Advice: OC 680 SLI vs stock 290's Xfire

[Unoid]

It's winding down time for my mining rig. with difficulty only going up, I will soon be making 0$ by mining. (@ around 8000 diff) I've already made up my hardware costs..

I game at 1440P at 96hz. Most games currently, I can keep 70-100fps if I cut back on msaa, shadows, etc, like in BF4.

My two 680's are at 1200mhz cores and +250mhz on memory.

Based on stock clock reviews of the 290's in xfire, they would only really be 10-15% better FPS. Not exactly worth it.

So in the next months, Sell the 290's for msrp ($400 USD each), or keep them at stock clocks with reference coolers and sell my two 680's (reference MSI's 2gb. $225-250 each)

I'm leaning towards selling the 290's and waiting for Big chip 20nm single card.

 

[spat55]

I agree with you. I have never had SLI so can't say from hands on experience but Nvidia are meant to have better drivers and you can get a lot of money off those R9 290' and wait for the 20nm GPU' to come.

I want a R9 290 myself but am waiting for 20nm, also my HD 7850 crossfire isn't to bad for the little £180 it is worth now

 

[Silverforce11]

So you're saying you go with 680s and if you cut back on MSAA, shadows, etc...

why?

CF R290 don't have to cut back on IQ at 1440p.

 

[Capt_Kob]

I have crossfire 290s and can play BF4 on ultra getting between 90 and 120 FPS on an 2560x1440 Tempest.
I mean, you HAVE 2 290s in your possession, right? drop them in your gaming rig, fire up BF4 and see for yourself.

 

[DooKey]

So you're saying you go with 680s and if you cut back on MSAA, shadows, etc...

why?

CF R290 don't have to cut back on IQ at 1440p.

I think he's asking whether he should sell the 290s and make get money for slightly less FPS until 20nm or sell the 680s for less money and get slightly more FPS until 20nm.

IMO he should sell the 290s and wait for 20nm cards.

 

[96Firebird]

Honestly, I would try the 290s in CrossFire and see if the noise and heat bother you. If it does, easy decision to make. If it doesn't, bit of a tougher situation. What games do you play?

 

[Attic]

I'd sleep easier with the 290's in xfire. But you gotta try em out first before anything.

 

[KaRLiToS]

Sell those GTX 680. Probably the worst card I used the past two years at 1440p resolution. Even a HD 7970 was way better than that at 1440p. Sell them while they still have some value and keep the R9 290s until 20nm.

The R9 290 will still have good value by this time. Also, no one knows exactly when 20nm will release or how they will perform.

The R9 290s will be good for at least another year and a half at single 1440p monitor. In my case, I will wait for the second generations of 20nm .

 

[thesmokingman]

It doesn't compute? Why would you sell the 290s with Mantle getting better and better. Also 20nm isn't coming around the corner so in all likelihood you could be waiting a long long time and I'd rather not be stuck with the 680s imo.

 

[el etro]

R9 290 SLI still is stronger than the two 680s@1200Mhz, but if you are planning to game at 1440p with good details the 680SLI may fit your needs.

 

[dragantoe]

I consistently have problems with SLI, but I've had a couple crossfire setups and the driver support experience always seemed much smoother. But at 1440p, two gtx 680's would suffice... I'm waiting for Maxwell to upgrade myself

 

[guskline]

Because I like to tinker with water cooling and since your sig says you also custom cool, if you have the rad space, sell the 2 680s, throw water blocks on the R9 290s in CF and have a blast. Especially since your resolution is 1440p

 

[Unoid]

Because I like to tinker with water cooling and since your sig says you also custom cool, if you have the rad space, sell the 2 680s, throw water blocks on the R9 290s in CF and have a blast. Especially since your resolution is 1440p

I'm cooling only my CPU atm. It's not worth it to me to buy 200$ of waterblocks, and add a 100+ radiator to handle the 600+watts. + need PSU if I want to overclock any. PC P&C 850watt.

I'm sure I'll throw int he 290's soon her and see how they work in my games. I expect the 10-20% fps boost (or at least less lowering of options)

two 680's can last me another year for 20nm if required. Or use the 290's and upgrade at 16nm? The question at hand is that I can get above the $400 MSRP I paid for my cards NOW. (a potential benefit of $300) or so. I guess in the end it hardly matters what I end up doing.

 

[Attic]

I think with the 680's you will be tied into an upgrade from now going forward.

The 290's you can enjoy not worry about upgrading for IMO 2years and also enjoy better gaming experience starting today.

I don't think you will get $400 for used 290 reference. $350-$375 is what i'd expect.

Selling the 680's still gets you some scratch.

Bottom'd lined, yep, no matter what you do you are going to have a beastly setup wither either cards.

 

[_UP_]

Sell those GTX 680. Probably the worst card I used the past two years at 1440p resolution. Even a HD 7970 was way better than that at 1440p. Sell them while they still have some value and keep the R9 290s until 20nm.

