Gadgetory

welcome back to harbor unboxed today we have many benchmarks I know you guys always enjoy that last weekend I did look into whether or not NVIDIA GPUs were gimping AMD rising performance as alleged by a few viewers some were claiming that I should retest with the fury ax to see if horizon performs better then say it did with the gtx 1070 now typically i test at 1080p using quite high quality settings and throw in something very high-end like the titan XP as this removes at least for the most part the GPU bottleneck this isn't true though for the gtx 970 or fury x however i didn't want to change up the testing method for that video if nvidia where indeed hampering risings gaming performance it would clearly show as the fury x would pull ahead of the gtx 1070 instead we ended up seeing what you would expect to see the fury x was slower than the gtx 1070 and at no point indicated there was any kind of gimping going on still in that video we saw a couple of really interesting things invidious DirectX 12 performance has greatly improved and as a result under realistic conditions the GTX 1070 looked great in relation to the fury X when testing with games such as ashes a singularity escalation Deus Ex mankind divided and hitman for example that's that we found some interesting results in rise of the Tomb Raider when comparing the r9 295x2 to the Titan XP in fact there was some kind of intense bottleneck going on here on the horizon 7 platform with the 1800 X CPU as the Titan XP was limited to GTX 10 70 levels of performance this limitation either seemed to be imposed by the arisin platform or Nvidia Zone display driver the next step was to take these graphics cards and test them at a low resolution with lower quality settings to remove the GPU bottleneck so this is what I've done well sort of a good number of you requested that I use the gtx 1060 and RX 480 for this testing so that's what we've done and they should be powerful enough to get the job done at the proposed quality settings those settings being a resolution of 1280 by 720 using the lowest quality presets available I'm well aware these aren't settings gamers use and while that isn't the point of this test in total we have 13 games 5 of which were tested using both DirectX 11 and DirectX 12 the latest AMD and NVIDIA display drivers were used and once again we have the 7700 K and 1800 X both of which have been clocked at 4 gigahertz please be aware I'm not trying to show which is the fastest CPU here so cool your jets over the KB Lake down clock again that's not the point of this video and I won't be addressing comments relating to this both CPUs have also been tested using ddr4 3200 memory I have some nice new CL 14 stuff from g.skill which works great with the Rison processes okay so I think that's about everything you need to know for now let's start with the titles that support both DX 11 and DX 12 looking at rise of the Tomb Raider we first see that the 7700 K was just 5% faster than the 1800 X when both are paired with the gtx 1060 that said though if we compare the minimum frame rate the 7700 k is 14 percent faster now moving over to the RX 480 results we see that for the average frame rate the 7700 K is now 9 percent faster than the 1800 X the minimum also extends slightly to 18% so what this means is using dx11 Rison actually performs worse in relation to the 7700 k we see our X 480 now changing to the DirectX 12 API we find a very different ball game starting with the gtx 1060 results we find the 72 700 K is 14% faster when comparing the average frame rate swapping the GeForce graphics card for the RX 480 and now the 7700 K is just 7% faster so you could argue that we might be running into a GPU bottleneck with the 7700 K here at 172 fps the fact that there's an 18% variation in minimum frame rate performance using the rx 480 and just 15% with the GTX 1060 certainly does suggest this for those wondering these results are based on the can benchmark that's it I did take figures from the geothermal valley and they do line up with what I was seeing from a custom pass using present Mon I can include those results later if need be but I'll probably wait till Vega or ice before touching on this again moving on we have total war Warhammer and using DirectX 11 we find the 7700 K does slightly better with the GTX 1060 while the 1800 X was a fraction faster with the RX 480 these aren't exactly decisive margins though so let's check out the direct x12 figures well here we see that risin looks significantly more competitive when using the RX 480 this is the most extreme example we've seen yet the 77 RK was 28% faster for the average and 27% faster for the minimum when using the gtx 1060 and this certainly explains our Titan XP results seen previously moving to the RX 480 and now kb lake is just 9% faster on average and just 5% faster for the minimum it's a seriously huge reduction in margin though I should note that the minimum frame rate was slightly higher with the gtx 1064 the horizon processor the division was tested using dx11 and DX 12 but first let's check out the DirectX 11 numbers using the GTX 1060 the 7700 K was 11% faster for the average and 7% faster for the minimum swapping to the RX 480 the average grew ever so slightly to 12% while the minimum blow out to 12% so when using dx11 in this title Rison actually compares better using highway from the Green Team that said we find a drastically different story when moving to the DirectX 12 numbers once again using the GTX 1060 the 7700 K was faster than the 1800 X this time we saw 14% for the average and 11% for the minimum drop in the RX 480 and we find the 1800 X match in the 7700 K with roughly the same performance that's a pretty interesting result right there but let's move on and see what else we find Deus Ex mankind divided that provided some very interesting results indeed and we're going to look at these a little differently what's interesting to note here is that the 7700 K with both graphics cards and the 1800 X with the our X 480 all provided much the same performance regardless of the API used so focus your attention on the 1800 X with the gtx 1060 now let's transition to the dx12 graph see that if you go back and forth you can clearly see that it's really just the 1800 X with the gtx 1060 that changes by any kind of significant margin here this might make it difficult to claim that this is an nvidia driver issue given that the 77 hurricane doesn't suffer the same decline of performance of course the 1800 ex works perfectly fine with the RX 480 so this could point to an Nvidia driver issue and I guess I'd say it's down to invidious DirectX 12 implementation offloading items to the CPU in a way that's fine for Intel but not optimized Verizon moving on ashes of the singularity escalation is an important game