Gadgetory

what's up guys Jase $0.02 here we're gonna build upon the last couple of videos that we've been doing because when I take on you know experiments like how cheat can we build a computer it's funny how many things like sort of pop up as that series progresses right we had part number one where you guys didn't like the parts I chose maybe the part number two where I found out that the bottleneck and we'll talk about bottleneck to in a second here but the the results that we saw were actually being caused by something other than the CPUs when we were doing a CPU test so yeah here it goes part number three I guess dude great this is the best part right here it watches watches guys it's new Celsius x36 from fractal design yeah so but do you know what it does yeah it makes things cool yeah cool hey guys and today I am super excited to be doing a video with the one and only next two cents yo I am excited man so what are we doing today we're gonna review the Celsius s 36 I get that a lot today what's up sure I could give you some workout tips how's tomorrow sound awesome later and serger the new Celsius as 36 can make almost anything cooler almost so I discovered in my last video that even though we had two identical msi gtx 1050 graphics cards right here is Nayan 1050 Ti it's not like the arrow it's not like a special card it's just there I TX PCB with a single fan doesn't even have any power connectors on anywhere gets its power only from the motherboard we had two identical cards here are the boxes for those right here identical part numbers to that we were seeing variants and performance between both of those cards unfortunately the comments section in the last video some people called me out saying that nope they're the identical cards something else is happening here because they should be performing within like 1 FPS of each other that's not what's happening here now before you guys start saying J stop addressing the haters no there's points to be made here and this is good conversation to have because what I what I found was when we swapped the cards the poor performance followed the card so that's the point of today's video it's not about whether or not AMD is bottlenecking or Intel's bottlenecking or something was wrong in system a versus system be the problem followed the card which told me the card is where we have to investigate so this is the G 4560 system right here that we used in that benchmark and the card that's currently in there right now is the one that was originally in the AMD system the one that we realized was actually not boosting as far as we'll say we'll say GPU number one right here we'll call this GPU two hopefully we can keep track of that counting is hard especially for me you can see right here we've had this test running for a while now it's made quite a few loops the temperatures have equalized it's sitting at 62 C you can see we got the GPU running at 99% load that's at 17 hundred and thirty-four megahertz that's GPU boost 3.0 that speed is faster than what the Box advertises this bathroom manufacturer advertises because GPU boost 3.0 from Nvidia is a self overclocking utility it's gonna take into account things like how much power is being drawn from the card how much Headroom there is between max TDP allowed for total board more power and what are the temperatures if it sees that we have power left over and temperature hit target left over then it will self overclock which is what you're getting here right now so based on the 63 see the power draw its overclocking itself to seventeen hundred and thirty-four megahertz now if this had been a card that had its supplemental power like a PCI Express power connector on it you would actually be able to move the power target slider in our case we can't do that and this everything you're seeing right here is stock MSI Afterburner is just reporting what it sees and that's allowing us to just kind of monitor what's happening here so we're only using heaven here because the experience is that I'm about to show you carried across in all of our tests and all of our benchmarks especially ones where CPU were not that important because then the GPU is responsible for everything and the higher the FPS the the more CPU potential bottlenecking will have here and that's a whole nother video I'm gonna revisit the whole idea of bottlenecking in another video I haven't talked about that in a while so you saw we actually we've done in 1696 for a second there that's just that one scene it doesn't use a lot of GPU but you can see 63 see seventeen hundred and thirty-four FPS or megahertz is where we basically capped out and I'm gonna take a picture because I think my brain has reached the maximum capacity of what I can actually remember now and I will remember those numbers so now I'm gonna do the official benchmark run and let us get a number here and then I'll swatch the swap the GPU and show you the difference that way you guys can see there was no shenanigans here but I'm not trying to figure out what's happening here anyway we'll come back so the test is complete we got a 16:57 as our score and to kind of recap where we were when we did the video the other day with this GPU again the same GPU we got a 1665 and the other GPU we had a 1708 member these are the H 1/10 scores not the H 270 which was the original that we had used so the score is pretty consistent 16:57 HCI scores of 17 34 and again temps kind of chilled out around 63 if we take a look at MSI Afterburner you can see as well that we didn't really we didn't have any throttling really I mean it came down a little bit but wouldn't call that throttling that's just based on the temp because you understand the way it works it has these states right the BIOS basically says the state is temperature range between here go to here temperature range between here as its lower go higher so as the temperature does this the frequency does that right so making airplane curves here that's what basically happens so some folks would call that throttling that's not throttling the actual definition of throttling is saying that it's slowing down the speeds because the temperatures I had reached max so throttling specifically refers to max temperature max temperatures have been achieved therefore we have to slow down the cpu to keep it from going above that maximum temperature so although people would say yeah this was throttled because it slowed down based on temperature no that's a curve this is a curve throttled the official definition of throttled now is have to argue people about this the official definition of throttling refers to saving a PC component from overheating this is not overheating by any means but that's what it looked like you can see it was pretty damn consistent right so there is our GPU usage though every all those dips are between scenes here's our CPU temperature this was me restarting the test right there to check for a consistency and yeah so there's everything else right there CPU usage you can see that this is why I'm using this test because it allows the GPU to push itself as hard as it can this is CPU usage and you can see we're not maxing out the G 45 sixes whatsoever so now that we've got those scores right there a 16:57 someone take a picture of this so we can reference and yeah we're gonna swap the card now and do the same test again all right so we swing and switch things over gotta make sure everything's identical you only have to do is close one draw if you ever notice you an uninstall one drive and Windows keeps putting it back every single update that's really really freaking annoying show you guys once again there's no overclocking applied whatsoever it's