Radeon RX 570 vs. RX 580 vs. GeForce GTX 1060 3GB vs. GTX 1060 6GB

[Julio Franco]

It's time to evaluate the Radeon RX 570 all over again, but this time the contenders will be the RX 580 and the GeForce GTX 1060 on 3GB and 6GB flavors. We're going to put all these mainstream GPUs head to head in 36 games at 1080p and 1440p to see which comes out on top in terms of performance and value.

Read the full article here.

https://www.techspot.com/review/1780-geforce-1060-vs-radeon-580-vs-radeon-570/

 

[NightAntilli]

What annoys me of this article is;

1) The GTX 1060 6GB is put on top, as if it's the faster one, despite the RX 580 being faster in more games
2) Exactly that. The RX 580 is faster in most games (9 out of 12 to be exact), but the few outliers ruin the average framerate so both the GTX 1060 6GB and the RX 580 end up at 82 fps as if they're equal.
3) No chart was added with a direct games list comparison between the GTX 1060 6GB and the RX 580.

It almost seems as if the article is trying to hide that the GTX 1060 6GB is slower than the RX 580 in most games. At least the final recommendation is in line with the numbers.

 

[Cycloid Torus]

Nice article in an increasingly complex arena. Worth noting RX570 has an 8GB variant which is priced between RX570 and RX580. It may not add much for the current crop of games as tests show highly similar results. If games get more memory hungry, this may be a dose of future proofing.

UEFI vs BIOS is an open question. I have seen some cards which indicate dual settings compatibility. However, I have seen reports of AMD R series cards not running on an older non-UEFI motherboard. If you have a legacy BIOS motherboard as I do, you should confirm that these cards will actually work with your system.

 

[Nightfire]

What annoys me of this article is;

1) The GTX 1060 6GB is put on top, as if it's the faster one, despite the RX 580 being faster in more games
2) Exactly that. The RX 580 is faster in most games (9 out of 12 to be exact), but the few outliers ruin the average framerate so both the GTX 1060 6GB and the RX 580 end up at 82 fps as if they're equal.
3) No chart was added with a direct games list comparison between the GTX 1060 6GB and the RX 580.

It almost seems as if the article is trying to hide that the GTX 1060 6GB is slower than the RX 580 in most games. At least the final recommendation is in line with the numbers.

I feel like those were all 3 the same reason: To show that the RX580 came on top.

It's nice to be able to more easily compare the Rx 580 to both GTX 1060 cards. If the Rx580 was on top, you would not be able to compare them as easy.

Steve's Nov 15 article compares those cards directly.

This seems like a waste of time explaining and a VERY minor nitpick.

 

[Nightfire]

Steve, you must have the most patient wife in the world. Cheers to her for letting you work so much!

We are all really excited for your RTX 2060 content!

Did you ever get an offer to review a GDDR5x GTX 1060? These look to be hidden gems as the extra bandwidth would really help newer dx12 games such as Forza.

Idea for a less than 10 game comparison. RX 590 vs. GTX 1060 GDDR5X, both overclocked.

"Max Performance at $250"

 

[VitalyT]

What is the point testing against GTX 1060 when nVidia has been trying to phase it out, and they say the stocks have finally been depleted? Very soon you won't be able to buy one.

 

[VitalyT]

What is the point including GTX 1060, when nVidia has been phasing it out, so you won't be able to buy one soon? They are already saying the stocks have been fully depleted.

RTX 2060 is being brought in massively as the replacement.

 

[Fobus]

What is the point including GTX 1060, when nVidia has been phasing it out, so you won't be able to buy one soon? They are already saying the stocks have been fully depleted.

RTX 2060 is being brought in massively as the replacement.

2060 is a not replacement for 1060. More like 1070/Ti. Nvidia currently has no budget refresh.
Besides, second hand market will be alive for some time, so it is relevant.

 

[dirtyferret]

Demand drives up some of the Nvidia prices and even though I have a GTX 1060, I would go with AMD at the lower price points if that card needed replacement.

I find Nvidia's current price structure, disturbing.

 

[Footlong]

What is the point including GTX 1060, when nVidia has been phasing it out, so you won't be able to buy one soon? They are already saying the stocks have been fully depleted.

RTX 2060 is being brought in massively as the replacement.

The used Market should not be ignored. RTX 2060 is not an entry card, at least not in the price aspect. People who are conscious of their budget should choose between the older GTX 1060 or better yet the AMD RX 580. For 1080p gaming on PCs everything else is overkill. Budget oriented machines should invest on better SSD if possible or one good freesync screen.

