Date: July 8, 2019
Author(s): Rob Williams
Mid-range gamers who are overdue for an upgrade might want to pay attention to the latest options to just hit the market. Following NVIDIA’s launch of GeForce SUPER last week, AMD is now taking the veil off of its Navi-codenamed GPUs, based on the brand-new RNDA architecture. With three resolutions on tap, let’s explore the performance of the $349 RX 5700 and $399 RX 5700 XT.
It’s been an action-packed summer so far for PC hardware, so we hope you’ve been able to keep up. On the graphics card front, NVIDIA released its GeForce SUPER variants of the RTX 2060 and 2070 last week, with promises of a 2080 SUPER coming soon. With the just-released Radeon RX 5700 series, AMD is helping divide the green side’s launches up nicely.
It’s hard to believe that we’re finally able to talk about what Navi, and ultimately, its resulting RX 5700 graphics card series, brings to the table. The Polaris architecture has felt long in the tooth in some ways for a little while, especially when Vega has shown some huge strengths in comparison – especially on the compute side of things.
Navi is a lot more than a codename. It’s the first codename to represent the brand-new RNDA architecture, the first major step beyond GCN – an architecture which has by many opinions stifled real improvement for AMD’s mainstream efforts in recent years. AMD doesn’t just have its sights on the desktop. Navi is going to become the star of the show when next-gen consoles arrive.
Whereas Vegas was designed for the heaviest computational loads, Navi is built for mass distribution and mass consumption. What this means for top-end future GPUs, we’re not sure, but the X700 branding leaves a lot of room at the top.
Given that we’re juggling so many launches at once, we’re not going to cover everything we likely should on these GPUs right now, but once things settle down, we’ll explore a bit more. That’s especially the case with workstation-type benchmarks. As we did with both the RX Vega and Radeon VII launch, we’ll be kicking off our WS GPU test suite shortly, generating fresh results with updated software. We’re really interested to see where Navi will land in use cases other than gaming.
Here’s a quick overview of AMD’s current lineup:
|AMD’s Radeon Gaming GPU Lineup|
|Cores||Base MHz||Peak FP32||Memory||Bandwidth||TDP||Price|
|Radeon VII||3840||1400||13.8 TFLOPS||16 GB 4||1 TB/s||300W||$699|
|RX 5700 XT Anniv.||2560||1680||10.1 TFLOPS||8 GB 1||448 GB/s||235W||$449|
|RX 5700 XT||2560||1605||9.75 TFLOPS||8 GB 1||448 GB/s||225W||$399|
|RX 5700||2304||1465||7.95 TFLOPS||8 GB 1||448 GB/s||180W||$349|
|Vega 64||4096||1546||12.6 TFLOPS||8 GB 4||484 GB/s||295W||$499|
|Vega 56||3584||1471||10.5 TFLOPS||8 GB 4||410 GB/s||210W||$449|
|RX 590||2304||1576||7.1 TFLOPS||8 GB 3||256 GB/s||225 W||$279|
|RX 580||2304||1340||6.1 TFLOPS||8 GB 3||256 GB/s||185W||$229|
|RX 570||2048||1244||5.1 TFLOPS||8 GB 3||224 GB/s||150W||$179|
|RX 560||896||1175||2.6 TFLOPS||4 GB 3||112 GB/s||80W||$119|
|RX 550||640||1183||1.2 TFLOPS||2 GB 3||112 GB/s||50W||$99|
|Notes||1 GDDR6; 2 GDDR5X; 3 GDDR5; 4 HBM2|
Architecture: Radeon RX 500 = Polaris; Vega 56/64 & VII = Vega; RX 5700 = Navi
Most of the RX 500 series was launched two years ago, yet they remain the current option at their respective price-points. The lowest-end Navi chip so far is the $349 RX 5700, so it’s not exactly for desktop gamer “masses” quite yet, but those lower-end GPUs will surely come soon enough.
A first for Radeon is that these new Navi-based GPUs make the jump to GDDR6. As with the older Vega cards, the Radeon VII uses HBM2, which is quite a bit more expensive than GDDR6. That makes us thankful HBM didn’t get wrapped into the Navi picture somehow, or else AMD would have run into another situation of it being really hard to have workable margins.
While we can’t cover as much as we want to right now, we can’t go straight into performance without first talking about some of AMD’s developments with gaming technologies. As covered a few weeks ago, AMD is introducing FidelityFX, an open-source toolkit for adding new features to games that will work on all GPUs, not just AMD’s own.
