$560 MSI GTX 980 Ti Golden Edition

[tential]

Edit: Reading fart

 

[Arachnotronic]

What? Why!?

Step up program!

 

[tential]

Step up program!

Guskline said his card had been solid. I thought he "sold" it.

 

[Arachnotronic]

Guskline said his card had been solid. I thought he "sold" it.

Referring to SexyK's post.

 

[guskline]

Arachnotronic: You have some pretty nice machines in Gaming #1 and Gaming#2. Do you plan on Pascal or Polaris or one of each?

 

[Arachnotronic]

Arachnotronic: You have some pretty nice machines in Gaming #1 and Gaming#2. Do you plan on Pascal or Polaris or one of each?

Pascal since they are both paired with G-Sync monitors. That is unless NV lays a turd with Pascal or AMD really blows Pascal away, then I'll get a free sync monitor for one of the systems.

Hoping for enough performance in the 1080 Ti (whatever full GP104 is called) to need only one card per system. They were previously 980 SLI and 980 Ti SLI.

 

[SexyK]

By way of update, I decided not to risk it and just grabbed an EVGA 980 Ti. If I can ultimately set up to a next-gen card with equal or greater performance in the next three months, great. If not, at least I won't be stuck with/trying to flip a completely impotent card. Full mITX VR/living room build now in sig.

While progress has certainly slowed across the board, I'm still impressed that his tiny cube build is as fast or faster than my water cooled hex core build, and also almost as quiet. The platform is much improved as well.

Just to close the loop on this, EVGA already has three 1080s available in the Step-Up Program: the Founders Edition ($699), their 'vanilla' 1080 ($609), and the ACX 3.0 ($619).

Seems unlikely to me that they will be adding any of the premium versions into the mix any time soon, so I went ahead an requested a Step-Up from the 980Ti to the 1080 ACX 3.0 at no cost (aside from $11 shipping). Overall, pretty happy with this result.

 

[RussianSensation]

Just to close the loop on this, EVGA already has three 1080s available in the Step-Up Program: the Founders Edition ($699), their 'vanilla' 1080 ($609), and the ACX 3.0 ($619).

Seems unlikely to me that they will be adding any of the premium versions into the mix any time soon, so I went ahead an requested a Step-Up from the 980Ti to the 1080 ACX 3.0 at no cost (aside from $11 shipping). Overall, pretty happy with this result.

Congrats! I thought they'd only allow step-up from older cards if the card you are stepping up to is of greater value? I am guessing the reason they let you step-up is because your 980Ti cost less than $609? Good perk if true since it means for the future I can recommend someone buy an NV flagship card 2 months out from the release of new cards if I know the step-up is this easy

 

[tweakboy]

That is as rip off , with a new card out 1080 GTX ,,, their 50 to 75 dollars more for amazing peformance ,, thanks gl

 

[96Firebird]

You can step up (or down) with a higher priced card, you just don't get any money back in the end.

What happens when I Step-Up to a cheaper product?

You will only have to pay for the return shipping at the time of paying for your Step-Up. EVGA will not credit you the difference.

http://www.evga.com/support/stepup/ (about halfway down the page)

Theoretically, a person who had just bought a Titan X for $999 a month ago could step-up to the 1080. Of course, they could probably sell a used Titan X for more than $619, maybe?

Anyways, good to hear it worked out for ya SexyK. :thumbsup:

 

[HannooFX]

It is $499 after rebate ($529 before)

http://www.newegg.com/Product/Produ...127912&cm_re=gtx_980ti-_-14-127-912-_-Product

 

[swilli89]

It is $499 after rebate ($529 before)

http://www.newegg.com/Product/Produ...127912&cm_re=gtx_980ti-_-14-127-912-_-Product

Amazing that with both a new architecture and a vastly improved node (with TSMC absorbing the lions share of start up costs), that nvidia actually released a 1080 that DECREASES the value per dollar.