 

[fps4ever]

It's been well known for a while the RX 580 is what you seek for 1080P. Especially the deals you can get in the used market.

 

[codgerface]

What annoys me of this article is;

1) The GTX 1060 6GB is put on top, as if it's the faster one, despite the RX 580 being faster in more games
2) Exactly that. The RX 580 is faster in most games (9 out of 12 to be exact), but the few outliers ruin the average framerate so both the GTX 1060 6GB and the RX 580 end up at 82 fps as if they're equal.
3) No chart was added with a direct games list comparison between the GTX 1060 6GB and the RX 580.

It almost seems as if the article is trying to hide that the GTX 1060 6GB is slower than the RX 580 in most games. At least the final recommendation is in line with the numbers.

The results are based on a 36 game test as the article states several times and the graph in question specifies. Techspot just didn't post all 36 graphs in the article for obvious reasons. This is a public patreon post, so you should be able to see all of the results here:
https://www.patreon.com/posts/radeon-rx-570-or-24037399

The 1060 6 GB is "faster" than the 580 in ~17 out of 36 games, and the two cards tie in ~5 games.

Go easy on the accusations. The author has praised the AMD RX 570 in these last two articles, and the thing you take away is that maybe he's trying to secretly inject an Nvidia spin? C'mon... lighten up.

 

[jbrsci]

Nice comparison update. It's definitely helpful to have more options in the mainstream category of GPUs. It will be interesting to see if the 2060s are considered valuable for the money.

The main problem I've got with the RX 580, not mentioned in this article, is its RIDICULOUS power consumption versus the GTX 1060. Both cards will give you similar performance, but the RX 580s consume about 70% more power to do it. GTX 1060s average 125W in gaming, while the RX 580s average 210W! Maybe a small point, but that is a lot of waste heat and power doing nothing for you.

Show a frames-per-watt graph and you'll see the NVIDIA cards pull far ahead.

 

[ET3D]

Good work, Steve!

Would be nice to have the same summary graphics (averages and comparisons) but for the 1% results instead of average FPS.

 

[Dsaltos]

Nice work Steve!

 

[EClyde]

Framerates....5 will get you 10

 

[Swagger 4lyfe]

What annoys me of this article is;

1) The GTX 1060 6GB is put on top, as if it's the faster one, despite the RX 580 being faster in more games
2) Exactly that. The RX 580 is faster in most games (9 out of 12 to be exact), but the few outliers ruin the average framerate so both the GTX 1060 6GB and the RX 580 end up at 82 fps as if they're equal.
3) No chart was added with a direct games list comparison between the GTX 1060 6GB and the RX 580.

It almost seems as if the article is trying to hide that the GTX 1060 6GB is slower than the RX 580 in most games. At least the final recommendation is in line with the numbers.

Where is the fact that RX580 consumes a lot more power? The TDP is higher in the AMD card at 185 watts while the GTX 1060 6Gb is only 120 watts. And ,in spite of that, it goes head to head with AMD RX580.
You only know how to complain, but praise, ufff... that's really hard to do, isn't it?
And then comes the conspiracy theories that nvidia buys people to have better reviews than AMD, because of course that's the only explanation...
They are both great cards and they do a great job at 1080p and 1440p.

 

[ZackL04]

What annoys me of this article is;

1) The GTX 1060 6GB is put on top, as if it's the faster one, despite the RX 580 being faster in more games
2) Exactly that. The RX 580 is faster in most games (9 out of 12 to be exact), but the few outliers ruin the average framerate so both the GTX 1060 6GB and the RX 580 end up at 82 fps as if they're equal.
3) No chart was added with a direct games list comparison between the GTX 1060 6GB and the RX 580.

It almost seems as if the article is trying to hide that the GTX 1060 6GB is slower than the RX 580 in most games. At least the final recommendation is in line with the numbers.

Are they not arranged by price? Take your tin foil hat off and just read the information. The article plainly spells out “get the AMD for your needs”

 

[NightAntilli]

The results are based on a 36 game test as the article states several times and the graph in question specifies. Techspot just didn't post all 36 graphs in the article for obvious reasons. This is a public patreon post, so you should be able to see all of the results here:
https://www.patreon.com/posts/radeon-rx-570-or-24037399

The 1060 6 GB is "faster" than the 580 in ~17 out of 36 games, and the two cards tie in ~5 games.

Go easy on the accusations. The author has praised the AMD RX 570 in these last two articles, and the thing you take away is that maybe he's trying to secretly inject an Nvidia spin? C'mon... lighten up.