FidelityFX is an extension of the company’s efforts to use open-source to encourage wide-spread adoption, and it’s really one of the more ambitious moves the company has made. Ultimately, FidelityFX is meant to mimic GameWorks and have a number of different features built into it, with developers able to pick and choose which ones to implement.
Right now, FidelityFX has a single feature available, called Contrast Adaptive Sharpening. In essence, this is a method of upscaling textures in important areas of a scene to improve overall crispness. Upscaling will come into play for ideal use, such as using 1440p resolution in a game while upscaling to a 4K monitor. The goal is to deliver 1440p performance at what looks like 4K resolution, or at least a close representation of it.
More “FX” will come in time, but so far, adoption for what’s here has been quite good around the industry, so we look forward to seeing what else could be added. The potential here is big, especially when the features are not going to be exclusive to any vendor.
Other features include Anti-Lag, which is meant to decrease the amount of latency which occurs between a mouse click or button press and the action on the screen. This is a feature that’s really targeted at those who are really sensitive to latency, which is to say, eSports and competitive gamers. The ultimate value on this feature is yet to be seen, but we’d be interested in hearing more from the gamers who’d actually be able to appreciate such a feature.
When Anti-Lag was announced, NVIDIA immediately retorted with the fact that “Maximum Pre-Rendered Frames” has been an option in its driver for over a decade. We’re not knowledgeable enough on the true specifics of each technology to make a solid claim one way or another, but AMD remains steadfast in the fact that the technologies are not the same. If you’re a competitive gamer, you’d definitely want to enable the setting regardless of game, just to see how you fare. If it’s not well, you can simply disable it from that point on. AMD admits that it won’t benefit every game, but it’s worth testing to see what differences you can perceive.
Right now, there doesn’t seem to be a real killer feature on Navi, at least in the same vein as something like NVIDIA’s RTX real-time ray tracing capabilities. The new Radeon features are honestly great, though, so hopefully many developers will latch onto things like FidelityFX and not just take advantage, but potentially contribute back to it, as well.
With that, let’s move along to a look at performance, but not before having a quick gander at our testing setup:
|Techgage Gaming GPU Test PC|
|Processor||Intel Core i9-9900K (3.6GHz Base, 5.0GHz Turbo, 8C/16T)|
|Motherboard||ASUS ROG STRIX Z390-E GAMING|
CPU tested with BIOS 1005 (April 10, 2019)
|Memory||G.SKILL TridentZ (F4-3400C16-8GSXW) 8GB x 2|
Operates at DDR4-3200 14-14-14 (1.35V)
|AMD Graphics||AMD Radeon RX 5700 XT (8GB; July 4 Beta Driver)|
AMD Radeon RX 5700 (8GB; July 4 Beta Driver)
AMD Radeon RX Vega 64 (8GB; Radeon 19.6.3)
AMD Radeon RX 590 (8GB; Radeon 19.6.3)
|NVIDIA Graphics||NVIDIA GeForce RTX 2070 SUPER (8GB; GeForce 430.86)|
NVIDIA GeForce RTX 2060 SUPER (8GB; GeForce 430.86)
NVIDIA GeForce RTX 2070 (8GB; GeForce 430.86)
NVIDIA GeForce RTX 2060 (6GB; GeForce 430.86)
|Storage||Kingston SSDNow V310 960GB (SATA 6Gbps)|
|Power Supply||Corsair RM650x (650W)|
|Chassis||NZXT S340 Elite Mid-tower|
|Cooling||Corsair Hydro H100i V2 AIO Liquid Cooler (240mm)|
|Et cetera||Windows 10 Pro (64-bit; build 18362)|
All GPUs have been tested with up-to-date drivers, and for the first time for our gaming suite, in the updated Windows 10 May Update. Our OS install is kept as clean and optimized as possible to reduce possible benchmark interference, ensuring accurate results. V-Sync, G-SYNC and FreeSync are disabled at the monitor and driver level.
A total of ten games are included in our current test suite. Recent additions include Tom Clancy’s Rainbow Six: Siege and Counter-Strike: Global Offensive for some super-high FPS eSports testing, as well as the new F1 2019, Metro Exodus, The Division 2, and Total War: Three Kingdoms. Meanwhile, Battlefield V, Far Cry 5, Monster Hunter: World, Shadow of the Tomb Raider, and the usual assortment of synthetics make a return in our updated suite.