The R9 290 will still have good value by this time. Also, no one knows exactly when 20nm will release or how they will perform.

The R9 290s will be good for at least another year and a half at single 1440p monitor. In my case, I will wait for the second generations of 20nm .

I think with the 680's you will be tied into an upgrade from now going forward.

The 290's you can enjoy not worry about upgrading for IMO 2years and also enjoy better gaming experience starting today.

I don't think you will get $400 for used 290 reference. $350-$375 is what i'd expect.

Selling the 680's still gets you some scratch.

Bottom'd lined, yep, no matter what you do you are going to have a beastly setup wither either cards.

To add a bit to KaRLiToS and Attic, think of it this way. Let's go with best case scenario for keeping the 680s, you get $800 for you 290s now, but by the time 20nm comes out you're looking at $150 for each 680, at most. So that adds up to $1100.
Let's go with worst case scenario for the other case, you get $225 for each of your 680s now, and only $200 for your 290s when 20nm comes out. Adding this up is $850. That is not a big difference and you'll have a worse experience in the meanwhile. And, this is comparing best case and worse case. Like Attic said, you'll probably get a little less for your 290s now, and more than $200 later. So the delta will be smaller. For better performance.
I'd sell the 680s no doubt, not to mention they only have 2gb RAM which might be a problem in new games about to come out.

 

[Subyman]

It is kind of a hard decision actually. The R9's are going to be extremely noisy especially in crossfire, but you already know that. 680's are plenty fine for most games. I like nvidia drivers more. A buddy has R9 290's in crossfire and he's run into some crossfire issues. If you really just want to hold yourself over until 20nm, then sell the 680's and one R9 290. A single R9 290 should be plenty for most games.

 

[Insomniator]

If you can get 800 for the 290's then do it.

If all you have to sacrifice with 680's is AA/AF then why get rid of them in favor of louder, more expensive cards?

Of course everyone here will say keep the newer, better performing tech... But I'd rather save the money now and buy a single GPU monster in the next year.

 

[TreVader]

20nm won't be out until Q3 2014 at the earliest, and if the maxwell release is anything like Kepler then "Big Die" 20nm won't be around until 2015. It's unlikely GM104 will provide any performance increase over 290 CF anyway.


There are no games out right now that will stress 2x 290 @1440p. You should be fine.

 

[guskline]

I'm cooling only my CPU atm. It's not worth it to me to buy 200$ of waterblocks, and add a 100+ radiator to handle the 600+watts. + need PSU if I want to overclock any. PC P&C 850watt.

I'm sure I'll throw int he 290's soon her and see how they work in my games. I expect the 10-20% fps boost (or at least less lowering of options)

two 680's can last me another year for 20nm if required. Or use the 290's and upgrade at 16nm? The question at hand is that I can get above the $400 MSRP I paid for my cards NOW. (a potential benefit of $300) or so. I guess in the end it hardly matters what I end up doing.

I understand the hesitancy on cooling blocks. My thought was the R9-290s have double the Vram so they will really help in 1440p resolution with all the eyecandy.

No doubt custom water cooling can add up in $$$. It's just that the Hawaii chip being a real heat producer can be effectively cooled and OC'd with water pretty well.

 

[spat55]

If you can get 800 for the 290's then do it.

If all you have to sacrifice with 680's is AA/AF then why get rid of them in favor of louder, more expensive cards?

Of course everyone here will say keep the newer, better performing tech... But I'd rather save the money now and buy a single GPU monster in the next year.

Not everyone as I agree with this I would rather hold on with aging tech then feel more of a jump going to the next thing.

 

[3DVagabond]

Keep the 290's. @ 1440p 680's are becoming bandwidth constrained, VRAM limited, and don't generally scale as well as crossfire Hawaii. BF4+Mantle you are going to be able to enjoy better IP and still keep your frame rates where you like them. Plus, it's exciting messing around with new features like Mantle and TrueAudio. Both techs are in their infancy and should be very interesting and entertaining for the next year or two. I don't think we are going to see 20nm any time soon, if at all. It's starting to be rumored that we might be on 28nm until 16nm FF 2nd half 2015. D:

 

[Silverforce11]

Why do you need waterblocks and a custom loop? It's simply excessively expensive for what it offers. Ghetto water it for dirt cheap.

 

[MeldarthX]

Hmmm........680s are nice cards; but those that say keep them over the 290s for BF4; completely ignoring mantle effect.

That's not a 15-20% upgrade.....normally would be bit more like 25%; with mantle; specially mins; easy 50% upswing.....

That isn't anything to sneeze at. These will hold him for at least 2 years instead of an year....he won't be bandwidth contraint; or will he hit the vram wall; which is something we are starting to see....and will see with new games coming out later this year and next year.