Gadgetory

we finally got around to reviewing the RX 598 the XFX Fatboy it's got some thermal problems we'll be talking about that later but also not a whole lot of motive to read to review this quickly because it's a RiRi fresh there's the R X 480 then the 580 came out it's a refresh of the 480 it's a pre overclock the 590 came out which is a pre overclocked 580 which is a pre overclocks 480 so it's it's not that inherently exciting it's not new architecture it's there's not a whole lot new going on but it's still worth reviewing because even though it's a refresh as we saw with the 580 it might have a pretty good place in the market so the 580 despite being a refresh a pre overclocked one did turn out to be a decent buy at its price point and that's what we're gonna look at today with you or axe 590 specifically the Fatboy model that we have here and I will talk thermals power and gaming performance before that this video is brought to you by coursers vengeance RGB memory Corsairs vengeance RGB Ram uses pre-screened ICS for better overclocking Headroom and tighter timings courses Ram also has the benefit of wide region motherboard support by landing on qualified vendor lists for motherboard makers and trained wide compatibility between boards learn more at the link in the description below for the specs on this if you want to know I'm watch our our X 480 review from like two or three years ago and then watch the 580 review those are the specs it's a higher clock so we've got that covered beyond that pricing is probably the most relevant thing to notice or for perspective the 1070 TI cost about 360 - $450 depend on which model you're buying when you buy it where you buy it the non blower models the ones that are actually worth buying are closer to 430 to 450 us that's the 1070 TI the 590 the RX 590 costs about $260 for a model similar to this one I think this is 270 right now about the same thing no 270 to 60 so you're looking at 170 dollars more for performance that well we'll talk about that later but the 1070 Ti is a different class of performance not the best comparison that a better comparison might be something like the 1060 or maybe the 1070 and because the RX 590 does fall between them and performance in a lot of instances the pricing should also be between them so 1060 s these days if you haven't looked the the 1066 Giga Byte the real like the legitimate one the original 66 ka not to be confused with the other five that came out that's about two hundred ten bucks in the US for for one of those and they go up to 240 250 so not a ton of price movement but a bit since it came out and that puts it a little cheaper than this one or you get the 1070 for something like 300 plus or minus 20 on average 300 320 so that's where you looking at for the price stack up so let's go through the game benchmarks we'll talk about gaming FPS frame times talking about thermals power consumption and a bit more for the XFX rx 590 fat boy Sniper Elite 4 gives us a look at a well implemented DirectX 12 title with close to hardware programming including asynchronous compute support we'll start with our 4k results as those are the more complete and give us a good picture of scalability versus other hardware remember the idea here is that we can see how the RX 590 compares relative to other devices not necessarily to see absolute performance well look at other results with lower resolutions for that angle at 4k using Sniper Elite as sort of a synthetic test for low-level api's we see the XFX rx 590 fat boy runs at 43 FPS average one stock or about 46 FPS average one overclocked lowest for both results are good with 1% at 41 fps for the OC and 37 fps for the stock result this is indicative of fluid frame times something we'll look at in our frame time plot next the frame rate isn't where we want it but again that's all relative to graphics settings what's important here is that frame time consistency is there absolute numbers will rise with lower settings momentarily as for the rx 580 the real one the msi RX 580 gaming x not the data land card which is a 570 that AMD respawn is a 580 we end up at 39 FPS for the 8 gigabyte gaming X which is the same as an overclocked rx 570 ie dat'll and RX 584 clocking the RX 582 1450 5 megahertz gets it to 41 FPS average roughly tied with the stock rx 590 that's because an RX 590 is an RX 580 but pre overclocked so this all makes sense comparatively the gtx 1070 SC ends up at 48 FPS average with the gtx 1066 gigabyte gaming ax at 37 FPS average just to be clear that's the 1066 gigabyte not to be confused with the 1063 gigabyte with one fewer SM nor the 