To evaluate just how good the Radeon RX 5500 XT is, we have one card from MSI and one from Gigabyte. From MSI we have their ‘Gaming X’ model and Gigabyte sent their ‘Gaming OC’ version, both look very good.
The RX 5500 XT GPU uses a tiny 158mm2 die packing just 1408 cores and 88 texture units, almost 40% fewer than the RX 5700. There’s also half as many ROPs, which has reduced the memory bus from 256-bit down to 128-bit wide. This sees memory bandwidth halved to 224 GB/s. AMD has kept the GDDR6 memory and you can purchase the 5500 XT with either 4 or 8GB of VRAM. We also have a 5500 XT 4GB card on hand and we’ll provide a detailed 4GB vs 8GB comparison soon, but for this review our focus will be on the 8GB version.
Getting back to the cores, they’re clocked at 1670 MHz for the base with a boost of 1845 MHz, which means they’re clocked 7% higher than the 5700. However, because there are a lot less, the TDP has been reduced from 180 watts down to 150 watts. The 8GB version of the 5500 XT comes in at $200 MSRP and frankly that seems a tad too ambitious for the hardware specs. The 4GB model will cost $170, and again that seems like a lot of money for a 4GB graphics card in 2019. Either way, we’ll review the price to performance offered by the 5500 XT towards the end of this review, for now let’s check out those benchmark figures.
Our GPU test system is powered by a Core i9-9900K clocked at 5 GHz with 16GB of DDR4-3400 memory. We exclusively used the Gigabyte RX 5500 XT Gaming OC for testing the 5500 XT in this review. We have 12 games to look at, all tested at 1080p and 1440p using medium to high quality presets or settings.
Benchmarks
First up we have Shadow of the Tomb Raider and with the high quality preset enabled at 1080p, the 5500 XT was good for 77 fps on average. Not bad performance, but not great, given the RX 590 was a little faster and the 1660 Super almost 20% faster.
Even at 1440p we see much the same and despite the extra VRAM, the 5500 XT can’t gain on the 1660 Super.
We see that the 5500 XT does manage to edge out the RX 590 in Assassin’s Creed Odyssey with an average of 71 fps, making it 4% faster, or a mere 3 fps. The GTX 1660 Super was 11% faster, though the 5500 XT was 8% faster than the GTX 1650 Super. At 1440p we see similar margins, the 5500 XT is still on par with the GTX 1660 which also puts it on par with the RX 590 and GTX 1650 Super, so not a great result for AMD’s new $200 offering.
Using the high quality preset in Red Dead Redemption 2 might have been a mistake with these lower-end GPUs as frame rates are quite low, also the 3GB 1060 didn’t work as the texture setting was too high and you can’t override memory restrictions in this title. So rather than give the 3GB 1060 an advantage with the medium texture setting, we opted to skip it for this one. Performance at 1080p from the 5500 XT is disappointing, it only matched the RX 580 and GTX 1650 Super. Things weren’t much better at 1440p and using the high quality settings the game wasn’t really playable, so you’ll need to drop down to medium if you hope to play at this resolution with anything less than an RX 5700 or RTX 2060.
The 5500 XT delivered a smooth 106 fps average at 1080p in World of Tanks, but that was only enough to see it positioned as a mid-field runner in our testing, matching the 3GB 1060 and putting it on par with the RX 590. Still a pretty unexciting result. The 1440p result for the 5500 XT isn’t much better, here it’s a fraction slower than the RX 590 and no better than the much cheaper GTX 1650 Super.
Performance in Far Cry New Dawn is typical of what we’ve seen so far, the 5500 XT is a smidgen quicker than the 590 and ~6% faster than the 1650 Super which frankly is a pretty miserable result given they want to charge $200 for this thing. Not much different at 1440p: the 5500 XT and 590 are neck and neck while the $230 GTX 1660 Super is well out in front delivering nearly 20% more performance. This isn’t great for AMD given Nvidia is charging just 15% more.
Frame rates in Rainbow Six Siege are strong at 1080p using the very high quality preset, one notch down from ‘ultra’, though we are manually cranking the render scale back up to 100%. The GPU delivers good performance overall, though the 5500 XT was still slower than the RX 590 and not a great deal faster than the cheaper ($160) GTX 1650 Super. Much the same is seen when testing at 1440p, the 5500 XT only manages to keep pace with the GTX 1650 Super and RX 580.
The Star Wars Jedi: Fallen Order performance at 1080p is average. Here the 5500 XT is slower than the GTX 1650 Super and the old GTX 1060 6GB. Basically it finds itself situated between the 590 and 580. Moving to 1440p allows the 5500 XT to match the 1650 Super and 3GB 1060, so not great.
Using the high quality preset at 1080p in Battlefield V allowed the 5500 XT to average 83 fps in our test which sadly meant it was only able to match the RX 580 and render 3fps more than the 1650 Super on average. The more limited 4GB VRAM buffer appears to hurt the 1650 Super at 1440p and now the 5500 XT is able to deliver 13% more performance, with 62 fps on average, but the Radeon does cost ~25% more.
The Call of Duty: Modern Warfare results are very similar to what we’ve seen so far, the 5500 XT is basically on par with the RX 590 and GTX 1650 Super. Even at 1440p the 5500 XT can’t get away from the 1650 Super, though it does slip behind the RX 590.
