GTX 1080 now or wait for Vega?

Should I get a GTX 1080 now or wait?


  • Total voters
    151
Perhaps I should wait until Deus Ex is released later this month and see if I can play it at a reasonable fidelity level. If so, I'll wait a little while longer to upgrade. If not, then I get a 1080. By then the 1080 stock should be more stable. Thoughts?
 
In 2005 I bought a 6800GT. But the 7800GT was about to come out, so I should have waited. I vowed never to make the same mistake again.

So when the 8800GT came out, I was smarter than that. I waited. I didn't bite on the 9800GT, the GTX 260, the GTX 470, 570, 670, 770, 970, and I sure as shit won't be getting the 1070. No, I know that something better MUST be around the corner. I'll wait for that.

Bottom line - when a product is out that meets or exceeds both your performance and price requirements, it's time to buy. Only wait if there's something about the current product that doesn't quite line up with your requirements/expectations.
 
Fyi i went from a 7970ghz to a (2) 290x crossfire to a 980ti, with my main monitor being 2560x1440, I would say that the 980ti is way more beast mode than when the crossfire gave me trouble. Point is the deals seam to be in the 980ti stuff with the channel trying to expel these in the $350-385 range. I am pretty sure that they are almost on par with the 1070's (minus power requirements), and would give you 4k with goodies stuff turned down, or 1440p for sure. Basically the best bang for the buck right now with the $420 is pricing on the 1070's. Also might give you a nice holdover until the 1080's price at least gets to msrp, or Vega / 1080ti's come out. But there is the caveat that all technology $$ drops so you will probably take a hit in the resell market by the time this comes to pass. Just my 2c ;)
 
I would recommend just getting something used or new at a reduced cost from the last generation. A Fury X, Fury or even a 980 Ti would do you fine. The reason I recommend this is because you do not game all that much anymore. Maybe even a 1070 is you can get one at MSRP or on sale. Otherwise, if you are looking to spend that money burning a hole in your pocket, buy the one you can afford the most now and do not look back.
 
I'm waiting for the ti or vegans to drop. It's going to be a while isn't it lol?
 
Vega is just gonna be another disappointment, just like every single AMD releases.

I love how AMD fanbois always anticipate some kind of "change", always hoping for a positive surprise. Fury, Polaris, you name it. All ended in disappointment, but hey, some people never learn.

And btw OP, good luck playing your waiting game. There's always something better at the horizon.
 
TBH, the best move for you for 4K gaming would be to get the Pitan X, as it's currently the only video card that can handle almost any game using 4K@60Hz with all details turned up (especially with an overclock, even one as mild as turning up the power/temp limits). I'm planning to get one (under serious consideration if AMD doesn't release anything to match it this year), but I will be buying my new build as well, so my tentative purchase plan timeframe is fall/winter 2016.
 
Vega is just gonna be another disappointment, just like every single AMD releases.

I love how AMD fanbois always anticipate some kind of "change", always hoping for a positive surprise. Fury, Polaris, you name it. All ended in disappointment, but hey, some people never learn.

And btw OP, good luck playing your waiting game. There's always something better at the horizon.

That's nice, you continue to make love with your Nvidia GTX whatever. :D I am an AMD fanboy and have no issues using the hardware I have purchased. Sorry if I do not like what you like but, then again, I am not sorry.
 
TBH, the best move for you for 4K gaming would be to get the Pitan X, as it's currently the only video card that can handle almost any game using 4K@60Hz with all details turned up (especially with an overclock, even one as mild as turning up the power/temp limits). I'm planning to get one (under serious consideration if AMD doesn't release anything to match it this year), but I will be buying my new build as well, so my tentative purchase plan timeframe is fall/winter 2016.

I personally have determined that anything more than $300 for even a 4k card is not worth it too me. $1200 is just way to expensive for that card for me, maybe not for others though. I like to enjoy whatever I have when I have it, I have just gotten tired of chasing the so called last, greatest when it is not worth the cost.
 