Here’s the full list of tested synthetic benchmarks, games, and developer allegiances:
As covered in our look at NVIDIA SUPER, we had plans to include a Vulkan test, but failed to get a game to cooperate with us in time. We started with Rage 2, and ended up finding some real-life rage as it kept crashing after a patch. We moved onto World War Z, and somehow suffered the fate of different APIs being benchmarked throughout testing, and as a game that won’t even let you screenshot its results screen… our recent Vulkan pursuits have just ended in vain. We do plan to rectify this in the near-future, as World War Z is an AMD partner title and would even out the above list a bit better.
Note: You can download all of the tested setting images at once here (ZIP, 7MB).
With that all covered, let’s get right to it, starting with Battlefield V on the next page:
There are a handful of interesting comparisons that can be made in any one of the graphs throughout this article, but the two most notable ones are the RX 5700 XT against the RTX 2060 SUPER, and the RX 5700 against the original RTX 2060. Both sets share the same SRPs, $399 and $349, respectively, so we couldn’t ask for better match-ups. It is worth pointing out, though, that the RX 5700 includes an 8GB framebuffer within its price tag, while NVIDIA’s stuck at 6GB.
With the ultrawide resolution of 3440×1440, something wonky is going on with Radeon. We first saw this bizarre issue in our SUPER article the other day, but once we plugged the 5700 series results into Excel, we had to dig deeper. Being capped at 60 FPS, you’d imagine that V-Sync was enabled – but that wasn’t the case.
Even with V-Sync disabled inside of the driver and the game itself, Radeon cards seem to cap at 60 FPS in this game, but only at this ultrawide resolution. Before launch, we were told that there were some ultrawide-related bugs in the beta driver, but this one wasn’t on the company’s radar. Funny enough, this was the absolutely only ultrawide bug we encountered.
Given that the 0.1% results are really low for Radeon at the ultrawide resolution, we hope that whatever fix comes along to tackle the frame cap will also fix the somehow impacted minimums. Other than ultrawide, both of the 5700 cards deliver great performance up to 1440p, with even the non-XT surging 24 FPS beyond 60 FPS.
Once again, both of the RX 5700 cards are driving some great performance here, with even the non-XT model hitting 60 FPS at the ultrawide resolution. For those who like even higher framerates, the XT has what it takes to breach the 100 FPS mark. If you want to sacrifice some detail, you could likely inch your way towards taking full advantage of your 144Hz monitor with either of the 5700 cards.
In the competitive match-up, NVIDIA edges out the RX 5700 XT at all resolutions, much to the red team’s chagrin, we’re sure, as it’s the one with a logo on an actual Formula 1 car right now.
It’s worthy of noting that the RX 5700 has consistently been beating out the last-gen Vega 64, something that makes us wonder if the same scaling would be seen in an area where Vega really excels: compute. We’ll be tackling that benchmarking this week, so stay tuned for our report on that.
Far Cry 5 was a game heavily promoted by AMD, and it also happens to have AMD’s logo on the splash screen, so it’s not too surprising to see Radeon perform so strongly here. But this strongly? The XT actually manages to surpass the RTX 2070, which is not something you’d expect to happen – but it does, at all three resolutions.
Another interesting angle is that the non-XT keeps right up to the RTX 2060 SUPER. NVIDIA found a way to eke another frame or two out of the performance there, but overall, there’s some seriously good performance from the 5700 series here.
Where Far Cry 5 is clearly marketed around AMD hardware, Metro Exodus has been marketed around GeForce. That’s thanks in big part to the fact that the game can take advantage of real-time ray tracing, a feature currently exclusive to NVIDIA’s RTX line. If you are really wanting to try next-gen gaming with ray tracing, AMD unfortunately does not have an answer yet.
If you don’t care about ray tracing until performance improves, we understand where you are coming from. Fortunately, you won’t be lacking anything truly important but ray tracing if you go with AMD, as the performance delivered by the 5700 series once again is very impressive. This is an NVIDIA title, yet the $399 5700 XT outpaces the $499 RTX 2070 (but not the new $499 RTX 2070 SUPER).