all exactly the same temp limit same power limit like I said you can't even add power limit right a hundred is the max this is the bad GP I say bad this is the lesser performing GPU we're now running the exact same settings again but I want to show you guys do I remember how where it went to on the previous one it was sitting there like seventeen thirty something of the 1739 watch it watch where this one's gonna go remember same GPU 1810 immediately now it's gonna come down right but you see how it's already 60 megahertz faster then it then this guy was now 60 megahertz matters trust me it matters on a core or on a GPU like this this is what I was trying to explain that we finally realized what's happening because before we were just running our tests and going through them and we're thinking wow same cards same memory even the same series of motherboards we're applicable when we did the first video we tried to make it as seriously concise and and specifically matched as possible some folks even suggested that the power supply was to blame but if that was the case he would see right now we wouldn't see the GPU performing better out of the box versus this one right because now it's in the same system and this is the way we should have done it initially because what we actually introduced here was a variable that we didn't foresee being a problem that actually ended up making our results look very differently than what the reality actually was so by trying to create two identical systems as close as possible I actually added flaw in the video and yeah so as you can see right now we're still sitting at 1785 temperatures are starting to come up a little bit but what I'm gonna do right now is I'm gonna let this run for a minute try and balance out and figure out where the max temps are gonna be where the max frequency is gonna drop and we'll run the test again you guys will see score and how that's actually performing better with that GPU and why my claims of the graphics card were the problem were actually accurate so it looks like things have stabilized interesting note though is heaven although it is a GPU benchmark and stress test it doesn't hammer the card as hard as something like Ghost Recon does Ghost Recon has to go on low settings and still maxes out the memory on this but we are seeing at least identical temperatures here it looks like we hit 64 for a second you can see so paper to come up this scene right here is really interesting in that it doesn't leverage the GPU as much see how it came back up to 17 72 on this card remember it dropped down into the 1690s and then came back up to 17 30 but what I want to show you now is by running this test again that's actually 1c hotter than this one now go figure anyway we'll just throw temperature out the window because 1c is not going to make a big difference in terms of the curve as you can see but I'm gonna run the benchmark again remember what was our previous score see this is why I took pictures cuz I'm telling you I will have to tell you I don't remember a damn thing all right 16:57 is what the previous score was but we're also gonna look at the minimum and maximum because that that's actually how the score is generated min max average so yeah we're gonna let this test run and you'll see now with the same exact system and switching the graphics card to an identical graphics card how much performance difference there really was so the test is done and we just got a 1704 and the previous test right was a 16 57 now let's compare that to what we got in the video the other day so ramp my spreadsheet here and I take a look at what we got 17:08 with this same card on the last video so a difference of four which is huge but let's talk about the min max right so the average FPS here was sixty seven point seven on the previous tests right before this was sixty five point eight a minimum of twenty nine flat over here twenty five point seven on the on the previous GPU and a max of 130 point eight the previous test was a max of one twenty seven point nine so as you can see it's pretty obvious that even though we have a truck backing in and its backup beeper making noise even though we have two identical cards right here there are clearly differences between the cards and that's gonna fall down it's going to come all the way down to a six all right a six is a really fascinating subject if you actually takes the time to research Asics exists and anything that have a circuit board but it has to do with a six and anyone who understand a six knows the difference it makes when it comes to overclocking not specifically in the frequency but the voltage in the way the voltage interacts with the core the amount of voltage leaks and the amount of power that you can actually give it it makes a huge difference in terms and I know I'm really simplifying this but it makes a huge difference in terms of overclocking and I think that's what we have happening here I think we have two different Asics that are causing a significant issue here what I should have done actually before I swap the card was check the ASIC quality let me see if I can do that real quick well I just tried to read the ASIC quality on the card it looks like the Pascal cards are still not showing up in gpu-z when it comes to ASA quality that's too bad cuz I think that would have told it's a little bit of a story here Asics do not a sequela does not guarantee an overclock it is not like the higher the ASIC quality the better the overclock in fact some people would argue it's the opposite lower the ASIC quality the better the overclock but I seeing a huge difference in value here where at least given us some place to look I'm not too concerned about the card itself having anything physically wrong with it I think it's just that I think it's just variants and core quality I'm starting to wonder if maybe this is something that's present more so on the lower tiered cards o than the higher tiered cards I have a lot of 1080s a lot of 10 70s a lot of 1080 eyes and they all perform within a ridiculous margin of each other I have for ten atti founder's cards here right now for various builds I have tested all of them and they all pretty much run identical the only thing that seems to vary between them is they get a little better over time because the drivers improve but then if you take another card and throw it in they pretty much are the same score not massive differences like this guy right here so I would have loved to have put in a high tier graphics card and test this the problem is I'm using the budget built I mean maybe we'll maybe this would be something we test in the future on the test bench and we'll start with the low tier and then use a couple of identical cards and the high tier and see how much of a difference that actually makes but obviously here this video shows that my dog graphics cards are created equal but that said guys I'm going to go tell me what you thought of today's video what other experience you think we should do I might revisit this in the future with using the test bench and testing it on very low end cards that are identical and very high-end cards that are identical and I think it'd be cool to get them from two different sources because I've always had this theorem mind that some places seem to get lower tiered or lower quality parts I know it's really hard to discern but things like stepping on CPUs I've always noticed that the CPUs I get from Micro Center never overclock as good as CPUs you get from somewhere else and I want to test that and see if that's true anyway it looks I'm gonna be spending more money time to go guys thanks for watching as always I'll see you in the next one