AMD Radeon RX 580 vs. NVIDIA GeForce GTX 1060 OC: Mid-range GPU shootout

Battle for the masses

Let’s get one thing out of the way first. The AMD Radeon RX 580 is not a new card. It’s really a rebranded Radeon RX 480 with some tweaks under the hood, and is based on the same Ellesmere GPU. This really comes as no surprise, as AMD has time and again slapped new names on old products that have been re-jigged with higher clocks and the like.

But while the AMD Radeon RX 480 lost out to NVIDIA’s GeForce GTX 1060 last year, the company is hoping to reassert itself in the sizeable mainstream gaming market with its second-generation Polaris cards. This is Polaris rebooted, and in a way, one could argue that it is what Polaris should have been in the first place.

The Radeon RX 580 sits on top of its mainstream GPU line-up, and it goes head-to-head with the GeForce GTX 1060 in chasing after casual gamers. And to keep things fresh, it’s bringing things like higher clock speeds and new power saving features.

As it turns out, NVIDIA also just released versions of its GeForce GTX 1060 (and 1080) cards with overclocked memory, dubbed the GeForce GTX 1060 OC. As you can tell, this is an attempt to make NVIDIA's offerings more competitive against an updated Polaris architecture. Clearly, the tables have turned. Going by the numbers we got in the following pages, the GeForce GTX 1060 could probably use the boost.

If you’ve got 500-odd dollars to spend today and are looking for a card that will run any game today at 1080p, the Radeon RX 580 and GeForce GTX 1060 should be right in your crosshairs, more so the new higher clocked GTX 1060 OC edition.

In this shootout, we’ve pit the MSI Radeon RX 580 Gaming X+ 8G against the MSI GeForce GTX 1060 Gaming X+ 6G (this is version with the overclocked memory) to find out which card deserves your money more.

Going head-to-head

As a member of MSI’s expansive Gaming series of cards, the Radeon RX 580 Gaming X+ 8G is a familiar sight. We’ve seen the red-and-black cooling shroud countless times before, and even on previous-generation cards, so hardly anything has changed when it comes to appearances.

Under the hood, there are minor refinements coming from the Radeon RX 480 in the form of a refreshed Polaris architecture based on a third-generation 14nm FinFET process, higher clock speeds, and more efficient power consumption. 

More specifically, you should find that power consumption is lower while watching video or running multi-monitor setups. This is because AMD added a mid-power memory clock state for applications that require higher memory clocks, such as video decoding and multi-monitor operations, can operate at instead of ramping straight up to the highest and most power-hungry state. This helps save power, as the memory doesn’t run faster than it needs to.

However, as we’ll see in the results section, overall power consumption is still higher than the previous generation because of the more aggressive clock speeds. This is reflected in the TDP figures as well, where the Radeon RX 580 has a 185 watt TDP versus the 150 watts for the Radeon RX 480.

But overall, the Radeon RX 580 only nets a small jump over the Radeon RX 480, so it is not really intended as an upgrade option over that card. Instead, AMD is going after those users who are running far older hardware such as the Radeon 200/300 cards or NVIDIA GeForce GTX 700/900-series GPUs. At its heart, the Radeon RX 580 is still using the same Ellesmere GPU as the RX 480, so this is yet another instance of AMD re-branding an old product for a new generation. 

The MSI card is clocked at a 1,431MHz base clock and 8,000MHz effective memory clock in the default Gaming mode, up from the 1,303MHz on the older Radeon RX 480 Gaming X 8G. In OC mode, the base clock ramps up slightly to 1,441MHz, and the effective memory clock speed goes up to 8,100MHz.

It features a 256-bit memory bus, which when combined with the 8GB of GDDR5 memory, gives you a total of 256GB/s in memory bandwidth. In comparison, the GeForce GTX 1060 Gaming X+ 6G has a narrower 192-bit memory bus and just 6GB of memory, which makes for 216.7GB/s in memory bandwidth.

Still, that’s up from the 192.2GB/s on the original Founders Edition card (and 194.6GB/s on the Gaming X version in OC mode), thanks to the faster 9,026MHz memory (it was set at 8,008MHz before). The higher bandwidth should give it a small performance bump, but it's limited to just a few percentage points.

