Gadgetory

the Intel i7 7700 K found its place in our bench this past week tested in both the MSI Z 270 gaming Pro carbon and gigabyte Z 270 m7 motherboards the i7 7700 K runs a base clock of 4.2 gigahertz with turbo up to 4.5 while running on similar architecture to skylake the results is that gains are primarily derived from frequency but changes to speed shift cross point and low-level overclocked tuning it means there's a lot to consider this Intel i7 7700 K review is brought to you by thermal takes p3 case a $100 enclosure that can be leveraged as a wall-mounted test bench p3 is easy to work on for enthusiasts who regularly change components click the link below for more information before an architecture overview let's quickly go over the seventh-generation kb Lake SKUs for the new family there are 40 of them in total that includes the why u H and s Series CPUs the most relevant ones of course are the i7 7700 km 9k the i-5 so d 600 K and the i3 70 through 50 K which is a KS q i3 that we'll be looking at hopefully late January and in total the TDP for all 40 SKUs ranges from 4.5 to 91 watts so it is a pretty big range and the most popular SKUs are those three that I mentioned the case use respectively the 1000 unit pricing for those CPUs will be $340 for the eye 7077 100k 242 for the 7600 k and 168 for the 7350 KB 7700 9 k will be three hundred three dollars and the cheapest i-5 will be 65 watt TDP i5 2400 at 182 dollars the cheapest relevant I 3 is the i3 sony 100 at 117 1k you we've put the three case Q CPUs into a specs table here with focus on the 7700 K that's on reviewing today that's a traditional 4 core 8 thread CPU but with a higher frequency than previously this is at least partly achievable by refinement of the 14 nanometer process something that Intel is now calling 14 nanometer plus the physical changes are almost entirely at the Finn level with Intel widening the gate pitch and making fins taller and this allows for increased frequency and higher overclocking potential somewhere around an extra 200 megahertz minimally for the stock shipping clock rates are upwards of about 400 megahertz one of the most critical aspects of the KB Lake architecture is that the Intel CPUs are now leveraging speed shift in a more effective capacity and that's primarily by increasing the frequency at which the clock rate can change in a given second and it's up to a thousand Hertz now a thousand times per second the KB Lake CPUs can switch their frequency to a lower or higher value as needed depending on the workload this means a few things one of course is that you improve battery life in notebooks and things like that when you don't need to be running a higher clock rate for trivial tasks for the CPU to handle the other is that you can get boosted performance when there are those really bursty high requirement periods of an increased clock to handle more instructions and although we've asked into what the minimum granularity is for the clock switching at that 1,000 Hertz interval I haven't heard back yet but it does change out a thousand times per second and that's bigger news for speed shift just like everyone else these days intel is now moving to better be CLK awareness by providing a volt frequency curve that doesn't mean the motherboard manufacturers are leveraging that to the full potential but they are adding a volt frequency curve so voltage and frequency will adjust in step with one another this is a step toward resolving some issues with the higher voltages than what's required but motherboards are still dictating a large part of that especially with BIOS on things like the gigabyte board that we'll be reviewing shortly the KB Lake boards we've tested so far do have large differences in thermal performance from one to the next with the same CPU and the same clock rate just because the BIOS modifies what the running voltage is for more architecture discussion of that interest you check the link in the description below to our full written review which has all the block diagrams for the new chipsets including the Z and H and B series chipsets and some additional information on how the architecture has changed with KB Lake and a lot of that is kind of a recap with skylake but you can learn more there let's get into the testing now so first of all a huge note because we're doing a CPU review it means that we're changing CPUs which of course means we're changing it platforms for a lot of those as we change platforms there might be to change things like memory or the memory clock especially as we get into older generations like Sandy Bridge for all the platforms and test benches again link in the description below if you're curious what memory frequency was used or what motherboard was used things like that and while we do want to add the i-5 and the i3 and of course the Xen CPUs at some point when they are available for benchmarking today we're focusing on the ice evans frankly that was enough work to do for now with CES around the corner the i5 and i3 will be added shortly thereafter for this truncated version of our thermal benchmarks for the i7 7700 K we're looking primarily at the performance difference that voltage can make we tested using both the MSI gaming Pro carbon and gigabyte RGB version of the m7 boards each of which has a different auto configuration for voltage and then we also threw in manual voltage tuning for a better understanding of the 7700 k's temperatures compared to manually tune to 6700 K CPUs these tests use the kraken X 62 cooler which was chosen so that we could monitor the temperature Delta between the liquid the actual cooling liquid in the tubes and the CPU package note that all these tests use the same settings a is T is