Gadgetory

and Vidya just launched its GTX 1060 video card including the founders Edition and the AIB partner models all today and in theory the price floor stretches down to $250 though we've seen how that's worked with Pascal as far and the founders Edition is priced at $300 the new gtx 1060 uses the GP 106 GPU to slightly cut down version of the GP 104 200 and 400 chips that we've seen on the 1070 and 1080 and before getting to this review this coverage is brought to you by msi and their new x99 a gaming motherboards for Broadwell ii including the x99 a titanium board with pretty serious power phase design when the gtx 1060 initially launched we'd already published estimated specs that turned out to be pretty much dead on that included speculation on TMU county pcs SMS and g pcs and just as an FYI we're able to extract all of these numbers by using simple math so each Pascal GeForce GPU hosts 128 CUDA cores per SM so by knowing the total kuduk or count of 1280 cores on the gtx 1060 we can divide by 128 and find out the GP 106 GPU has 10 SMS split into clusters of five between 2g pcs or graphics processing clusters you can learn more about that on the website we also know that HSN hosts ATM use or texture map units yielding a DTM use on the gtx 1064 more on the low-level specs on the drill-down hit the link in the description below for the article and you can learn more about that stuff as far as top level stuff the GTX 1060 founders Edition this is clocked at 1506 megahertz base and it's got a 1708 megahertz boost clock and the MSI card comes in a bit above that hitting somewhere in the range of 1809 megahertz or there abouts when we're running it with its OC mode enabled the memory is gddr5 and it runs at 8 gigabits per second or if you prefer 8 gigahertz basically the same thing for this type of measurement and that is g5 not gddr5 X or g5 x memory bandwidth is at 192 gigabytes per second calculable by using another simple equation its memory bus width divided by eight to convert bits to byte multiplied by the memory clock than by two for DDR and then by two again for gddr5 the GTX 1060 it has a TDP of 120 watts it uses a single six pin power header note that AMD non-video TDP numbers cannot be linearly and directly compared as they are generated using different standards and it's more of a measurement of what's required to cool the thin than anything we use our own power draw test later to compare the two the table on the screen now shows our bench components check the full article for complete test methodology drivers used and more architecture depth we have more than a thousand words dedicated to test methodology so I'm not gonna go through all of it here we're gonna start with thermals because it reveals some pretty interesting traits of the founders Edition cooler especially when compared against an aftermarket cooler like the twin froze or six cooler which by the way is identical on the MSI 1060 1070 at 1080 for the most part there's a few heatsink differences there's some missing or slimmed down heatsink some things like the PWM but basically the same design so we look at those two with endurance and short-term testing with thermals let's start with peak average thermals and then move on to throttling analysis the GTX 1060 founders edition card runs at 40 3.6 C delta T over ambient it's your channel for an explanation on what that means and when under full load that's what it's hitting but idle it's hitting 6.15 C to compare the msi gtx 1060 gaming X which we've also got here with trend froze our six cooler and noteworthy a higher clock rate lands at thirty 7.36 Celsius and that for both of these cards is a good deal lower than what we saw in the 1070 and 1080 the GTX 1070 fe operated at fifty 3.86 Celsius load making the GTX 1060 EFI a full 10 C cooler than the 1070 compared between the 1070 gaming X and 1060 gaming X again pretty comparable similar cooler from MSI we see the GTX 1060 gaming X runs 4.4 C lower with identical ish coolers and the GT X 1080 EFI runs at 50 7.51 Celsius or if you add in ambient it's hitting an 82 C a hard stop throttle which we explore in our 1080 review if you're curious either way you take it GP 106 is a considerably cooler chip than what we saw with GP 104 especially when looking at the reference design cooler or Founders Edition or whatever they want to call these days and that thermal reduction is a direct result of the lower performance which demands of course less power to sort of fuel the card here's the chart as it appears over time at listing only a few cards the full chart is on the site our thermal testing is fully automated so a test execution is exactly the same each time and this is somewhat useful but if we look at this chart another way we can add in an endurance metric so now we're looking at thermals and clock rate of the reference or founders