Why could they list the other comparisons? We have the whole list for;
RX 570 vs 1050 Ti
RX 570 vs 1060 3GB
RX 570 vs 1060 6GB
RX 570 vs RX 580

Those are all viable comparisons. I guess technically one can infer the data of the RX 580 vs the 1060 6GB based on the last two comparisons. But those two are direct competitors and I simply felt that it was missing.

Where is the fact that RX580 consumes a lot more power? The TDP is higher in the AMD card at 185 watts while the GTX 1060 6Gb is only 120 watts. And ,in spite of that, it goes head to head with AMD RX580.
You only know how to complain, but praise, ufff... that's really hard to do, isn't it?
And then comes the conspiracy theories that nvidia buys people to have better reviews than AMD, because of course that's the only explanation...
They are both great cards and they do a great job at 1080p and 1440p.

Power consumption only matters when it's AMD. When nVidia had inferior power consumption for the same (or even worse) performance, everyone still bought nVidia. To me, power consumption is an excuse, not an argument. Not until the one complaining about it can show how it is indeed relevant.

Are they not arranged by price? Take your tin foil hat off and just read the information. The article plainly spells out “get the AMD for your needs”

I did say that the conclusion lines up with the numbers, didn't I?

 

[Nightfire]

@ NightAntilli - Ok you made your point known. Nobody cares but you and you are now just trolling the thread. Way to show gratitude for the work done by Hardware Unboxed.

 

[Vulcanproject]

What annoys me of this article is;

1) The GTX 1060 6GB is put on top, as if it's the faster one, despite the RX 580 being faster in more games
2) Exactly that. The RX 580 is faster in most games (9 out of 12 to be exact), but the few outliers ruin the average framerate so both the GTX 1060 6GB and the RX 580 end up at 82 fps as if they're equal.
3) No chart was added with a direct games list comparison between the GTX 1060 6GB and the RX 580.

It almost seems as if the article is trying to hide that the GTX 1060 6GB is slower than the RX 580 in most games. At least the final recommendation is in line with the numbers.

Presumably founder's edition (reference) GTX1060 6GB versus an RX 580 with an overclock/partner cooler enabling higher boost, because there is no official reference model of that card.

Anything other than a founder's GTX1060 is going to be faster than the performance shown here as it will be clocked higher/boost higher. That's what most people will buy/have bought. The founder's is literally the slowest GTX1060 6GB available.

Just a notation.

 

[Nightfire]

Presumably founder's edition (reference) GTX1060 6GB versus an RX 580 with an overclock/partner cooler enabling higher boost, because there is no official reference model of that card.

I would PRESUME he using using the MSI Gaming X+ version with 9GB/s that is pictured here and used as well in his Nov 15th test.

Nice try though.

 

[Vulcanproject]

I would PRESUME he using using the MSI Gaming X+ version with 9GB/s that is pictured here and used as well in his Nov 15th test.

Nice try though.

Clarification on models would be nice, it really does matter if people are splitting hairs over average frames.

In that November test the MSI was the same performance as the Aorus RX580 over the 21 game average.

Also the Aorus typically overclocks on the memory further, but has next to nothing on the core speeds left to give. It's pushed to the absolute limit. The MSI model usually has decent gains left in it if you overclock, like most GTX1060 models. There is a wide range of GTX1060 models, the vast majority are capable of exceeding 2GHz sustainable boost speeds comfortably.

Not sure what you mean by 'nice try.' It doesn't sound too useful to the discussion.

 

[Nightfire]

In that November test the MSI was the same performance as the Aorus RX580 over the 21 game average.

And it was the same performance here as the RX 570 was 14% slower than both.

Not sure what you mean by 'nice try.' It doesn't sound too useful to the discussion.

It was a nice try casting doubt on the validity of the test even though there was no grounds to it.

 

[Vulcanproject]

And it was the same performance here as the RX 570 was 14% slower than both.

Drivers and games change even in a short space of time. We are told the driver versions but not the card models, which is rather important. Not to mention the CPU and memory configuration is not identical to the November test you speak of.

It seems you didn't grasp this.....

It was a nice try casting doubt on the validity of the test even though there was no grounds to it.

Nice try? What on earth are you still babbling on about my noobish fellow?

A clear indication of the models used in the text would be preferable. The powercolor RX580 is priced and named, but the picture is of the Aorus. No confirmation of any of the other models is given, except a photo of a GTX1060. I don't know about you but tests like these would be of a massive benefit outlining clearly all the precise GPU models used, especially if the comments below start to split hairs over framerates.

As such a wide variety of GTX1060 6GB SKUs with a massive array of potential clock speeds can make quite a difference. As I highlighted to the person first complaining about this performance comparison.

Here endeth the lesson.