Wait for Pascal or upgrade to 980ti

[Subwayeatbig]

Hi all,

I am currently running a gtx 570 and I recently just purchased a new 1440p panel. My system is clearly showing its age of course.

This leads me to the dilema, should I purchase a 980ti or wait for Pascal while using a 970 as a stop gap?

Thanks,

Alvin

 

[happy medium]

I dont think a 300$ card is a stop gap. WHats the rest of your system specs like?

but for 1440p? I guess its good.

I'd grab this one.
I think there is a gtx970 in the hot deals section for 275$?shipped
http://www.ebay.com/itm/MSI-GeForce...857366?hash=item43e2619f16:g:LGcAAOSwgNRV8Ya6

http://forums.anandtech.com/showpost.php?p=37807664&postcount=3

 

[Subwayeatbig]

I dont think a 300$ card is a stop gap. WHats the rest of your system specs like?

but for 1440p? I guess its good.

I'd grab this one.
I think there is a gtx970 in the hot deals section for 275$?shipped
http://www.ebay.com/itm/MSI-GeForce...857366?hash=item43e2619f16:g:LGcAAOSwgNRV8Ya6

http://forums.anandtech.com/showpost.php?p=37807664&postcount=3

Well,

I have a 2600k oc'd at 4.5ghz
samsung 850 evo 512gb?
16gb 1633 mghz ddr3
24 inch 1080p monitor -> 27 inch 1440p monitor with g sync
I thought I had a 650watt psu but apparently not lol
http://www.newegg.com/Product/Product.aspx?Item=N82E16817207001

When I had the 570, I thought id just keep it for a couple of years but that was not the case as its been more than 4 or 5 years?? My thought was buy the 980ti and enjoy it and maybe sell it and buy a Pascal video card when that will come out or buy a 970 and enjoy that and keep it maybe, but then again I might just wait and ends up being 4 years later lol.

 

[happy medium]

27 inch 1440p monitor with g sync

buy a gtx980ti and enjoy it. sell it later but I doubt you will need to.

on sale for 599$ and a free game. Hurry.
http://www.newegg.com/Product/Produ...4500378&cm_re=gtx980ti-_-14-500-378-_-Product

Next you can do a full system upgrade with Pascal refresh. In 2018?

 

[Fallen Kell]

Personally I am waiting (but I have a 670GTX so am another generation newer than your card which still works for "most" games, especially since I play at 1920x1200).

 

[moonbogg]

If you have the monitor now, i'd get a 980ti now and enjoy. After Pascal comes out, you'll still have enough power to be comfortable until the next pascal wave comes around.

 

[Mopetar]

I'd honestly wait (I'm in a similar boat where I could stand an upgrade) because I think Pascal is going to be a massive performance leap over what we have available now, which is going to kill resale value for high-end cards.

Going from 28 nm to a 14/16 FinFET is going to be massive and even if NV doesn't go all out at first, I suspect some monstrous performance gains.

 

[Mondozei]

You shouldn't have gotten that 1440p monitor yet, is the unpopular but truthful statement that has to be made.

Pascal will be a rare event: node + uarch at the same time. Last time it happened it was from fermi to kepler and the performance increase was around 90% over previous generation.

Even if we assume a more "conservative" 70%, that's still 2X of what we've seen recently(30-35%). So getting a 980 Ti means that a mid-range GP104 Pascal will most likely be slightly faster than it in just one year's time, instead of full fat Pascal which will simply crush a 980 Ti by a fairly massive margin. (I don't expect GP100 or full fat Pascal to come out next year for consumers en masse, because NV needs something in 2017. A Titan card at most).

You often hear, "there's always a faster card around the corner". That's true, but these are also special circumstances.

No way a 570 is going to be sufficient for 1440p unless you're playing slow turn-based games and/or other games at low, low settings. I'd return your 1440p monitor if it was possible and wait for full upgrade next year.

If you really can't do that - or don't want to do that - then I'd say the only option out is a 970, and I'd pay for a used one to save cash, since you'd upgrade to Pascal within 12 months anyway, why splash on a new card?