Here’s a snapshot of the specifications of both cards:

Because both cards are both Gaming X+ models from MSI, their feature set is practically identical, so we won’t be discussing them separately. Instead, we'll focus on highlighting certain design differences, such as their dimensions, power requirements, and heatsink layouts. 

For starters, they both utilize MSI’s Twin Frozr VI cooler, which relies on a nickel-plated copper base plate that connects to 8mm-thick heatpipes to channel heat away from the GPU. However, the VRMs and GDDR5 memory on the Radeon RX 580 actually have additional heatsink shielding placed directly above them, but the GeForce GTX 1060 has just a single large GPU base plate.

The dual Torx 2.0 semi-passive fans will stop spinning when the card isn’t working that hard, which translates to a temperature threshold of around 60°C. Above that, the fans will start up, but the good news is that they remain relatively quiet and never become overly bothersome.

In addition, these are double ball bearing fans that should last much longer than their sleeve bearing counterparts. The latter might suffer from performance degradation over time as the oil dries out from high temperatures and heavy use, and double ball bearing fans will hold up much better under the conditions that GPU fans are subject to.

We should point out that these are both fairly large and wide cards for the performance they offer. At 140mm across, they are the same width as the MSI GeForce GTX 1080 Ti Gaming X, so you’ll want to take note of this if you have an extremely narrow chassis. Having said that, the Radeon RX 580 card is actually quite a bit thicker than the GeForce GTX 1060, as much as a centimeter more as opposed to the NVIDIA card and it requires slightly more than the standard 2-slot allowance. This is probably due to its higher power draw, which necessitated a beefier heatsink design. If you refer back above, you can see that that the Radeon RX 580 has a more complex heatsink array, more cooling base and more heatpipes than the heatsink solution designed for the GeForce GTX 1060 OC. Is this a sign of AMD pushing the limits of the Polaris architecture? 

As a result, while the GeForce GTX 1060 has just a single 8-pin PCIe connector, the Radeon RX 580 has two of them. That actually seems like quite an onerous requirement for a card of its caliber, and power efficiency is still the GeForce GTX 1060's strongest suit. 

Both cards also feature a metal backplate for extra structural reinforcement. There are small design differences between the two, with the Radeon RX 580 Gaming X+ 8G featuring a more subtle black dragon decal (the dragon is a lighter gray on the GeForce GTX 1060).

The final difference is the selection of display output connectors – the Radeon RX 580 has one DVI-D port, two HDMI ports, and two DisplayPort connectors, while the GeForce GTX 1060 swaps out one of the HDMI ports for a third DisplayPort output. This won't matter too much to most folks, but the Radeon RX 580's implementation may prove slightly better for those who want to connect an HDMI display and VR headset at the same time (without converters). 

Test Setup

The detailed specifications of our current graphics card testbed system is as follows:-

• Intel Core i7-6950X (3.0GHz, 25MB L3 cache)
• ASUS ROG Strix X99 Gaming
• 4 x 4GB Corsair Vengeance LPX DDR4-2133 (Auto timings: CAS 15-15-15-36)
• Samsung SSD 840 Pro 256GB SATA 6Gbps solid state drive (OS + benchmark + games)
• Windows 10 Pro 64-bit
• ASUS PB287Q, 4K monitor

For comparison cards, we went with the PowerColor Radeon RX 480 Red Devil to see how much performance has improved over the last generation, and the Aorus Radeon RX 570 4G to gauge how much more performance the flagship Polaris card will get you.

The full line-up of graphics cards and their driver versions are listed below:

• MSI Radeon RX 580 Gaming X+ 8G (Radeon Software Crimson ReLive Edition 17.5.2)
• MSI GeForce GTX 1060 Gaming X+ 6G (ForceWare 382.33)
• Aorus Radeon RX 570 4G (Radeon Software Crimson ReLive Edition 17.5.2)
• PowerColor Red Devil Radeon RX 480 (Radeon Software Crimson ReLive Edition 17.2.1)
  
   

[hwzcompare]

[products=600477,600484,600479,591215]

[width=200]

[caption=Test GPUs compared]

[showprices=1]

[/hwzcompare]

Benchmarks

Next up, here’s a list of all the benchmarks used:

• 3DMark (2013)
• VRMark
• Middle-earth: Shadow of Mordor
• Deus Ex: Mankind Divided
• Tom Clancy’s The Division
• Ashes of the Singularity
• Hitman

We used the Fire Strike Extreme benchmark and stress test in 3DMark (2013) for our temperature  and power consumption tests respectively.

