Radeon 5770, 6870, 7870, 270X, 380 RX 480 compared [Computerbase]

[Bacon1]

https://www.computerbase.de/2017-01/radeon-hd-7870-vs-r9-380-rx-480-vergleich/2/

 

[poofyhairguy]

I don't get why those cards were chosen. Why not all x80 or all x70?

The 470 would have still killed it.

 

[Bacon1]

Not sure, guessing price + what they had available to test with

5770: $159 (5850 @ $259 would be better)
6870: $239
7870: $350 (7850 @ $250 would be better)
270x: $199
380: $199
480: $239 (8gb)

 

[Stiliyan]

I don't get why those cards were chosen. Why not all x80 or all x70?

The 470 would have still killed it.

From 5xxx series to 6xxx there was a shift - the new 69xx replaced the 58xx, and 68xx are replacement to 57xx. Similar thing happened with 7870 and the 270x. I don't know the latter is in the test.

 

[RussianSensation]

https://www.computerbase.de/2017-01/radeon-hd-7870-vs-r9-380-rx-480-vergleich/2/

In 12/14 games RX 480 got > 60 fps at 1080p. In Rise of the Tomb Raider it almost got 60 fps -- mild overclocking would have gotten there. Just goes to show that for 1080p 60Hz gaming, there is no need to buy $300-400 GTX1070/980Ti/Fury X level cards. It's better to buy a $160-200 mid-range card, and upgrade again in 2-3 years. Too many gamers "future-proof" with $400 cards for 1080p 60Hz, and once again history shows it makes no sense to buy a high-end card like that to future-proof for 1080p.

HD5850/HD6850/7850/RX470 owners would have done quite well. It's pretty rare for just 1 SKU tier up from the same architecture to significantly outlast the tier below over 2-4 years of ownership. Hence why I went GTX1070 this round and not 1080. The more expensive GPUs become and the more AAA console ports we get, the more it makes sense for most gamers to just get "good enough performance"; and then just upgrade the card in another 2-3 years [of course another option is to resell every new gen].

RX480/1060 are solid cards for 1080p 60Hz, and are a massive improvement from GTX950/960/380. If next gen 2018 $200-300 cards can come close to GTX1080 level of performance, it's going to become very hard to recommend anything above that for 1080p 60Hz gaming. Seems the disconnect between high-end cards and 1080p resolution is going to grow even more in the next 1-2 years and yet so many PC gamers are still using budget monitors with 1080p 60Hz and below specs.

 

[Stuka87]

In 12/14 games RX 480 got > 60 fps at 1080p. In Rise of the Tomb Raider it almost got 60 fps -- mild overclocking would have gotten there. Just goes to show that for 1080p 60Hz gaming, there is no need to buy $300-400 GTX1070/980Ti/Fury X level cards. It's better to buy a $160-200 mid-range card, and upgrade again in 2-3 years. Too many gamers "future-proof" with $400 cards for 1080p 60Hz, and once again history shows it makes no sense to buy a high-end card like that to future-proof for 1080p.

HD5850/HD6850/7850/RX470 owners would have done quite well. It's pretty rare for just 1 SKU tier up from the same architecture to significantly outlast the tier below over 2-4 years of ownership. Hence why I went GTX1070 this round and not 1080. The more expensive GPUs become and the more AAA console ports we get, the more it makes sense for most gamers to just get "good enough performance"; and then just upgrade the card in another 2-3 years [of course another option is to resell every new gen].

RX480/1060 are solid cards for 1080p 60Hz, and are a massive improvement from GTX950/960/380. If next gen 2018 $200-300 cards can come close to GTX1080 level of performance, it's going to become very hard to recommend anything above that for 1080p 60Hz gaming. Seems the disconnect between high-end cards and 1080p resolution is going to grow even more in the next 1-2 years and yet so many PC gamers are still using budget monitors with 1080p 60Hz and below specs.

Actually, all you need to do is undervolt the 480, and its performance will go up. Especially on the reference cards which will throttle at 100% load. Not as much of an issue on non-reference cards though.

 

[crisium]

Not sure, guessing price + what they had available to test with

5770: $159 (5850 @ $259 would be better)
6870: $239
7870: $350 (7850 @ $250 would be better)
270x: $199
380: $199
480: $239 (8gb)

It is an odd choice. The 270X, being an overclocked 7870, is completely unnecessary to include. You're right, 5850 and 7850 would better fit the prices. A 380 is a very light OC 285, and that was a $250 debut, so with those 2 changes they'd have a very good spread of $240-$260 AMD cards from 2009-2016.

I also would have liked to see some Medium or possibly even Low settings to see if VLIW is still playable at all. Was there ever any doubt that mid-range VLIW couldn't run Ultra in modern games?

Bizarre test.

 

[Bacon1]

I also would have liked to see some Medium or possibly even Low settings to see if VLIW is still playable at all. Was there ever any doubt mid-range VLIW couldn't run Ultra in modern games?

Agreed. I hate the "Max everything" settings that most places use, as usually just dropping one or two settings 1-2 notches for almost no IQ loss will double frame rates.

I mean for this test they are trying to show off how much the arch has changed (1000% difference in some games), but having some tests with playable settings would be great.

 

[happy medium]

Mabe because the 370 was slower than 270x? WHere is the 470?
The 380 and 480 don't belong in this test.

 

[AtenRa]

Mabe because the 370 was slower than 270x? WHere is the 470?
The 380 and 480 don't belong in this test.

Why ??