AMD Radeon RX 590 Crossfire and the State of Multi-GPU Expertise

We are always surprised at how often we receive requests for Crossfire and SLI benchmarks. Although readers haven't been investing in either technology for years, interest still seems to be high. It is all the more surprising that AMD and especially Nvidia have made no secret of the fact that they use multi-GPU technology.

In the end we were stubborn and basically refused to check it over the past year. But recently two RX 590 cards came up to us and we thought, why not?

The last time we did a multi-GPU comparison at Catrachadas was in 2015 when we tested a GeForce GTX 980 Ti SLI configuration against Radeon R9 Fury X Crossfire. The Fury X cards prevailed in the 10 games tested at 4K with a small lead of 4%. After that, we found fewer reasons for a comparison and tried to achieve playable frame rates at 4K.

So we tested two GTX 1080s in SLI and later paired two Titan X cards in 2016, which we called the ridiculous graphics test because the cost of the graphics cards alone was over $ 2,400. We didn't know that GPU prices went crazy shortly after.

For today's review, we have a dozen modern games and we'll see how well two RX 590s compare to single cards in 1080p and 1440p. For comparison, we have nine other graphics cards, including high-end models such as Vega 64 and RTX 2070. Our test bench system consists of a Core i7-8700K, which is built into the Corsair Crystal 570X with 16 GB DDR4-3400 memory.

Benchmarks

Battlefield V was the first title we tested and we don't see a Crossfire in this title. No additional power from the second card at 1080p, and that also applies to the resolution of 1440p.

That was really surprising when Crossfire was working in Battlefield 1.

Here we see a 33% gain for the average frame rate at 1080p. Not great, but a big improvement over nothing, and scaling improves at 1440p, where the RX 590 in Crossfire boosted performance by 46%.

The best example of Crossfire scaling we came across in this new series of games was Strange Brigade. Here the Crossfire RX 590s increased the average frame rate by almost 90% … and even better, the frame rate performance was still very good.

At 1440p we see a scaling of over 90% at 94%. This is an exceptional result for the Crossfired 590s.

Radeon GPUs perform poorly in Assassin's Creed Odyssey, so it's not surprising in this title that Crossfire support isn't there. While the 590s in Strange Brigade were faster than the RTX 2070, they are 44% slower here.

If you move to Hitman 2, another title that doesn't support Crossfire and therefore runs with technology enabled will slightly slow performance. This can also be seen at 1440p.

Shadow of the Tomb Raider brings us a 48% performance increase at 1080p, and with it the RX 590 alongside the GTX 1080 and Vega 64. The scaling is dramatically improved at 1440p, and now we see a performance increase of 61%, which brings up the 590s the level of the RTX 2070. A solid result for this game.

Forza Horizon 4 is another title that does not support Crossfire. Therefore we could not find any gains at 1080p or 1440p. In fact, we saw a slight performance regression.

To my surprise, Monster supported World Hunter with Crossfire, although the frame time performance at 1080p was somewhat sketchy. We saw a 32% increase in the average frame rate, but only a 14% improvement in frame time performance.

A jump to 1440p helps fix this problem, but even then the scaling is below 40%, which is weak given the investment and hardly justifies a second graphics card.

Here's an example of a title that "looks good" if you focus on the average frame rate, but the experience was actually terrible. Despite an average of 100 fps in Star Wars Battlefront II at 1080p, the frame time performance was shocking and fell well below the result of a single 590. The game was basically unplayable and buggy when Crossfire was activated, and in this case offer constant 30 fps a nicer gaming experience.

The problem with frame time persisted at 1440p. We didn't get very far looking for a fix, but Crossfire doesn't work right away, so you're better off with a single RX 590 in this title.

The performance in frame time was also somewhat sketchy in Deus Ex Mankind Divided, although not nearly as bad as in Battlefront II. The problem was a bit more noticeable at 1440p and although the average frame rate was improved 590 with a second, it was 590 the overall experience is not so and due to the mismatch between the average and the 1% low result when using Crossfire we would rather play this title with a single 590.

Far Cry 5 also had a bit of stuttering, which was present at both 1080p and 1440p.

Project Cars 2 was another title that saw performance gains, but frame time performance wasn't great and much worse than that of high-end single GPU graphics cards.

power consumption

All graphics card configurations except the Crossfire cards were tested in Crysis 3 to measure power consumption. However, since Crossfire doesn't work in Crysis 3, we used F1 2018 as it was one of the better games for scaling. Typically, using a second RX 590 is expected to increase overall system performance by approximately 60%, resulting in fairly brutal consumption.

Here the Crossfire 590 increased the total system power to 576 watts, and I observed a little more than 600 watts in the Strange Brigade. This is far more than just a Vega 64 Liquid graphics card and almost twice as much as a single RTX 2070, so pretty horrible stuff when it comes to power consumption.

Put everything together

Compared to a single RX 590, we were able to increase the average frame rates by 14% at 1440p, but (and that's a big one, but) this number alone is misleading. The frame time performance in Star Wars Battlefront II and Deus Ex Mankind Divided was so bad that I preferred to play with a single card. Stuttering was also an issue in Far Cry 5, Project Cars 2, and The Witcher 3.

I usually also test with DiRT 4, but this title had serious graphic problems with Crossfire enabled, so I had to remove it from the series of games tested.

So would you better buy a faster single GPU graphics card or two cheaper graphics cards?

If you haven't figured that out yet, here's your answer: The RTX 2070 costs $ 500, while two RX 590s each cost $ 280 a little more. Even if you compared the RTX GPU to a pair of $ 200 RX-580, the result would be similar. Get the more expensive high-end graphics card.

If everything is in order for the multi-GPU plan, the RX 590 killed it and surpassed the RTX 2070 by a whopping 25%. But of the 20 games we tested, we only saw this margin once. The next best result was a 14% win in F1 2018, 9% in Prey, 5% in Project Cars 2, but frame-time performance was poor, as was Deux Ex and The Witcher 3.

closing remarks

Three years later, we find that multi-GPU technology on paper seems like a good idea again, but in practice it's a small mistake. In our opinion, SLI / Crossfire only makes sense for those who have money to burn. For example, RTX 2080 Ti SLI graphics cards are currently the only sensible multi-GPU configuration. If you are at the end of the street and there is nothing faster but want more and can spend a lot of money, you can buy a second RTX 2080 Ti.

However, it does not make sense to use RTX 2080 SLI cards, for example. If you buy just one RTX 2080 Ti, you'll get smoother performance on most titles.

For the RX 590 in Crossfire, we would rather have a single Vega 64 graphics card. It is extremely rare for two 590s to offer higher frame rates than a single Vega 64 while still offering smooth gameplay.

If you're just playing a game like F1 2018 that really supports Crossfire, it's hard to get two RX 570s for $ 300. But who buys a graphics card to only play one or two games at a time?

Other disadvantages that are also part of this conversation are the heat and power consumption. These two RX 590s have released so much heat into the Corsair Crystal 570X case that you might justify spending more money on fans, and even then you'll get hotter because of the way the cards are stacked. You also lose power. The RTX 2070 works fine with a 500 W device, and 600 W would be more than enough. However, the Crossfire 590s require an 800 watt unit, 750 W would be the minimum.

Ultimately, it's the poor software support that wipes out these multi-GPU setups, so no one should bother with SLI or Crossfire today.

Purchasing links:

Leave a Reply

Your email address will not be published. Required fields are marked *