Gadgetory

everyone welcome back to another ask Jia and this is our third and final one we're shooting for this week's series we have one on the patreon channel if you go to patreon.com/scishow section below for next time and for this one we have a couple of questions about very specific thermal questions actually on SSDs or opt in specifically have questions about liquid metal testing and performance between cool labs conductor na and others and then we've got a couple of others in here as well like a question about the evolve ATX before that this video is brought to you by Lian Li is at LAN Cole one case the land coal one uses a mix of aluminum and steel to create a unique front bezel design that still manages to allow air through at the artistically cut panel fan mounting options are everywhere in the case and ease of installation features that make the case trivial for building a system the land coal one is competitively priced while still being constructed with high-quality materials and a sturdy chassis learn more at the link in the description below it's just before getting started as always our store has been recently pretty fully restocked the full logo shirts are on the store now to store documents Nexus dotnet for those and we also have the mod mats available on backorder keychains are still there if you're interested in the GM keychains they're a pretty affordable item and then we have our video card Anatomy posters it's actually one of my favorite things we have right now just because it's a really cool and useful design so if you're interested in that stuff it's on stored I game is next to Scott and that the glasses will be coming in again the next couple weeks first question cazador de Wolfe says why isn't this color corrected I don't know what you're talking about next question metal shark says what is the ideal temperature range for obtained memory chips appreciate and likes it warm 50 to 70 degrees Celsius question mark and the controllers like it cool but cannot find much on obtain reportedly the only difference between the Intel 900p and 480 gig or 480 gigabyte version and the 905 P for 80 gigabyte beyond lights was a firmware upgrade will this firmware upgrade made available to 900p owners if not does the 905 PL ow firmware upgrades so that a theoretical 9/10 P offering more performance due to firmware alone doesn't obsolete the 905 P and our verb so this is out of my area of expertise but I was able to approach Alan mal Ventana from PC perspective who is in my opinion the leading SSD reviewer in the space and asked him because we don't really follow opt in so Alan's an expert I asked him got his answer and I'm gonna read that to you but you can go to PC Pro com2 check out their review he's also got an editorial on how 3d crosspoint chain or a phase change memory works if you're interested in more of that so starting off with his answer here's what he said flash sales are written more easily at higher temperatures because the electrons can more easily pass through the insulating material resulting in less wear 3d crosspoint works in a completely different manner passing current through a very small 20 nanometer phase change material quickly heating it to 300 degrees Celsius to 650 degrees Celsius the temperature changes are so rapid and on such a microscopic scale that the temperature of the rest of the dye is mostly irrelevant we can then fall back to the general solid state component guidance that cooler temperatures result in lower failure rates it's also worth noting that opt in SSD operating temperatures are rated for 0 to 70 degrees Celsius while NAND based products from Intel are rated at 0 to 55 degrees Celsius so it appears that opt in parts care less about temperature within reason limiting the NAND parts to a lower maximum is likely to meet the end-of-life data retention spec of three months per jeddak enterprise specs higher temperatures equal greater leakage and voltage shift of sort data generally speaking electronic failures are complicated with multiple mechanisms at play you have some parts that fall closer to the 10 degrees Celsius equals 1/2 rule every 10 C increased doubles the failure rate and you have other parts like flash memory cells that do the opposite since the right mechanic involves injecting electrons through an insulator where you would want to be at those higher temps to help reduce breakdown of the insulator data stored in 3d crosspoint Sall should be relatively immune to temperature until you approach the temperature is required to erase the stored data greater than 300 degrees Celsius I wrote an article covering how all that works lots of eye candy for b-roll as well and you can check that article out will link it below but it's on PC / comm and it's one of Alan's editorials and hopefully we can show some of the butyl he's talking about while I'm talking about that otherwise check out the article Thank You Alan very much for contributing by the way next question is from Daniel Imboden who says two questions do you compare have you compared different liquid metals is it normal for soldered CPUs to lose thermal conductivity over time and what is the problem I deleted an old e 8500 and and applied liquid metal which got much better OC results but why so for the first question have we compared them yes we have we've compared thermal courtesy conductor not with two of the cool lab solutions I think it's ultra and pro are the two most prevalent so we compared those this is a while ago it is not something that we've published but it's something that we did just for internal testing and internal purposes so tested - cool lab solutions conduct a nought and there was one other one I can't remember the name of but we tested one other liquid metal as well and the three brands and four total solutions were all within one ish degree Celsius I think the maximum difference the max delta was about two degrees and