off turbo is off frequency is set to 4.5 gigahertz fixed on all CPUs and C states are off fan and pump speeds are at max settings so 1800 rpm on to 140 millimeter fans with a max pump rpm for the first test we're looking at the Gigabyte motherboard being tortured with Fourier transforms for half an hour and the i7 7700 K is constantly in the 90 C range occasionally hitting 90 for ambient was 21 to 22 C for all these tests so you can do that calculation yourself if you want at our temperature range is plus or minus 1 C this was the first board we tested so the concern initially was that kb lake would run way too hot because the same stock configuration on the board made a be quiet cooler a 50-dollar be quiet cooler operate at about a hundred C with the CPU now further investigation revealed that it was less of an issue with kb lake and more of an issue with the motherboard which is what we'll talk about in our other board reviews this week the board was pushing voltages as high as 1 point 4 volts at times which is wholly unnecessary that maasai motherboard meanwhile produce an auto configured voltage of about one point two eight to one point three two volts with a corresponding temperature range of eighty to eighty to see max about ten C lower this is with a package power that's reduced around twenty seven Watts from gigabytes but with exactly the same frequency performance we then dropped the gigabyte board down to one point two thousand five volts manually giving it a range of one point one eight eight to one point two seven five volts as controlled by the motherboard though with our manual guidelines this broad temperature down from ninety four C with the auto out of box configuration this is straight from the factory although with an updated BIOS prior to release and brought it down of 94 C to a seventy C Max massive difference the next question is whether this is any better or worse than the previous generation and so enters these 6700 K with the same setup same configuration same frequency all that stuff and using the MSI gaming m7 motherboard with a z1 so V chip set these 6700 K is producing a temperature of about 70 to see on the package with the Auto V core out of box and the V Corps was about 1.3 to with this configuration this is already cooler than the comparable MSI gaming Pro carbon benchmarks despite running a higher reported wattage in fact when custom tuning these 6700 K to operate at exactly the same voltage as our manually tuned 7700 K we're seeing a difference of approximately six to seven Celsius this is potentially because these 7700 K samples we have two of them are having trouble transferring heat to the IHS which of course would mean that your liquid cooler is less effective but it will require more testing to validate now one thing keep in mind that a lot of this is on the motherboard vendor and that's something we'll talk about more but basically you should check your voltages when you buy boards for this platform because it will drastically impact your temperatures aside from thermals the blender benchmarks were some of the most exciting to conduct our resident video producer and 3d animator Andrew Coleman built a custom blender scene for gns render benchmarks and it's got a mix of motion blur different materials and material properties ray-tracing things like that so it's not an easy scene to render relying on the CPU for crunching the scene the intel i7 lineup dating back to Sandy Bridge though skipping Ivy Bridge shows the march of progress from 2011 to 17 we see significant scaling compared to 2011 si7 2600 kcp you resulting in render time differences of upwards of 32 minutes if it's your job to render animations and you're using an older CPU to do so it's probably time to upgrade to nearly anything else of course this is ignoring the fact that CUDA and OpenCL exist skipping the Ivy Bridge we see the devil's canyon 4790k rendering the same scene in about 49 minutes this has improved upon by the 6700 K skylake CPU which does the work in about 45 to 46 minutes and that's an improvement around three minutes per frame from Devil's Canyon sky lake moving to the 7700 K at its stock clocks that time has cut another three minutes to 42 23 seconds or about 7% shorter in total render time this scaling has been fairly linear for three generations now obviously with a big jump between the Sandy Bridge CPU and what follows out of curiosity we ran the same render benchmark with hyper-threading disabled on the 7700 K resulting in a render completion time of 60 minutes worse than the 4790k blender takes advantage of the additional threads here and is a real-world demonstration of what gamers often miss out on which is job management engines that more properly task out all the threads in the system overclocking the 7700 K to 5.1 gigahertz which is trivial work on this platform speeds up our render times by an additional three minutes over the stock configuration Cinebench is the next in our line of tests this one is a synthetic benchmark but is similar in its objective to our custom-made blender animation it's still rendering it uses more Universal scoring though that allows for better comparison between Hardware we're seeing a performance output of 988 points using the full stock configuration of the 7700 K with hyper-threading a single core receives a score of 196 for an MP ratio 5.05 disabling hyper-threading drops us pretty heavily down to 767 points and a fairly trivial overclock to five point one gigahertz with hyper-threading enabled and a 1.3 7v core on the msi gaming Pro carbon results in an 1122 CPU score the single core performance for the same overclock is at 222 and the MP ratio remains 5.05 last generation 6700 K performs at 9:41 