980 vs 290x/390x any new set of benchmarks?

[positivedoppler]

Gets debated here all the time that AMD ages better. The 980 originally cruised passed the 290X on the majority of the benchmarks on initial release. It's been over 2 years since and with all the focus on the new generation of finfet gpus I wonder if things are still the same with the old 28 nm cards. Thoughts?

 

[tviceman]

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080_Ti/30.html

http://gamegpu.com/test-video-cards/opredelenie-naibolee-bystrykh-i-optimalnykh-reshenij-2016-goda

Compared to release reviews of the 390x, not much has changed. They're pretty close at 1080p and 1440p (the same as when the 390x came out), but the 980 has WAY more OC headroom and WAY less power consumption. Both cards are still great 1080p, adequate 1440p solutions, and insufficient at 4k.

 

[caswow]

actually the 390x = 290x and its competition was the 780ti.
290x 5. Nov. 2013
780ti 7. Nov. 2013

i would much rather put the 980 against the r9 nano.
980 19. Sep. 2014
r9 nano 10. Sep. 2015

prices were at least in europe pretty much the same.

each of them had similar power consumption.

 

[tviceman]

actually the 390x = 290x and its competition was the 780ti.
290x 5. Nov. 2013
780ti 7. Nov. 2013

i would much rather put the 980 against the r9 nano.
980 19. Sep. 2014
r9 nano 10. Sep. 2015

prices were at least in europe pretty much the same.

each of them had similar power consumption.

Yes, its the same chip but the 390x was nearly 10% faster than the 290x because of the significantly faster vram and core clock bump. Also, the 390x was released AFTER the 980 came out, after the 780 TI had been discontinued, and it's performance was about equal to the 980 in stock vs. stock comparisons. The 390x's release date was closer to the 980 than Nano's release date.

The Nano was a $650 USD card when it came out while the 980 was $500 USD at the time of it's release. http://www.anandtech.com/show/9621/the-amd-radeon-r9-nano-review You're either mis-remembering prices OR European prices were whack.

 

[caswow]

i doenst matter what their msrp or benchday price was. streetprices only thats it. and no i remember right because you can check the price on geizhals.at for its lifetime.

and no the release date for the 390x was 18. Jun. 2015 while the r9 nano came 9 days prior to the 980. depending on the 980 they were ~50€ appart in their cheapest times in their first months.

 

[Rannar]

i doenst matter what their msrp or benchday price was. streetprices only thats it. and no i remember right because you can check the price on geizhals.at for its lifetime.

and no the release date for the 390x was 18. Jun. 2015 while the r9 nano came 9 days prior to the 980. depending on the 980 they were ~50€ appart in their cheapest times in their first months.

You might want to recheck that. Nano came out sept. 2015 and 980 in sept. 2014. 980 main competitor was 390x

 

[IEC]

Both are pointless in today's market when you can get a RX 480/580 or GTX 1060 with similar performance for less money and less power consumption. In North America at least, the 4GB RX 480 has been as cheap as $140 after rebates.

 

[OatisCampbell]

Both are pointless in today's market when you can get a RX 480/580 or GTX 1060 with similar performance for less money and less power consumption. In North America at least, the 4GB RX 480 has been as cheap as $140 after rebates.

BUT the question was "Did the 290X age better?" which we've all seen asserted. Given some slop for difference in tested games then and now, and small difference in resolutions, I'd say the 290X gained ground. Here's why:

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_980/26.html

980 vanilla 13% faster than 290X at 1600p, the highest res you would use either card.

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080/26.html

980 vanilla down to 4% faster at 1440p, a lower resolution. Definitely some gain by the 290X, and 390X 1% faster than vanilla 980.

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080_Ti/30.html

Latest benches vanilla 980 claws back to 1% above 390X (which was 5% above 290X) so we might extrapolate the 980 now 6% better than 290X.

So the 290X halved it's performance deficit over the last three years.

What does all this mean?

In 2014, ponying up your $500 got you 13% better 1600P and a whopping 19% better 1080p.