1065 the gigabyte and most certainly not the 1066 gigabyte with nine gigabits per second memory and finally and most definitely absolutely not the gtx 1060 gddr5 x6 gigabyte card just the gtx 1066 gigabyte nvidia has ensured that all future mentions of the gtx 760 must be accompanied by this statement of clarification the point is the 1060 is beaten by the 590 which is functionally an overclocked 580 which wasn't overclocked to our X 480 the 1060 with an overclock ends up at 40 FPS average within striking range of the RX 590 but still outperformed by 7.8 percent the gtx 1070 outperforms the stock rx 590 by about 12% with a 1070 TI at 57 FPS average or leading by 32% when comparing only stock to stock the frame time Delta is about seventeen point six seven milliseconds bursts of 23 milliseconds on the 590 so about 5.6 milliseconds different frame times from the previous chart looked like this the gtx 1070 drew the most frames of course and also managed to keep consistently tight frame times to the frame to frame interval variants of plus or minus one millisecond this is incredibly good performance and is commendable of the car remember frame times are all about consistency of delivery being lower on the chart is better but being consistent frame the frame is best the 1070 is both but it's also more expensive drawing the RX 580 line next we see the frame time plot trends higher closer to 25 to 30 milliseconds rather than 18 to 24 of the 1070 frame the frame consistency is still good with the frame to frame deviation of plus or minus 2 milliseconds on average that is good performance and completely the RX 590 Fat Boys next now again this is 4k resolution so what we're looking at is the consistency of frame delivery not the outright frame rate the Fat Boy plots with the line competitive to the gtx 1070 and frame time consistency its frame rate is lower and the count of frames rendered is reduced as we saw in the previous chart but it's consistency is reliable this is good because frame rate can be resolved by reducing graphic settings but frame time consistency is difficult to fix the 590 does this well as do the other cards and so finally we'll plot the gtx 1066 gigabyte for perspective we see similarly commendable frame time consistency but overall reduced frame rate the higher placement on the y-axis means that the gtx 1060 runs slower and frame throughput than the other cards on the chart despite competitive consistency of those fewer frames the RX 590 ends up between the 1066 and the gtx 1070 cards in both frame rate and frame time performance and up for k numbers let's look at a truncated chart with just 1080p we haven't done as many lower end devices and sniper so this chart is more limited in quantity of cars but we'll make up for that in the next few games these cards of them the RX 590 leads the gtx 1060 and somewhat meaningfully outperforms the RX 580 the RX 570 matches the RX 581 overclocked which is as a reminder what the data land rx 580 represents so if you missed the previous content the data land rx 580 is an RX 570 it's just AMD called it an RX 580 2048 made a new SKU out of the 570 that's what that card is that's why there scare quotes so the scare quotes around the 580 are because it's 570 the distribution doesn't have too much spread against these devices but the 590 is in a better position than the other cards on the chart as before the 1070 would outperform the 590 but the 590 outperforms the 1064 f1 2018 we look at the counterpart to Sniper Elite 4 by benchmarking with a prime example of a well-built DirectX 11 game there are more API abstraction layers than with DirectX 12 but this particular game handles them well at 4k the RX 590 ends up at 38 FPS average stock roughly with the overclocked variants 40 FPS flat the gtx 1070 ends up at 46 FPS average and the RX 588 gigabyte gaming ex runs 35.7 FPS average when stock and 38 FPS average wid overclocked the overclocked result here puts it about on par with the RX 5 90's stock although the overclocked 590 again pulls marginally ahead 1440p is a more reasonable resolution for the 590 soul moved to that chart now for case not really the Forte of the RX 590 at 1440p and f1 2018 the RX 598 gigabyte card ends up at 64 FPS average with 1% lows averaging 47 FPS overclocking didn't do much for us gaining an additional 3 fps and the GTX 770 is the next card somewhat distant ranking at 78 FPS average and with low similarly paced as the RX 590 the ARS 590 outperforms the gtx 1066 gigabyte card when stock running 57 FPS average and overclocked running 61 FPS average with the plus 100 core offset for reference to 1066 