1080 is 40% more expensive for 20-30% more performance. I'm sure the apologists will come in and say technology isn't cheap and the top end card is always worse price/perf, but how programmed have you really become? This isnt 1) a new high end card and 2) totally different generation. Thins are supposed to improve each generation! Moore's law about microchip cost modeling! A more efficient architecture and vastly smaller and more efficient node should allow nvidia to keep profit margins steady while providing MUCH more value to their customers. Yet here we are with $700 300mm chips....

 

[LTC8K6]

1080 is 40% more expensive for 20-30% more money.

Sounds like a great deal...

 

[RussianSensation]

1080 is 40% more expensive for 20-30% more performance. .

The 1080 FE with max PowerTune and fan turned to 100% (1080 Max on the Computebase chart) is only 15% faster than an MSI Gaming 980Ti.
http://www.computerbase.de/2016-06/inno3d-ichill-geforce-gtx-1080-x3-test/3/

Don't worry though, NV's Pascal drivers will make sure this lead extends over time. The real story here is that $400 AIB 1070 will make both the $500 980Ti and $700 1080 FE look like overpriced turds. It should be possible to buy 1070 SLI for barely more $ than a 1080 FE. :thumbsup: 1070 for $380-400 is a real star of the show as I wouldn't pay $350-400 for a used 980Ti knowing I might be SOL when it comes to warranty in case the case fails over the next 2 years.

Inno3D Chill = 38 dBA under load
Asus Strix 1080 OC = 42.5 dBA under load
1080 FE = 48 dBA under load to prevent throttling
http://www.computerbase.de/2016-06/inno3d-ichill-geforce-gtx-1080-x3-test/4/

Even without any competition, the FE is a fail within AIB 1080 product stack. In fact, even if the FE was $599, I still wouldn't buy it over a $650 AIB 1080.

What's interesting is a non-throttling 1080 uses way more power than there reference card. Thermal throttling FE 1080 in a system ~ 246W, but PT+fan 100% 1080 = 291W (almost a 50W increase in power usage).

Inno 3D 1080 system uses 304W, which is higher power usage than a gaming rig with a GTX780Ti, GTX980Ti or even the R9 290! SO much for those 180W TDP claims. Pure marketing gimmick.
http://www.computerbase.de/2016-06/inno3d-ichill-geforce-gtx-1080-x3-test/4/

I personally don't care about power usage for a desktop system but I find it hilarious when people are defending the 1080 FE while ignoring both its noise levels and power usage once it's performing in non-throttling states. It's also interesting how power usage in overclocked states is completely ignored when 1080's power usage exceeds 230-250W cards such as the 980Ti, 780Ti and R9 290!

 

[swilli89]

The 1080 FE with max PowerTune and fan turned to 100% (1080 Max on the Computebase chart) is only 15% faster than an MSI Gaming 980Ti.
http://www.computerbase.de/2016-06/inno3d-ichill-geforce-gtx-1080-x3-test/3/

Don't worry though, NV's Pascal drivers will make sure this lead extends over time. The real story here is that $400 AIB 1070 will make both the $500 980Ti and $700 1080 FE look like overpriced turds. It should be possible to buy 1070 SLI for barely more $ than a 1080 FE. :thumbsup: 1070 for $380-400 is a real star of the show as I wouldn't pay $350-400 for a used 980Ti knowing I might be SOL when it comes to warranty in case the case fails over the next 2 years.

Jesus really? I was actually trying to be fair to 1080 owners with my performance assumption.

 

[RussianSensation]

Jesus really? I was actually trying to be fair to 1080 owners with my performance assumption.

At 2Ghz boost, a reference 1080 FE is using 220W.
https://www.overclockers.ru/lab/76273_4/obzor-i-testirovanie-videokarty-nvidia-geforce-gtx-1080.html

That means a 2.1-2.2Ghz 1080 is a 230-250W videocard. 245W R9 290 was criticized as if it needed 750W PSUs, remember? Even the 7970Ghz used 238W peak and was hailed as hot and power hungry.

 

[sontin]

Are you serious? A 1080FE @ 220W will be nearly 2x faster than a 290 and 3x faster than a 7970GHz.