Edition card over time you can see that the clock rate remains stable throughout the majority of the tests and is able to sustain temperatures shown here and there actual non Delta values around 67 to 72 Celsius there are a few sudden spikes that plunge the clock rate to a couple hundred megahertz and that manifests itself as sudden frame drops in gameplay these last a fraction of a second and are noticeable but aren't frequent enough to get mad about in general the GTX 1080 had bigger issues when running this OC endurance test on the FE card and again we look at that in that review let's zoom in on this chart though this version shows the temperature scale blown up in the time scale blown up so it just basically turns it into more of a microscope as to when the clock plummets and the frequency falls in step with temperature hitting 71 C which seems to be the throttle point of the GTX 1060 and it's auto stock settings including the auto fan this can be worked around by increasing fan rpm which we also measured and is shown on this chart you can see the fan rpm was capped at 60% which is probably a noise consideration that we'll talk about in a moment but what about MSI is cooler let's see if the aftermarket cards are able to compensate for the occasional frequency drops with a better cooler and it does sort of anyway first of all the overall clock rate is higher as a result of the pre OC applied by Emma size were at nearly two gigahertz here it's hitting nineteen hundred something depending on what game we're running and notice that the line is almost perfectly flat that's what you want the GTX 1060 EFI has a spiky line with constant fluctuation something we've seen on most reference cards from both vendors and this has been resolved by more advanced coolers with the after aftermarket Adhan board partners this creates a perfectly stable and repeatable frame rate throughout the looping Metro test we're still hitting those big spikes though if you look around the 100 minute mark and the burnin it's showing that we definitely don't have a perfect solution here with just an aftermarket cooler so we'll put this card under liquid like we did for the previous cards and see if it actually fixes anything or if it's some sort of lower level voltage or electrical thing as for our PM MSI starts at 0% fan speed shown here and relies on the heatsink for cooling at lower than 60 Celsius temperature and that's when the fan clearly kicks in so we have cross bars on the screen now to demonstrate what's happening here it's it's below 40 percent rpm even when under full load for in-depth thermals check the article for more as always but let's get into noise so for a quick look at noise we've got the chart on the screen now we measure in idle noise which is captured at the desktop we measure the auto fan noise which is captured after about a five-minute gaming workload burnin and then after auto we switch between a 50 percent at a 100 percent fan speed test so you can see what it looks like in an absolute worst-case scenario and all of that is represented here now GPU fans will basically never hit 100% unless it's in some sort of really constricted environment like a mini ITX case generally and this is true for these cards as well you're gonna be in the 50% range for fan speed at its max and the MSI card was below 40 percent almost all the time the founders Edition card was about 50 to 60 percent depending on the workload so that's what we got for noise levels as for power we're still working on more in depth power testing for the future but for now we're still using a wall meter and we're doing total system power draw that means you're basically looking at deltas between devices because we're measuring the total system draw not just for the card so for what we've got here we measure in a parent power draw which is represented as both amps or VA and we're seeing the gtx 1060 efi run 72 volt amps idle when with the rest of the build of course and that system with a 10-6 TFE runs at two 18.6 of old amps under load this makes it the least power-hungry card currently on our rebuild bench though that will change as we introduce things like the rs4 7460 and potentially a future gtx 1050 the msi variants is slightly more power-hungry at 98 volt amps idle and 222 load and overclocking sees the cards of course boost in their power draw the msi card is able to run a slightly lower voltage with its overclock and sustain the clock rate with greater stability than the founders edition card and that does impact the power draw here or a parent power Joe I should say time to get into the frame rates we ran nine or so games through the bench but only have a few of them here at link in the description below for all of those games and as a reminder we convey our metrics as average FPS 1% and point one percent lows which helps showcase the poor