Gadgetory

welcome back to harbor a box for another Vega related video this time we're overclocking and under vaulting to see what kind of impact this has on Vega 56s performance shortly after my 32 game benchmark comparing the entire Vega line up to the GeForce 10 series I started getting spammed with messages from viewers asking me to do some under vault testing as they suspect this really helped to improve Vega sufficiency in fact the graphs I was sent showed up to a 22% increase in frame rates while reducing total system consumption by 20% what kind of magic is this this made little sense to me but I was keen to see what kind of improvements could be made to Vega with a little tinkering so for now I've played around with the air called reference design Vega 56 graphics card to see if it can indeed become a GTX 1080 found edition killer or is that just a fairy tale at the very least we should get a good idea of what will be possible with Vega 56 want some more efficiently cooled board partner cards become available but before we get into the benchmarks and the settings used a quick shout-out to videoblocks for sponsoring this video so for those of you who aren't familiar with videoblocks they have one of the fastest-growing and largest stock video libraries on the internet with 3 million videos After Effects templates and motion backgrounds this includes the only contributor marketplace that gives 100 percent of the Commission back to the artists videoblocks is an amazing resource for anyone creating videos whether you're a professional or just someone looking to have fun and create some neat content they have plenty of footage to spice up your videos there are basically endless options and you'll have no trouble finding stock footage to suit your needs I myself spent how is just enjoying the nature section and importantly all of the footage comes with a royalty-free agreement so you can't get hit with copyright claims right now if you sign up you get a 7-day video blocks trial so you can try it out and get access to their massive royalty free of video library for free go to videoblocks comm /youtube or click the link in the video description below to start downloading and get seven days of videoblocks for free okay I'm done watching nature footage at least for now because it is time to fine-tune Vega 56 first up I started by reducing the voltage from 1200 millivolts down to 1017 millivolts and this was an 11 percent reduction in voltage this alone did reduce power consumption by around 10% but also reduced frame rates by 5 to 10% as well the next step then was to slightly raise the frequency from 1592 megahertz to 1600 and 12 mega Hertz and then increase the power limit by 50% finally the HBM 2 frequency was boosted from 800 megahertz to 930 mega Hertz and this is the highest frequency I was able to stabilize the memory at you'll see what kind of impact this had on performance in a moment then for the gtx 1070 fe card i raise the base frequency from 1500 and six megahertz just seventeen hundred and six megahertz and this resulted in a GPU boost 3.0 frequency of around 2000 and fifty megahertz the memory was also increased from 8 gigabits per second to nine gigabits per second finally the core voltage slider was maxed out at plus 100% and the power limit was set to 112 percent all right so with AMD's reference model and in videos found edition model both quite heavily overclocked how do they compare well let's go find it out first that we have battlefield 1 and here the GTX 1070 enjoyed a 15% increase in average frame rate once overclocked though this only allowed it to loosely match the stock vega 56 graphics card overclocking vega 56 increase the frame rates by a further 12 percent and this does indeed allow AMD's more affordable value graphics card to match the gtx 1080 at least for the average frame rate moving on we have the dawn of all three results and here the overclocked gtx 1070 boosted frame rates by just 7% vega 56 did slightly better as an improved the average frame rate by 9% and that placed it roughly on par with vega 64 and well ahead of the gtx 1080 so a solid result here for Vega it's not a massive 20% increase but we'll take it nonetheless Vega is already proven to be mighty impressive in dirt 4 and here we see once overclocked vega 56 is able to match a vega 64 with an impressive 1 119 FPS and that makes it much faster than the GTX 1080 and world's faster than the gtx 1070 meanwhile the GeForce graphics card only saw a boost of 6% once overclocked moving on the gains in Doom are fairly mild the GTX 1070 was 6% faster once overclocked and Vega 56 was 7% faster even so the overclock did allow Vega 56 to catch the stock GTX 1080 in this test so not bad but not amazing either that's it with frame rates in excess of a hundred and thirty FPS at all times I'm not sure how much it matters anyway the gains seen in Far Cry primal were mild at best - though better than those seen previously in Doom the gtx 1070 enjoyed a 9% bump in frame rate