here's the problem our tolerance for error is minimally one degree so we could not detect any meaningful difference and for the most part in terms of thermals they looked about the same now the theta with liquid metal with any paste or compound or anything is that the aging characteristics are really important and hard to account for in this type of testing so what we don't know what we did not test is how do the other solutions so conduct a naught versus cool Labs versus that third brand I can't recall how do they aged over a year of use what is the percentage of gallium in each one for example so we have a video that should have gone live by now about actually what happens to materials when you use different compounds on them so we've got a video on that but what we don't know is for each mixture they're all some form of gallon stand for the most part so it's going to be gallium indium and tin but how much gallium is there in each one typically it's about two-thirds the exact percentage fluctuates between brands so there could be some impacts performance there if you have more gallium like the two-thirds number it'll be a bit more friendly with friendlier with the copper yhs --is that are not nickel plated and migration is less of a concern electron migratory on migration which can cause this sort of plating effect but you don't get much corrosion much pitting it's kind of service-level pitting doesn't really impact performance but the point is the different mixtures of gallons down the different percentage of gallium indium and tin will impact the longevity how frequently you need to replace stuff if it caused any damage but for the most part all of these are pretty close they're really not that different just like most other things in the hardware industry so I have not done longevity testing on all of them just on conducting out for now but to answer your question the thermal difference just tendeth applied and then tested within a day is effectively zero there was effectively no difference now does that change over time I can't answer that right now it's not something we've tested but I can give you the that is there any difference immediately and the answer is no and as for your other question removing solder and replacing it with liquid metal should not really improve stuff for you so I don't know enough about the e 8500 I didn't work with it I don't I don't know anything about its interface it's thrown liner face material so I would think that no that shouldn't happen but I don't know why that did with your case maybe maybe I had micro fractures or something next one ZX z AK says as a newcomer to your channel though I've known of you in the tech space for years how would you sell your channel to a new viewer I'm subscribed and looking forward to future content but how much of your library of content is worth of you I don't mean to sound elitist or pretentious but rather that with the way tech changes and evolves I wouldn't necessarily need to watch benchmark videos for all the cards I definitely get what you're saying and I completely agree with you it's something that we we it's kind of interesting so a bit of a side note but running the site for ten years what I've learned is that I kind of figured out like figured out a lot of those problems early on like the one you're talking about and then you sort of forget that you figured it out and you rediscover it again when a viewer brings it up like you just did so to answer your question there is definitely still content that's worth viewing and I'm not gonna pitch it to you because it's not a marketing person but I'll give you a few examples factory tours if you search the channel for factory tour they should pretty much all come up those are interesting they don't really age because the same idea it's I mean maybe they're making older stuff but the tools don't change so factory tours would be very interesting pieces that you might like feature pieces are good that can teach you a lot about hardware so in my mind when when we when we brand something as a feature piece it is basically a very targeted test so rather than a full review a feature piece might be when we did Vega 56 and we brought the power limit up to 240 percent or whatever insane number it was so that was a fun one when we did Titan V testing we did Titan V scalability with two to Titan V's versus one and the one application that supported it we did Titan the clock scalability we've done Vega 56 versus 64 clock scalability we've done BIOS flashing so anything that's kind of a feature where the data in a year from now when no one's buying the card that we worked on in that piece the data is not applicable to your purchasing decisions anymore but the interaction of how Hardware behaves is applicable and will still matter so that's another what was another good one we did we did a lot of stuff with frontier Edition so here's a good one we did Vega frontier edition clock matched with PG with the fury X so clock for clock what's the performance Delta because otherwise the they're not too different on the surface of things and so that's really interesting one because you can see architecture scalability and it changes over time as drivers mature but the the underlying idea is we can look at cards in this instance by clock matching them and start to better understand how much of this performance improvement is from architecture versus how much of it is just from driving the clocks up with ton and so that's really interesting content the reviews I agree with you you don't you don't need to go back and watch our ZOTAC 1080 TI review if you're not buying it I mean like the thermal sections may be a little interesting but is it teaching you something that will be helpful maybe for the most part though you just kind of wait for the one that is relevant to the thing you're buying and watch that instead so I agree with you those aren't necessarily