This year you're probably looking at a 5-6% difference at the 1440p and the 980/390X holds pretty steady at 3-4% 1080p.

So was the 980 a good deal?

Depends how you look at it. 19% is almost generational difference these days, and 13% nothing to sneeze at. A year of that level of difference has value.

However; these days all of these cards pretty have yawn worthy performance that is pretty close, so guys like me who held onto our 290s like they were the last breasts on the planet got some redemption through driver updates and game changes.

 

[caswow]

You might want to recheck that. Nano came out sept. 2015 and 980 in sept. 2014. 980 main competitor was 390x

yes you are right i might have overseen this. we actually got a little bit ot there.

 

[tviceman]

BUT the question was "Did the 290X age better?" which we've all seen asserted. Given some slop for difference in tested games then and now, and small difference in resolutions, I'd say the 290X gained ground. Here's why:

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_980/26.html

980 vanilla 13% faster than 290X at 1600p, the highest res you would use either card.

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080/26.html

980 vanilla down to 4% faster at 1440p, a lower resolution. Definitely some gain by the 290X, and 390X 1% faster than vanilla 980.

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080_Ti/30.html

Latest benches vanilla 980 claws back to 1% above 390X (which was 5% above 290X) so we might extrapolate the 980 now 6% better than 290X.

So the 290X halved it's performance deficit over the last three years.

What does all this mean?

Hawaii stopped gaining ground by the time 390x came out and it's been relatively stable since.

 

[Mercennarius]

My 390X has gained almost 10% in 3DMark benchmarks in just over a year...almost every new driver has improved my performance. At stock clocks my graphics score is now over 13,500 in 3DMark. Still plays the newest games on high settings great.

 

[ocre]

BUT the question was "Did the 290X age better?" which we've all seen asserted. Given some slop for difference in tested games then and now, and small difference in resolutions, I'd say the 290X gained ground. Here's why:

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_980/26.html

980 vanilla 13% faster than 290X at 1600p, the highest res you would use either card.

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080/26.html

980 vanilla down to 4% faster at 1440p, a lower resolution. Definitely some gain by the 290X, and 390X 1% faster than vanilla 980.

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080_Ti/30.html

Latest benches vanilla 980 claws back to 1% above 390X (which was 5% above 290X) so we might extrapolate the 980 now 6% better than 290X.

So the 290X halved it's performance deficit over the last three years.

What does all this mean?

In 2014, ponying up your $500 got you 13% better 1600P and a whopping 19% better 1080p.

This year you're probably looking at a 5-6% difference at the 1440p and the 980/390X holds pretty steady at 3-4% 1080p.

So was the 980 a good deal?

Depends how you look at it. 19% is almost generational difference these days, and 13% nothing to sneeze at. A year of that level of difference has value.

However; these days all of these cards pretty have yawn worthy performance that is pretty close, so guys like me who held onto our 290s like they were the last breasts on the planet got some redemption through driver updates and game changes.

You know, its funny to see all the different ways history gets rewritten. It depends entirely on how the person decides they want it to go.

So, 2014.....
You might be surprised to hear that people were paying astronomical prices for the 290x. You forgot the 290x launch price was $550, but that fact is long forgotten. But, lets talk about 2014. The retail 290x was shockingly expensive, and it went on for months. There are articles that record, people were paying $900 for the retail 290x.

In 2014, ponying up $900 for a retail 290x could have got you a 15-18% slower card that would have cost 170% as much.

But, see, as absurd as it may sound, it shows how things are skewed to fit whatever narrative.
The numbers and year is all real. You just chose to look at things from a very tiny pin hole. Reality is much more complex than its ever repeated...more vast and complex than comprehended, much less recollected.

The expensive 290x was a huge deal. I remember it well. This certainly hampered the excitement, especially considering that there was a very formidable and abundant 780ti out at the same time.

