When is Volta going to be released?

[Kaapstad]

If you keep them bellow 41C all of them can, regardless the brand.

Don't think so as I have seen hardly any hit those speeds on the bench threads.

 

[tyler_jrb]

Not many 1080s do 2190mhz, like almost none.

Pascal Titan @2160p is nearly 50% faster than a 1080 in graphics heavy games.

Even at 4K most of the benchmarks are showing on average around a 25% improvement stock-stock and oc-oc.

Haven't seen a single bench showing anywhere near 50%.

 

[Kaapstad]

Even at 4K most of the benchmarks are showing on average around a 25% improvement stock-stock and oc-oc.

Haven't seen a single bench showing anywhere near 50%.

You asked.

Check out the 4 card scores




2160p


1 GPU

1. Score 727, GPU 980 Ti @1391/2005, CPU 4930k @3.4, gazzaman2k Link
2. Score 725, GPU TitanM @1494/2002, CPU 5960X @4.5, Kaapstad Link
3. Score 710, GPU 980 Ti @1560/2000, CPU 3450 @3.5, Joe! Link
4. Score 471, GPU 780 @1263/1852, CPU 4770k @4.5, whyscotty Link

2 GPU

1. Score 1450, GPU 980 Ti @1531/2002, CPU 4930k @4.0, Kaapstad Link
2. Score 1321, GPU TitanM @1455/1900, CPU 4790k @4.8, TTomax Link
3. Score 1319, GPU TitanM @1380/2002, CPU 4930k @4.7, Kaapstad Link
4. Score 1171, GPU 980 Ti @1395/1900, CPU 4790k @4.7, Tommy_Here Link
5. Score 1118, GPU TitanM @1216/1752, CPU 5960X @4.6, Cosimo Link
6. Score 936, GPU TitanK @993/1751, CPU 3930k @4.625, Gregster Link
7. Score 700, GPU 970 @1478/2030, CPU 5930k @3.8, Clov!s Link

3 GPU

1. Score 1292, GPU 290X @1230/1625, CPU 4930k @4.5, Kaapstad Link
2. Score 1251, GPU TitanK @1097/1753, CPU 3930k @4.4, whyscotty Link
3. Score 1214, GPU 290X @1135/1500, CPU 3970X @4.4, LtMatt Link

4 GPU

1. Score 3990, GPU TitanP @2063/2764, CPU 6950X @4.4, Kaapstad Link
2. Score 2755, GPU 1080 @2012/2750, CPU 6950X @4.4, Kaapstad Link
3. Score 2682, GPU TitanM @1480/2002, CPU 5960X @4.0, Kaapstad Link
4. Score 2000, GPU Fury X @1140/500, CPU 5960X @4.0, Kaapstad Link
5. Score 1759, GPU TitanK @981/1788, CPU 3930k @4.8, Kaapstad Link
6. Score 1702, GPU 980 @1472/1962, CPU 5960X @4.0, Kaapstad Link
7. Score 1682, GPU 290X @1230/1500, CPU 4930k @4.8, Kaapstad Link
8. Score 1546, GPU RX480 @1340/2200, CPU 6950X @4.4, Kaapstad Link
9. Score 1382, GPU 290X @1000/1250, CPU 3970X @4.9, AMDMatt Link

 

[tyler_jrb]

You asked.

Check out the 4 card scores

What I'd expect with 4 cards vs 4 cards. I was talking about single card vs single card and in games .

 

[Kaapstad]

What I'd expect with 4 cards vs 4 cards. I was talking about single card vs single card and in games .

It is actually harder to do with 4 cards as the CPU can bottleneck the more powerful cards.

If you want more proof go and check out the Firestrike Ultra bench using the graphics scores.

https://forums.overclockers.co.uk/showthread.php?t=18629665

The 384 bit bus on the Titans only really starts working at high resolution.

 

[mahius]

Not many 1080s do 2190mhz, like almost none.

Pascal Titan @2160p is nearly 50% faster than a 1080 in graphics heavy games.

Kaap, normally you're the sensible one (despite having quad TXP in your current build), but mate... you need to lay off the drink. A 1080 ranges in the 50fps when trying run games at 4k ultra and TXP might hit 60 in a few, but narrowly misses on most.

Does this look 50% faster to you mate?

https://www.youtube.com/watch?v=ZwV91khyoO0&t=515s

If TXP was enough to maintain 4k60, I'd have one by now. Alas it's not quite enough and it's not very good value for money anyway, especially when comapred to a 1080. And let's not forget... most folks can't afford even a 1080, let alone over £1k for a GPU. TXP alone costs as much as the rest of most people's PCs. It's not a card that one can recommend to just anybody.

