Gadgetory

surveying Vega 56 back with the power mod the 242 percent power offset this card doesn't have any stand of ability that we had previously if you remember the power offset it blasts the card with way more power than any stock or aftermarket configuration will allow for and in doing so allowed the vagin 56 card to surpass Vega 64 when we tested it originally the test today is whether or not it can beat the r-tx 2070 with the power mod because it's 370 to $400 for this thing as 500 to 600 for the 2070 so can we mod this to a point that it surpasses the r-tx 2070 before that this video is brought to you by US and the limited-edition foil Graf logo shirt this four color foil shirt is the iconic GN Graf logo with average one percent and point one percent bar colors it's printed on a soft high-quality and custom made 100% cotton shirt and is available on stored on Karen's Nexus net until stock runs out once it's gone we will not be making more of these shirts we sold out within two weeks of our previous limited-edition shirts so click on the link below to preorder now here's what we're doing if you missed it originally there is a mod you can do not even a difficult one for the Vega cards and that mod is to change the registry and you can do a registry entry change for power-play tables on these cards that allows for the power target offset to exceed the initial 50 percent stock up to whatever you want we set 242 percent we work to build Zoid on this probably about a year ago and they came out and doing so a year ago allowed us to fairly easily match Vegas 64 and even beat it when it was overclocked in some scenarios you could do the same thing to Vegas 64 so it's not head-to-head but you get the idea and then there are other mods like just straight BIOS mods be BIOS mods where on the card you could more or less unlock some of the frequency voltage targets that you get on Vegas 64 though we didn't apply that here because we weren't a hundred percent positive if the BIOS options we found online we're going to be trustworthy and also we have to send this card back anyway so what we did instead is just use the powerplay target and that allowed us to get so much power into Vega 56 that it can be the RT X 2070 in some scenarios the only trouble is the power consumption is massive so we have some charge for you today that are going to be gained for forints and then after the gaming performance you can see what the power cost is and I will note this as well if you want to do this mod on your own technically you know if it works then it does it does save a bit of money for the same performance level ignoring the power aspect what we don't know is does pushing an extra couple hundred watts into the core kill it over time and if it does how quickly so we wouldn't really recommend doing this necessarily because we don't know the long-term impact for your hardware but it's still a really cool experiment to see if you can force Vega 56 in the beam of the r-tx 2070 which is significantly more locked down you would have to do hard mods to get a similar over power target that we did on this with just a registry entry so let's get into it will do the charts as always there will be a link in the description below for the full testing methods and the article if you prefer that but let's start with Sniper Elite Force Sniper Elite 4 is one of the best modern API implementations and leverages asynchronous compute command queuing to use Hardware more efficiently without getting stuck waiting for as many synchronous instructions at 4k high we found that the stock Vegas 56 red dragon card performed around 55 FPS average with the stock our TX 2070 at 64 FPS average that's a market lead but one which we can take away by allowing an uncapped power target of 242 percent on the Vega 56 card and by pulling down significantly more power we have performed the r-tx 2070 by well it's within margin of error they're about the same really at sixty three point eight FPS versus sixty three point six FPS average that is well within our error margins but we can still call it a victory we've minimally managed to match the our TX 2070 with Vega 56 the latter of which costs about 370 to 400 while the our TX 2070 cos 500 to 600 for reference Vega 64 ended up at around 62 FPS average one stock and so we're able to perform around that level or passed it with 56 as well everyone 2018 is next our 4k ultrahigh this one places the reference Vega 56 at 49 FPS average when stock compared to the stock our TX 2070 at 60 FPS average that's a lead of about 22% the power color red dragon model operated faster though at 52 FPS average closing the gap marginally already and when overclocked in power colors of Vega 56 card with a massive power offset of 242 percent maximum power with about a 1710 megahertz plus or minus 10 on the core we surpassed the r-tx 2070 card by about 5.2 percent reaching sixty two point seven FPS average that's a tremendous jump of 28.5 percent over the stock Vega 56 card that's the reference model or 20% over the red dragon stock card possible only through this mod for perspective the Vega 64 card performs head around 59 FPS average and left the stock settings are DX 2070 does outperform 56 when both are overclocked to be fair reaching a new height of 60 FPS average but this gap is significantly smaller than the stock the stock tab of the RTX 2070 and making fifty six cards originally even though the RTR 2070 is able to outperform Vegas d6 with an overclock we still have to commend the performance of the Vega 56 part further mods like Tobias might gain us an additional overhead on the memory primarily and again equalize the score but