I personally say to go ahead with the 1080. The waiting game has never worked for me because things always get delayed and I end up waiting and waiting for something, hoping to save some cash, when I could've just gone ahead with my original plan and been having fun. How many times has Vega been delayed? Is nVidia going to actually release a Ti version of the 1080 if they don't have any competition? Has nVidia ever acknowledged a 1080 Ti is even in the works?

The prices on 1080's are settling down now as well. I bought my 1080 FE for the MSRP and had no issues with finding it in stock. Sure some AIB's are outrageously priced but there are several Zotac cards in stock on amazon right now in the $600-$700 range and you can find other brands in stock at that range elsewhere, you just have to do some work to find them.
 
I personally have determined that anything more than $300 for even a 4k card is not worth it too me. $1200 is just way to expensive for that card for me, maybe not for others though. I like to enjoy whatever I have when I have it, I have just gotten tired of chasing the so called last, greatest when it is not worth the cost.

I respect that. Just keep in mind that you'll have to wait another 2-3 years for the midrange cards to hit 4K@60Hz solidly, all the more so as there aren't too many die shrinks on the horizon (due to electron leakage). So don't expect the future midrange cards to have massive increases anytime soon. Going SLI/Crossfire is always an option, depending on what games you play. Of course, going the used route is great, too. ;)
 
I would go for the 1080. 1080 Ti is only a rumor at this point and even if/when it comes out, the price is likely to be high since there is no competition from AMD.
 
I just gave up and preordered a 1080 now that Amazon has some at decent prices. This means that the 1080ti will be announced just after my 30 day return window ends.

Yea, stock is finally catching up and now that there are choices... I find it's hard to wait for vegans or ti. Must fight the urge.
 
Here is the route I went....got an EVGA 1070 to see what it will do at 4k for me. Yes, I do not have all the bells and whistles on right now but it is 50-60 fps in GTA V on my 4k TV, which is good enough for now. I went with EVGA simply for the 90 day upgrade option. I still may pay to bump it up to a 1080.
 
Here is the route I went....got an EVGA 1070 to see what it will do at 4k for me. Yes, I do not have all the bells and whistles on right now but it is 50-60 fps in GTA V on my 4k TV, which is good enough for now. I went with EVGA simply for the 90 day upgrade option. I still may pay to bump it up to a 1080.

Yeah, that sounds like a good idea. Love EVGA's options to step up.
 
A couple points to consider:
  • No way in holy hell Vega going to be out by October. Even AMD just said it won't be available until 2017, and they have every motivation to have people think it's coming earlier. In general, one can take the earliest release date rumor for AMD cards and add 6 months to it (which is, not uncoincidentally, the length of time AMD marketing believes it can keep people hyped about a product i.e. they start to "leak" undercurrent-rumors around this time).
  • 1080 ti will likely arrive sometime October-December/Q4. It will likely be priced at "MSRP" around $800, but we all know that NVDA MSRP holds little weight apart from looking good on reviews and press releases. I'm betting actual street price for 1080 ti will be near $850-900. Remember, this isn't Maxwell where 28nm was a very mature process: this is Pascal on a new node where yields of the fully-enabled chips are low -- NVDA is not going to do any favors on pricing with the 1080 ti.
  • Stock is starting to loosen up for 1080, with a few available right now for around $650.
  • Neither Dishonored 2 (Gamebryo engine, if not mistaken) nor Deux Ex will be graphics killers.
All in all, I'd say go for a GTX 1080 now.

Edit: I do agree, in part, with SnowBeast's assessment above. There is no reason whatsoever for NVidia to release 1080 ti now. There is no competition and yields are too low on Pascal for profitability of a fully-enabled mass market chip priced below $1000. They will release 1080 ti when the yield and profitability intersect which won't likely happen until later this year.