while vega 56 did hit double digits with an 11% increase now pushing over ninety FPS on average vega 56 is quite impressive though not a gtx 1080 killer here when testing with fir on a-- we did see a very healthy 15% boost in performance for vega 56 and it is now delivering air-cooled vega 64 like numbers the gtx 1070 wasn't quite as impressive and overclocked it did fall further behind vega 56 both the gtx 1070 and vega 56 GPUs saw a measly gains in Mass Effect Andromeda around 5% more frames on average were seen here overclocked the margins between these two GPUs remained much the same and they did both turn in results that were lower than that of the GTX 1080 in this title moving on to the Mirror's Edge catalyst results the gtx 1070 was boosted by just 4 percent once overclocked meanwhile vega 56 spat out 8 percent more frames and was seen delivering vegas 64 light performance and wasn't much slower than the gtx 1080 here overclocking vega 56 does certainly help with performance in overwatch and while it trails the gtx 1070 in the margin is slightly reduced at least when looking at the average frame rate the 1% low results are still concerning but at well over 100 FPS it's probably not really a big deal testing with prey the results are much the same once overclocked though vega 56 is 1% low result is more impressive now and match that of the air-cooled vega 64 GPU overall we did see an 8% increase in frame rates of vega 56 once overclocked both GPU see a 6 percent increase in average once overclocked when we're testing with Resident Evil 7 and this was enough of a go 56 to match the GTX 1080 a stock GT X 1080 I should add but still an impressive result all the same the gain seen in Titan folds who were very mineral for both the GTX 1070 and Vega 56 GPUs if we'd seen more results like this though we'd have to conclude that it's really not worth the time invested to overclock these GPUs nearing the end of the benchmarks we have rise of the Tomb Raider and we see an 8% increase for Vega 56 and just 6% for the gtx 1070 certainly not mind-blowing results I'll admit but Vega 56 is now able to match the air-cooled Vega 64 graphics card finally we have Ghost Recon wildlands and again some more mild gains can be seen and they're not really worth getting excited over so then I'm gonna wrap up the FPS testing here and let's move on to check out the power consumption figures ok I can't say these results are at all surprising but given the graphs many of you are showing me it might surprise some of you so out of the box Vega 56 pushed total system consumption a little over 30% higher than that of the GTX 1017 certainly not a good result but since it was generally a little faster it's not the end of the world either however overclock the gtx 1070 sucked down just 11% more power and that minute still consumed a lot less than the stock vega 56 graphics card under vaulting and then overclocking vega 56 did certainly improve performance by anywhere from 5 to 15 percent depending on the title but it also increased total system consumption by an insane 35 percent that explains why the blower fan was screaming like a jet okay so based on that it doesn't really seem like under vaulting slash overclocking Vega really changes position against the competing g-force 10 Series products overall Vega 56 saw an eight percent boost in average frame rate from our tweaks while the gtx 1070 saw a seven percent boost so really much of a muchness there however probably more importantly while the gtx 1070 only saw an 11 percent increase in power consumption vega 56 saw a 35 percent increase so while we succeeded in making both faster we did make the gtx 1070 slightly less efficient while making vega 56 miles less efficient since the aim is to improve efficiency here at least I believe that's what many are getting excited about when it comes to Vega under vaulting it's not really the case there's no doubt room to improve on efficiency here at least slightly while keeping the performance of Vega 56 pretty close to what you get out of the box of course gains will vary from card to card though and I certainly can't see you slashing power usage by 20% while maintaining stock performance still if we look at the results seen in battlefield 1 Dawn of War 3 and dirt fall for example then there is a chance Vega 56 could come out well on top down the track of course it's yet to be seen but in terms of outright performance I'd bet Vega 56 winning out pretty comfortably in the long run as for efficiency I am Not sure that's a battle that can be or ever will be won but for most gamers it's probably not really an issue anyway that's gonna do it for this one I hope you guys found the results interesting slash useful and hopefully they will tie you over till we get some board partner cards on deck and finally a big thanks to videoblocks for supporting our work and offering viewers a free 7-day trial remember the link is in the video description so check it out I'm your host Eve see you next time guys