gigabyte card it costs about $200 to $250 and depending on the model and the current sales the RX 598 runs about 260 to 280 depending on model will talk value more later but that gives you an idea of where they land it's fully capable of 1440p though and ranks between the 1070 1066 gigabyte at 1080p the RX 590 ends up pushing 82 FPS average when stock or 85 FPS overclocked once again overclocking is hardly useful for this GPU the RX 590 performs about the same as the RX 580 gaming ex when it's overclocked to fourteen fifty five megahertz with both cards averaging 80 FPS the 1060 overclocked is around the same performance although gains against the baseline 1060 are noteworthy the 1070 maintains a significant advantage of the 590 in these charts party at 101 FPS average of the stock version far cry 5 gives us a look at ubisoft dunia engine pulling the title with more geometric complexity and screen space reflections than some other testing we've gone through so far for this one the RX 590 Fatboy ends up at 30fps average with an overclocked pitifully pushing it to 31.5 FPS so no real gain the gtx 770 runs at 36 FPS average while the overclocked 580 gaming ax i card at 1455 megahertz runs the same frame rate as the RX 590 since you know it's an overclocked RX 580 none of these are particularly playable so let's move on to 1440p at 1440p with the higher settings that we're using designated on the charge satan's become reasonably playable the RX 590 stock GPU operates an average frame rate of 55 which is ahead of an overclocked to 1066 gigabyte card by about 3.9 percent we're outside of standard deviation tests to test but not by much the difference is imperceptible to the user for a fairer comparison of stock to stock testing the gtx 1066 gigabyte stock without an overclock ends up at 50 FPS average still within reasonable range of the RX 590 the 590 doesn't hold a stron of an advantage here as it has in some other titles the gtx 1070 SCS average frame rate of 67 FPS post the lead of 21% over the 590 stock GPU with the 580 roughly tied with a 10 61% and 0.1% lows are both well times across all of these configurations the most important thing to learn from our multi-resolution testing as we move into the 1080p chart is whether performance scales linearly across all resolutions if it doesn't that could be indicative of a pipeline limitation ROPS for example or maybe video memory or texture map units at 1080p and far cry 5 the RX 590 maintains similar spacing against the gtx 960 and 1060 overclocked cards it's tied with the overclocked to 1060 and leading the stock by about 6.7% overclocking the RX 590 gets it to 80 FPS average but there's just no more room left in this card to overclock any meaningful amount at least not without mods we see gains from overclocking of 2.4 percent over the stock rx 590 and finally the 1070 maintains a significant lead post in a 96 FPS average of the 590 78 FPS average that's also accompanied by a price increase of course and will point out that Vega 56 does pretty well in this title also posting a 104 FPS average with the power color stock model the Vega 56 cards average about four hundred fifty to four hundred seventy dollars so buyers of the 590 are likely not willing to stretch this high-end price shadow of the Tomb Raider is still relatively and is meant to be one of the leading r-tx titles although our benchmark scene is standardized without any r-tx features Shadow the Tomb Raider is on a modified Crystal engine and uses DirectX 12 for its API one note for this one we had trouble getting the overclocked to stabilize in this game for the 590 but eventually got it running at 1650 to 1660 megahertz core rather than 1660 to 1670 and other titles 4k isn't very friendly to this class of device when using the settings we were using placing the RX 590 Fatboy at 28 FPS average with an overclock doing effectively nothing the 1070 also struggles down at 30 FPS average with a 1070 TI at 35 fps so let's move on right away we can't learn much here at 1440p the RX 590 ends up at 50 FPS average with the overclock taking us only 250 1.4 FPS not exciting sadly and about what we've been seeing thus far that most high rx 580 gaming that except 1450 5 megahertz hits 48 fps average position it itself close enough to the 590 that the differences are imperceptible the gtx 1070 holds a 55 FPS average marking shadow of the Tomb Raider as the title where the RX 590 strikes the closest to the gtx 1070 to the point that the value for this game best favors the 590 in terms of percentages the 1070 holds just the lead of 8.5% of the 590 which holds a lead of roughly 