3DMark (2013)

Unsurprisingly, the Radeon RX 580 was only quicker than the RX 480 by around 11 per cent. That’s not too paltry an amount, but it still isn't enough to justify an upgrade from the perceived model number upgrade. That said, the Radeon RX 580 managed to edge out the GeForce GTX 1060 in the majority of the 3DMark benchmarks, although the difference was mostly limited to just a few percentage points.

VRMark

VRMark is a relatively new benchmark used to assess whether a certain hardware configuration is ready for high-end headsets like the Oculus Rift and HTC Vive. Scores are derived from the average FPS, where the passing frame rate is 109. We’ve included a separate graph with the raw FPS figures as well to help make better sense of the VRMark scores.

All the cards here passed the test, which means they’re ready for high-end VR headsets, but the Radeon RX 580 once again topped the GeForce GTX 1060 here. That could be due to its higher memory bandwidth as VR content can be very memory intensive, but we should point out that the difference is actually rather inconsequential, in the ballpark of just 3 per cent.

As in 3DMark, the Radeon RX 580 is faster than the Radeon RX 480, but not by a significant margin.

Middle-earth: Shadow of Mordor

In Shadow of Mordor, we see the Radeon RX 580 inching ahead of its NVIDIA counterpart again, although the GeForce GTX 1060 did gain the upperhand in the 4K benchmark. Having said that, the differences are so small that for all practical purposes, both cards are essentially neck-and-neck here.

You do see what performance advantages you’re getting by opting for the RX 580 over the RX 570 though, and the former was roughly 20 per cent quicker at 1440p.

Deus Ex: Mankind Divided

The latest title in the Deus Ex franchise is probably one of the most demanding games you can run right now. Based on the Dawn engine, it features just about every trick to make your game look pretty, including things like volumetric and dynamic lighting, screenspace reflections, and cloth physics.

Happily enough, today’s top mainstream cards are more than up to the task of handling any game at 1080p, and both the Radeon RX 580 and GeForce GTX 1060 do reasonably well at a FHD resolution on both High and Ultra settings. However, the performance gap between the two was slightly larger here, with the Radeon RX 580 coming in at roughly 8 per cent faster at 1080p.

That said, both cards began to struggle once the resolution ramped up to 1440p, and they both took a huge performance hit. At above 40fps, things are still reasonably playable, but here’s where Mankind Divided really begins to test the cards.

Things then plummet into unplayable territory when you try to run the game at 4K, and that’s probably where you’d want to draw the line.

Tom Clancy’s The Division

The Division uses Ubisoft’s own Snowdrop engine, and the game gives the lead to the Radeon RX 580 as well, where it was approximately 8 per cent quicker than the GeForce GTX 1060 at 1440p (DirectX 11). However, it seemed like the AMD card benefited a lot more from the move to DirectX 12 here, and the performance gap between the two widened to around 15 per cent.

The Radeon RX 480 even comes perilously close to the GeForce GTX 1060 in DirectX 12, so AMD has the clear upper-hand here.

Ashes of the Singularity

Ashes of the Singularity has long been the poster child for the performance benefits a low-level API like DirectX 12 can bring. It is based on the Nitrous engine and can be extremely punishing thanks to the huge number of onscreen units and the sheer level of detail accorded to each unit. This means it requires a high number of draw calls, which is exactly what DirectX 12 is supposed to handle more efficiently.

In the DirectX 11 benchmarks, the GeForce GTX 1060 came ahead of the Radeon RX 580 for the first time, where it was up to 22 per cent faster at 1080p and High settings. However, its lead shrunk as the display resolution and game settings became more demanding.

But after switching over to DirectX 12, both cards were virtually tied, and they traded blows demanding on the graphics settings and resolution. There was hardly a clear winner here and they were always within a frame or two of each other.