to include as it's been recently updated to better utilize the Rison processes not only that but once again we have some interesting results to check out here we see that regardless of which graphics card you use the 1800 X is faster than the 7700 K now please note I'm comparing the average framerate figures from the normal and heavy batches from the very intensive built-in benchmark frankly this can benchmark does a much more extensive and better job of measuring in-game performance than I could so we see that when it comes to the minimum frame rate the 1800 X and 7700 K are very evenly matched that said for the average framerate the 1800 X was 8% faster with the gtx 1060 and 11% faster with the RX 480 not really a significant difference either way certainly margin of error stuff so I think it's safe to say that Rison is being represented correctly with either GPU battlefield one has a 200 FPS frame cap as you can see the gtx 1060 had no trouble reaching that with either processor the 1800 x does slip behind with the RX 480 but for the most part the results were pretty uneventful battlefields DirectX 12 performance is still a bit sketchy so I skipped that testing for now there isn't too much point reading into the firaon results using the 1800 X we see that the gtx 1060 and our X 480 deliver much the same performance again certainly margin of error stuff so whatever reason we are hitting a CPU bottleneck with the 1800 X and this allow the 7700 K to deliver almost 30% more performance with the GTX 1060 which obviously doesn't seem right summarizing optimisation needs to be done here that said under any kind of realistic conditions this is primarily a GPU bound game so a bit of a non-issue f1 2016 is a surprisingly good game for testing a CPU performance here we see when using the gtx 1060 that on average the 7700 k is 15% faster-moving CRX 480 though the margin is reduced to 9% that said don't jump up and down just yet the RX 480 is limiting the performance of the 7700 K here and a forwarding 1800 X the ability to catch up so this certainly isn't evidence of Verizon performing better with an AMD GPU Grand Theft Auto has been included because I know you guys want to see the results so what do we see here using the gtx 1060 the 7700 k was 9% faster on average and 6% fastest the RX 480 so again we are hitting a possible GPU bottleneck looking to the minimum frame rate this certainly seems to be the case as the 7700 K was 5% faster with the GTX 1060 and 6% faster with the RX 480 not much in it though so fair to say you are going to extract the maximum performance from either CPU in this title using either an AMD or Nvidia graphics card another title included at the request of you guys is the Witcher 3 Wild Hunt here we see once again at the RX 480 is limiting the performance of both CPUs and we once again have to turn to the minimum frame rates doing so we see the 7700 K is 8% faster using either the gtx 1060 or our X 480 back to some more interesting results mafia 3 always seems to deliver those now as we've seen in the past with the Titan XP testing the 1800 X is able to best the 7700 K in this title even when overclocking the kV like CPU to the max here with the GTX 1060 the minimum frame rate was much the same though the 1800 X was 5% faster on average however if we test with the our X 480 quite shockingly the 1800 X falls behind and the 7700 K is now seen delivering 6% more performance on average and 15% more for the minimum so the our X 480 delivers quite there below for AMD zone 1800 Xu capping things off we have our final batch of fascinating results now these figures are very very shocking indeed and really amplify the effect just seen we're testing with mafia 3 using the GTX 1060 the 1800 X was just 6 percent slower than the 7700 K and given the difference in IPC performance that's a great result however once we swap over to the rx 480 something goes horribly wrong for whatever reason the 18 under next fall's 25% behind the 7700 K or 33% if we look at the minimum frame rate I'm honestly not sure what's going on here it's certainly very strange that the 1,800 X looks so competitive with the 1060 but so extremely underwhelming with the our X 480 so are the NVIDIA GPUs limiting risings gaming performance well once again I'd have to say no they're not yes there were a few dx12 scenarios where rosin did perform better in relation to the 7700 K when paired with the RX 480 but if you're going to use that as evidence to somehow support the argument that video is handicapping rising performance then you would have to also acknowledge that AMD is handicapping risings dx11 performance at least in some titles so once again it seems everyone has it out for Eisen seriously though we found some interesting stuff here and it's things I'll be keeping in mind when benchmarking the rise in 5 CPUs in games for games such as Deus Ex mankind divided or total war Warhammer for example I'd be sure to include an AMD GPU in the testing for now though I'll probably just drop these games until we have a more powerful single GPU graphics card from AMD likewise I won't be testing with rise of the Tomb Raider though I never had any intention of using this game anyway as a side note I have to admit it really annoys me how much fuss was made over low resolution testing recently it's quite obvious to me at least anyway that this is how reviewers should test and yet the backlash was quite extreme granted it was just a vocal minority making all the fuss and based on the feedback I received from regular viewers it was quite clear you guys understood the importance of removing a GPU bottleneck for testing CPU gaming performance the fact that we had to test at 720p using low quality settings just to reveal why Rison has been struggling in DirectX 12 titles when using the Titan XP or other high-end Nvidia graphics cards just goes to prove this point additionally you can't blame reviewers for using the most powerful GPU available for testing CPU gaming performance it's just common sense anyway what we found here really only impacts DirectX 12 performance and for the most part I had focused on DX 11 games for my rise in testing anyway that's all for this one guys it's certainly very interesting stuff I'm keen to revisit once we have a more powerful GPU for AMD once Vega arrives speaking of interesting stuff Kim's review of that insane 240 Hertz a soos ROG monitor will be on the channel tomorrow so keep your eyes out for that one and following that the next day we will have an in-depth look at rhizomes memory performance then on Monday you can look forward to another video from Tim as you reviews Kingston's HyperX revolver s headset and finally that would be followed by our official horizon 5 coverage so that should keep you guys busy I'm your host Steve see you again soon