worth watching but the factory tourists feature pieces we had a series we work on TL DR that we've kind of done away with the name of but those videos are interesting they talk about how stuff works so we talked about like how air coolers work and we have custom 3d animations in them that's really cool how NAND works is a good video we did one on how liquid coolers work that's a really good video with custom animations that we made in-house lots of stuff like that and I there's more - but but that's kind of that's that's what I've got in my head right now but yeah hardware news and reviews probably out the most interesting our disappointment build was really fun though and then like the sort of seminal reviews like the H 500 P but it's like a classic so you got to watch those definitely watch the classics but yeah I don't know I like the question I'll think more about how we can maybe categorize our content to split aside stuff that's like hey I'm new here but I don't want to dig through a bunch of reviews and news that's a year old what can i watch so well we'll try and figure something that maybe did playlists the feature pieces or something next one Maj so sbk says hi can you please explain how is really how related how our SOC voltage and ram speed and voltage related and rephrase your question of it so mister sbk let's get into this one you can think src voltage and si or system agent voltage to memory as CPU voltage is to clocks so it's it's kind of a microcosm of what you're doing on the CPU side with V core and a CPU frequency same idea we're with it driving a higher frequency on the memory let's just assume there there are a lot of factors at play here so let me outline those to drive a high frequency you need a good die on the memory so that Samsung or high next or whatever you a good die in the memory that's number one so you kind of been in a bit there number two you need a good IMC or integrated memory controller on the CPU that's really important you can have memory that's rated for forty six hundred megahertz and we do but if you're IMC just can't go above 3600 there's nothing you can do and then you also need a board that can support it and a couple of other things but those are the two major ones it's the I am see quality and it's the die of the silicon quality on the memory and then you also do have a concern of the CPU core itself so now you're you're trying to get a CPU that has both a strong core because you obviously want to overclock the core pretty high if you're already into memory overclocking and it has a strong IMC so it's asking a whole lot once you get into memory as for SOC voltage V SOC and VSA a VCCS a system agent voltage on Intel what you're looking at there is a way to help stabilize when you have memory failures so if you're starting to overclock your memory let's say you go to maybe 4,000 Duggers it's pretty good one so your system might be able to sustain 3600 with no changes and you got kind of a have kind of a black hole in between there at 4000 megahertz you might need some more voltage and pushing DRAM voltage is fine that'll certainly get you a lot of the way there it's a relatively safe thing to do up to an extent anyway but at some point the imc the integrated memory controller on the cpu so it's it's within the die of the cpu on the same substrate and the same guy that IMC is going to have trouble keeping up and will need assistance via voltage to do so so you're pushing voltage for the memory to the memory clocks up and then at some point if that's not good enough anymore you have to push voltage to the system agent or SOC so that the CPU can drive its IMC in a way that will support the memory overclock so hopefully that helps quick note here be really careful when messing with sa IO and SOC voltages they are the easiest way to degrade IMC quality if you aren't careful if you don't know what you're doing and I'm not gonna go through all the numbers for all the different architectures but for modern architectures like let's say desktop parts you definitely definitely want to be below 1 point 4 volts on Intel and AMD for SI or SOC for SOC you want to be closer to 1.2 on anthe processors I think officially rise in 2000 I think officially it is 1.3 as a safety now but I'm not positive I haven't checked in a while when Rison one originally launched it was recommended to stay at 1.2 volts for V SOC and I think it's 1.3 now but double checked me because I don't keep it on my head but it's in our content out there somewhere and then si you're looking at closer like 1.2 1.3 and if you really don't care about the quality and you're just benchmarking for a short time and you need it you new 1.4 we've done it but if you run it that way for long enough you will degrade the IMC quality and lose the ability to maintain high frequencies at low voltages so be very careful about that and don't just take the numbers I said here as as gospel for anything because changes for every architecture and I don't keep 100% of it in my head go research on overclocking forums and stuff like that too next one RR says curious about your software setup for your test benches specifically your Windows setup which settings have you changed to minimize variants in your results how frequently do you allow Windows 10 to update I ask because it's may be useful info for anyone trying to optimize their own systems so we we disable Windows Update but we'll update Windows manually for really for like milestones we disable Windows Update because it's important for us to know if it updates because I've seen it seeing systems turn on when I'm just like I'm facing the other way the system's been off for two days and I just hear it spin up like there's a ghost in the room you turn around and windows updates itself so that's a big problem because yes we want to keep windows up-to-date within reason we don't need every small security patch I should say small