Bonus is in soooo a GTX 680 or an AMD 7970?

[Axon]

No Skyrim, It will be for Tera Online and Diablo 3 mainly. Also a smattering of other MMOs like Wow and DDO from time to time.

If you're in for purely gaming, you can easily get by with a mid-level GPU. If you MUST spend now, my recommendation would be a used 7970. People are unloading them for laughable prices, sometimes as low as $450 because of the 680's rep.

 

[Jacky60]

I'd go for the 680 if I were you unless you can get a 7970 $50 cheaper which I doubt. 680 is quicker and cheaper and the extra 7970 memory doesn't seem enough to tip the scales at multimonitor or very high resolutions.

 

[ElFenix]

this thread is fully of funny

congrats on posting after 12 years of lurking. that's probably a record catch

 

[BallaTheFeared]

No Skyrim, It will be for Tera Online and Diablo 3 mainly. Also a smattering of other MMOs like Wow and DDO from time to time.

Nvidia has much better performance in WoW, and AO support in Diablo 3.

 

[blckgrffn]

The cheapest quiet 7970 is $550. You can almost get a GTX680 and a used GTX460 for folding on the side for that much! If you are gaming at 1080P, keep in mind that your CPU will bottleneck you in some games such as SKYRIM. GTX570/HD7850 are good cards at $250-260 for 1080P unless you need the fastest BF3 performance or want to treat yourself to the best single-GPU card

This is actually probably your best bet if you have the PSU for it.

Should double your GPU folding capacity with just the GTX 460.

And for the x games that use it, PhysX card

As others has said, a GTX 580/570/560 Ti 448 is also a massive step up from your 5770 and you know what you are getting in terms of folding. Honestly, spending ~$270-$300 on a 570 or nicely cooled 560 Ti 448 makes a lot more sense than going for the big fish. That's a big premium you are paying!

In short, I voted "other" because the 7970 and the GTX 680 don't provide the utility their price would require (obviously, very much IMHO.)

It doesn't sound like you are driving a 30" monitor or something where the difference would even be noticeable.

 

[f1sherman]

It may look close, but really it's NOT.

PERFORMANCE: GTX 680 is a faster card.

It's faster at 1080p, a tad faster at 2560, and at multimonitor resolution.
Overclocking is a luck-based attempt, which generally closes the gap , but then you are left with not so stellar perf/W and the jet-like noise from 7970.

7970 is faster card only in cases where bandwidth _is_ an issue for 680,
where there are no CPU bottlenecks whatsoever, and CPU is largely irrelevant - Metro 2033*** being the perfect example.
So right away you can say that 680 will rule the day in any reasonably well coded multiplayer,
because of clients overhead, and 680's graphic workload optimized scheduler.

(***BEWARE: developer's beta of Metro: Last Light has already been seen running on Nvidia hardware and featuring some impressive PhysX/Apex Clothing effects)

PRICE: Wash!

GTX 680 is $50 cheaper, but 7970 comes with 1GB of VRAM, 1 VRM phase more, is sturdier built, has better PCB, and larger bill of materials.

POWER CONSUMPTION: 680 wins again. It depends on a game, but on average 680 will consume something in between of Tahiti Pro/XT.
They are pretty much the same on Desktop, with 7970 winning the day if you tend to leave your PC idling.

NOISE: Nvidia has far mor experience with fighting the heat, and this shows on cooler design.

10db of load difference as measured by Techreport is already a double of GTX 680's perceived loudness.
Imagine what happens when you overclock your Tahiti...

SLI/CF: 680 edges once again.

Nvidia has far better multi GPU support. It won't work always, but it's supported in bigger number of titles then CF.
Also, subsequent drivers do not tend to brake what's already working.

Automatic profiles downloading, and Frame Metering simply do not exist on AMD.
So even in cases when Crossfire scales better, you will be left with choppier game-play (BF3 @ [H])

FEATURES: No contest. 680 hands down!
A quick rundown on the Nvidia's exclusive bag of goodies which is getting heavier by the day:

PhysX, CUDA, GPU Boost aka overclocking for the dummies, hw acceleration in large number of applications, active sensor based TDP/OVP protection (as opposed to AMD empiric TDP limiting)
hardware-based H.264 video encoder, Ambient Occlusion, Nvidia Inspector, temperature and FPS limitter, far more robust anti-aliasing, compatibility bits, superior Full-screen antialiasing support, downsampling, improved driver FXAA, TXAA (soon to rule like the boss!), infinitely better 3D support, automatic profiles downlading, nearly shimmer-free industry leading anisotropic filtering(yeah that includes better ingame image quality - 7970 masks the AF problem via image blurring).... a working Forum, Driver Feedback and Support

Feature list is a main reason why I can't even consider buying AMD at this point in time.

I will also mention far better OpenGL support, traditionally better driver team, faster problem tackling, and a much closer relations with number of developers.

Having 3 distinct architectures in a very short period of time (VLIW4, VLIW5 and GCN), and having an imperative to support all three,
means some tough times are ahead for both AMD driver team and their user base.

In my opinion 7970 should be bought only if you can't find GTX 680, you can't live without OpenCL, or just want to make a statement.