Also a GTX 1080 is about as fast as 480 Crossfire. Certainly a monolithic 450-500mm die will be better than two 230mm cards in crossfire, but you can already see AMD's problem. Perf per watt limits their potential performance--RX 480 is already 150W, a die twice that size is going to 250-275W at only 1200MHz. Unless they are going to make a Fury X Version 2 and it put it on water and letting this thing go to at least 350W/1600+MHz they aren't coming close to GP102/or a GTX 1080Ti.

Unless their plan is to release a 300W furnace to compete with the GTX 1080 in 1H 2017, in which case Vega would almost certainly have to undercut GTX 1080 MSRP to be competitive.
 
Wait game never ends. Pay for your experience. You can always sell the dam thing here in a few days and upgrade again. you can't replace time you can always replace card! If you are considering a 1080 then obviously you are not a budget buyer. Buy the thing and frag on! You can always jump on vega if it's faster. For now get what will make your experience better!
 
Also a GTX 1080 is about as fast as 480 Crossfire. Certainly a monolithic 450-500mm die will be better than two 230mm cards in crossfire, but you can already see AMD's problem. Perf per watt limits their potential performance--RX 480 is already 150W, a die twice that size is going to 250-275W at only 1200MHz. Unless they are going to make a Fury X Version 2 and it put it on water and letting this thing go to at least 350W/1600+MHz they aren't coming close to GP102/or a GTX 1080Ti.

Unless their plan is to release a 300W furnace to compete with the GTX 1080 in 1H 2017, in which case Vega would almost certainly have to undercut GTX 1080 MSRP to be competitive.

You're assuming the Vega is a doubled up Polaris 10. Fiji wasn't a doubled up Hawaii. Indications are that Polaris was a reworked Tonga or Hawaii, while Vega will likely be a reworked Fiji.
 
also all the major retailers are price gouging right now with the 1070/1080...I would wait for things to stabilize
 
Also a GTX 1080 is about as fast as 480 Crossfire. Certainly a monolithic 450-500mm die will be better than two 230mm cards in crossfire, but you can already see AMD's problem. Perf per watt limits their potential performance--RX 480 is already 150W, a die twice that size is going to 250-275W at only 1200MHz. Unless they are going to make a Fury X Version 2 and it put it on water and letting this thing go to at least 350W/1600+MHz they aren't coming close to GP102/or a GTX 1080Ti.

Unless their plan is to release a 300W furnace to compete with the GTX 1080 in 1H 2017, in which case Vega would almost certainly have to undercut GTX 1080 MSRP to be competitive.

You are assuming too much. You looked at power usage of fury x or fury? It was hell of a lot competitive to 980ti then hawaii was. So what makes you think that if they reworked the die and shrunk it down to 14nm and not get the same thing done. You keep assuming that it will be double the polaris 10 and keep using power performance from 480. It is totally irrelevant, it is a different die and different architecture. So to assume all that is not even remotely fair. If you look at hawaii vs fiji. Power improvements were significant.
 
You are assuming too much. You looked at power usage of fury x or fury? It was hell of a lot competitive to 980ti then hawaii was. So what makes you think that if they reworked the die and shrunk it down to 14nm and not get the same thing done. You keep assuming that it will be double the polaris 10 and keep using power performance from 480. It is totally irrelevant, it is a different die and different architecture. So to assume all that is not even remotely fair. If you look at hawaii vs fiji. Power improvements were significant.

perf/watt metrics for the 480 is worse when compared to pascal than the fury to the 980ti, much worse almost 1.5 times, and this puts it at level of Hawaii and Granada to the 980. (in relationship terms not absolute terms) Granted this is no way to compare Vega to GP102, since Vega will have HBM, so they might not need to water cool it this time, but I don't expect them to catch up to gp102 at all, they might get close like in the 10% range but that is also being optimistic. Remember Vega, Polaris, were already well into design before any hints of what pascal or even Maxwell 2 was capable of in this metric.