11% over the stock 580 and the overclocked 10 60s around that range to running 1080p instead the RX 590 ends up at 72 FPS average which is completely playable while using high graphics settings the overclocked is once again pitiful foods Tina is only at 1.9% that's disappointing and this really not much to be gained here because the cards already a pre overclocked spree overclocked card so there are X 580 with an overclocked is again close to the RX 590 stock GPU not that exciting and predictable but still an improvement of sorts GTA 5 is last using the rage engine which is redundant by Rockstar Games the benchmark is intensive on draw distance particularly for geometric detail and shadow detail at long ranges 4k is almost entirely unplayable with our very high and ultra mix of settings on our standardized test bench and the that's for the rx 590 the iris 590 is positioned with results of about 25 FPS average for the stock 590 Fatboy or 23 fps for the 580 gaming x stock let's skip right over to 1440p at 14:40 PG ta positions the RX 590 Fatboy at 51 FPS average tied exactly with the overclocked rx 580 gaming X and marginally ahead of the rx 580 sort of boring if we're being honest but that's the nature of a reef refresh if you already knew where the RX 580 landed when stock and overclocked you could figure out where the RX 590 would be and where it is in this game the gtx 1060 leads is performance over the RX 590 if GTA is the only finger plane and if you can find a 1066 gigabyte for about 200 to 220 dollars it'd be a better buy for this title branching out to other games would change that prioritization and GTA is pretty old so you also have to question how long you'd be playing only that game these results are consistent with other sites and are a bit unique to GTA 1080p for GTA gives us our last game test the RX 590 and gtx 1060 maintain the same spacing as before with a 1066 gigabyte card leading the 590 by large margins marking 83 FPS average stock versus 76 FPS average stock the 1070 polls even further ahead landing at 106 FPS average and the overclocked msi RX 580 is about the same as the 390 X and there rx 590 when the 580 is overclocked power consumption is last for this we do a total system power consumption logged over time so this is not for the individual card but for the entire test bench all system components are heavily controlled with voltages even minor voltage rails also heavily controlled to ensure data accuracy the RX 590 stock GPU total system power consumption average is about 320 watts with the msi RX 580 gaming X system averaging 20 to 30 watts lower on average and that's without having the overclock applied so that's stock the GTX 1066 by total system power averages 240 watts when the 1060 is overclocked with a plus 100 megahertz core offset as shown in previous charts the 590 is still a heavy drinker of power as the previous rx 500 series cards have been for thermals we first tested the card with the stock V BIOS configuration which is with the switch set toward the PCIe connectors XFX says this is the faster fan speed v bios were the more aggressive v bios but that was clearly inaccurate the GBO core hits about 79 to 80 degrees Celsius at steady state which is the temperature target for this v bios this is the stock setting under these settings the V Ram module we measured climbs to a hide ninety one point eight degrees Celsius this is nearing the specification limit of gddr5 and once accounting for measurement error and the fact that we're looking at T case naught T Junction might even be hitting this back limit it's close enough to be disquieting the MOSFET we measured is at 85 degrees this is hot and uncompetitive but is technically well within the operating temperature specs for MOSFETs most FETs can take 125 to 150 degrees Celsius so we're not in any danger here it's just not competitive with other coolers and is indicative of inefficient design and use of an otherwise large cooler it's just it's not efficient for how big it is fan speed for this V bios hits around 1,500 rpm to maintain eighty degrees on the GPU but the trouble is that the neighboring components run hot with the left v bios switch engaged instead not the stock position the GPU core runs 69 to 70 degrees Celsius with the vram now at 81 degrees definitely more acceptable and the mosfet at 74 degrees considering fan speed is now 2,000 rpm up from 1492 1500 this remains uncompetitive with others on the market but is more acceptable than the stock measurements you'll want to switch into this mode for normal use especially if in a hotter case and one that might push ambient