 

[Olikan]

i voted in "other"....... because you should buy the cheapest one,
the 7970 is cheaper than the 680 in many places out-side U.S.

 

[PUN]

I'd go for the 680 if I were you unless you can get a 7970 $50 cheaper which I doubt. 680 is quicker and cheaper and the extra 7970 memory doesn't seem enough to tip the scales at multimonitor or very high resolutions.

Jacky60's Sig:

2x 6990 @ 880
Hanns G. 28inch @ 1920x1200

epic fail

seriously, get rid of one 6990 + some cash and get yourself a 2560x1600.

 

[JAG87]

It may look close, but really it's NOT.

PERFORMANCE: GTX 680 is a faster card.

It's faster at 1080p, a tad faster at 2560, and at multimonitor resolution.
Overclocking is a luck-based attempt, which generally closes the gap , but then you are left with not so stellar perf/W and the jet-like noise from 7970.

7970 is faster card only in cases where bandwidth _is_ an issue for 680,
where there are no CPU bottlenecks whatsoever, and CPU is largely irrelevant - Metro 2033*** being the perfect example.
So right away you can say that 680 will rule the day in any reasonably well coded multiplayer,
because of clients overhead, and 680's graphic workload optimized scheduler.

(***BEWARE: developer's beta of Metro: Last Light has already been seen running on Nvidia hardware and featuring some impressive PhysX/Apex Clothing effects)

PRICE: Wash!

GTX 680 is $50 cheaper, but 7970 comes with 1GB of VRAM, 1 VRM phase more, is sturdier built, has better PCB, and larger bill of materials.

POWER CONSUMPTION: 680 wins again. It depends on a game, but on average 680 will consume something in between of Tahiti Pro/XT.
They are pretty much the same on Desktop, with 7970 winning the day if you tend to leave your PC idling.

NOISE: Nvidia has far mor experience with fighting the heat, and this shows on cooler design.

10db of load difference as measured by Techreport is already a double of GTX 680's perceived loudness.
Imagine what happens when you overclock your Tahiti...

SLI/CF: 680 edges once again.

Nvidia has far better multi GPU support. It won't work always, but it's supported in bigger number of titles then CF.
Also, subsequent drivers do not tend to brake what's already working.

Automatic profiles downloading, and Frame Metering simply do not exist on AMD.
So even in cases when Crossfire scales better, you will be left with choppier game-play (BF3 @ [H])

FEATURES: No contest. 680 hands down!
A quick rundown on the Nvidia's exclusive bag of goodies which is getting heavier by the day:

PhysX, CUDA, GPU Boost aka overclocking for the dummies, hw acceleration in large number of applications, active sensor based TDP/OVP protection (as opposed to AMD empiric TDP limiting)
hardware-based H.264 video encoder, Ambient Occlusion, Nvidia Inspector, temperature and FPS limitter, far more robust anti-aliasing, compatibility bits, superior Full-screen antialiasing support, downsampling, improved driver FXAA, TXAA (soon to rule like the boss!), infinitely better 3D support, automatic profiles downlading, nearly shimmer-free industry leading anisotropic filtering(yeah that includes better ingame image quality - 7970 masks the AF problem via image blurring).... a working Forum, Driver Feedback and Support

Feature list is a main reason why I can't even consider buying AMD at this point in time.

I will also mention far better OpenGL support, traditionally better driver team, faster problem tackling, and a much closer relations with number of developers.

Having 3 distinct architectures in a very short period of time (VLIW4, VLIW5 and GCN), and having an imperative to support all three,
means some tough times are ahead for both AMD driver team and their user base.

In my opinion 7970 should be bought only if you can't find GTX 680, you can't live without OpenCL, or just want to make a statement.

Wish I could sig something this long. Because it's sig worthy.

I vote for sticky to prevent all future AMD vs. NVIDIA threads.

 

[Vengor]

Yeah, so far the GTX 680 is a head early in the race I do have a board capable of SLI so I was thinking of getting two older boards like the 69XX or 5XX. But I didn't want to add yet another option to the poll and figured it would be covered by other.

 

[Axon]

EVGA 680s are in stock as of 6:48 pm est.

 

[hyrule4927]

That's not a logical conclusion because you have automatically assumed that compute performance = Double precision performance. Folding @ Home benefits from many fast CUDA cores, not gobbles of DP performance. If this was Collatz Conjecture, MilkyWay@Home or PrimeGrid, it would be different. What's the score that HD7970 gets in F@H? 7000 points at stock clocks? That would put it around 9000-9500 @ 1200mhz using that data point. The added compute performance doesn't seem to benefit HD7970 for this program unless there is a specialized OpenCL client that takes advantage of its double-precision capabilities?

7000 PPD? That doesn't sound right. My little 6850 can manage 6500 PPD . . . :hmm:

 

[MrK6]

LOL yep lurking for 12 years does that make me one of the biggest lurker ever caught so far?

We keep an informal, internal record. At 12 years and 1 month between registration and first post, you are the current record holder
-ViRGE

This is awesome, congrats Vengor!

