 500 bucks you can buy a case labs case and never buy another case ever again but they are expensive up front and hopefully get you set up for a long time they're also better for water cooling so that's kind of that's what I got in mind we look at a lot of cases positive I'm forgetting full tower cases I know I am so for those of you who keep a good eye on full tower cases please post them below so that George can take a look through your suggestions but George those are mine it's basically looking at case labs just so you know what's out there but don't go in planning the buy one because they're expensive look at the dark Bass Pro 900 if you don't like it that's fine but I actually did quite liked that case personally the c7 harpy is okay but it's got some problems and I'll think about others but if you 71 maybe that's what I got for you right now I think alternatively you go mid tower and get like a small box or build another small system and do an ass and just do network attached storage and then you could put it on a gigabit or 10 gigabit line 10 gigabits probably bit expensive so maybe gigabit locate it in your room with the router and you've got network attached storage and now you can have a smaller case that's an alternative for you to think about there are NASA's that are sold for relatively cheap that are fairly all-in-one most of them won't have the failures we had recently last one goes to pilot how often are you able to make it to a bike track do you bring your own bike or rent do you have a favorite and is there one you are looking forward to visiting so Whistler is probably one of my favorites I got to go so when this there's a season for downhill season and it's typically like May maybe June through like October and at that point all the cherry lips closed and it becomes winter parks except for a couple of parks that are actually some not far from us that are year-round so I try to bring my own whenever I can if it's on the East Coast and I can access it by car I definitely do that I ride local jump lines I don't know if we'll put a picture in but I have photos of like giant jump tracks we have locally and stuff like that so those are great but they don't have trailers so push the bike up write it down to the jumps or local mountain lines so I get out pretty regularly for just like trail riding locally and just and jump tracks locally a couple times a week I try to do for like an hour or so each time hour or two just to keep saying it really helps me stay focused on the site if I go out there and just do a bunch of jumps for like an hour and just do nothing but session jumps it clears my head completely and I can think about what I want to accomplish for the day for the site so as often as I can downhill riding requires driving or flying but I do get out a good amount of weekends through the summer and I bring my I bring two laptops I'll edit a video or schedule them for posting and write articles while I'm there so I I still worth like four to eight hours a day but that's a lot less than normally so it's it's like a mini break but not that much of one and buying or renting I try to bring my own bike if I can if it's in like Whistler like where I did the recent video then I rent there because it's just it's too expensive and there's damage waiver so I don't have to worry about damage because not my bike and there's damage waiver so yeah favorite probably Whistler in Canada I've ridden in Taipei after Computex I've gone twice now and that's pretty fun I like in Taoyuan I like mammoths in California snowshoe Bryce Bailey I don't like that much but yeah those are my those are my top picks so anyway that's it for this one as always subscribe for more go to patreon.com/scishow cameras exits get the bonus episode and go to store it on cameras nexus net the mod mats are in stock and shipping now the these shirts are in stock keychains are finally back in stock crystals are finally back in stock you guys like bought everything thank you very much we finally restocked them all really happy to see that people love the stuff and bar buying it and then the posters are brand new and in stock as well so I thank you for the continued continued support we sincerely appreciate it thanks for watching I'll see you all next time