 

[Greebo]

It never slipped Volta* wasn't and isn't aimed at GeForce despite what people generally assume (that isn't to say we'll never see Volta in consumer cards). Amongst other factor being stuck on 28nm so long has kind of upset the progression of things.



Its a misconception - at the presentation they talked about Volta as a "future" process and didn't specify it in a timeframe or that it actually came directly after Maxwell - they actually talked in vague terms in regard to what came immediately after Maxwell - if you look at the slides you'll notice there isn't a date there even though it falls into the slot for "2016".


* There has been some renaming, etc. which confuses things as well.

Was pretty sure the old routemaps went straight onto Volta with no mention of Pascal.

They only created Pascal because they would have had a two year gap with no new cards out if they had to wait for Volta.

 

[maonayze]

You asked.

Check out the 4 card scores


4 GPU

1. Score 3990, GPU TitanP @2063/2764, CPU 6950X @4.4, Kaapstad Link
2. Score 2755, GPU 1080 @2012/2750, CPU 6950X @4.4, Kaapstad Link
3. Score 2682, GPU TitanM @1480/2002, CPU 5960X @4.0, Kaapstad Link
4. Score 2000, GPU Fury X @1140/500, CPU 5960X @4.0, Kaapstad Link
5. Score 1759, GPU TitanK @981/1788, CPU 3930k @4.8, Kaapstad Link
6. Score 1702, GPU 980 @1472/1962, CPU 5960X @4.0, Kaapstad Link
7. Score 1682, GPU 290X @1230/1500, CPU 4930k @4.8, Kaapstad Link
8. Score 1546, GPU RX480 @1340/2200, CPU 6950X @4.4, Kaapstad Link
9. Score 1382, GPU 290X @1000/1250, CPU 3970X @4.9, AMDMatt Link

Since when has 3990 been 50% more than 2755? I would agree that the 4xTXP is 50% more than the Fury X but not the 1080. Sorry dude your maths is out a bit there.

 

[TwsT]

*checks crystal ball*

don't know but after the 1080ti.

 

[nashathedog]

Everything concerning the 1080ti depends on Vega, For starters, Do Nvidia have exact performance figures for Vega or are they in the dark too? If they know the numbers and it is not a TXP competitor they may release a 1080ti just before Vega releases as they did with the 980ti and Fiji cards, If they don't know they may wait until afterwards and then if it is slower than a TXP release a 1080ti with less cores and aftermarket cooling so it can match the TXP, If Vega is faster or as fast as the TXP they may release a fully chipped version as they did with the 780ti and Titan Black when the 290x came out matching the original Titan, The question then will be are they continuing as 10 series cards or will they push out a faster clocking refreshed range as is often mentioned, Personally I'd be very surprised if they could get the current Pascal card to clock further but I'm no expert.

 

[Sc00p007]

Since when has 3990 been 50% more than 2755? I would agree that the 4xTXP is 50% more than the Fury X but not the 1080. Sorry dude your maths is out a bit there.

Unless my maths it out, i make that 45% not far off the claimed 50%. Isn't it more like 100% faster than the Fury X?

 

[Kaapstad]

Not many 1080s do 2190mhz, like almost none.

Pascal Titan @2160p is nearly 50% faster than a 1080 in graphics heavy games.

Since when has 3990 been 50% more than 2755? I would agree that the 4xTXP is 50% more than the Fury X but not the 1080. Sorry dude your maths is out a bit there.

Not my maths only your reading.

 

[Kaapstad]

Kaap, normally you're the sensible one (despite having quad TXP in your current build), but mate... you need to lay off the drink. A 1080 ranges in the 50fps when trying run games at 4k ultra and TXP might hit 60 in a few, but narrowly misses on most.

Does this look 50% faster to you mate?

https://www.youtube.com/watch?v=ZwV91khyoO0&t=515s

If TXP was enough to maintain 4k60, I'd have one by now. Alas it's not quite enough and it's not very good value for money anyway, especially when comapred to a 1080. And let's not forget... most folks can't afford even a 1080, let alone over £1k for a GPU. TXP alone costs as much as the rest of most people's PCs. It's not a card that one can recommend to just anybody.

I think the guys who did the youtube video should lay off the drink lol.

When comparing the Titan and 1080 @2160p running the same clocks the difference is between 30% and 40% for most games.

Lets see what happens when you run both on identical systems at identical clockspeeds. Fortunately I have some 1080s too.

CPUs 6950Xs @4.0
Clockspeed used for both cards 2101/2772
2160p max settings
DX12


Hitman 47

MirrorQuality is not an option available to set but it is also something the 1080 did not use.





ROTTR






Warhammer

 

[Panos]

@Kaapstad

By any chance do you sell one of your furyX? You have 4
(seriously want one)

 

[TheRealDeal]

I think the guys who did the youtube video should lay off the drink lol.

