Gadgetory

welcome back to Harper unboxed you know what I was thinking just the other day there's something we haven't done enough of on the channel we really haven't done enough comparisons of him the geforce gtx 1650 and the Radeon rx 570 okay bitter sarcasm there I almost couldn't I couldn't really keep a straight face during that one we have done a stupid amount of gtx 1650 content over the past few weeks but every single test just seems to lead into another one and well there's been one test that you guys have been requesting ever since i initially tested the gtx 1650 and that is an under volting comparison with the RX 570 so take the RX 570 and then see if we can starve it down to GTX 16 50 levels of power consumption so this is a more for science type test we do those every now and then they're a bit of fun but yeah just to be clear this is not a buying guide and I'm not trying to find a way to justify recommending the RX 570 over the GTX 1650 we already recommended the RX 570 over the 1615 our day one coverage and so did something like 99% of all other tech media outlets so yeah this is just a for science test and I'm really just doing it at the request of you guys therefore I whipped out my gigabyte horas RX 574 gigabyte so along with a little copper insert in the back that looks really fancy but anyway I grabbed this guy and I started doing some tinkering so the aim here was to try and get this down to GTX 16 50 levels of performance so basically under clock this card till it's performing similar to this card and then tune the voltages and power target to try and get it well give it as efficient as possible while delivering the same level of performance so that was the aim and yeah that's what we're doing this video but I had a few challenges along the way because the asset quality of this particular horas rx 570 isn't great fact it's a it's pretty poor really so yeah I didn't win the silicon lottery with this particular graphics card gpu-z reports the asset quality beat is sixty five and a half percent that means it's only better than four-and-a-half percent of all our x5 70s in their validation database so that means it is a bit of a stinker but for this test I think that's actually good news at least from my perspective anyways it means whatever I achieve with this card will be achievable by almost all our x5 70s so I'd rather test with pourcel that can oppose to a model with an asset quality of I don't know 90% or better as that's not really representative of your typical rx 570 I did just recently buy a new 8 gigabyte rx 570 and it has a reported asset quality of 72 percent making it better than 30% of the RX 5 seventies in the gpu-z database so again not great silicon here but it is better than my original model anyway since all the testing for this video was conducted before I made that purchase I've just stuck with those results and I'd rather show a worst case scenario anyway so everything you're going to see in this video will be based on the gigabyte horas model using AMD's watt man I set the power target to minus 40 the voltage to 1010 millivolts at state 7 with a base clock of 12 hundred megahertz now if I left the power target unchanged the RX 570 was still very power hungry and I even found reducing it down a minus 30 had no impact on power consumption it was like no changes were being made it wasn't until I hit minus 40% of the power consumption was reduced and quite substantially so was the performance I was also able to get the voltage down to 980 millivolt but the system wasn't 100% stable here and would occasionally crash to the desktop at a thousand millivolts did appear to be 100% stable but for a little reassurance I left at 1010 millivolts I have seen reports of rx 570 owners who are getting down to the 960 to 990 millivolt ranging frequencies between 1250 and 1350 megahertz but they're also reporting an asset quality of over 80% so that was more yeah top quality silicon additionally I've also seen a lot of reports from rx 570 owners that could only achieve what my Oris model does so obviously just like overclocking your mileage will vary and under vaulting success is very much silicon dependent anyway at 1010 millivolts a minus 4t power tower GPU power consumption dropped down to 88 what's that means a stock rx 570 consumes 72 percent more power however even under vaulted the RX 570 was still consuming 24% more power than a stock GTX 1650s so this level you still couldn't run it solely from PCIe power still this is a massive improvement of what the rx 570 sees out of the box but how much impact does this reduction in power consumption have on performance to answer that I first fired-up farcry new dawn and here we saw the RX 5-7 he basically matched the stock gtx 1650 so here it was consuming 24% more power for the same level of performance not great but it's certainly not bad for a three-year old architecture and of course stock the RX 570 is 24% faster than the gtx 1650 though it does use twice as much power to achieve that performance uplift then with both GPU is fully unleash the RX 570 is still 23% faster in this title Forza horizon 4 is a strong title for AMD so it comes as little surprise that the under vaulted and overclocked RX 570 is able to beat the stock gtx 1650 in fact it even matched the overclocked 1650 and that actually makes the undervolt at RX 570 more efficient in this title then when we throw it efficiency at the window and overclock them both for maximum performance the RX 570 is 21% faster Resident Evil 2 as a more balanced title and here the undervolt at RX 570 was 5% slower than the stock gtx 1650 so for this game the GeForce GPU is far more efficient but if all you care about is performance at overclocking both to the max ease the RX 570 win by a 30% margin making it the vastly superior GPU in terms of FPS performance finally I took a look at the most recently released title it we've added to our battery of benchmarks and that title is World War Z here the undervolt at RX 570 was a whisker faster than the stock gtx 1650 but that still makes the GeForce GPU more efficient but this time we do find we're looking at maximum performance the overclocked rx 570 smoked the overclocked GTX 1650 by an almost 40% margin well there you have it is technically possible for the RX 570 to get much much closer to the gtx 1650 in terms of efficiency but unsurprisingly it can't match it after all it is a three year old product so if AMD were to make a Polaris based 75 watt Radeon it would be quite a bit slower than the GTX 1650 but I suppose if they did so and offered it at $100 it would actually end up being a better product again it's the price that's the ultimate issue here some of you think I flat-out hate the GTX 1650 and that's simply not the case I really do admire what NVIDIA has been able to achieve here with a 75 watt Capet is truly impressive the issue as I've plainly stated time and time again is that at $150 u.s. it makes no sense to buy it's unsuitable for low powered budget OEM pcs is it's just simply too expensive and in any PC with an ATX power supply the rx 570 is a vastly superior option while also saving us some money at the same time of course if there was a $150 low-profile version of the GTX 1650 then I could understand the appeal of that product for higher-end small form-factor pcs but short of that it's useless in my opinion and you know the second msi gigabyte or whoever else makes a low-profile version it'll cost around $200 making it an even tougher sell anyway not going to rehash that conversation any further I think really just the point of this video was to see how efficient we could make the Radeon rx 570 and as I said it was just a for science type test and really nothing more than that so yeah I found the results interesting hopefully you guys did as well and if you did you can give us a big old thumbs up that would be much appreciated you can also subscribe if you're not already to make sure that you catch our other content and you can support us on patreon because that allows us to buy graphics cards like the the eight gigabyte rx 570 that we needed for certain testing baby weight that's gonna do it for this one thank you very much for watching I'm your host Steve I'll see you again next time