Judging by your OP, it seems like you want a good gaming card that will also fold. I think the GTX 680 is fine for your needs, grab an eVGA and don't look back. As far F@H goes, it seems unlikely that nvidia's architecture change would completely abandon its advantage there.

 

[Smartazz]

Unless the 7970 drops below the 680 in price I say 680 all the way. If they were the same price, I'd personally get the 7970, since I'd be willing to OC to the limit and the 3GB of VRAM would be nice for mods.

 

[ss284]

Where is the 680 in stock?

 

[bryanW1995]

Yeah and i would believe that however Nvidia is working with Stanford on the new client.

...We are working with Stanford on a new Folding@Home client that takes maximum advantage of our new Kepler architecture. Stay tuned for more info!


Amorphous

So when there is a direct comment on that very thing from Nvidia I have to wonder just what is up their sleeve.

It doesn't matter how much they "work on it", gtx 680 sucks at DP. I voted for gtx 680 b/c it's clearly the best gaming card, but after reading your thoughts/comments I think that 7970 better fits your stated desire of great gaming + great folding.

 

[BallaTheFeared]

It doesn't matter how much they "work on it", gtx 680 sucks at DP. I voted for gtx 680 b/c it's clearly the best gaming card, but after reading your thoughts/comments I think that 7970 better fits your stated desire of great gaming + great folding.

7970 is in the same boat, F@H performance is pretty low, well below that of Fermi performance.


CPU folding is still the most cost effective and highest yielding PPD option available, as well as the most efficient, they're not giving credit where it is due.

 

[Vengor]

So far it looks like the GTX 680 will be the clear winner here. Now where in the hell can I find one? No one has them in stock.

 

[Axon]

So far it looks like the GTX 680 will be the clear winner here. Now where in the hell can I find one? No one has them in stock.

You've got to be quick. Newegg had a handful friday night, but they were gone in minutes.

 

[JBT]

Since you don't have any card yet GTX680.

 

[Vengor]

LOL I had it in my basket and when I went to go pay it was already out of stock at that time.

 

[Magic Carpet]

Interesting, according to TPU, during actual gaming [Read the Peak chart], power consumption-wise, two cards are about the same. But 680 is lacking ZeroCore tech.

 

[Powermoloch]

Gtx 680

 

[mak360]

Interesting, according to TPU, during actual gaming [Read the Peak chart], power consumption-wise, two cards are about the same. But 680 is lacking ZeroCore tech.

People are calling it cooler (its hotter than a 7970 by allot), less min/average power usage (uses more power than 7970 on min & average but on peak it wins), silent vs 7970 (yes by -3db), faster (by a few fps in each twimtbp game & loses in amd favored games, also overclocking is a whole new ball game)

the numbers above are just from tpu and are worse for the 680 on some other sites.

in the UK and most of EU, the 7970 is cheaper by $65

 

[Haserath]

Some people don't seem to realize the GTX 680 over-volts itself during an overclock while many 7970's reach 1125 core without a voltage boost.

The 7970 loses in performance per watt at stock but gains during overclocking. Even the worst overclocking 7970's(read lower leakage/power) usually reach 1050mhz before needing extra volts.

Folding I'm unsure whether the 680 will do well or not. It has quite a bit more raw processing power than the 580, but it lost other parts of the hardware that helped in compute workloads. The 7970 should gain over previous generations thanks to new scheduling, but it needs a core that can make use of the GPU's new features.

In games the 680 wins at stock. Don't know about overclock performance; I've only seen one overclock review from an AMD biased site(at least it looked like it). I also don't know if 3GB will ever come in handy over 2GB.

That's not a logical conclusion because you have automatically assumed that compute performance = Double precision performance. Folding @ Home benefits from many fast CUDA cores, not gobbles of DP performance. If this was Collatz Conjecture, MilkyWay@Home or PrimeGrid, it would be different. What's the score that HD7970 gets in F@H? 7000 points at stock clocks? That would put it around 9000-9500 @ 1200mhz using that data point. The added compute performance doesn't seem to benefit HD7970 for this program unless there is a specialized OpenCL client that takes advantage of its double-precision capabilities?

GTX460 768 beats an overclocked HD6990 and is >2x faster than a single HD6970.

If Double precision performance mattered, this would have never happened.

GF104 chip has 1/12th the SP performance in double-precision.
Cayman has 1/4th of SP performance in DP.

GTX460 768mb = 113 Gflops of DP performance
HD6970 = 676 Gflops of DP performance (6x more)
HD6990 = 1,352 Gflops of DP performance (12x more than a GTX460)

Based on this estimate a 1200mhz HD7970 should roughly get 8000-10000 points, if that.

It looks like Folding @ Home primarily benefits from many fast CUDA cores and memory bandwidth. Therefore, GTX680 should crush a 7970. Considering a GTX460 > HD6990 despite 12x lower DP performance, I can't see how HD7970 can possibly outperform a 1536 SP @ 1058 mhz GTX680 with same memory bandwidth as a GTX580.

^Those must be old numbers. When I was folding, there were newly released cores for AMD that were getting at least 8k PPD on my 6950.