[Techspot] 6 generations of AMD Radeon $200 GPUs compared

[Mondozei]

Here it is.

 

[frozentundra123456]

Very nice improvement in performance per dollar, but terrible efficiency (for a die shrink) for the 480. Instead of the 2.8x promised, it is less than twice that of a 3 generation old card.

 

[ConsoleLover]

Very nice improvement in performance per dollar, but terrible efficiency (for a die shrink) for the 480. Instead of the 2.8x promised, it is less than twice that of a 3 generation old card.

Who the fuck cares? Who cares if it consumes 180W with 8pin and overclocked from AIB partners, GTX 1060 OC'd with 8pin from partners also consumes 30-40W more, so the difference between the top OC's partner cards power draw is about 30W, essentially you'll save $4-5 per year on electricity, big deal!!!!!!!

Its not like the difference is 100W at least, its minor difference and the performance it offers for $200 is amazing, hopefully though the price do come down, I actually can't find any $200 4GB 480, the cheapest one is $220 custom cards. But most 1060 6GB's are $280 to $330

 

[guachi]

Thanks for the link.

The article is basically why I finally got around to upgrading my card last month. The 480 is just a huge improvement.

 

[guskline]

I also enjoyed the article since I owned a Radeon 5830 and really liked it. My RX 480 at $239.99 reminds me of the 5830. It wasn't the fastest BUT a solid card that played games of the day well.

I have the RX 480 paired with an inexpensive 24" freesync monitor and it performs very well in DOOM, BF1 beta etc.

 

[DAPUNISHER]

I have a 7850 2GB still chugging away in my wife's boxen (Farmville 2 is the hardest workout it gets), a 380 4GB in a guest boxen, and a RX 470 in another boxen. I still consider it as having 3 of the cards on the list. I have Intel+GTX too, just sayin'

The 7850 was $150 back in '13 talk about value. That I could still fire up some modern games at 1080p med. settings I would not have expected. The 380 was a deal from GPU Shack, good guys. Gave me a better trade in price for a GTX 760 2GB than I could have gotten selling it, and they paid for the shipping.

 

[mohit9206]

They said they also did a similar article for Nvidia $200 cards but i can't find it.
Anyway a good article.RX480 on average 2.5 times faster than HD7850 at the same $250 price point.
Wish we could have seen the same kind of performance increase for sub $150 cards.

 

[Spjut]

It had been interesting to see if the 5830 and 6870 would have gotten to playable levels if they had played on low and at lower resolution. Aside from Overwatch, those games don't officially support the pre-7000 cards.

Speaking of older cards, I've been tempted to get Overwatch and give it a go on the HD 4870.

 

[Face2Face]

Very cool article. I love seeing older cards, and how they perform in newer titles.

 

[SPBHM]

I think they got it wrong with the 6870 the card had 32ROPs, not 16.

 

[kalrith]

They said they also did a similar article for Nvidia $200 cards but i can't find it.
Anyway a good article.RX480 on average 2.5 times faster than HD7850 at the same $250 price point.
Wish we could have seen the same kind of performance increase for sub $150 cards.

They said they did an article "comparing six generations of flagship GeForce graphics cards", not $200 GeForce cards. Here's the article: http://www.techspot.com/article/928-five-generations-nvidia-geforce-graphics-compared/

 

[sm625]

It is funny how the power consumption of cards in the $200 range has ballooned up to such a degree. Of course this article is mainly irrelevant since the RX480 is not anywhere near a $200 card. You will be lucky to find a 4GB version for $230. What was the true market price of a 7850 3 months after release? I am like 90% sure this one was much higher as well.

 

[SPBHM]

since they tested the 1GB 7850, I'm pretty sure 3 months after release the card was around $200 (the cheaper models)...

but yes, currently the 470 is the more realistic $200 card.

 

[Headfoot]

It is funny how the power consumption of cards in the $200 range has ballooned up to such a degree. Of course this article is mainly irrelevant since the RX480 is not anywhere near a $200 card. You will be lucky to find a 4GB version for $230. What was the true market price of a 7850 3 months after release? I am like 90% sure this one was much higher as well.

Within $30 is easily "anywhere near" $200. Ridiculous hyperbole.

 

[DAPUNISHER]

Within $30 is easily "anywhere near" $200. Ridiculous hyperbole.

It is a head scratcher how people compartmentalize their thinking.The way they value their purchases and rationalize which are ok or not is wacky. Heck, the review sites have been doing it. The mantra was the 470 is a good card with a bad price because it showed up $200 base instead of $175. All said with a straight face while wasting money drinking a $4 coffee and eating a $10 lunch. Heck go to the movies once with your s.o. or family, with popcorn and drink, and there is that money and more. All so they can complain how bad the movie was in ATOT.

Also, did I miss it or did the 3GB 1060 experience any of the wrong/overpriced propaganda when launched? Less ram, same price, go buy it, is how I remember it happening. No agenda at work there eh? Or maybe, just maybe, if they stop lazily running canned benches, meaningless 30 second custom loops, and actually play the games, they would conclude their reviews differently. Benchmarks are important, they let me determine if hardware is running properly or not. Beyond that, the typical spammed histograms are of little value to me. To paraphrase 5th element: Histograms not important. Only gameplay experience important.

And it is just obfuscation anyways. These articles are always based on MSRP, as they should be.

