Nvidia Geforce RTX 2080 Super Review: Just A Bit Of A Boost

Following the recent release of the GeForce RTX 2060 Super and 2070 Super, we have the final piece in Nvidia’s Super puzzle. Coming in at the same $1249 price point as its predecessor, the new RTX 2080 Super version will offer some performance increases, though we suspect we won’t see anything too dramatic considering the proposed specs and the fact that Nvidia doesn’t need to cannibalise sales of the 2080 Ti, nor does it have any direct competition at this price point.

The GeForce RTX 2080 Super adds two extra SM units for a 4% increase in cores, meaning this is now a fully enabled TU104 silicon. This is the same silicon used by the standard RTX 2080, as well as the 2070. There’s also a 9% frequency uplift for the base clock, and a 6% increase for the boost clock. Due to the way GPU boost works though, we anticipate this will be closer to a 3 to 4% increase in frequency.

The GDDR6 memory also sees a boost and now we’re looking at a peak data transfer speed of 15.5 Gbps, up from 14 Gbps, an 11% improvement in memory performance. These upgrades have increased the TDP by 16% from 215 watts up to 250 watts.

Nvidia says the RTX 2080 Super delivers a robust 4K experience along with real-time ray tracing. For this review we’re going to look at the 4K gaming experience, while skipping over ray tracing for now as our opinion of the technology in its current state hasn’t changed since we covered it many months ago.

We do plan to revisit ray tracing on a roundup piece with the Super graphics cards soon though, so if you’re interested in the technology that will no doubt be worth a read.

For testing we’ve used our usual gaming rig comprised of an Intel Core i9-9900K clocked at 5 GHz with 32GB of DDR4-3400 memory. The latest drivers available at the time of testing have been used. We have 13 games to look at before the usual performance breakdowns and cost per frame data.

Benchmarks

The newest game in our battery of benchmarks, F1 2019, sees the 2080 Super deliver 75 fps on average making it — drum roll please — 4% faster than the standard 2080 and 17% slower than the 2080 Ti.

Thankfully the margin is a little more noteworthy in Battlefield V. Here the 2080 Super offered an 8% performance uplift over the older 2080 and that basically put it on par with the newly discontinued Radeon VII.

Rainbow Six Siege sees a minor 3% boost to the average frame rate over the standard 2080, though we do see a 6% improvement for the 1% low performance. This time the 2080 Super was 15% slower than the Ti model.

Can we interest you in 1-2 fps more performance? Good, because that’s all you’re going to get in Metro Exodus. That’s to be expected given the specs.

A 6% performance boost in Resident Evil 2 may be cause for celebration (for the same price if we’re totally fair), taking us to 66 fps on average at 4K, for some nice smooth gameplay.

We didn’t quite hit 60 fps in Shadow of the Tomb Raider but given the 2080 Ti just falls short that’s not surprising. The extra memory bandwidth does allow the 2080 Super to beat the vanilla 2080 by an 11% margin in this title.

Memory bandwidth isn’t that important in Fortnite, so a small 3% performance uplift over the standard 2080 for 72 fps on average at 4K. Still it was a nice smooth and very crisp looking gaming experience.

When testing with The Division 2, the Super provided us with a 4% performance boost over the standard RTX 2080.

DiRT Rally 2.0 gets a 3 fps performance boost. Ignore the 1% low result and we see a breathtaking 4% performance uplift.

Forza Horizon 4, like Shadow of the Tomb Raider, is a game that benefits from some extra memory bandwidth at 4K and here we see a more significant 7% performance boost, taking the 2080 Super to 92 fps on average.

We also see a 6% performance boost in Far Cry New Dawn which allowed the 2080 Super to just edge out the Radeon VII.

The 2080 Super provided a 4% performance improvement from the standard 2080 in World War Z.

Finally we have Assassin’s Creed: Odyssey for a 3 fps performance uplift, though that does equate to a 6% increase in frame rate.

Power Consumption and Temperatures

The RTX 2080 Super consumes 5% more power than the standard 2080, which sounds right given the specs and what we just saw performance-wise. This means it consumed 6% less power than the Radeon VII, but 24% more than the new 5700 XT.