security patch as in like not spectrum out down will grab those but if it's just like a small some kind of small update maybe they add more USB device support or more nic support or something like that who cares don't hate it it just messes us up but we need to know when it updates and for what reasons so we manually control it so we disable Windows Update you can do it through the services tab if you don't like it and like just now we just rebuilt all of our test benches for CPU and GPU with the move because we're doing all new content retesting everything so now is a good time where we just update it we install fresh windows OS we update it to the newest version and then we disable update lock it there complete the tests for this content and once this content is done we can either decide to update that os keep it or blow it all away and do a new image depending on kind of the quality of the integrity of the image if it's had any if it's gotten messy or degraded from all the drive around resolves reinstalls it's important for us to monitor that stuff but basically it answer your question when we are working on a single content piece we disable all updates so that we don't end up in a scenario where we have half the test data with one version of Windows and then it updates without us knowing overnight or something and then half the test data with a different version of Windows because then that interferes with the quality of the data so yep that's that's all we do it's all manual next one Vernon says we're removing portions on the top and bottom of the shroud on a GPU to help with cooling it when the heatsink fins are oriented from motherboard to case a side panel that is by giving warmer air and a easier exit route so this is actually a great point and that if you ever want to know which way a video card exhausts you can kind of grab the card and this is not the best example because the circuit heatsink but you can look down and look at the fins and see how they're oriented and if they run this way so it's long ways that way with gaps in between then it's the air when it pushes down through an axial fan not a radial fan pushes down from an axial fan the air will follow the channels so it's gonna go in that example it'll go towards the front will go towards the back not a ton through the top if it's oriented up and down like this user is suggesting it'll go up and down so to answer your question yes removing a if the shroud is really covering it that's poor design but removing it would certainly help with ensuring that the air gets out where it needs to get out and it's stuff we've done in the past but not for a while huh this sets pretty cool so yeah if you modify the shroud so that the air can actually is follow the channels and escape like out towards the case panel then it it'll help a bit it's just how much you'd have to test it if they fully a clue or obstruct the the fins at the edge then that's just bad design and yet you should modify it but if it's just kind of like a nameplate and it's taking like a couple millimeters of depth and it's a couple inches long it's not really gonna change anything you're just mutilating your card for not a whole lot of gain so it depends on how much it includes that the finned exit next one Nura says what happens if you use two sticks of RAM with different amounts of storage on it Congrats on 300,000 thank very much so two different two different capacities it'll it'll mess up the timings probably if they're different timings as well so you'll end up with additive capacity so you have eight and four it'll be twelve but which is not exciting but if the eight has like see all 15 times and the four as you know something else 14 16 then you might want to go in there and manually tweak the times to make sure they're all roughly the same because otherwise you can start encountering just blue screens corruption and stuff like that it's it's a good idea to keep the memory times as close as possible next to our last one Angie Agnes in says if you reviewed a case that made an EVGA FTW GPU hit 90 degrees would you call it a perfect case you've ever tested in response I didn't watch the hardware connects video I can't comment on it I'm sorry so I'm not going to comment and can't comment on on whatever they did because I haven't seen it but I will answer the question though of will you be reviewing that if antics evolve X our plan is to yes review it but we have to buy it and I don't think it's available for consumers yet and by the time I heard that it was up for pre-order all the pre-orders hold out and although we tell you not to preorder hardware in this instance it makes sense for us to preorder it so that we can tell you if you should buy it or not but I didn't get in on time so we're gonna keep an eye out if any of you out there bought the Evolve X and you're okay with parting with it for about a week let us know you can tweet at us that's probably easiest way at gamers Nexus let us know if you want to loan it out once it comes in otherwise we will be buying on as soon as they are available and testing it reviewing it we tried to meet up with fantex at Computex this year and last year and a fan maybe CES one year contacts last two years though and fantex was unable to accommodate a meeting with us they didn't want to they weren't able to find time to talk to us I guess so they didn't send us one but that's fine I'll buy one and we'll see how it does it's not sure when so keep your eyes open for that subscribers always leave your questions below if you have them go to store documents access net to pick up one of our mod mats the shirts the poster I mention we've got the teardown cubes back there in the corner and the plenty of other stuff I'm stored at games access dot net you might find something interesting otherwise patreon.com slash gamers access to grab the bonus episode thanks for watching I'll see you all next time