points or 185 single core the 4790k two generations old receives 8 9 eight points or 180 for single core performance finally the Sandy Bridge i7 2622 score with a 130 point score for single core performance we're moving on to 3d mark next another synthetic test then we'll get into game benchmarks 3d mark tests use the new time spy and the 1080p version of fire strike just a quick note here 3d mark results do you have some variants in them we see as much as 50 points of variance from one past the next so this system gets more accurate as the performance between two devices gets farther apart the stack up is about what you'd expect based on previous tests the stock 7700 K receives a score of 18,000 685 points with physics score entirely CPU base and probably the most important at 14 four seven eight points overclock in the 7700 K to five point one gigahertz results in a score of 19 five one eight points or sixteen for three one for physics benchmarking so to recap the two important numbers here 14 4 7 8 for the stock 7700 K physics score and 16 for 3 1 when using the overclocked version to put some perspective in the benchmarks we can speak strictly to the synthetic test performance improvement and that's approximately 13% in the physics testing or 4.5% the total score the 6700 K posts a score of 13 64 8 for physics or about 5.7 percent slower than the 7700 K at stock clocks and real-world terms the physics performance of the 7700 K posts an FPS about 46 with the overclocked version at 52 and the 60 to 100 K at 43 fps this is of course ignoring things like frame times we'll get into with the game benchmarks 4790k is at 40 fps and moving onto times by our cpu score is 5852 on the stock 7700 K or 6371 overclocked to 5.1 the 6700 K rests at 55 oh nine points and the 4790k at four eight five eight and then the 2600 K is at 32 42 points so we can see some of the disparity but need more perspective from games to get a better idea that's enough synthetic testing for now blender gave us a real world look at performance for rendering and now we need one for gaming the main way to do CPU tests for gaming workloads is to run game benchmarks that are more cpu-bound rather than GPU bound of course because you want to root out those differences however because so many games are mixed workload or GPU heavy we've thrown a couple of those in here as well like watchdogs too just to give a better more realistic idea of how a high-end single GPU will scale or limited scaling when working with even an i5 and i7 based on how the game is optimized but of course we've got the CPU heavy stuff as well just for a more hard numbers look at the differences battlefield 1 has received several patches since our initial Battlefield 1 cpu benchmark and we're also using new components for the testing so the numbers are not comparable using DirectX 11 only here because DX 12 still has some problems in bf 1 we're testing our suite of i7 CPUs with 1080p ultra settings and a 96 degree horizontal fov with these settings we're seeing the 7700 K KB like CPU operate at 141 FPS average stock with the GT X 1080 FTW followed by 115 fps or 1% lows and 104 fps 0.1% lows overall the CPU allows for consistent frame times at the low end while enabling the 1080 FTW to operate effectively to its full potential disabling hyper threading keeps our average the same but brings the frame times down measurably the 1% lows are now 102 0.1% now 84 and then the 6700 K pushes 147 FPS average effectively the same with lows at 105 and 87 no perceptible difference here but certainly a measurable one and more CPU bound games will help illustrate the differences further in theory to compare it to a popular CPU from two generations ago the 4790k operates at around 140 FPS average 107 fps 1% lows and 94 0.1% lows the marginally improved frame times are slightly a result of higher turbo clocks and this is something we saw in our initial 67-yard k review as well you can check that out if you're curious about why it's happening our 2600 K instantly eats performance and creates a CPU bind where it was previously more of a GPU bind we're at 118 FPS average 78 1% load in 65 0.1% loes ashes of the singularity has a built-in CPU benchmark when operating DirectX 12 and provides more of a CPU centric look at performance with 1080p high settings full resolution textures that we're seeing a performance throughput of 41.5 FPS average 33 1% Lausanne 31 0.1% blows disabling hyper-threading which is obviously not something this is really built for brings us down nearly 10 FPS and averages overclocked we don't get much additional performance certainly not a noticeable amount the 6700 K it stock settings operates about 1 FPS lower than the 7700 K about 3% slower in percentages if you prefer and the 4790k pushes a 36 FPS average with about 85 FPS reduction in low framerate the 2600 K puts things into perspective a bit better approaching two times slower than the overclocked to 7700 K moving on to GTA 5 the chart is expectedly topped by the OC dat 7700 K at a 151 FPS average with low sustained near 100 FPS there's a few FPS reduction and performance to the stock 77 honored K with nearly equal performance with a 67 RK with this configuration the 4790k posts a drop of about 7 to 8 FPS from the 7700 K or 5.2 percent slower in averages the 2600 K pulls our GPU performance down quite a bit - 104 FPS average with 0.1% low values now at 65 FPS nearly a 50 FPS reduction in performance thanks to 2011's Sandy Bridge our second-to-last title is Metro last light just for something a bit known and because we've seen reasonable differences in CP performance in the past we're seeing an average FPS of 145 on the kb lake ksq cpu with stock settings with lows at 108 fps