To the OP depending on what you are going for I would get the gtx 1080 that should be a great card till next gen, in you need 4k gaming wait for the 1080 ti. (Vega won't be cheap it will be priced higher, just has to be, we saw with Fiji, there is little margin for HBM products, and with a new version of it, it will be the same, if it ain't priced higher than the gtx 1080, then its performance will be about the same as the gtx 1080 so either way you are pretty much set)
 
Last edited:
You are assuming too much. You looked at power usage of fury x or fury? It was hell of a lot competitive to 980ti then hawaii was. So what makes you think that if they reworked the die and shrunk it down to 14nm and not get the same thing done. You keep assuming that it will be double the polaris 10 and keep using power performance from 480. It is totally irrelevant, it is a different die and different architecture. So to assume all that is not even remotely fair. If you look at hawaii vs fiji. Power improvements were significant.

You realize a lot of that perf per watt was just vaporware due to the inability for Fijis pipeline to feed all of its shaders right? Fiji had 45% more cores than a 390X yet on average was only 20% faster. AMD ran out of die space on 28nm which resulted them in overloading on shaders. Whereas big pascal is scaling nearly 100% clock for clock with its extra cores, about 35-37% faster on average at the same clockspeed with 40% more cores.
 
+
perf/watt metrics for the 480 is worse when compared to pascal than the fury to the 980ti, much worse almost 1.5 times, and this puts it at level of Hawaii and Granada to the 980. (in relationship terms not absolute terms) Granted this is no way to compare Vega to GP102, since Vega will have HBM, so they might not need to water cool it this time, but I don't expect them to catch up to gp102 at all, they might get close like in the 10% range but that is also being optimistic. Remember Vega, Polaris, were already well into design before any hints of what pascal or even Maxwell 2 was capable of in this metric.

True. I don't expect vega to match the gp102, just won't happen due to architecture difference. But I suspect vega to have similar ratio as fury to hawaii.
 
I hope they will be able to get more performance out of it then what they got from Hawaii to Fury lol, if that's all they get, they will be in trouble that won't match the 1070......
 
I hope they will be able to get more performance out of it then what they got from Hawaii to Fury lol, if that's all they get, they will be in trouble that won't match the 1070......

I think 1070ish performance is what to expect from Vega. R&D matters.
 
I would wait for the 1080 price to stabilize to a reasonable point and then grab one. If you can wait for a 1080ti, then wait for it.
 
You realize a lot of that perf per watt was just vaporware due to the inability for Fijis pipeline to feed all of its shaders right? Fiji had 45% more cores than a 390X yet on average was only 20% faster. AMD ran out of die space on 28nm which resulted them in overloading on shaders. Whereas big pascal is scaling nearly 100% clock for clock with its extra cores, about 35-37% faster on average at the same clockspeed with 40% more cores.

We all know clocks are limiting factor on amd side. This all depends on where vega lands. i am not arguing all the other stuff. I am talking about the power efficiency with fiji given how big the chip is. I don't concentrate too much on rumors. But we will see what we get. All I am saying is you can't take power efficiency of Polaris and translate it to vega. AMDs most efficient card is probably the Nano to date. There is no denying fiji is much more efficient chip than hawaii or fiji. All I am saying is if they can get clock speed improvements and make tweaks to vega it might be pretty close to pascal, within 10% or so.

Look at the graphs below. average wattage is in the 180s around 40 more than gtx 970 but it also does give you more performance as well. Would someone like a nano 2 running around 1300-1400 clock speed from average of 875 in games and similar power output. That would be damn sweet. It really depends on what clocks amd can get out of vega. I am keeping my expectations low. But I expect it to be more effecient than polaris regardless.

upload_2016-8-6_11-55-0.png

upload_2016-8-6_11-55-29.png
 
Last edited:
I would wait for the 1080 price to stabilize to a reasonable point and then grab one. If you can wait for a 1080ti, then wait for it.