We also measured total system consumption from the wall and here we see the faster 2080 Super increased total draw by 12% over the standard 2080. There are a few reasons for this increase, the GDDR6 memory will be consuming more power given the 11% overclock, and the CPU will also be working slightly harder given the increase in performance. What’s interesting is that the 2080 Super does push system consumption higher than that of the Radeon VII.

We won’t go into overclocking as with all Turing GPUs you’re likely looking at a ~10% performance boost at best through manual tinkering. The Founders Edition model we tested maxed out at 72 degrees in a 21 degree room and there it maintained a core clock speed of 1905 MHz with a fan speed of 1820 RPM, making this graphics card rather cool and quiet after an hour-long test in F1 2019.

Performance Summary

The RTX 2080 Super looks to be the least “super” graphics card from Nvidia yet, but before we get too bummed out with this release, let’s take a look at some performance breakdown graphs and cost per frame data.

RTX 2080 Super vs. RTX 2080 @ 4K

On average the 2080 Super was 5% faster than the standard 2080. For testing we used the MSI RTX 2080 Duke which comes with a modest overclock that sees it maintain 1905 MHz under load, the same operating frequency as the 2080 Super FE card.

RTX 2080 Super vs. RTX 2080 @ 1440p

For those of you gaming at 1440p on average you’re looking at a 6% performance uplift, similar to what we saw at 4K. Those of you who were planning to jump the gun anyway, will see some extra frames for the same price, so there’s that.

RTX 2080 Super vs. RTX 2080 Ti @ 4K

On average the RTX 2080 Super was 16% slower when compared to the 2080 Ti. This is not a massive difference and it is around 30% cheaper, so if you’re not after the most premium 4K experience possible, the 2080 Super looks to be a decent option.

RTX 2080 Super vs. RTX 2080 Ti @ 1440p

For gaming at 1440p the 2080 Ti looks to be a bit of a waste as the more affordable RTX 2080 Super was just 12% slower and never more than 19% slower worst case.

RTX 2080 Super vs. Radeon RX 5700 XT @ 4K

When compared to AMD’s new $649 Radeon RX 5700 XT, the 2080 Super was on average 26% faster at 4K. Do consider that the 2080 is considerably more expensive, yet the GeForce was never more than 47% faster. Evidently the higher you go, the less value you get so you’ll have to decide if paying that kind of price premium for the ability to run at slightly higher quality settings is worth it.

RTX 2080 Super vs. Radeon RX 5700 XT @ 1440p

When gaming at 1440p the RTX 2080 Super is less attractive as it was just 17% faster than the 5700 XT on average.

Cost per Frame and Conclusion

Starting with a MSRP comparison and using the 4K gaming data, we see that the RTX 2080 Super comes out at a cost of $US9.85 ($14) per frame. Given it comes in at the same $1249 MSRP as the original 2080, that makes it roughly 6% cheaper per frame. Compared to the 5700 XT you are paying 41% more per frame which is a massive price premium, but that’s just what you’re faced with when going up a level in performance.

The 1440p results are less favourable, again looking strictly at MSRPs. We see that the 2080 Super costs $US5.73 ($8) per frame, or 5% cheaper than the standard RTX 2080. It’s also 52% more costly than the 5700 XT per frame, so unless you need to go from around 106 fps on average to 122 fps, the Radeon GPU is a more sensible purchase.

But what about current market pricing? The only major change in the analysis would be that the RTX 2080 Super will keep its price, while soon to be extinguished RTX 2080 cards have dropped down to $899 for some models. In that case the standard 2080 offers a better value given the small change in performance.

Earlier this month we got the RTX 2070 Super and 2060 Super. The 2070 Super provides 12% more performance than the standard model and we liked the boost enough to kill off most competitors at this price point, though the Radeon RX 5700 XT comes very close in performance for less money. The 2060 Super, on the other hand, gave us nothing. It costs 18% more than the standard 2060 and was just 8% faster on average. We’re seeing a similar story with the RTX 2080 Super, where the performance lift is even less significant.