and 101 FPS disabling hyper-threading as we've always seen with Metro last light kills performance but it's also not a standard use of an i7 and it's probably not developed for moving to the 6700 K we see about a 2% reduction from the 7700 k's results the 4790k falls to about 137 FPS average that's a performance loss of about 6% from the 7700 K and a 2600 K it drags it all down to about 111 FPS average with lows now in the 70s watchdogs tube provides a look at a modern mixed workload title the game has some scaling between CPUs but is also pretty abusive on the GPU this means that you're getting a fuller picture at gain limits when changing CPUs as dictated by the games selves but also seeing that a multi-generational jump will still greatly improve performance let's Shawn most obviously when referencing the i7 2670 4 with lows at 54 and 48 even just skipping Ivy Bridge and jump into Devil's Canyon we see more than a 20 FPS gain jump in another generation we land on these 6700 K is 1 10 FPS average with lows at 83 and 67 fps 0.1% lows these 7700 case stock CPU provides performance throughput of 112 that's 7 FPS average with lows at about 88 and 76 this gain isn't something you'll notice as a user between the 60 770 700k CPUs but it's noteworthy nonetheless we're still looking at a jaunt to about 2.5% before either hitting other barriers or running out of the overhead provided by the higher clock rate overclocking gets us another one FPS or thereabouts but that's it in the immediate future once CES is over and that's this week we'll be revisiting the i-5 the i3 and adding FX and Zen CPUs to the benchmarks so definitely stay tuned for that and we're hoping to add a few more games as well like Total War which is absolutely cpu-bound that'll be in there for those benchmarks and maybe some more unique things as we roll but for now blender and these games have us covered for the 7700 K which proves to be an incredibly easy overclock or that much is for sure it's hitting 5.1 gigahertz easily on the gaming pro carbon with about a 1.3 7 voltage if you let it auto control but I could definitely get it lower at least a bit by doing manual control and we'll have more info on that and the motherboard reviews which are going live today and later this week performance overall lands us around where we'd expect from a new generation i7 out of Intel and that's a couple percent in games meaning that owners of last generation the 6700 K should probably just stick with that Intel crosspoint obtain an h2 65 support I certainly value ads with the new platforms but unless you have a hard need for one of those or these small improvements we've shown it's best for the 6700 K owners and maybe even 4790k owners to stick with what they've got thermals are certainly a huge challenge for the KB Lake 7700 kcp you you'll want to check the motherboards that you buy to make sure that BIOS isn't unnecessarily blasting voltage with the CPU like the gigabyte board we tested which was pushing 1.3 to 1.4 volt jest with these stock configuration totally unnecessary that was with Auto voltage out-of-the-box the age of the hyper 212 for something like an i7 as a lazy cooler when not overclocking is also over the case cues we tested two of them even with voltages around 1.18 on the msi board we're still hitting around 70 C with a crack in X 60 to 280 millimeter liquid cooler and that's with the turbo clock fixed to 4.5 gigahertz so we were operating at that clock rate for the duration of the test it was a heavy workload test yes but these CPUs are definitely hotter than previous ones and if you're planning to be an i7 owner it's probably time to move on from the $20.00 CPU coolers at least get something that's a little bit better but again there's a lot more thermal discussion in the article link the description below so if you want to see how it performs in say blender benchmarking which is a bit different than prime95 with large F of T's you can find that in the article and that will give more of a real-world look at a high workload scenario for temperature performance and the sort of end-all here is the same as with most of our Intel flagship reviews for the last two generations now which is basically if you've got the previous generation or even the generation before that in the case in this case I'd be the 47 98 6700 K it's not really worth upgrading unless you need one of those features that was previously mentioned like obtain support cross point things like that where you really want something fun to overclock because the 7700 K is a damn good overclocker but not everyone wants it for that reason for anyone on older generations Sandy Bridge especially Ivy Bridge although wasn't tested would fall between Sandy and Haswell architectures that we did test so we can extrapolate the performance anyone using those to sort of bridge platforms would probably do well to upgrade at some point whether that's to a 7700 K or Zen of course we'll have to wait until the benchmark then but for now the 7700 K is a big gain over Sandy Bridge the 2600 K that much is for sure for gaming that is especially true but if you're looking at something more recent then gaming benchmarks are only given a couple percentage points in performance improvement before running into either other walls or just running out of the headroom provided by the higher clock rate on this cpu versus the six 700k which is again a couple hundred megahertz so that is all for this review as always patreon link in the post roll video to help us out directly links to the description below for more information if you want to read the review with additional info on thermal benchmarks and other things subscribe for more I'll see you all next time you