1080 and 1070 prices seem to have stabilized for now. Plenty can be had around 650 and gtx 1070 less than 450. I think they are all around after market pricing set by AIB partners. Newegg has bunch of evga and other 1080s and 1070s going at MSRP that is mentioned by board partners or may be 5-10 bucks above it. I know alot of evga cards are at their announced price.
 
You realize a lot of that perf per watt was just vaporware due to the inability for Fijis pipeline to feed all of its shaders right? Fiji had 45% more cores than a 390X yet on average was only 20% faster. AMD ran out of die space on 28nm which resulted them in overloading on shaders. Whereas big pascal is scaling nearly 100% clock for clock with its extra cores, about 35-37% faster on average at the same clockspeed with 40% more cores.

So by that explanation. If all those shaders were being fed it uses more power but it also gets you more performance. So by that means the perf/watt remains the same. So it remains that fiji will still be more power efficient then hawaii.
 
NKD

Interesting point has been raised before, I'm wondering how Fury X power consumption behaves in AotS for example and compare to Metro Last Light which PCPER and TPU always use for gaming tests
 
NKD

Interesting point has been raised before, I'm wondering how Fury X power consumption behaves in AotS for example and compare to Metro Last Light which PCPER and TPU always use for gaming tests

Isn't efficiency just relative to performance anyways? I mean Let's say it gets more performance as it shaders are crunching and utilized more efficiently in AotS or other dx12 games, which most likely will use more watts but then you get more performance so it will just be same performanc/watt.

I am just wondering if this carries over to vega. If anyone can fuck it up its AMD but then again they also created the Nano. So who knows lol.
 
Isn't efficiency just relative to performance anyways? I mean Let's say it gets more performance as it shaders are crunching and utilized more efficiently in AotS or other dx12 games, which most likely will use more watts but then you get more performance so it will just be same performanc/watt.

I am just wondering if this carries over to vega. If anyone can fuck it up its AMD but then again they also created the Nano. So who knows lol.

Efficiency is relative to whatever performance metric you choose yeah, but there's the added scheduling logic being used (ACEs) ontop of the added power consumption from actually utilizing the whole shader array. I'm just curious to see how much it draws
 
also all the major retailers are price gouging right now with the 1070/1080...I would wait for things to stabilize

The major retailers are not. It's the third party storefronts that are price gouging. Amazon and Newegg are selling at MSRP when they have stock.
 
I generally like keeping my GPU for a while. I mean, I'm still running my 5-year old 7970 :p
If you need a card now get the GTX1080.
If you can hold out, see if a GTX1080ti is released.

I'm buying a ti if its released. If not I'll stick with my 980ti until a worthy upgrade appears, that might be another year+.
 
If you need a card now get the GTX1080.
If you can hold out, see if a GTX1080ti is released.

I'm buying a ti if its released. If not I'll stick with my 980ti until a worthy upgrade appears, that might be another year+.


980ti ended up comparing pretty favorably to the 1080, I'm framing mine when I'm done with it :p
 
Get a 1080 or Titan X and be done with it. Both are on the new node and won't go obsolete any time soon.

What's the worst case? A few months from now the 1080s drop $100 in price? To me $100 isn't worth waiting for.
 
4K won't be doable in the long term, even with GP102, I'm guessing some games that come out this year won't run at 4K60 max settings on a Titan X (even if you drop AA); those with waterblocks and overclocks "balls to the wall" ~2100mhz will be sitting on 15tflops!. Maybe that will let you run 4K60 for a year or so, with small compromises on settings, maybe. If you want to commit to 4K with no compromises in the long term, you have to commit to upgrading to the best every single time you can, and watercooling it - unless you want to opt for mgpu. 4K is a hassle. I would opt for 1440p, you can run 5K DSR when the game allows for it, or high refresh in others.
 
Back
Top