These results have proven rather fun so far, so let’s see if it continues:
Let’s kick this page of results off with a look the two main battles we’ve been watching. The red team once again performs great here, beating out the NVIDIA competition at every single resolution in both match-ups. What’s perhaps better to see is that this game can use such high-end settings, along with the high-res texture pack, and still hit 60 FPS easily on most of these GPUs at up to 1440p. It just smooths out as you go up after that point.
Even the ultrawide resolution works fine on most of these GPUs. When Monster Hunter: World first came out, it felt incredibly sluggish, but these results prove that it’s actually pretty optimized. Or, it at least gives the impression through these results. With the fact that ultrawide resolution was added a few months ago, it’s obvious that the developers are keen on taking better advantage of our PCs.
We can add Shadow of the Tomb Raider to the list of RTX-infused titles that run really well on AMD’s new RX 5700 series. We’re once again seeing the $349 RX 5700 keep super close to the $399 RTX 2060 SUPER at 1080p and 1440p, and managing to surpass it at 3440×1440. Not all games have scaled the same way, but AMD is definitely putting a price tag on NVIDIA’s RTX features in this game.
NVIDIA’s RTX 2070 SUPER runs away with the lead in The Division 2, with the original (pre-OC model) RTX 2070 trailing behind it a good distance. But right behind that is the RX 5700 XT. It makes perfect sense now why NVIDIA decided to completely replace the original RTX 2070 with the SUPER, along with its pricing.
The RX 5700 continues to emulate the performance of the last-gen Vega 64, once again reminding us that we still have a lot of workstation benchmarking ahead of us. With Navi surprising so often in the gaming results, we’re hoping to see some likewise pop up in our creator tests, too.
Three Kingdoms becomes one of the couple of games in this suite that both RTX 2070s run away with performance in. The 5700 XT falls behind the RTX 2060 SUPER, but thankfully not to a considerable degree.
A great thing about pretty much any Total War game is that you are given the opportunity to tweak many graphics settings so that you can fine-tune performance and IQ just to your liking. At 1440p and the given high-end settings, only the RTX 2070 SUPER proves truly worthy, while all of them fall well short of 60 FPS at the ultrawide resolution.
If you want to experience Three Kingdoms in its highest-detail glory at above 1440p, you will need beefier GPUs than those seen here. Luckily the RTX 2080 SUPER is just around the corner then, right? To be clear though, it’s not like reducing a few settings to hit 60 FPS is going to make a severe detriment to the overall IQ. You will have to choose your battles in Three Kingdoms wisely, even those with graphics settings.
As covered originally in our look at NVIDIA’s SUPER cards, CS: GO is not a game we ever expected to test, but with eSports popularity still growing at an incredible rate, so too are higher frame counts – the kind that make 60 FPS seem really limited. While it can be argued that there’s a point when there’s “enough” frames each second, monitors offering up to 240Hz panels make it really tempting to find games that can hit the appropriate frame rate.
In our look at SUPER, we found it interesting that we couldn’t inch past the 235 FPS mark, which gave an impression of a CPU bottleneck. Well, AMD’s RX 5700 XT managed to get past that barrier, but not by much, ending up at 238 FPS. As we’ll see in a minute, greater than 238 FPS in general is definitely possible in competitive games.
At 4K, every single GPU listed here will deliver suitable performance, with only the RX 590 being knocked out if you are going after a 144Hz panel. As always, settings can be changed to inch your FPS up higher if you are desperate for more frames.
Even at max detail, every single one of the listed GPUs can eat through Siege. That even includes the lowly RX 590, which still manages to go well beyond 144 FPS at 1080p. Those adamant about 144 FPS can rule out a GPU like the RX 590, but consider anything Vega 64 onward.
Actual hard work starts to hit these GPUs at 4K, a resolution that will still see ~60 FPS performance on even the lowest-end GPU here. Across the three resolutions, AMD and NVIDIA went back and forth on strengths, so ultimately, they are fairly equal overall.
A fun aspect of UL benchmarks is that you never quite know what to expect with scaling, and it often disagrees with the scaling we see in our regular testing. But, it’s still a good overall gauge of what’s possible, which in this case is apparently an RX 5700 XT that’s as fast as the RTX 2070 SUPER in the 1080p Fire Strike test, and not far behind at 4K.
Ever since Pascal, NVIDIA has had dominating performance in the DirectX 12 Time Spy test, and that reality continues here. Perhaps the most interesting result here is with the bottom-ranking RX 590, falling considerably behind the next step up in this chart.