frame times that sometimes occur and you can learn more about that on the channel where we have a video explaining those three numbers first shadow of mordor at 1080p ultra there's very tightly timed 1% point 1 percent low values you're seen on the GTX 1060 are largely a result of its generally stable clock rate and the other optimizations and architects are also helped of course but the GTX 1060 outputs 87.7 FPS average 60 8.7 FPS 1% low 66.1% lows and the our X 480 as comparison is at 78 FPS average the percent difference between the 2 1060 Fe and our X 480 reference is 11% with the 1% lows differentiated by 50.5% looking at the msi 1060 gaming ex pre overclocked we're hitting about the same FPS that's mostly a thermal advantage on FPS advantage and the difference between 2 1070 and 1062 ARDS is 31% 1440p has both the MSI and Fe GTX 1060 s at 63 FPS average landing them well within playability for Mordor which really doesn't demand that high of a framerate the our X 480 runs at 58 fps or a percent difference of 8 8% between the rx 480 reference and the EFI 1060 the 1070 runs at 77 FPS average though with repeatedly lower 1% low values and the 1070 is separated by a 20% difference from the GTX 10 60s at 1440p 4k is a serious struggle for these cards they're really not quite low end but they're definitely lower and and they're not really built for 4k either the 10 60 runs at 35 to 36 fps between the Fe and gaming X cards there are X 480 sets at 32 fps not a noticeable difference since the game becomes choppy at this point anyway but certainly a measurable one the gtx 1070 r9 390x are more capable cards than either of these running at 51 FPS average and 41 FPS average respectively GTA 5 remains one of our most demanding gaming benchmarks at 1080p with very high and ultra settings we're hitting 95 FPS average on the gtx 1060 f:e the 1% in point 1 percent low is remained reasonably tightly timed and above 60 FPS even for the dips the RX 40 is competitive at 85.3 fps marking a difference of ten point eight percent between the fe and four eighty reference cards this is consistent with mordor in terms of percent difference scaling and the gtx 1070 fe sits at 123 FPS or a difference of 23.5 percent from the 1060 comparing the two with percent difference formula moving on to 4k the 1060 predictably struggles at 33 FPS average though it still sits above the SSC version of the 970 we've only run a few of the most recent cards through GTA 5 at 1440p and just really curious we're getting 63 FPS average on the 1016 about the same on there are X 480 our full Mirror's Edge catalyst analysis is already on the channel but let's start with 1080p ultra here we see the FAU card hitting 82 FPS average while the msi card hits 86 or a difference of 4.7 percent from that prio see the rx 488 gigabyte card is at 74 FPS or about a 10% difference from the gtx 1060 FG again consistent with the previous few titles the gtx 960 SSC prio seed pretty reasonably sits at 51 FPS or a full 46.6% different from the gtx 1060 Fe that's a pretty large generational growth even actually especially considering the SSC model of the 964 teen 40-feet ultra the GTX 1060 Fe in gaming action main gaps similarly to 1080p both just below 60fps the RX 488 gigabyte struggles with 0.1% low values tipping the 20 FPS with the 8 gigabyte card while keeping a 47 FPS average you'd want to drop to high settings for a more fluid output here as those lows introduce a lot of stuttering and jarring motion 1080p hyper runs well on the 1060 s keeping near 60 FPS average and the RX 40s it's around 53 fps also pretty reasonable though the 4 gigabyte card struggles with its 18 fps 0.1% lows and 40 FPS average we explain this phenomenon and the RX 484 gigabyte vs. 8 gigabyte test 4k had the GTX 1060 s at or below 30 FPS with high settings depending on the card you can see more of that chart in the article black ops 3 tends to favor AMD cards a little more than some of our other games tested and it's also somewhat clock sensitive it depending on the instance the settings things like that at 1080p we're seeing the GTX 1060 Fe run at 120 2.7 FPS average and the our X 480 cards at 127 FPS average and 132 FPS average depend on if it's 4 or 8 gigabytes the MSI GTX 1060 gaming X is at 120 9.3 so we've got a couple FPS gain actually 7 FPS gain over the FE card and that marks a difference in averages of 5% the RX 488 gigabyte and gtx 1060 fer differentiated by 7.5 percent at 1440p all of these cards remain well within playable range even for a competitive FPS the 1060 fe is at 78 FPS average gaming acts at 80 2.3 FPS average so you got a bit of a 5 FPS game there and the RX 480 is at 83 FPS average so it's mostly tied for KSC is the FE 10 60 dip to 38 FPS average and the 4 gigabyte R X 480 suddenly becomes unplayable with its 12 point 7 FPS 0.1% lows the gaming X sits at 40 point 3 