at this point the power is way over 27 usage but we'll look at that after the game benchmarks at 1440p the stock power color of a given t6 red dragon landed at 88.5 fgs average a market gain over the stock vacant 56 reference card it's still significantly behind the stock r-tx 2070 which performs at 102 FPS average leaving the vacant 56 red dragon to be outperformed by about 15% stock to stock for reference new vegas 64 Strix ran out 96 FPS average pushing Vega 56 with our power play tables mod resulted in a frame rate that again surpass the r-tx 2070 card the modded Vega 56 push is 104 FPS average with significantly tighter frames I'm consistency than the r-tx 2070 a result of never running zap our perf cap and it pulled the head by about 2% that's a remarkable feat for a card that cost so much alas considering the r-tx 2070 cost 500 to 600 dollars again still overclocked in the RT X 2070 allows us again on by another 10 FPS to be fair but that's becoming difficult to notice for the end-user as frame rate is already so high but also power is more noticeable and again we'll talk about that soon we're not saying this is the best approach given the insane power draw and the mod we're doing but we are saying that it's pretty damn cool to be able to push a cheaper card this far ahead with some tuning for a perspective here's a look at frame time consistency at 1440p remember lower is better but more consistent is better than lower this kind of plot is far more valuable than frame rate as it gives you a true representation of what you're seeing we're looking at frame to frame interval or how long it takes each frame to be drawn after the previous frame Thank You misty six is significantly better in this particular title than the Nvidia devices which may just come down to game level optimization it's not always true this is true for both these stock and overclocked 2070 cards where you see that the increased frame time consistency on Vega 56 is represented by its lack of spiky lines that you're seeing on the 2070 and that's something that can be seen in gameplay to some extent as well finally at 1080p the Vega 56 card measures at 104 F he has average on reference or 115 FPS average for the power color model this allows a power color card to reach 1070 TI levels of performance although it's low frame time performance is measurably better the stock r-tx 27 he ends up at 136 FPS average and the massively overpowered rx fifty-six red dragon ends up at 134 FPS average it's not bad but does show that we're running into other limitations at 1080p a 4 Vega 56 frames I'm consistency it remains superior for the Vega cards here and somewhat noticeably so GTA 5 is less exciting at 4k the red dragon 56 card ran at 35 FPS average with the RDX 2070 at 48 fps average that's a lot of ground to gain and it just couldn't do it even with the power mod the Vega 56 card ended up at 40 FPS average still behind the 20 70s 48 FPS average over a clock in the 2070 out of fairness the gap widens to 53 FPS average looking instead at 1440p the Vega 56 card pushes 80 FPS when modded improved over the stock Vega 56 red dragon score of 70 FPS Ori TX 2070 still holds a lead at 89 fps though with the overclocked 2070 at 98 FPS average no hope for Vega 56 to pass the 2017 this older dx11 title let's move on to the next game Far Cry 5 is up next and at 4k high settings we measured the AR Fagen 56 red dragon stock card at 41 FPS average right around the 1070 TI and behind the 1080 FP TW modern with the power offset vacant 56 surpasses the AC Vega 64-46 FPS average by a couple of frames per second landing at 49 FPS average on the red dragon this allows power colors Vega 56 to beat the RT X 2070 by marginal gains at 49 fps verses 46 fps while still maintaining tight frame time consistency to be fair though the arc TX 2070 is equally smooth and its frame throughput in this game overclocking the 2070 does allow it to surpass making 56 once again but by similar margins to making 56 is passing of the stock 2070 it's at 51 FPS average vs. 49 again this is pretty damn impressive for the Vega 56 card although it is definitely way out of spec at 1440p the vega 56 red dragon measured at 74 FPS average immeasurable improvement over the 69 FPS average of the reference card the RT x 2070 ends up at 84 FPS average and is just barely beaten by our modern vega 56 again we have to be happy with the results of this mod power is another story one that's pending momentarily but the performance is impressive at 86 FPS average the vagin 56 mod we're looking at a significant uplift over the stock performance as expected the 2070 s overclocked allows it to retake the lead at 93 FPS average at 1080p far cry 5 becomes a CPU bound at the absolute high end of a dodge is this convocation with the parts we're testing today the Vega vent e6 reference card is decently behind the power color model 1492 FPS average instead of 104 power mod in bacon with d6 gets us to 120 FPS average a remarkable climb of just over 20% the our TX 2070 is within margin of error plotting 118 FPS average once again we've managed to minimally tie or marginally outpaced the our TX 2070 with modern Vega 56 and also once again if you're willing to mod the Vega 56 card you're probably willing to do so at least a simple overclocking area on the 2070 this overclock on the 27 he gets it to 128 FPS average but the gap has gone much smaller than previously shadow of the Tomb Raider is last it's a significantly newer title that is awaiting full RDX support still at 4k on high settings we measured the Vega 56 cards at 35 FPS average and 38 FPS average for the stock and overclocked variants respectively this