Metro Exodus was tested with the ultra quality preset as the 5500 XT and equivalent Nvidia GPUs averaged over 60 fps at 1080p. The 5500 XT came in behind the RX 580 and it was also 10% slower than the 590. Moving to 1440p saw the 5500 XT drop down to an average of 51 fps which meant it was a few frames slower than the RX 580, though at least in this title it was quite a bit faster than the GTX 1650 Super, especially when looking at 1% low performance.
Here we see that the 5500 XT once again trails the RX 580, this time in F1 2019 at 1080p using the high quality preset. While 121 fps on average is great gaming performance, it’s still less than you’d get with the 580, 590, GTX 1660 and 1660 Super. The 5500 XT does match the 580 at 1440p, making it 5-6 fps slower than the 590 and 11% slower than the GTX 1660 Super.
Last up we have Gears 5 and here the 5500 XT is able to match the RX 590 with 75 fps on average at 1080p using the high quality preset. The 1660 Super is 14% faster and given it costs 15% more, it looks like AMD has just slotted into Nvidia’s current pricing scheme, rather than compete with it.
Power Consumption
Here’s a quick look at total system power consumption in Gears 5 and F1 2019, As you can see on the graph below, the 5500 XT is a big improvement on the RX 580 and RX 590, reducing total system consumption by around 30%. The GTX 1660 Super used slightly less power but at this point it’s a non-issue and they’re more evenly matched. Overall power consumption is good, not amazing for a 7nm GPU, but it’s comparable to where Nvidia are at right now.
Performance Summary
Given the price, we’ve got to say the 8GB 5500 XT didn’t look too hot in most of those benchmarks. Before we jump into the cost per frame analysis, let’s just look at the average performance across the dozen games tested at 1080p and 1440p.
On average the 5500 XT was slower than the RX 590 and only ~3.5% faster than the RX 580. Moreover, it was just 5% faster than the GTX 1650 Super and 13% slower than the GTX 1660 Super.
Gaming at 1440p brings about the same conclusion, it’s a whisker faster than the old RX 580 and slightly slower than the RX 590. Let’s move on to check out the cost per frame data…
Basically the Radeon RX 5500 XT 8GB comes in offering the same level of value as the GTX 1660 Super, at $2.27 per frame based on our 12 game sample. That means you’ll be paying a ~4% premium per frame over the RX 590 at current prices. It’s not a bad value in general, but coming from other Navi-based GPU offerings where AMD has been aggressive and pushed the envelope, here it’s merely slotted in between current GeForce offerings, or worse. The 5500 XT comes in at almost 20% more per frame than the GTX 1650 Super, which seems to be very aggressively priced and works out great for Nvidia since we’re talking about a similar level of performance overall compared to the new Radeon. Granted, the 8GB 5500 XT packs twice as much VRAM, but it’s still not a good result for AMD and those seeking a budget graphics card. It’s also worth noting that the 8GB 5500 XT costs ~14% more per frame when compared to existing 8GB RX 580’s.
Misses the Mark
If you were expecting AMD to come in and rock the boat you will be dissapointed, as we were. The Radeon RX 5500 XT was never meant to blow your socks off from a performance perspective, but we do take issue with pricing. AMD’s made no attempt to undercut Nvidia, rather they’ve slotted the 5500 XT into Nvidia’s existing pricing structure. In other words, they’ve rocked up to the party late, and offered us nothing new. It’s kind of crazy, they were selling us a 232mm2 die for $200 or less and have been doing so for over a year now. We finally get an update that’s considerably smaller at 158mm2, a 32% smaller die that offers a similar level of performance, and it costs just as much. No doubt 7nm supply is playing a big role here, but we doubt many of you care about that, that’s more of an AMD problem to solve.
For a while, AMD has been very aggressive on pricing their mainstream GPUs in order to remain competitive using 3 year old Polaris GPUs. This started with the RX 480 back in June 2016, at $240 for the 8GB model. A year later it was refreshed as the RX 580 at a $10 discount, and by late 2018 the same GPU was retailing for well under $200. Putting AMD’s own products aside for a moment, the 5500 XT struggles to compete with what Nvidia’s already offering in the GTX 1650 Super and 1660 Super. You could also argue, Nvidia made all the right moves to secure the mainstream segment ahead of AMD’s release since the Super cards are both Q4 releases. Therefore we see no reason to wait for RX 5500 stock to arrive and then try and hunt one down at the MSRP. There are plenty of 1650 Super and 1660 Super graphics cards selling at or very near to the MSRP, so we’d just grab one of those or alternatively snag a cheap RX 580.
Power consumption of the 5500 XT has been greatly improved over the RX 580, but do gamers really care about that? It’s not like you can’t easily power the 580 or 590 with a cheap 400 watt power supply. Remember the power results in this review were recorded in our overclocked Core i9-9900K test system, so expect total system power usage to be considerably lower with a Core i5 or Ryzen 5 processor. So while lower power consumption is certainly a nice bonus and should result in cooler and quieter graphics cards, we feel most gamers looking to spend $200 or less on a graphics card, mostly care about price to performance and here the 5500 XT offers nothing new. Ultimately, we’ll have to wait and see if retail pricing moves anywhere in the coming weeks and months to make the 5500 XT a more attractive option.
Finally, a quick side note regarding AMD’s new Radeon software features found in the Adrenaline 2020 Edition drivers, they’re impressive and AMD should be commended for their work. Adding features such as Integer Scaling for all Radeon GPUs is great and Radeon Boost looks interesting, though we doubt we’d use it, but still a nice option to have and we’re sure some gamers will find it useful.
