AMD Radeon RX 5700 XT and RX 5700 Video Card Review

Great review. The 5700xt looks excellent. Great price point and only a little slower then the 2070 super while being a good deal cheaper. Building a new rig this winter and planning on a 1440p monitor, so seems like a perfect fit for me. Looking forward to it and seeing what amd has to bring in the future.
 
Great review Brent. I love the format and the way you lay out the charts by price.
This release for AMD looks like they are getting competitive again with the architecture and the pricing. I am thinking about upgrading my EVGA 1070 and now I have choices.
 
I was expecting the 5700 XT in a few scenarios, really holds it's weight against the 2070 Super especially considering it's price. Hopefully these sell like hot cakes for AMD.
 
I've been reading CPU reviews so far this morning and just started reading this but I had to pop in and say I am glad you put up a YouTube video.
Not a bad first video, but was that a ceiling fan reflection I saw?
*EDIT*
Just finished reading, another good review. That 5700 XT Is just what I was hoping for.
 
Last edited:
I was looking to upgrade my GTX 1080 .. maybe the 5700 is in the stars. For some reason I want to run an all AMD setup instead of just CPU. I like the efficiency of the 5700.

Great Review
 
Thanks for the review! 5700 looks like a great card. I too hope non reference cards can get those temps lower and clocks even higher. Happy to see a more efficient architecture over GCN from AMD. But would RDNA make for a good high end large die GPU? Either way it is a step in the right direction.
 
Beautifully done review - thanks!

I'm likely going to move on a GPU deal come black Friday/cyber Monday, and it's either going to be 5700XT, 5800/XT, or 2070Super.

I want to wait for the 5800 to be reviewed and let some time pass so that I can decide which is the best bang for the buck for my wants and needs.
 
Nice review. I am specifically interested in anti-lag. It would be great to see that covered with your follow up article(s). Thanks!
 
One thing I mentioned on the Hard|OCP forums that I'd like to bring up here: There is a severe lack of testing of the most popular PC games by reviewers. Titles such as R6:S (not just the canned built in benchmark but actual gameplay), Fortnite, Apex Legends, PUBG, League of Legends, World of Warcraft (go through some heavy dungeons and repeat) etc which are ignored because they require extra work.

Hardware Canucks did test PUBG/Fortnite/R6S and found Navi was really lacking in the 1% numbers for Fortnite/PUBG and overall slower in all 3 by a decent margin which is important to hundreds of thousands of gamers who are in the market for a new graphics card specifically for these games:

fortnite.pngpubg.pngR6S.png

Some tests would inevitably require multiple runs and an average would have to be taken from the 1%/avg but it is feasible and shouldn't be neglected in future reviews. Most PC gamers are playing the aforementioned titles and I'd argue they are far more relevant than testing games such as Hitman or Kingdom Come. While the latter may offer a consistent benchmarking platform, they don't tell you much about the cards you are testing outside those games. Game engines such as UE4, Unity, Crytek, Source etc all give wildly different performance metrics depending on the game and studio developing them so you can't even make a generalization about DX 11/12 performance based on a certain game as it may not be representative outside its limited scope. For example, PUBG being a UE4 title doesn't run so hot (never really has) but Fortnite has always been butter smooth despite both being UE4 and the same genre.

I hope my suggestion isn't brushed off, this is a problem I always had with Hard|OCP and I hope this new website you guys have founded starts going in a bit of a different and more fresh direction. I'd also like to see some video reviews from this website that go along with the actual written review and once you guys have the traffic/hardware, some FCAT numbers for regular 2D gaming and VR.

Personally, to me as a tech enthusiast, a more in depth review that has a written article + video is far more important than day one reviews. Of course day one is also important for growth and traffic but you could always call the day one review a "preliminary review" and follow-up with a lot more info a few weeks later. Hopefully you guys don't become beholden to AMD/NVIDIA/Intel NDAs and other corporate bs like many other tech sites/journalists although I understand there is always some politics involved when dealing with them.

Edit: I did see your YT video and its a good start. I'd suggest fashioning them in the future similar to what Eurogamer and Hardware Canucks do, I find theirs to be of high quality and quite enjoyable. GN is okay too but he usually takes too long to get his point across and loses the viewers attention.
 
Last edited:
Some tests would inevitably require multiple runs and an average would have to be taken from the 1%/avg but it is feasible and shouldn't be neglected in future reviews. Most PC gamers are playing the aforementioned titles and I'd argue they are far more relevant than testing games such as Hitman or Kingdom Come. While the latter may offer a consistent benchmarking platform, they don't tell you much about the cards you are testing outside those games. Game engines such as UE4, Unity, Crytek, Source etc all give wildly different performance metrics depending on the game and studio developing them so you can't even make a generalization about DX 11/12 performance based on a certain game as it may not be representative outside its limited scope. For example, PUBG being a UE4 title doesn't run so hot (never really has) but Fortnite has always been butter smooth despite both being UE4 and the same genre.

I hope my suggestion isn't brushed off, this is a problem I always had with Hard|OCP and I hope this new website you guys have founded starts going in a bit of a different and more fresh direction. I'd also like to see some video reviews from this website that go along with the actual written review and once you guys have the traffic/hardware, some FCAT numbers for regular 2D gaming and VR.

If you're not doing multiple runs for each test that you do, then you're reviewing it wrong.

Game selection is a challenge. They do take time, and our methodology is different from [H] in that we are doing a hybrid of real world gameplay and canned benchmark runs. This has allowed us to expand the scope of games that we can fit into a review, which is moving in the right direction. However, there's still a point where judgement calls have to be made. We typically look at titled based upon those that are pushing the quality of PC gaming ahead by challenging cards and taking a step forward in visuals. When you look at Fortnite, Pubg, et al, they simply don't meet that criteria on either front (improves PC gaming graphics or challenges cards).