 

[CakeMonster]

I have no faith that we will get a large jump. Maybe on average a "larger" increase in 2016 than 2015, but they are going to try to segment and intentionally delay as much as they can to milk this generation. So expect to be able to shell out $700 each year from 2016 to 2019 on probably the exact same process for gradual increases.

 

[jpiniero]

Pascal will be a rare event: node + uarch at the same time. Last time it happened it was from fermi to kepler and the performance increase was around 90% over previous generation.

The 680 was about 30-40% faster than the 580 in most titles at launch, nowhere near 90%. And that was on a chip that cut the DP performance. nVidia's had to shuffle things around due to the fab issues (notice how unified memory is gone even on Pascal) so I wouldn't be surprised if Pascal is simply Maxwell with better DX12 support and HW async compute.

 

[nvgpu]

http://www.geeks3d.com/forums/index.php/topic,4247.0.html

Pascal

-D__CUDA_ARCH__=600

-D__CUDA_ARCH__=610

-D__CUDA_ARCH__=620

Pascal is not Maxwell.

 

[Subwayeatbig]

Thanks for the advice!

Though, I have to say I am tired of waiting, because the games I want to play are out now. Waiting 12 months and doing a "full system" upgrade to me isn't in the cards because I don't see any huge benefits going to a new cpu. I mean the more I thought about it and if I can have the 980ti for at least two years then it would be a worth it purchase.

 

[RussianSensation]

Waiting 12 months and doing a "full system" upgrade to me isn't in the cards because I don't see any huge benefits going to a new cpu. I mean the more I thought about it and if I can have the 980ti for at least two years then it would be a worth it purchase.

There is no need to upgrade your 2600K OC unless you have 980Ti SLI. There are several good deals on 980Ti depending on where you priorities lie. I outlined two of them in this thread. If you kept your last card 4 years, chances are you'll enjoy the 980Ti and upgrade again in 2018-2019.

 

[MagickMan]

The 1080 TI, or whatever it'll be called, won't come out until 6 months after Pascal has dropped. That means it'll be here in Q1 of 2017.

 

[moonbogg]

The 1080 TI, or whatever it'll be called, won't come out until 6 months after Pascal has dropped. That means it'll be here in Q1 of 2017.

Yeah, Nvidia has to puke on us with the over priced mid range crap first. Then another year later they might finally release a decent chip, which will still be a cut down version, and then another cut down version, and then finally the actual, real full chip except with half the Vram. So, the way I see it, Pascal won't be here until 2018 sometime.

 

[CakeMonster]

Even later, there's at least one Titan in between.

 

[Arachnotronic]

get the 980 Ti. Life's too damn short to play the waiting game in perpetuity.

 

[RussianSensation]

I am going to update my suggestion with another option - a $350 EVGA B-stock 980.

EVGA GeForce GTX 980 GAMING ACX 2.0
http://www.evga.com/Products/ProductList.aspx?type=8

Here is my reasoning: since this card is priced well below current market price, you may be able to resell it with very little loss in resale value should you decide to upgrade to Pascal. If Pascal comes out and current cards tank in resale value, since you only purchased the 980 for $350, you would still be able to sell it for $250, losing just $100 in the process. At the same time the 980 is way more powerful than your 570.

BTW, you should rule out 970 for 1440P as 390 is superior for that resolution.

 

[Arachnotronic]

I am going to update my suggestion with another option - a $350 EVGA B-stock 980.

EVGA GeForce GTX 980 GAMING ACX 2.0
http://www.evga.com/Products/ProductList.aspx?type=8

Here is my reasoning: since this card is priced well below current market price, you may be able to resell it with very little loss in resale value should you decide to upgrade to Pascal. If Pascal comes out and current cards tank in resale value, since you only purchased the 980 for $350, you would still be able to sell it for $250, losing just $100 in the process. At the same time the 980 is way more powerful than your 570.