As with the DX12 Time Spy test, NVIDIA has some major strengths in VRMark’s tests, spread across both DX11 and 12. In the Cyan room, the RX 5700 XT beats out its direct RTX 2060 SUPER competition, while the roles reverse in the Blue room test, to a more considerable degree for NVIDIA.
The Blue room test represents future VR workloads, and based on the framerates delivered, it seems safe to say you’ll probably want GPUs higher-end than these here so that you can hope to hit 90Hz per eye guaranteed. But, it is important to note that this test represents truly strenuous workloads, and isn’t representative of typical gaming VR expectations.
Unigine helps wrap up our look at performance with similar scaling as seen in the 3DMark tests. The RX 5700 XT sits behind its 2070 SUPER competition, but it sure does come close to parity in the 4K “optimized” profile test. The RX 5700 and RTX 2060 non-SUPER go back and forth with these tests.
It’s a fun time for GPUs, with both AMD and NVIDIA doing their part to keep people talking. To simply say that AMD’s Navi-based GPUs have been hyped would be an understatement. We remember engaging in deep conversation with industry friends about the then-unnamed RNDA architecture two years ago during Computex, so it truly does feel like the RX 5700 series has been a long time coming.
It’s also a launch that’s caused NVIDIA to use this particular point in time to launch iterations on its original GeForce RTX GPUs, save for the top-end 2080 Ti. We looked at the RTX 2060 and 2070 SUPER variants the other day, and expect to have a look at the RTX 2080 SUPER when it arrives in a few weeks.
There’s no better example than what we’re seeing right now to show what a competitive market can do. AMD announced its RX 5700 series at E3 in mid-June with higher price tags than what we’re seeing today. The RX 5700 XT dropped from $449 to $399, while the non-XT’s price saw $30 shaved off, now putting it at $349.
Some have speculated that AMD “baited” NVIDIA into launching with lower prices, but we’re not entirely sure about it. The RTX 2060 SUPER has come as an addition to NVIDIA’s lineup with a $50 premium, so there’s no lower price there, and the other two SKUs replace the old ones in their same price slot. Whatever the reason for the current price-points, we’re glad to see them. There is a lot of gaming performance in these $349~$399 GPUs, and a notable boost with the RTX 2070 SUPER at $499.
When AMD announced Navi’s advance price drop, it admittedly gave us a bit of a bad feeling about what the results would look like. We figured even at the revised pricing, AMD would struggle overall. Instead, what we saw was some seriously competitive hardware. In many cases, AMD’s RX 5700 XT fought ridiculously hard against the card $50 more expensive, exceeding it on multiple occasions. That’s even in games that are geared more at NVIDIA, like Shadow of the Tomb Raider.
The RX 5700 has also impressed all-around, doing great battle against the RTX 2060 non-SUPER, which will continue as a product alongside the new SUPER. But considering the fact that the RX 5700 managed to beat the non-SUPER RTX 2060 more often than not, AMD wins that $349 slot, if we’re talking pure performance and not taking RTX features into consideration. We’d have to take the framebuffers into consideration, though, and at $349, AMD offers 2GB more than NVIDIA.
The choices at the $349 and $399 price-points are great right now, from both camps. You just have to choose which strengths from each you care about most. At $349, the RX 5700 won most of the battles against the RTX 2060, but the card of course misses out taking advantage of real-time ray tracing games, of which the collection of supported titles is growing. NVIDIA’s advanced screenshot tools Ansel and FreeStyle have growing industry support, as well, but won’t matter to everyone.
Similar pros and cons can be seen in the $399 battle against the RX 5700 XT and RTX 2060 SUPER. AMD again performed great overall, competing hard against the competition, managing to win many battles. Overall, the RTX 2060 SUPER can’t be called a clear winner, so AMD has delivered on its promise of giving the market a really competitive product.
That brings us to a thought of… if AMD had kept its original pricing, our outlook right now would be quite a bit different, because NVIDIA would have effectively won most of the battles, and tacked on those much-hyped RTX features. And really, those RTX features are the bulk what your choice comes down to right now. Both AMD and NVIDIA offer great performance for their respective price-points, but NVIDIA seems to carry a bit of a tax for its special RTX toys.
The ultimate message to come out of this competition is that the time to buy a mid-range GPU is right now. Or, at least as soon as these new GPUs from both sides pop up at etail.
Copyright © 2005-2019 Techgage Networks Inc. - All Rights Reserved.