That being said, we are certainly open to suggestions to balance things out and switch up the games over time....
 
Agree on the price chart - that is a nice small addition.
 
If you're not doing multiple runs for each test that you do, then you're reviewing it wrong.

Game selection is a challenge. They do take time, and our methodology is different from [H] in that we are doing a hybrid of real world gameplay and canned benchmark runs. This has allowed us to expand the scope of games that we can fit into a review, which is moving in the right direction. However, there's still a point where judgement calls have to be made. We typically look at titled based upon those that are pushing the quality of PC gaming ahead by challenging cards and taking a step forward in visuals. When you look at Fortnite, Pubg, et al, they simply don't meet that criteria on either front (improves PC gaming graphics or challenges cards).

That being said, we are certainly open to suggestions to balance things out and switch up the games over time....

Visuals is only part of the equation. You have to ask yourself as a reviewer and consumer: Why are people purchasing these graphics cards? Well the answer is obviously complex and multifaceted but online MP gaming is probably the biggest reason of them all. Specifically the titles I mentioned as they have millions upon millions of players and they do push graphics cards to their limit. Apex legends at 1440p with all eye candy turned up will drop a 2080 Ti down pretty far, especially in areas where there are a lot of structures. Even PUBG will bring a 5700/5700 XT to their knees when the graphics are maxed out in a busy area so it falls in line with your desire to challenge a video card. If you look at my post above, the 1% numbers for these cards are abysmal in PUBG/Fortnite and this information is nowhere to be found outside of Hardware Canucks which is a shame because there are a lot of consumers who are being misinformed about the real performance of these video cards.

Now take Kingdom Come or Hitman 2, what do they tell you about AMD/NVIDIA performance in general? Nothing much outside of those two games which hardly anyone plays if you go by sales metrics and hours played in 2019 vs the games I mentioned. You can't even take the numbers from those 2 games and make a generalization about performance for either brand because it doesn't hold up.

IMO, you would be doing a disservice to your readers if you neglected to benchmark those games and include them in future reviews.
 
Last edited:
A game like PUBG is incredibly hard to get a run-through done, especially consistently. Every round is different, the time is different, everything. Some games like that let you record your session, and play it back to watch it again, but even doing that it's not the same, and often game updates break or make the run not able to be viewed anymore, and they only last for a limited time. Then you have to do a whole new run which is not comparable to the old one. It takes time to do a review, and with the length of time it takes to switch out cards and test a run could end up not working anymore and now you have to do the whole thing again in a short amount of time, or just leave it out. It's problematic from a review standpoint and takes more time away from doing other things with the video cards. We could just play it and report our experience, rather than test performance and report numbers I suppose, but we'll have to look at if that's what people want to see or not.
 
A game like PUBG is incredibly hard to get a run-through done, especially consistently. Every round is different, the time is different, everything. Some games like that let you record your session, and play it back to watch it again, but even doing that it's not the same, and often game updates break or make the run not able to be viewed anymore, and they only last for a limited time. Then you have to do a whole new run which is not comparable to the old one. It takes time to do a review, and with the length of time it takes to switch out cards and test a run could end up not working anymore and now you have to do the whole thing again in a short amount of time, or just leave it out. It's problematic from a review standpoint and takes more time away from doing other things with the video cards. We could just play it and report our experience, rather than test performance and report numbers I suppose, but we'll have to look at if that's what people want to see or not.

Consistency is key. Its hard to do consistent runs in multiplayer games. Variations in each run can skew the results. Its bad enough trying to do the same things in a single player game for test purposes. It's nearly impossible when you have other players doing different things in a game.
 
Last edited:
Any word on when custom cooled partner cards will be out?
 
This review from all the ones I've read is the best one yet. Great background information was included which can be reference in the future. Good stuff indeed!

Looking at many reviews, the $399 price point AMD set basically gave roughly a 1080 Ti performance level with it. Now that is moving things forward by AMD. Pretty cool.

I've read that the 5700 looks to be OC limited by the drivers, as in power levels limited to 20% vice 50% and max clocks. Hope to see how these cards actually OC when more mature drivers hit.

There are some very unique design features, such as wave 32, that most likely have significant performance boost when optimize for. AMD with GCN had initial performance not much better than Nvidia but it blossom well with each driver update. May see that pattern again, I think it is most likely. AMD is not as fine tuned like Nvidia with optimizing drivers with new architecture.

The 5700 XT for just gamers makes the Radeon Vii look obsolete or over priced. The performance turned out better than I expected overall. I hope AMD sells plenty of these to nudge Nvidia to keep lower pricing in the future. As for partner cards, where are those leak images of them? With a better cooler, more aggressive clock speeds, we are looking at 5%-10% there as well, while starting at a much lower price point then Nvidia cards. This will definitely get very interesting.

As for PCIe 4.0 testing, maybe revisit mGPU testing, since AMD uses the PCIe bus for all transfers between the cards, might be very interesting if this benefits mGPU plus with AMD lower latency (claimed) may make mGPU more viable in the future.
 
Thanks for the review guys. Awesome and detailed.

Interesting to see how in some games the 5700xt would trade blows with the 2070 FE and the tables would turn between 1080p and 1440p. Much like the 2080TI the 5700xt seems to do better when it gets a real workout at the higher resoultion.

"Hot" and "Power Hungry". So sad to hear these words still apply to AMD gpu's but I'm at least glad they're showing up with some good cards again.
 
Become a Patron!
Back
Top