is a letdown it allows even Vegas de forage remain in the lead changing to 1440p we see the same performance stack Vega 56 ties with 64 which is good for other reasons but it does not meet the 2070 we've got a mixed bag for performance overall so here we go then we're getting into the power measurements here and for power we logged power during the entirety of our benchmark passes and this is important we're doing total system power here instead of current clamping the cables because there's power through the socket as well and AMD and NVIDIA have different policies for how much power they draw through the PCIe slot and that is why we're doing total system draw because we don't have a better way to clamp the PCIe slot so we're going to a total system draw here but it's logged and we synchronized the runs so that you can compare run to run the different SKUs of cards and see how they do and that should be pretty useful here so also all voltages on the board the motherboard are completely controlled so there's basically zero variance there it's it's all in the video card at this point for ashes of the singularity you can see our power consumption across multiple passes including the peak total system draw at the most intensive parts of the pass the Vega 56 mod pushes up to about 600 watts total system draw during this stressing part of the benchmark while the stock making 56 card peaks at around 400 watts we're pushing about 200 watts more with the power offset in the mod that's a huge amount of extra power going to the core as long as the vrm is well cooled and assuming the vrm is a decent one on the Vegas d6 PCB in question the vrm should be fine what we are unsure of is how much the Vega GPU can take for how long a period of time for this reason we don't necessarily recommend this mod it could kill vrm components if not done carefully and on good PCBs with these in cooling and we don't know what will happen with the GPU after a long period of time with this much power for the RT X 2070 that's around 300 to 350 noit's total system draw during the heaviest load of the tests because the card is so peril emitted at all times thanks to Nvidia including only a 14% power offset on the $500 model were testing the maximum power draw on the OC test never exceeds 357 watts on average it is closer to a 330 watt peak draw the rest of the games follow a similar pattern so we'll leave that here the r-tx 2070 clearly has a power consumption advantage when stock or overclocked with the overclocked version drawing slightly less power than Vega 56 although it is largely insignificant here at 397 peak versus 357 peak it's the modern 56 that clearly loses an all power consumption needs a period unless your goal to outperform the r-tx 2070 also includes heating the house and again while this isn't something we'd really recommend you do it's kind of impractical because we don't know the long-term impact of pushing an extra 200 watts into the core the V RMS can pretty much handle it if you don't buy a garbage card so that makes it partly impractical it's still really fun though the cost-benefit is convincing on 56 and mostly it's just fun it's a lot of overclocking so you get way more Headroom we're pushing over seventeen hundred megahertz on the core and it can go higher we in the past we've pushed it higher with liquid so in the past we did in excess of 1720 approaching seventeen fifty megahertz if I remember correctly for a liquid-cooled vacant fifty-six card so if you get a really well cooled 56 you have nothing to worry about on the vrm it's just that core and will perhaps leave build Zoid to speak to that if he has any thoughts on it but other than that vague 56 is capable of beating the r-tx 2070 stock and it can get pretty close to the overclocked 2070 when you blast it with over 200 watts more power than the 2070 and 200 watts more than the original Vega 56 which was already drawing more than the 2070 so whether or not that matters to you is part of the equation but it's more of the longevity either way though it's a lot of fun to do and it's cool to see Vega can perform so well right now is actually this goes back to our review of the 2070 where suddenly now that Vega is available again with the death of mining basically Vega is looking like a good option 64 is not a card we've ever really liked because you can do this 256 and get up to 64 levels of performance but it's still it's it's about the same price as the 2070 s on the low end sometimes it's cheaper so it's worth looking at the numbers and figuring out if you want to consider an D at this point we've been talking about exclusively Nvidia for months now because everything's come out at the high end and AMD gets no reference at the high end 2080 28 ETA 1080i doesn't matter and he does not compete there whatsoever for performance so Nvidia dominates that market and because they dominate that market you kind of start forgetting that AMD exists in all markets so anyway the 2070 did decently at $500 we have a review on it if you're curious but this is just to say you know maybe remember that Vega exists we did like 56 we never liked 64 though the price proposition was not there but if the 1070 TI start going away and were left with just the 27 he's done suddenly it might be worth considering so that's it for this one thank you for watching as always go to store documents access net to pick up our limited-edition Graff logo foil shirt once they're gone they're gone we're not going to reorder them and you go to patreon.com/scishow and exit ops that directly there that you were watching I'll see you all next time I think that was pretty good that's pretty convincing