Is it worthwhile to upgrade to GTX 780ti from Sapphire custom OC R9 290?

[phexac]

I recently got a 34" ultra wide screen, so the resolution I'm working with is 3440x1440.

I am looking at the GTX 780ti that's custom OC and actually offers 10%+ improvement even compared to the stock 780ti:

http://www.newegg.com/Product/Produ...yMark=False&IsFeedbackTab=true#scrollFullInfo

The demanding games I am interested in are Far Cry 4, Crysis 3, Witcher 3 and the like. The NVIDIA card is significantly faster (and the OC one is faster still).

There are two reasons to forgo this update:
1. the 290 I have can play these games well enough that I will not see a difference in actual gameplay.

2. There is a new generation of cards coming out soon that should eclipse the 780ti (or at the very least offer the same performance at a lower price point). Anyone know if something is in the works over the next few months?

Appreciate the advice.

EDIT: Yeah, I meant 980ti

 

[crisium]

780Ti? You are linking to a 980Ti.

 

[happy medium]

no that's not a good upgrade.

 

[Deders]

At that res every advantage helps. You'll also be able to enable gameworks features, but that might bring framerated back down in some cases.

How much is the TI?

 

[SPBHM]

I think there is enough gain from a 290 to a 980 Ti OC to justify it as an upgrade, but if you are worried about new products being a lot better, holding with the 290 seems like a logical decision at this point, 290 is still quite good, and the next gen of cards should offer a decent gain (new process and memory tech are expected, also more adapted to DX12)

(I'm going to assume 780 was a typo)

 

[happy medium]

GTX 780ti

? 980ti? that's a very good upgrade

 

[IllogicalGlory]

If your motherboard supports it, you could buy a second 290 and save yourself $350.

In the titles you posted at least, Crossfire seems to work rather well.

 

[phexac]

Yeah purely looking purely at performance, it looks like the GTX 980ti (especially with 10%+ factory OC, most of which seems to scale almost linearly with game performance) is a huge upgrade from R9 290. On average it looks to offer a 50% to 70% improvement in frame rates. I guess my main concern would be that the next generation of cards will offer higher/similar performance at a lower price point while also offering better DX12 support.

 

[Headfoot]

I was just at the point of deciding whether to upgrade from 290 to 980 Ti for an eyefinity set up (5760x1080). Ultimately I decided to get a second 290 instead, as I could get the second 290 for $175 used and get similar performance in most games for a heck of a lot less money. I will have to deal with dual card problems, but for saving nearly $400 I think its worth it. (I dont usually sell my old video cards immediately, but i swap it down to upgrade my guest machine and retire the old card there)

If you have the mobo and CPU grunt to do it, I'd just get a second 290 and wait for 14+16nm cards to come out in the spring/summer 2016.

 

[phexac]

I was just at the point of deciding whether to upgrade from 290 to 980 Ti for an eyefinity set up (5760x1080). Ultimately I decided to get a second 290 instead, as I could get the second 290 for $175 used and get similar performance in most games for a heck of a lot less money. I will have to deal with dual card problems, but for saving nearly $400 I think its worth it. (I dont usually sell my old video cards immediately, but i swap it down to upgrade my guest machine and retire the old card there)

If you have the mobo and CPU grunt to do it, I'd just get a second 290 and wait for 14+16nm cards to come out in the spring/summer 2016.

My issue with that I would also need to upgrade the power supply since my current one is a 620W Corsair. Outside of costing money for 800W+ to run dual R9 290s, it is also extremely annoying to rewire the entire interior of the PC . Plus there are the dual card issues that I don't really feel like dealing with.

I am leaning towards waiting for the cards based on the new manufacturing process though... Probably 75%+ of my gaming is actually League of Legends, and that can run on a calculator as far as graphics requirements are concerned. R9 290 has no problem maintaining the top allowable 60 FPS at max settings even at 3440x1440.

 

[ThatBuzzkiller]

Just upgrade to a 980 Ti ...

Don't expect a moderate increase in performance for price when these days we have to deal with rising transistor costs ...

 

[Erenhardt]

Just upgrade to a 980 Ti ...

Don't expect a moderate increase in performance for price when these days we have to deal with rising transistor costs ...

I second that. Buy now and next year. We need more people to upgrade if we are going to see another record breaking year from nv.

 

[boozzer]

Just upgrade to a 980 Ti ...