temperature is higher than the 22 plus or minus 1 degrees that we test in and just as a reminder we pointed all of this out when we did the teardown of the rx 590 it was a while ago but we'll cut to some footage of it just as a reminder instead of using a copper coal plate contacts and everything or even an aluminum one they're using a steel cold plate contacting the vram and that's connected via screws then connected to the rest of the cooler so it would have made more sense to do some kind of direct contact even with just just pads or paste or something between the vram and the heat pipes or whatever's there ideally you go with a larger single plate that covers the GPU in the vram or one that maybe has stepped down for the GPU for clearance purposes but with a design that was chosen where it's it's multiple connecting interfaces you lose efficiency every time you transfer heat through a new interface and there are a lot of them here for this card so it is inefficient and it runs hot for the vram as a result this is something that we we warned about when we tore the card apart but now it's more evident that it is a problem conclusively then the rx 590 is boring doesn't mean it's bad it doesn't mean it doesn't have a place in the market but it is genuinely a boring products who work on or review now that's okay because in terms of price to performance it does find is it's the same as by it's almost literally the same as buying an RF 580 and an overclocked unit so obviously there's no reason to upgrade if you have a 40 or 582 this if you don't have one of those you could buy the rx 590 if it's at 260 270 dollars a gigabyte model and performance is pretty good actually it's really not bad frame times were consistent overall there were no major problems with frame times we did not have major problems with drivers a lot of those seem to have been fixed lately with a and these efforts they're overclocking extremely uninteresting we'd recommend and said something like Vega 56 it's significantly more expensive to go that route but if you want something for overclocking for fun like that's why you're buying it not just for gaming then don't bother don't buy this for that Vega d6 is good for that route if you prefer to go that route and stay AMD and honestly 56 is a bit more fun to work with for overclocking than Nvidia anyway you have more freedom with it but as a card out of box it's fine it's just it's not exciting it is a fair replacement for the 1060 if you're looking at 10 60s as one of your options right now and you can't find them maybe in your region for the prices we have here like 210 bucks then this is superior at the same price for sure it's in some games superior at higher prices because it's never really more than 40 $50 more and depending on the game workload you're looking at the 590 does better enough that it's worth buying there are instances where that's not true GTA 5 and this is consistent another test we've looked at GTA 5 specifically does favor the 1060 over even the 590 so if that's the only game you're playing then don't bother but otherwise you're playing more things it seems fine 1070 you get I don't know anywhere from 7 to 12% more performance on average there are a couple games where that's wait less just depends on the title but 7 to 12% more performance looking at like $40 more for the low-end 1070 is 50 60 for the mid-range ones so I probably is just it's really granular right now and video has 18 billion skews of Pascal cards and Andy isn't intending to compete with just as many of Polaris cards so you end up with a product stack we're telling you which one to buy is very much going game by game cart by card and saying get this one for this game get this one for that game and so all I can say is look at the charts and I can tell you that this specific model probably not one we recommend necessarily the the thermal performance is a joke a la Thea I mean in one of those tests thermals were hitting like 90 degrees Celsius on some of the power components which is technically within spec but if you're gonna make a card this literally named the Fat Boy and being bigger than two slots make it cool so we don't recommend this specific model the 590 as a product itself seems fair it's fine but get a different one this one is just the cooling is just no so that's it for this one subscribe for more as always and get a store like here his exes that night to help us out directly hopefully we'll have some more exciting GPUs coming from AMD at some point in the future and as soon as we do we'll review those as well let you know how they do I'll see you all next time