BTW, you should rule out 970 for 1440P as 390 is superior for that resolution.

Great suggestion here.

 

[MrTeal]

The 680 was about 30-40% faster than the 580 in most titles at launch, nowhere near 90%. And that was on a chip that cut the DP performance. nVidia's had to shuffle things around due to the fab issues (notice how unified memory is gone even on Pascal) so I wouldn't be surprised if Pascal is simply Maxwell with better DX12 support and HW async compute.

This. The 90% argument makes sense if you compare the 580 to big Kepler (GK110), but Titan didn't come out until a year after the 680 (and wasn't 90% faster anyway) and the full GK110 die didn't ship in the 780Ti until a year and a half after the GTX 680.

Unless nVidia really changes the way they've done things the last few years, the first Pascal will probably be faster than a 980Ti on a ~300mm^2 die, but not by a massive amount. You might have to wait awhile to get the big die version that really blows away the 980Ti.

Do as RS suggested and pick up a cheap 980, or just buy the 980Ti now and enjoy it for a couple years until you need to upgrade.

 

[Erenhardt]

I am going to update my suggestion with another option - a $350 EVGA B-stock 980.

EVGA GeForce GTX 980 GAMING ACX 2.0
http://www.evga.com/Products/ProductList.aspx?type=8

Here is my reasoning: since this card is priced well below current market price, you may be able to resell it with very little loss in resale value should you decide to upgrade to Pascal. If Pascal comes out and current cards tank in resale value, since you only purchased the 980 for $350, you would still be able to sell it for $250, losing just $100 in the process. At the same time the 980 is way more powerful than your 570.

BTW, you should rule out 970 for 1440P as 390 is superior for that resolution.

No way in hell 980 will be worth $250. I expect when 16nm hits, we will have GTX980 performance in the lower mid range GPU for $250 or less. All new, with new features, with better power consumption.

980 will drop along with 970 and r9 290 to below $200. Every $ spent above 290/390 price will be totally lost as their resale value will be the same when much better 16nm products hit the market.

If you want to buy now, get the cheapest 290/390 and change it later to 16nm card, or go full out for 980ti and keep it longer.

 

[JimmiG]

Well,

I have a 2600k oc'd at 4.5ghz
samsung 850 evo 512gb?
16gb 1633 mghz ddr3
24 inch 1080p monitor -> 27 inch 1440p monitor with g sync
I thought I had a 650watt psu but apparently not lol
http://www.newegg.com/Product/Product.aspx?Item=N82E16817207001

When I had the 570, I thought id just keep it for a couple of years but that was not the case as its been more than 4 or 5 years?? My thought was buy the 980ti and enjoy it and maybe sell it and buy a Pascal video card when that will come out or buy a 970 and enjoy that and keep it maybe, but then again I might just wait and ends up being 4 years later lol.

That system is still really good, except for the GPU. The system is wasting away while you keep that 570 and wait, IMO.

I was in a similar situation this summer (running a GTX 670) and got the cheapest GTX 970 I could find. That way I can enjoy games like Witcher 3, Fallout 4, Anno 2205 etc. today without feeling too bad if Pascal ends up beating everything else and costing $250 (unlikely).

However I play at 1080p. At 1440p, I'm not sure the GTX 970 would be enough. Maybe look into a cheap GTX 980, a heavily factory OC'd GTX 970 or something on the AMD side.

 

[RussianSensation]

No way in hell 980 will be worth $250. I expect when 16nm hits, we will have GTX980 performance in the lower mid range GPU for $250 or less. All new, with new features, with better power consumption.

980 will drop along with 970 and r9 290 to below $200. Every $ spent above 290/390 price will be totally lost as their resale value will be the same when much better 16nm products hit the market.

If you want to buy now, get the cheapest 290/390 and change it later to 16nm card, or go full out for 980ti and keep it longer.

Well I am not a rocket scientist so I cannot predict the exact resale value on a 980 in 2016 but just looking at NV's pricing model, it took them almost 3 years to match 680's level of performance in a $199 960. 680 came out March 2012 and 960 came out January 2015.