Don't expect a moderate increase in performance for price when these days we have to deal with rising transistor costs ...

you sound like you work for amd or NV.

 

[phexac]

you sound like you work for amd or NV.

I love the online paranoid contingent that somehow thinks that internet forums are infested with disguised employees of tech companies.

 

[daveybrat]

There's no way i'd personally upgrade from a 290 at this point in time. Like the others have said, your card is still fine for today's games and Pascal should be a nice improvement over Maxwell.

Upgrade next summer

 

[Silverforce11]

Do you like having your expensive GPU made obsolete within ~6 months?

That's what it's like buying a top-GPU now as 14/16nm true next-gen looms on the near horizons.

It's akin to paying full price for a 780Ti ($750) last year, just before Maxwell landed and demolished it.

 

[RussianSensation]

I am leaning towards waiting for the cards based on the new manufacturing process though... Probably 75%+ of my gaming is actually League of Legends, and that can run on a calculator as far as graphics requirements are concerned. R9 290 has no problem maintaining the top allowable 60 FPS at max settings even at 3440x1440.

If 75% of your time you spend playing LoL where your existing card is great, why would you spend $600+ on a new card? Didn't you just answer your own question in this thread? BTW, what's your CPU as your signature appears outdated.

 

[RussianSensation]

If your motherboard supports it, you could buy a second 290 and save yourself $350.

His monitor is 3440x1440 though not 2560x1440.

A single 980Ti will struggle with The Witcher 3 or Crysis 3 maxed out. I guess if he is willing to spend $600+ on a new card, I'd recommend waiting until 14nm/16nm cards at this point. Even a 980 successor should beat 980Ti. It might also be possible to score 970 SLI successors for $700-800. If we look at 970 SLI vs. 980, it is clear as ever that 970 SLI users made the right choice vs. 980 and I expect 970 SLI successors on 14nm/16nm could be quite the price/performance combo!

I am leaning towards waiting for the cards based on the new manufacturing process though... Probably 75%+ of my gaming is actually League of Legends, and that can run on a calculator as far as graphics requirements are concerned. R9 290 has no problem maintaining the top allowable 60 FPS at max settings even at 3440x1440.

If 75% of your time you spend playing LoL where your existing card is great, why would you spend $600+ on a new card? Didn't you just answer your own question in this thread? BTW, what's your CPU as your signature appears outdated.

Just upgrade to a 980 Ti ...

Don't expect a moderate increase in performance for price when these days we have to deal with rising transistor costs ...

Why do people keep saying this? If the next generation brings 70-100% increase in perf/watt, that means a $649 card would demolish a 980Ti, while even a $449 one would beat Fury/980.

October 2013 R9 290X = $549
November 2013 GTX780Ti = $699
September 2014 -> $330-350 GTX970
November 2015 -> $250-280 970/390 can be found on sales or essentially half-price of a 290X or more than half-price of a 780Ti

Sure, we might not see immediate price wars and price drops as both AMD and NV have to launch cards but it's going to happen. Either way, older cards like 980Ti and Fury X will need to be cleared out. This always happens.

 

[boozzer]

I love the online paranoid contingent that somehow thinks that internet forums are infested with disguised employees of tech companies.

I think the snarky part of my post just flew right over your head.

 

[ThatBuzzkiller]

Why do people keep saying this? If the next generation brings 70-100% increase in perf/watt, that means a $649 card would demolish a 980Ti, while even a $449 one would beat Fury/980.

October 2013 R9 290X = $549
November 2013 GTX780Ti = $699
September 2014 -> $330-350 GTX970
November 2015 -> $250-280 970/390 can be found on sales or essentially half-price of a 290X or more than half-price of a 780Ti

Sure, we might not see immediate price wars and price drops as both AMD and NV have to launch cards but it's going to happen. Either way, older cards like 980Ti and Fury X will need to be cleared out. This always happens.

Perf/watt means jack when it comes to high powered systems like desktops. What chip designers want is higher perf/area to reduce manufacturing costs. A 16nm $650 part won't even come close to "demolishing" current top end parts. At best even with aggressive overclocks and a new microarchitecture it will only outperform it by around 20% assuming the ideal linear performance/transistor scaling ...

Price wars won't happen so often like they used to anymore when you can't get any cost scaling from from new high end transistor technology and I doubt that either AMD or Nvidia are interested in lowering their profit margins too for that to happen ...