 

[Rifter]

It is funny how the power consumption of cards in the $200 range has ballooned up to such a degree. Of course this article is mainly irrelevant since the RX480 is not anywhere near a $200 card. You will be lucky to find a 4GB version for $230. What was the true market price of a 7850 3 months after release? I am like 90% sure this one was much higher as well.

Its as close or closer, to its MSRP than a GTX 1070 is to its MSRP of $379.

Which is all irreverent anyways as this should be based on MSRP not whatever the current market prices are. Basing off MSRP is the best way to do a $200 roudup type of review IMO.

 

[frozentundra123456]

Who the fuck cares? Who cares if it consumes 180W with 8pin and overclocked from AIB partners, GTX 1060 OC'd with 8pin from partners also consumes 30-40W more, so the difference between the top OC's partner cards power draw is about 30W, essentially you'll save $4-5 per year on electricity, big deal!!!!!!!

Its not like the difference is 100W at least, its minor difference and the performance it offers for $200 is amazing, hopefully though the price do come down, I actually can't find any $200 4GB 480, the cheapest one is $220 custom cards. But most 1060 6GB's are $280 to $330

Apparently AMD does since they hyped up the supposed 2.8 x improvement in perf/watt and ended up with only about half that.

 

[frozentundra123456]

It is a head scratcher how people compartmentalize their thinking.The way they value their purchases and rationalize which are ok or not is wacky. Heck, the review sites have been doing it. The mantra was the 470 is a good card with a bad price because it showed up $200 base instead of $175. All said with a straight face while wasting money drinking a $4 coffee and eating a $10 lunch. Heck go to the movies once with your s.o. or family, with popcorn and drink, and there is that money and more. All so they can complain how bad the movie was in ATOT.

Also, did I miss it or did the 3GB 1060 experience any of the wrong/overpriced propaganda when launched? Less ram, same price, go buy it, is how I remember it happening. No agenda at work there eh? Or maybe, just maybe, if they stop lazily running canned benches, meaningless 30 second custom loops, and actually play the games, they would conclude their reviews differently. Benchmarks are important, they let me determine if hardware is running properly or not. Beyond that, the typical spammed histograms are of little value to me. To paraphrase 5th element: Histograms not important. Only gameplay experience important.

And it is just obfuscation anyways. These articles are always based on MSRP, as they should.

Are you kidding? The 3gb 1060 was bashed mercilessly in this forum, despite the fact that it is cheaper than the 6gb and offers competitive performance in the vast majority of games. In fact it was repeatedly called "trash" and "DOA" .

 

[frozentundra123456]

nvm

 

[96Firebird]

Its as close or closer, to its MSRP than a GTX 1070 is to its MSRP of $379.

Wrong, cheapest GTX 1070 is $400 at Newegg...

 

[DAPUNISHER]

Are you kidding? The 3gb 1060 was bashed mercilessly in this forum, despite the fact that it is cheaper than the 6gb and offers competitive performance in the vast majority of games. In fact it was repeatedly called "trash" and "DOA".

FTFY.

BTW, I was referring to review sites. Message boards are a different beast.

 

[SpaceBeer]

This is the first "test" which shows RX 480 consumes more power than R9 380 and 270X. Also, all these cards should consume more or less the same amount (± 10-20W). So I find these results completely irrelevant, at least those for power consumption. I don't know if the FPS measurements are good.

 

[eRacer]

This is the first "test" which shows RX 480 consumes more power than R9 380 and 270X. Also, all these cards should consume more or less the same amount (± 10-20W). So I find these results completely irrelevant, at least those for power consumption. I don't know if the FPS measurements are good.

The power consumption results look fine to me. Other reviews show RX 480 power consumption around that of 380/380X, and the RX 480 can consume 40+ watts more than 270X in gaming benchmarks. The RX 480 is pushing the CPU and rest of the system harder than that of the 270X, which is why system power consumption of the RX 480 is more than ~40W higher than that of the 270X system.

 

[SpaceBeer]

Variations are huge when you compare results from several reviewers (AT, HC, TPU, Guru3D, Computerbase...). For example, in CBs test RX 480 consumes ~40W more than R7 370. And we know R9 270X consumes ~30W more than R7 370. In Guru3D test, RX480 consumes less then R9 270X
https://www.guru3d.com/articles_pages/gigabyte_radeon_rx_480_g1_gaming_review,5.html

So I find all these tests useless

 

[ElFenix]

Variations are huge when you compare results from several reviewers (AT, HC, TPU, Guru3D, Computerbase...). For example, in CBs test RX 480 consumes ~40W more than R7 370. And we know R9 270X consumes ~30W more than R7 370. In Guru3D test, RX480 consumes less then R9 270X
https://www.guru3d.com/articles_pages/gigabyte_radeon_rx_480_g1_gaming_review,5.html

So I find all these tests useless

performance per watt is a weird test. playing a game isn't like transcoding a movie. transcoding a movie is a set task that has a set number of frames to process, so the energy to complete the task can be measured directly. and that benchmark tells you exactly what you need to know for comparison purposes - what's the total energy used to recode this movie?

but we don't game that way. if you measure power consumption using a canned benchmark, hurry up and wait comes into play. but more FPS doesn't get me through the level faster. it's going to take me 10 minutes regardless.

imho, testing for performance per watt (or even better, watt per performance) should be done with locked performance. set vsync and go at it. test at different FPS levels. note which cards can't hold what levels and for what percentage of the test.

this also has the benefit of locking the processor's performance and many of the other system components. eliminate variables!