FPS so you ideally dip the settings for faster competitive play if desperate to use any of these cards at 4k but 1440p is a much better fit ashes of singularity shows Nvidia's DirectX 11 optimization focus and Andy meanwhile shows that they've done better with DirectX 12 and their DirectX 11 performance is pretty bad at 1080p high the GTX 1060 runs at forty 5.3 FPS average for dx11 that's the EFI version about 47 fps for DX 12 the rx 40 performed significantly worse with dx11 at 26 FPS average but is effectively tied when using DX 12 and the scaling continues to 4k high though the rx 40 actually pulls marginally ahead of the GTX 1060 about 0.8 FPS when using DX 12 let's look at frame Layton sees in milliseconds as the average frame latency here the GTX 1060 efi runs an average millisecond at frame time that's the time between frames of 21.2 nine milliseconds width DX 12 verses 20 2.07 4dx 11 slight improvement the RX 480 8 gigabytes it's at 22 to average time between frames and milliseconds for DX 12 versus 39.2 640 X 11 that is a huge performance almost 2x 1070 is at 15.4 3 at milliseconds average time between frames with DX 12 and the 390x just for comparison to last gen is at 20.4 6 milliseconds we've got more ashes charts on the website including latency percent change and several other games are also on the site including doom assassin's creed syndicate in some light fashion we've also got Vulcan and OpenGL tests note doom the division metro tiles principle and some other stuff so hit the length of description below for all of that overthought can the gtx 1060 is the same as the previous pascal cards we are bypassing the voltage frequency curve it was really a cool idea on paper but I find it very annoying as someone who has an idea of what I'm doing and I'm sure that's true for a lot of folks out there so we're still using the slider manual approach to overclocking and as for results we've got this stepping table that shows progression of the efi versions overclock the peak clock just shows the spikes and can be somewhat ignored core clock is the column you want to look at this is the one that we measure after some burnin and find that it seems to be about where the clock sits at once stable we managed hit a stable core clock of approximately 20 88 megahertz on the FE card with peaks at 2150 1 point 5 megahertz the clock stabilized at 2088 during the endurance burnin since we're showing here the memory was overclocked by 450 mega Hertz offset so that puts that 44 55 megahertz and that's what the power target maxed out at 116 percent and the fan at about 2,000 rpm the overclock is a 250 megahertz offset from Corp as an aside notice that in that second step in our table for overclock progression we see an increase in the core clock stability just by increasing power target and nothing else so that's something that's worth learning about as for MSI's card we hit about the same overclock frequency and struggled to push much beyond what's shown on the screen Emma size card was able to sustain a tire o'seas with a lower voltage though which is always beneficial to power draw and if you want to see overclock thermal stability charts check the article but here's a quick look at the frame rate charts GTA 5 at 4k sees a slight boost from 33 FPS average to 36.7 FPS average or a 10 point 6 percent difference as the efi card and at 1080p it posts the gtx 1060 at 95 FPS average with the stock clock and then 104 dot 5 for the OC variants another difference of about 10 percent 9.5 actually shadow of mordor at 1080p post the jump of about 9 FPS on the GTX 1060 EFI with the overclock the RX 4 80 reference it's at 85 FPS with its 13 40 megahertz clock rates a gain of about 7 FPS average 1440p tests pushed the 1060 efi from 63 FPS to 70 FPS average and more on this in the article along with Mirror's Edge catalyst overclocking results and that leads us to the conclusion so the recent video card releases from both Andy and Nvidia have pretty much shown that 1440p gaming is now a real possibility even at those higher settings when looking at the sort of mid-range card market so previously it was not really possible for a 200-dollar video card to run 1440p games with higher graphics settings and that's shown pretty well in our charts where we do have the 960 SSC presence I know it's not all of them but we have it in some charts and you see a 40 to 50% difference between the 10 60 Fe and the overclocked 960 which was the SSC version from EVGA this that's all we got so 40 to 50 percent gains pretty big or I should say 40 to 50 percent difference not percent change a percent difference and that's what allows 1440p gaming to become more of a reality the RX 4 80 C's similarly large gains from the r9 380 and even the 380 X so this is a thin now that's possible it's it's time really for a lot of the budget to mid tier gaming setups