Since 980 came out September 2014, I think my estimate of $250 by September 2016 is fairly reasonable, if not too aggressive given recent times. If you think the depreciation will be a lot more aggressive, I cannot really argue against your point since I cannot exactly predict the future , but I am just using historical info as some gauge.

Also, look at AMD's position -- they can hardly compete on price/performance and they actually raised market prices of 290/290X when they introduced 390/390X. I am going to assume that AMD will still re-use Fiji in much the same way they reused Hawaii. That means Nano/Fury would move into the $329-429 price bracket. 980 isn't that far off that level of performance so why would we have a card way faster than a 980 for $249 next year?

I have my doubts.

Let me use another viewpoint. NV took more than 1.5 years to go from GTX760 to GTX960 and that added just 14-16% more performance (!).

Right now, 980 is an astounding 100% faster than a 960.

So what do you think are the chances NV will serve us a $199-249 Pascal x60 card that matches 980's level of performance in 2016? Have you NOT seen NV's GPU strategy since 2012? This isn't the good old days where you get 80-100% more performance every 2 years at the same price. NV is going to milk it for as long as it takes before Volta comes out.

Why would NV release a card for $249 that matches 980's performance in 2016 when they can release cards 30-40% faster than the current line-up and then in 2017 another set of cards 30-40% faster than those? That's exactly what they did with Kepler and Maxwell. This way NV would have 2 flagship, one in 2016 and a new one in 2017, maximize profits by again selling a mid-range x60 card as an x80 card, and maximize revenue by getting gamers to upgrade twice (680->780/780Ti or 980->980Ti). And AMD? They are not in a position to dictate much so they would MUCH prefer NV to slowly trickle down newer cards rather than release a card 80-100% faster than a 980Ti and wipe the floor with AMD.

 

[Fallen Kell]

Don't forget that this process node change isn't the typical stepping difference that we have had in the past. This is a 2 step node process change, not a single step. It is the size ratio difference of going from Nvidia Tesla series (55nm) to Nvidia Maxwell (28nm).

 

[RussianSensation]

Don't forget that this process node change isn't the typical stepping difference that we have had in the past. This is a 2 step node process change, not a single step. It is the size ratio difference of going from Nvidia Tesla series (55nm) to Nvidia Maxwell (28nm).

I am confused. 55nm to 28nm is a half-node + full-node (1.5 full node steps):

55nm -> 40nm (half node)
40nm -> 28nm (full node)

28nm -> 20/22nm (half node)
28nm -> 14/16nm (full node)
http://www.decryptedtech.com/leaks-...f-node-and-move-straight-to-16nm-for-new-gpus

How are you getting 2 steps? It's the same as moving from 40nm to 28nm (Kepler to Maxwell) or 1 full node. Is it not?

TSMC's description also coicides with just a 1 full node shrink:

"TSMC's 16FF+ (FinFET Plus) technology can provide above 65 percent higher speed, around 2 times the density, or 70 percent less power than its 28HPM technology."

A 2 node step would result in more than 2X the density increase and more than a 70% reduction in power usage/more than 65% increase in transistor switching speed. This coincides with rumours that next gen flagship chips will have 16-18B transistors, which is about double of the existing GM200/Fiji chips.

1.5 node steps would be 28nm -> 10nm (14/16nm is full node + 1/2 node to 10nm), while 2 full node steps is 28nm -> 7nm.

---

GTX285 (55nm) had 1.4B transistors.
GTX780Ti (1st gen 28nm) had 7.1B (5X 285's)
GTX980Ti (28nm) had 8B (5.7X).

Granted, there will be other benefits such as HBM2/GDDR5X (higher memory bandwidth) and possible improvements in IPC/compute/DX12 capability via a newer architecture. I don't think NV will launch a consumer $699 card 80-100% faster than a 980Ti in 2016. Why would they when they can just split the generation into parts or first release a Titan X successor for $1K+?