When you talk about 2014, it seems long forgotten that the gtx980 launched with high praise and near universal glow. Its price was a shock. The 290x was not this extremely cheap card you remember, those low prices you remember are not how the story began. AMD slashed, slashed, and slashed prices in response.

Your statement is strange, see the 290x was over $500 in 2014. For the first quarter, it would cost well over 500, upwards past $800 and towards a grand. The 290x finally got back down to 550 but even after the 980 launch, it would cost you $500. AMD barely responded, dropping the price a tiny bit.

Here is the bad part.. At that time, the 330 dollar 970 was all over the 290x in performance. Yet the 290x remained priced $500, significantly higher. AMD dragged their feet for a good while until eventually the fire sales rolled in as the stock and shelves were piled up.

In 2014, when the 980 launched...you had the choice between a slower card that used considerably more power or a new overclocking monster that at stock was power sipping and up to ~20% faster. AMD lowered prices to 500 bucks in response to the 980. Its a far different scenario than gets painted years later. People act like the fire sale 290x prices that came way later was how things were all along. But its far from the truth.

You know, the 980 came with a huge splash. There was so much overclocking left on the table with the 980. Its almost always ignored but surprisingly enough an overclocked 980 was pretty close in performance to the fury non x that launched at almost $600 way way later. People who bought a 980 and overclocked it, they had fury non x performance in 2014.....in a card that was cheaper than the Fury would be when it finally launched.

People see things strange looking back. The hawaii was always a great chip but it didnt launch at 200 bucks. The fire sales were reactionary to a market that had lost interest in that chip. There were a lot of factors which played a hand, none of them were a fault in the chip itself. There is no need to rewrite all these different versions of history. Anyone can take moments of different times and blur them together to tell a story, but reality is much more complex and has so many factors its almost impossible to keep up and completely impractical properly discuss here. Its much more complex if you really want to go that many years ago.

 

[Shmee]

The way I see it, getting a 290 on launch or near after would have been best bang for buck. 290X not really worth it. If one got their's at ~$400 before scrypt mining boom happened, they are set till Pascal IMO. Or possibly longer, if they want to wait for Vega.

I got my 290 at launch, a reference Powercolor card. Paid $399 for it, and much more than that back on it through mining. I still have it in an old X58 rig. Of course, once the scrypt boom happened, prices skyrocketed. Fast forward, I picked up 2 290 Tri-X cards for crossfire. Got them on ebay, $290 each. Not bad, for nearly double performance. And then came the eth boom. Again, cards easily paid for themselves and then some. Then pascal launched. I got an MSI 1070 Gaming X soon after launch, and it has worked great. About the performance of my 2x290s, but less power usage and a single card. It has also helped earn its keep in eth and Zcash mining, as well as gaming.

The above poster is right, there is a complex history, but if one bought a 290(x) at the right time, sometimes getting a used one on the cheap, they were great cards to get, at least until Polaris launched. Optimally, one would have bought a 290 at launch or soon after, or would have got a good used one once prices came down after the mining boom. At this time, prices would start at around $150 for a used 290, and I believe they are still going for around the same today. Now of course, there are better options to buy at that price range. But before Polaris and before Lower end Pascal (1060 and down) there was not really a better bang for buck at that price range.

Hawaii has certainly had a good run, and is still a good option today, if one already made the purchase.

 

[Bacon1]

So, 2014.....
You might be surprised to hear that people were paying astronomical prices for the 290x. You forgot the 290x launch price was $550, but that fact is long forgotten. But, lets talk about 2014. The retail 290x was shockingly expensive, and it went on for months. There are articles that record, people were paying $900 for the retail 290x.

The reason people were paying those high prices for the cards is because they made massive amounts mining off of them... so we can either call them $900 or maybe we should label them as positive $500 cards since you could make lots of money off them. You are right it makes it easy to rewrite history though

 

[OatisCampbell]

You know, its funny to see all the different ways history gets rewritten. It depends entirely on how the person decides they want it to go.

I wasn't trying to re-write history and remember the great reviews the 980 got. I did forget the price gouging on the Hawaii chips, probably because I never mined.