 

[aaksheytalwar]

I expect the next gen to be 30-60% faster at their respective price points, out of the gate. 12-24 months from launch, they'll probably be close to 60-100% faster compared to cards at the same price right now. They may be faster still in fact.

 

[boozzer]

I expect the next gen to be 30-60% faster at their respective price points, out of the gate. 12-24 months from launch, they'll probably be close to 60-100% faster compared to cards at the same price right now. They may be faster still in fact.

nah 30-50% is the max. because we will be stuck with 16nm for a long time just like 28nm.

 

[RussianSensation]

At best even with aggressive overclocks and a new microarchitecture it will only outperform it by around 20% assuming the ideal linear performance/transistor scaling ...

Ok, we'll just agree to disagree here. You also didn't account for next gen games taking advantage of more demanding features (Asychronous compute, Voxel global illumination, etc.) that will run faster on newer architectures. Sure, if you want to derive your averages based on 1000s of games, then you may be correct since we have had many years of straight up console to PC ports using outdated game engines. Of course it's not realistic to expect the new GPU architecture to run older game engines 80-100% faster since they don't have the latest tech and forward looking game code. Secondly, if you focus on 1920x1200 or below benchmarks, sure newer GPus won't be as dramatically faster since they'll be CPU bottlenecked.

What about newer games? 780Ti gets killed by Maxwell.

Don't you think that games released in 2016-2018 will use more demanding graphical effects, perhaps next gen effects that run much faster on newer architectures? You are also assuming that NV will focus just as heavily on Maxwell driver optimizations in 2016 and beyond which is a bold assumption after what happened with Kepler. Driver optimization alone could give Pascal a 10-20% advantage over Maxwell in 2016 and beyond titles.

nah 30-50% is the max. because we will be stuck with 16nm for a long time just like 28nm.

Some of you guys are too pessimistic. The move from Maxwell to Pascal should be as good as Fermi to Kepler, if not better.
- New node (just like Fermi to Kepler)
- New architecture (just like Fermi to Kepler)
- 1TB/sec HBM2 (memory bandwidth increase of 3X over Maxwell while GTX580 to 780Ti was moving from 192GB/sec to just 336GB/sec).

Sure, NV/AMD might spread the increases over the course of 2 years but we should have a card 80-100% faster than the 980Ti before Volta generation launches in 2018. GTX680 was 33-35% faster than the 580 on day 1.

 

[boozzer]

Some of you guys are too pessimistic. The move from Maxwell to Pascal should be as good as Fermi to Kepler, if not better.
- New node (just like Fermi to Kepler)
- New architecture (just like Fermi to Kepler)
- 1TB/sec HBM2 (memory bandwidth increase of 3X over Maxwell while GTX580 to 780Ti was moving from 192GB/sec to just 336GB/sec).

Sure, NV/AMD might spread the increases over the course of 2 years but we should have a card 80-100% faster than the 980Ti before Volta generation launches in 2018. GTX680 was 33-35% faster than the 580 on day 1.

I don't think you understood my posts rs. the point is if the move to 16nm results in 100% increase in performance, it will be spread out over 3 gens of gpus. that was my point.

 

[RussianSensation]

I don't think you understood my posts rs. the point is if the move to 16nm results in 100% increase in performance, it will be spread out over 3 gens of gpus. that was my point.

Ya, I don't agree with that. I think Pascal will give us 80-100% and Volta another 50-70%. Pascal gen could last from Q2 2016-Q2 2018 where Volta will pick up from Q3 2018-Q3 2020. By August 2020, I expect a single chip flagship card to be 2.7X-3.4X faster than 980Ti (and I am leaning more towards the 3.3-3.4X side, with 2.7X being a conservative estimate).

To put it into perspective, November 2010 GTX580 (1.0X) vs. May 2015 GTX980Ti (3.3X faster) and 980Ti OC is 3.63X faster. So it took about 4.5 years to increase performance 3.3X. From May 2015 (980Ti's launch) to August 2020 is more than 5 years. If testing GPU performance at non-CPU limited resolutions such as 3440x1440 or 4K, there is little doubt in my mind that in 5 years we'll have a GPU 3.3-3.4X faster than the 980Ti.

From now on, AMD/NV will have to focus on improving performance from an architectural point of view as they cannot rely on node shrinks every 18-24 months as was the case for decades.