Hitman

Hitman is one of the more forgiving games in our benchmark suite, and the Radeon RX 580 and GeForce GTX 1060 were more than capable of handling the game at pretty much any resolution. Even at 4K and Ultra settings, the game is still relatively playable, although you won’t get a butter-smooth experience.

That said, the edge went to the Radeon RX 580 again, although the gap was once again restricted to just a handful of frames.

Temperature and power consumption

Since both the cards we’re reviewing today use the same MSI cooler, their temperature figures should be quite comparable. The Radeon RX 580 ran 5°C hotter than the GeForce GTX 1060, which is probably unsurprising given its much higher power draw and TDP (the NVIDIA card has just a 120 watt TDP compared to the Radeon RX 580’s 155 watts).

It also posted a total system power consumption figure of 300 watts, higher than the Radeon RX 480 as well.

As it turns out, these results are actually testament to how efficient NVIDIA’s Pascal architecture is as it managed to run cooler and consume less power while offering similar performance.

Overclocking

We used MSI’s Afterburner utility to overclock the cards, and here’s a table summarizing the clock speeds we managed to achieve.

The Radeon RX 580 overclocked surprisingly well, and this was a nice change considering the lackluster results we had with the Radeon RX 480. What’s more, the Radeon RX 570 took quite well to overclocking as well, so AMD clearly has made some much-appreciated tweaks in its second-generation Polaris architecture.

Ultimately, we achieved between 5 and 6 per cent increase in Fire Strike for the GeForce GTX 1060 and Radeon RX 580 respectively. The proportionate increase was similar in Fire Strike Extreme as well, so there’s still a tiny bit of performance to squeeze out of these cards with overclocking.

Having said that, the Radeon RX 570 deserves special mention, as we managed to get a 12 per cent performance boost in Fire Strike after overclocking, which is quite impressive for a card of its class.  

And the winner is…

At the end of the day, it all comes down to price. The MSI Radeon RX 580 Gaming X+ 8G costs US$280 on Newegg, while the GeForce GTX 1060 Gaming X+ 6G retails for more than US$300.

On the local market, the difference is even more stark. The NVIDIA card sells for S$549, or S$40 more than the older Gaming X model without the overclocked memory. And while we don’t have the local price for the Radeon RX 580 Gaming X+ 8G, the Radeon RX 580 Gaming X is going for S$469. Even slapping on a S$40 premium for the Gaming X+ model would only bring the price to just over S$500, still cheaper than its NVIDIA counterpart.

Other brands have also priced their Radeon RX 580s quite competitively. The exception is ASUS, where the ROG Strix RX 580 costs a mind-boggling S$559. That aside, the Aorus Radeon RX 580 XTR 8G costs S$429, while the Sapphire Nitro+ RX 580 8GB costs S$445, so the MSI card isn't the cheapest on the market.

Whichever way you parse it, the Radeon RX 580 is generally faster and more affordable, which makes picking it over the GeForce GTX 1060 almost a no-brainer. Yes, the performance difference is small, and the GeForce GTX 1060 does come ahead in select titles. But the Radeon RX 580 takes the lead more often, and a tiny performance lead is still a lead, and why would you ever pay more for something that performs worse?

Finally, there's the inconvenient fact that the GeForce GTX 1060 still does not support SLI, which seals the path for future upgrades in that direction. On the other hand, the Radeon RX 580 supports up to 2-way CrossFire, so you'll always have that option available.

To be sure, there are still reasons to pick the GeForce GTX 1060. Its low power consumption is still one of its strongest points, and the Radeon RX 580 simply cannot match up to how efficient it is. If you're working with a more compact casing or stricter PSU requirements, the NVIDIA card is a better bet. And if you own a G-Sync monitor, you’ll have no choice but to go with NVIDIA.

But barring these factors, the MSI Radeon RX 580 Gaming X+ 8G is still a very good card for its price, especially when the competition costs more. It is thicker and more power-hungry, but those are small trade-offs to make for what it offers. That said, we wouldn’t be surprised if NVIDIA responds with price drops to become more attractive to gamers, which would certainly be a welcome move.

All things considered, if you’re using an older card from three or four years back, the Radeon RX 580 would make for a great upgrade. The GeForce GTX 1060 is a good card in its own right, but it sadly falls short when it comes to price.