to be looking into monitor upgrades I would imagine if they're still in 1080p now stand alone let's ignore the comparison the obvious one to the RX for ready for a moment and just talk about the gtx 1060 as a product it runs reasonably cool with the founders edition cooler and this is really a big change from what we saw with the gzx 1080 1080 wasn't awful but we saw some thermal throttling on it because it was hitting that 82 C threshold really easily and that's just a result of a higher TDP chip I'm not going to say a high TDP chip but certainly higher than the 1060 under the very basic blower fan cooler and we resolved that by putting it under water and showing how much liquid actually can improve things or just better air if you have something more standard like one of these twin frozer or a CX or windforce or whatever one of those Strix all those will improve over the founder's edition and reference cards and that's something we've said since the 1080 intense OD reviews GP 106 is a big part of this so GP 106 is a lot cooler it's lower TDP and lower measurable power draw then the 1070 and 1080 that's a big reason we're seeing the thermal reduction the stability isn't bad clock rate stability is generally acceptable on the FE card which is the FE card is pretty stable we see some clock rates spiking where you get that effect in the chart but it's not big spikes there's a couple big spikes and they are pretty large they're from the the operating frequency to a couple hundred megahertz but those are rare and they're spaced out enough that even though you can see a very quick and short stutter in frame output it's not something that's going to repeat with great regularity it's not too annoying and it's not honestly outside of what we see with pretty much every card on the market except for the higher-end liquid-cooled ones like the hybrid and things like that or the hybrid we built I should say which is this one here on the table as for the head-to-head with AMD the RX 480 is the obvious competition here it comes in two flavors four gigabytes and eight gigabytes the gtx 1060 is strictly six gigabytes it does not have an eight gigabyte model does not have a four gigabyte or two gigabyte model so you're looking at a card that is priced at $300 for the founders edition version and 250 as the bottom and recommended price for AIB partners that does not mean they will be available there but they should be eventually it's normally how the market works immediate launch they may not be though so you look at 250 to 300 for a six gigabyte gtx 1060 that on average in our test that we've shown in this video performs about 10% different than the RX for 80 improvement over the 480 versus a $200 for gigabyte card that performs fairly comparable to the 1060 in a lot of cases but not all Assassin's Creed syndicate the division are good examples of games where you really do actually see a real deficit with that 4 gigabytes for the 8 gigabyte model for 240 so let's just look at those two 8 gigabyte to $40 card verse is a 6 gigabyte which of course there's more to it than that because architecture but 6 gigabyte - $50 - $300 card if you can find a 10 64 $250 it would be a very good consideration at that price point if you're already looking at a 480 especially if the 480 uses the reference cooler because we found that really quite honestly the reference color was trash if you can find an aftermarket cooler ie from an AIB partner for the RX 484 around the 240 to 260 mark versus and aiv partner gtx 1064 around the 250 260 mark then that's the real comparison to be making you'll want to look at the specific review for that product but in terms of framerate performance general considerations of overclocking aside the difference is about 10% in most games that difference is certainly measurable in some games it may be perceptible if you're pushing just barely past whatever framerate threshold you're targeting but for the most part they are both reasonable performers at 1080p and at 1440p do not expect 4k from either of these cards so that's that's most of what I got for you they are both decent cards the 1060 Effie does cool better but it's $300 so we really can't consider to directly as a comparison that's all the data though so I've given you all the data you need hopefully you can make a decision from that I'm not going to make one for you we try not to do that for anyone if you have questions post them below do our best to answer article in the description below for full in-depth details including additional games testing and some more of the opinion side that we don't have here and of course patreon you can post all video to help us out directly you subscribe as always thank you for watching I'll see you all next time you so by knowing the tooth total Bluett again another fYI is calculable and that's probably a noise config