I got my IceQ 290 for around $300 and it came with a ton of games too, remember thinking it was the best deal I'd gotten on a card. I must have bought it after 980 launch, seems like a million years ago these days.

 

[sm625]

No amount of aging voodoo is going to overcome the perf/watt gap between those two cards. 980 will be faster until the day both cease getting any driver optimizations.

 

[2is]

980 was the better card back then, it's still the better card today.

If this were talking about the 7970 vs 680 it would be a more interesting topic.

 

[Headfoot]

980 was a turd for the majority of its lifespan due to outrageous pricing. It was never worth $550. Only after 980 Ti came out and we were nearly to Pascal did the pricing actually come in line with the small speed bump the 290x/390x -> 980 upgrade got you. And even then, the 980 Ti was a much better buy.

 

[crisium]

The 980 was less viable after the Ti came out, even considering the price cut, imo.

When the 980 was a "flagship", it had its role as regent to the throne. Firesale 290 series offered far better performance-per-dollar once this begun (took some months). And for power efficiency and/or Nvidiaphiles, the 970 was a radically better value. But "flagships" do what they do.

After the Ti though...

$330 970 vs $500 980 vs $650 980 Ti (and firesale 290 series below $300 until 390 launches)

I have no doubt that there were plenty of customers with a budget of at least $500 but less than $650. But I always recommended those people save money on a 970/390/290 series or find a way to stretch to the 980 Ti. Don't forget that the 980 actually has the highest stock clocks of three Nv cards, so for max OCers the 980 Ti pulls away even more and the 970 edges slightly closer.

It is a bit of a shame that the fully enabled GM204 was not so viable though. It was a great card, it just didn't have its role. Even at stock, a 980 Ti offered better performance-per-dollar than a 980 which is bizarre. It's supposed to get worse the more you pay. I think in the year between 980 Ti and 1080, 980 could have been recommended a lot more if it were $430 like the 390X.

 

[IEC]

Which card you like depends on your own preferences in terms of metrics.

If perf/$ is your primary metric, then the 290 non-X was the clear winner.
If perf/W is your primary metric, then the 980 wins, no contest.
If mining performance is your primary metric, then the 290 non-X was the clear winner both from perf/$ and perf/W standpoints.

Obviously, if it did better for your applications, then you will be inclined to view any benchmarks in light of your own favorable experiences with the card (whether consciously or subconsciously).

If building today, 10/10 times I would choose a RX 5xx/4xx or GTX 10-series card over either. I actually did just that, choosing a GTX 1060 over throwing in an R9 290 for a friend's build. For that particular application, modern codec support, modern connectivity, and a small footprint (power and space-wise) was more important than bang for the buck. The R9 290 is relegated to my test bench, simply because it's the only card I have that isn't being used for anything else.

 

[Headfoot]

Which card you like depends on your own preferences in terms of metrics.

If perf/$ is your primary metric, then the 290 non-X was the clear winner.
If perf/W is your primary metric, then the 980 wins, no contest.
If mining performance is your primary metric, then the 290 non-X was the clear winner both from perf/$ and perf/W standpoints.

Obviously, if it did better for your applications, then you will be inclined to view any benchmarks in light of your own favorable experiences with the card (whether consciously or subconsciously).

If building today, 10/10 times I would choose a RX 5xx/4xx or GTX 10-series card over either. I actually did just that, choosing a GTX 1060 over throwing in an R9 290 for a friend's build. For that particular application, modern codec support, modern connectivity, and a small footprint (power and space-wise) was more important than bang for the buck. The R9 290 is relegated to my test bench, simply because it's the only card I have that isn't being used for anything else.

All true. Though its been hashed over and over, there are very few real, material reasons to buy higher perf/w in desktops outside of coin mining, where the 290 was still more efficient due to higher throughput. Buying based on Perf/w 9 times out of 10 was a smokescreen for wanting to buy nvidia out of brand loyalty but not wanting to admit it. And now we see who doth protest too much...