When comparing the Titan and 1080 @2160p running the same clocks the difference is between 30% and 40% for most games.

Lets see what happens when you run both on identical systems at identical clockspeeds. Fortunately I have some 1080s too.

CPUs 6950Xs @4.0
Clockspeed used for both cards 2101/2772
2160p max settings
DX12


Hitman 47

MirrorQuality is not an option available to set but it is also something the 1080 did not use.





ROTTR






Warhammer

So around 35%. Remember yours is a best case Scenario for the Titan as it's watercooled. With the stock air cooler it won't hold those boosts unless you are cranking the fan up. The gtx1080 with aftermarket coolers will. So you are probably losing some performance on the Titan unless watercooling. 4k is also a best case scenario as at 1440 you can take another 5% off. The original Titan was around 40% faster than the gtx680 at 1600p and cost £800. That seemed like silly money back then but Titan Pascal makes it look like a decent move

With Nvidia now not supporting 4 way sli will you be buying 4 of there next big card.

 

[Silent_Scone]

When I say so.

My wallet can't currently take it.

 

[Dave2150]

With the 1080 Ti looking to come out in Spring, is Volta even going to be released this year?

I'm looking for more power than what my 1080 offers at 3440x1440 but I also want a cool and quiet experience that only hybrid and water-cooled cards can offer. Given how abysmal the aftermarket 1080 launch was, the AIO 1080 Tis are unlikely to be available until the end of the year, in which case it might be better for me to wait until Volta is released.

But when do we think it's actually going to be released?

1080ti will have excellent availability from all AIB partners upon release. NVIDIA have been sitting on it for months after all.

That said, it would be wise to wait for Vega to release before buying a 1080ti, unless you're a die hard NVIDIA lover.

 

[Ash0787]

Heres why I wouldn't advise buying a Vega - look at AMDs current best card the Fury X,
its easily beaten in most cases by the 1070, especially in VR where the difference is even higher than you see in typical benchmarks such as this one http://www.tweaktown.com/tweakipedia/116/fury-vs-gtx-1070-battlefield-dx11-dx12/index.html

Fury X also downclocks itself for no reason and you have to install third party software to fix it, also lacking various features that nvidia has such as shadowplay, SMP etc,
and its power consumption is approx twice that of the 1070 ( 2x 8pin connector vs 1x 8 pin on the 1070 ). Also don't forget that Fury X cost £500 to £600 at launch, whereas the 1070 was around £350 to £500.

Wouldn't surprise me if a similar thing happened with Vega too.

I wouldn't mind a Volta architecture 1080 equivalent or even the full size chip but it sounds like we still don't know when that is likely to materialize.

 

[TheRealDeal]

Heres why I wouldn't advise buying a Vega - look at AMDs current best card the Fury X,
its easily beaten in most cases by the 1070, especially in VR where the difference is even higher than you see in typical benchmarks such as this one http://www.tweaktown.com/tweakipedia/116/fury-vs-gtx-1070-battlefield-dx11-dx12/index.html

Fury X also downclocks itself for no reason and you have to install third party software to fix it, also lacking various features that nvidia has such as shadowplay, SMP etc,
and its power consumption is approx twice that of the 1070 ( 2x 8pin connector vs 1x 8 pin on the 1070 ). Also don't forget that Fury X cost £500 to £600 at launch, whereas the 1070 was around £350 to £500.

Wouldn't surprise me if a similar thing happened with Vega too.

I wouldn't mind a Volta architecture 1080 equivalent or even the full size chip but it sounds like we still don't know when that is likely to materialize.

The worst post of 2017 currently goes to Ash0787.

What the majority of this has to do with Vega i have no idea. Fury X is last Gen on the 28nm process and the gtx1070 is this gen on 14nm. You failed to mention that the gtx970 was way cheaper than the gtx1070 at launch. Fury X was a top end card where as the gtx1070 is not. You have no idea on Vega's performance or price.

Amd's feature set just got a major bump and is just about on par with Nvidia's these days.

 

[Dave2150]

The worst post of 2017 currently goes to Ash0787.

What the majority of this has to do with Vega i have no idea. Fury X is last Gen on the 28nm process and the gtx1070 is this gen on 14nm. You failed to mention that the gtx970 was way cheaper than the gtx1070 at launch. Fury X was a top end card where as the gtx1070 is not. You have no idea on Vega's performance or price.

Amd's feature set just got a major bump and is just about on par with Nvidia's these days.

No thank you-schizo

No-one knows Vega's final performance numbers. Even if someone is planning on buying a 1080, or 1080ti, Vega could have a big affect on their prices, due to competition. Notice I use tentative language, since it might not affect the prices at all.

With the 1080ti and Vega looming on the horizon, IMO it's currently a very bad time to buy any high end GPU. Best to wait for both to launch and decide from there