Overall, the RTX 2080 Super is a solid premium-priced GPU, it’s just not particularly exciting. Officially discounting the RTX 2080 to $899 would have been worlds more thrilling in our opinion. Put in other words, the RTX 2080 Super is ~9% faster on average than the GTX 1080 Ti at both 1440p and 4K. So we’re getting a performance boost, at the exact same price point, almost two and a half years later. Without a doubt, GTX 1080 Ti owners can rejoice at the news.


Comments

    What a depressingly underwhelming revision.

    RTX 2080 - SUPERUNDERWHEIMG.
    - eBay for GTX 1080TI it is :)

    Ha! They gimp the 2080 Super just so they can justify the ridiculous pricing of the 2080 ti. Nvidia will never change.

      That's how product tiers work, for most companies in most industries. Cheaper models are deliberately given less power than more expensive ones. The 5700 is deliberately weaker than the 5700 XT, the 3600 is deliberately weaker than the 3600X. Not sure why you seem to think it's a uniquely Nvidia thing.

        No shit captain obvious. But technically it's been proven that the 3600 & 3600x are pretty much identical and you're really paying an extra $70 just for the X in it's name.

        I'm really criticising Nvidia's whole Super lineup being nothing special whereas the 5700 NT and 5700 are more impressive because they are a whole lot cheaper and offer near identical performance to both the 2070 Super & 2060 Super. And I'm still mad at how over-priced 2080 ti. The 1080 ti was a a fan favourite because it was sensibly priced for it's performance and it sold extremely well. But we have to pay a couple hundred more for Ray-Tracing (which is only available in a handful of games atm & drops fps by a significant amount when turned on) & DLSS (which is still useless to this day, especially when Battlefield V at 4k DLSS looks more blurry than 1440p native). You're better off buying an RX 5700 XT/5700 (which both support Image Sharpening which is similar to DLSS but it actually works & has almost no performance hit) or even a 1080 ti on ebay because the price to performance is much more justified than a 2080 ti or this Super lineup.

          I've seen this RIS vs DLSS comparison going around and it's weird, the two things are completely different and solve different problems. RIS is a post-process sharpen filter, basically ReShade built into the driver. Nvidia's equivalent to that is called Freestyle. RIS can't do what DLSS does without supersampling, which has a much bigger performance impact than DLSS does.

          I can appreciate being frustrated with the 2080Ti's price. It's probably little consolation that the 2080Ti is not actually the successor to the 1080Ti, and that Nvidia quietly rearranged their product line this generation so the 2080Ti fills the same niche that the Titan did last generation (high-end enthusiast), while the new Titan is wholly targeted at professionals. They didn't do a great job of communicating that, unfortunately.

          The 1080Ti was an amazing card and I agree it's still a good option for people not looking for the tip of the edge and the price premium that comes with it. Personally I think the 2080 (and the 2080 Super since it's just a refresh) is a better choice if you're coming in from anything prior to Pascal though.

            I'll always be a firm believer of bang for buck, price to performance when it comes to PC Hardware. So paying $400+ more than a RX 5700 XT for what is about 16% better performance in a 2080 sounds like you're paying a high price for such a small jump in performance (aka. a huge rip off). Better yet let's fork out $700+ more for a 2080 Super & get about 24% performance boost against the RX 5700 XT while were at it. The price to performance math just doesn't add up to me.

            Last edited 25/07/19 10:04 am

            I'd also recommend you check out the new video from aussie tech youtube channel Hardware Unboxed where they analyse & compare Radeon Image Sharpening, Nvidia Freestyle & Reshade. They concluded all three have the strength & weaknesses but they felt RIS was the most balanced in sharpening with no artefacts. Also RIS had the best fps performance when used while Nvidia Freestyle had the biggest fps drops & doesn't support all games.

              Sure, I'll check it out sometime. I don't actually use any of them so it's not an important feature for me, I'm more just confused that RIS and DLSS are being compared.

              I'm not anti-AMD, in case you think I've come across that way. I have hardware from both AMD and Nvidia in my PC right now (3900X and 2080Ti). I try to be as objective as I can be with this kind of stuff, but as a former game dev with engine experience I really like what DLSS does, and I think some of those 'RIS vs DLSS' comparisons floating around are really doing the latter an injustice. It's really good tech.

Join the discussion!

Trending Stories Right Now