 

[Bacon1]

Are you serious? A 1080FE @ 220W will be nearly 2x faster than a 290 and 3x faster than a 7970GHz.

Well I'd hope so. Its $350 more (comparing launch pricing, $500 more using current pricing), 3 years later and with a node shrink.

His point was people blasted the 290's power consumption, yet here a new "power efficient" card comes out and has the same consumption yet its not mentioned as being power hungry.

 

[sontin]

Because the power consumption was bad in comparision to the competition.

A GTX980TI uses 225W. So i guess we should have complained about that, too. Right?

 

[swilli89]

Because the power consumption was bad in comparision to the competition.

A GTX980TI uses 225W. So i guess we should have complained about that, too. Right?

Well if you believe the PR sping nvidia is spouting off, "they no longer even compete with AMD but with their previous generation". And in THAT light, no, 1080 is underwhelming in power consumption compared to a 980ti. It improves watts/fps by 15%? That's almost a comically bad amount on a new node and the new architecture.

 

[sontin]

What?! A 160W GTX1080FE is 30% faster than a 225W GTX980TI reference.
Perf/watt improved around 70% over the previous cards.

The GTX1080FE is even 10% faster than a nearly 300W Gigabyte GTX980TI custom card...

 

[MagickMan]

Don't worry though, NV's Pascal drivers will make sure this lead extends over time.

The architectural differences aren't nearly as great between M & P as they were between K & M. GP104 is largely just a cut down and shrunk Maxwell at a substantially higher clock (and added VR code). There's little they can do keep them from continuing to scale upwards together, as unified drivers improve and mature. They'd need to keep them permanently split to tank Maxwell (very unlikely).

 

[jackstar7]

What?! A 160W GTX1080FE is 30% faster than a 225W GTX980TI reference.
Perf/watt improved around 70% over the previous cards.

The GTX1080FE is even 10% faster than a nearly 300W Gigabyte GTX980TI custom card...

How do you keep the 1080 from boosting so that the power stays low?

 

[Magee_MC]

The architectural differences aren't nearly as great between M & P as they were between K & M. GP104 is largely just a cut down and shrunk Maxwell at a substantially higher clock (and added VR code). There's little they can do keep them from continuing to scale upwards together, as unified drivers improve and mature. They'd need to keep them permanently split to tank Maxwell (very unlikely).

In DX11, I would think that this would be true. And while I agree that we probably won't see the same level of falloff of performance from Maxwell compared to Pascal that we saw from Kepler to Maxwell there are a couple of areas where NV could put more focused effort into Pascal that would potentially result in lower Maxwell performance.

The first is in DX12. Since Pascal has much more fine grained preemption than Maxwell, I could see them pushing this and Maxwell would have no way to improve comparatively. The second is in VR. Pascal is more tuned to VR than Maxwell and you could see improvements in Pascal where Maxwell is relatively left behind.

 

[Qwertilot]

At 2Ghz boost, a reference 1080 FE is using 220W.
https://www.overclockers.ru/lab/76273_4/obzor-i-testirovanie-videokarty-nvidia-geforce-gtx-1080.html

That means a 2.1-2.2Ghz 1080 is a 230-250W videocard. 245W R9 290 was criticized as if it needed 750W PSUs, remember? Even the 7970Ghz used 238W peak and was hailed as hot and power hungry.

Well yes, and they frankly deserved criticism. Not the basic engineering behind the cards, but because they shipped what could, even should, have been much more universally attractive graphics cards at base clocks that pushed the power draws up so high, and the efficiency so low.

It was distinctly counter productive and mildly baffling really - it works much better to ship the reference cards at a level calibrated to produce a sane power draw (cf the ref 1080/the saner clocked AIB cards) - and let the AIB's explore the clocks past that point in special editions and the like.

Not that I think its a great idea to really push the clocks on the 'medium' branches of the NV's chips (cf 980 etc). It strikes me as distinctly missing the point of these chips, but since they're the fastest things currently in existence some people will pay for it.....