Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

Watch Dogs 2 CPU Benchmark - When an i7 Matters

2017-02-20
it's rising around the corner we wanted to publish a full CD benchmark of watchdogs 2 in our test course as we've recently found the game to be heavily thread intensive and responsive to CPU changes more than anything else a game even posts fairly sizable games for some overclocks like on the i-5 2500 K and establishes a real-world platform of one CPU choice it matters that's not always easy to find it is also easy to bottleneck GPUs with watchdogs to making it particularly relevant for CPU benchmarking we've already published reviews of the i-270 250k the i-5 7600 k and the i7 7700 k and revisit of the high five 2500 k most of those contain watchdogs to data we've been iterating as we go because it's a good game for benchmarking CPUs and further published a CPU optimization guide for the game so if you're curious about in-depth granular settings configurations for wash dogs 2 or just games in general because it does kind of apply elsewhere check that guide out but in that testing that perhaps unsurprisingly we found that geometry and extra detail most heavily impacted CPU performance will put some of those charts back on the screen now we need to become a CPU limited in watchdogs 2 geometry and extra detail should be at first on the list for reduction as shown in these charts they are the most impacting these settings are almost entirely CPU bound and that makes sense since increasing the poly count means we're also increasing the primitives that require draw calls to be made between a CPU and the GPU other settings tends to be more GPU bound and if you're not on an i7 already it's unlikely that you are GPU bound with watchdogs to the game is so heavily thread intensive that it really paves the way for CPU benchmarking for the year we've even noticed that watchdogs to it seems to care more about threads then a frequency so that's not to downplay the FPS increase we saw on the 2500 K when overclocking the 4.5 gigahertz but that's getting ahead of the content for today's benchmark we are going to be running through our standardized benchmark course and in doing so we'll be looking at 1080p with the high preset and its performance with the various CPUs with this course we're using mostly a standardized bench the G P remains fixed the power supply's SST all this stuff and you can find all that defined in the article below in the testing methodology section as always if you're curious about what's used but there are of course some differences with things like memory because as you change platforms you change the memory specification or compatibility from one motherboard and architecture to the next so that does change but again article below for all of that if you are really curious about it also defined below are the testing methods and these at 1% and 0.1% low values which most of you know at this point so probably don't need to look that up but it's there if you need it and there are some other notes from technician Patrick lace and as well who performed this testing now before you ask why didn't you overclock X why did you not overclock insert CPU here everything else you've done is now irrelevant the answer is because we do that incrementally so I answer this my comments every time but just put it in the video it is literally two times the amount of work to overclock every CPU there is a one to one gain and the amount of work required and there is a point as a business that it becomes unsustainable to perform every single type of test so it's done incrementally if we don't have overclocked for the thing that you want I'm sorry it'll probably be added in a revisit that's coming up likely sooner than you think especially with the rise in reviews inevitably around the corner for the new tests we're focusing on just getting a comparative benchmark across most of the Intel lineup while also adding in an dandy CPU as we work toward revention the FX series for the inevitable arrival again of Rison this is the first MVC view we've benchmarked this year been a while since we've done them so the FX 83-70 is making it read a beYOU on the watchdog 2 chart all CPUs tested performed near 60 FPS average as shown in the data here but some dips below in averages or in 1% low and 0.1% low metrics interestingly every single i7 CPU on the bench evenly aged i7 2610 Sandy and it's above 54 FPS or greater at their one percent load in terms of averages the 2600 K managed to maintain 74 FPS average with the newer 6700 K and 7700 ka CPUs reaching an obvious limit around 113 to 114 fps none of the i5 CPUs can break 85 FPS with C 7600 K coming closest and we're clearly encountering a CP bottleneck on our GT X 1080 FTW with the i-5 in this game we're also seeing the i3 CPUs dragged behind but the overclocked 73 50 K does well to keep up with an i5 4690k stock CPU landing at 68 FPS average and was the same at closely timed lows as other nearby CPUs sky lakes I 3 on the bench the i3 is 6300 at stock frequency sits at 54 FPS average with 41 FPS 1% to lows this is the slowest CPU on the bench and is behind both the i5 2500 km 2011 and the AMD FX 8350 1.7 FPS average and 46 FPS 1% load note that the 83-70 is technically an eighth core chip but keep in mind that Intel and AMD have very different definitions of what a core is a core on Intel is not equal to a core on AMD so if you're wondering why the ATT 7 doesn't perform the same as my 7 the answer is architecture with the AMD FX series we're looking at two integer units sharing an FPU each which would mean that performance is in optimal and games that rely on these types of floating-point calculations which is a lot of them there are other changes too of course cache performs differently prefetching all of that stuff performs differently we'll revisit the age of T 17 more depth eventually as we ramp into the rise in review or check back to the rise in review for an overview of the news and architecture overall this tells us that watchdogs to is an interesting and good benchmarking platform for CPU reviews because again this thread limitation is not a common thing with games that are CPU limited we often see that they are more frequency tied than anything these I 7 CPUs don't really get to stretch legs like they'd get to do in watchdogs to battlefield one for instance shows these things a lot closer together to the point where I sevens don't feel like they matter so much at least not unless you want 144 FPS or something like that then the extra few frames matter to anyone who is in that situation but that's not normally the case so this is a unique game for that reason we're keeping it on the TV bench for that unique and interesting reason it's a real-world game that actually uses threads and these are not completely absent in the market but they are rare so that's a good reason to include it now GTA 5 we recently ditched from our CPU benchmarks the watchdogs to kind of replaces it the reason for ditching it is because there's a huge game engine issue we're at about 165 to 187 FPF seems to be the magic number it is massive stuttering so as we approach these higher skew CPUs you see stuttering and performance metrics that don't look at all representative of the hierarchy for the CPUs the pricing or the performance normally that's all explained in the previous video where I worked with Patrick Latham who worked on this benchmarking to talk about the GTA 5 issues if you're curious you can learn more there but that's been removed from the benchmark because of that issue and because of the natural need to limit to 1080p and things like that to create a CG bottleneck so while 60 replaces it and this helps us set forth the expectations from you all as to how the game should be performing in the future next content coming up we're getting the AMD CPUs reacted to the bench including FX 83-70 which is here now the FX 6300 is on the to-do list and we'll be adding the KB like Pentiums pretty soon as well because I know everyone's interested in those and again Rison yes I know we all want the benchmarks don't have them yet but stay tuned for that subscribe for more as always link the description below for the full article testing methodology all that stuff patreon.com slash gamers nexus if you'd like that helps out directly or click the link in the post roll video thank you for watching I'll see you all next time it's been a while since we've had it do
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.