$560 MSI GTX 980 Ti Golden Edition

[Silverforce11]

CF or SLI is a non starter with so many games lacking support.

You can count DX12 games into this because few devs are good enough to handle DX12 multi-GPU or even if they could, the publishers won't bother to waste time/$ on it.

Then there's UWP...

 

[thesmokingman]

Probably not a mistake, they are starting to clear out the inventory. It'd be a bad idea to buy such an expensive card though right now given what happened with Kepler. Best to wait for Pascal.

Yeap not a mistake. It's official the maxwell line has hit the end of the line.

http://www.techpowerup.com/221783/nvidia-reportedly-stops-production-of-certain-maxwell-gpus

 

[SexyK]

Have a Titan X in my main gaming rig in the sig. Building a new 6700K mini-ITX system for room-scale VR in the living room. Any reason not to pick up an EVGA 970 now then step up to a 1080/1080ti once they launch? Never used the step up program before. Was contemplating just sucking it up and going with a 980ti, but I'd rather not "step down" if the 1080 ends up costing $550 rather than $650 and eating the difference.

 

[Kenmitch]

Have a Titan X in my main gaming rig in the sig. Building a new 6700K mini-ITX system for room-scale VR in the living room. Any reason not to pick up an EVGA 970 now then step up to a 1080/1080ti once they launch? Never used the step up program before. Was contemplating just sucking it up and going with a 980ti, but I'd rather not "step down" if the 1080 ends up costing $550 rather than $650 and eating the difference.

Sure there is. What if they launch after your step-up window closes. No firm real launch date nor performance numbers to see. Best to wait it out if you can.

 

[thesmokingman]

Have a Titan X in my main gaming rig in the sig. Building a new 6700K mini-ITX system for room-scale VR in the living room. Any reason not to pick up an EVGA 970 now then step up to a 1080/1080ti once they launch? Never used the step up program before. Was contemplating just sucking it up and going with a 980ti, but I'd rather not "step down" if the 1080 ends up costing $550 rather than $650 and eating the difference.

Nv is playing the cut supply game to lower inventory in order to artificially keep prices from bottoming out. What that means is right now there's a too much supply and not enough demand. Given that, I'm not sure when is a good time to buy, but if its not much of a concern, it doesn't matter if you're just going to step up later. However, I'd check the specifics of the program as its not a free for all, there are limits like 90 days to do it.

 

[SexyK]

Sure there is. What if they launch after your step-up window closes. No firm real launch date nor performance numbers to see. Best to wait it out if you can.

Nv is playing the cut supply game to lower inventory in order to artificially keep prices from bottoming out. What that means is right now there's a too much supply and not enough demand. Given that, I'm not sure when is a good time to buy, but if its not much of a concern, it doesn't matter if you're just going to step up later. However, I'd check the specifics of the program as its not a free for all, there are limits like 90 days to do it.

Yes - point taken. My understanding is that the window is 90 days. That puts us in mid-to-late July depending on when I order, so I think the safe money is on the product being launched at that point. Obviously no guarantees though. Clearly, the longer I can wait the better.

Also, I believe you can only step up to "vanilla" cards rather than custom versions. Not a big worry from my perspective, but some may care.

 

[thesmokingman]

Yes - point taken. My understanding is that the window is 90 days. That puts us in mid-to-late July depending on when I order, so I think the safe money is on the product being launched at that point. Obviously no guarantees though. Clearly, the longer I can wait the better.

Also, I believe you can only step up to "vanilla" cards rather than custom versions. Not a big worry from my perspective, but some may care.

You also have to wait in the queue for your turn. Think how many others thinking the same are in the queue as well. Then the early low stock issues, especially with a brand new node and the teething that comes with it.

 

[Arachnotronic]

Have a Titan X in my main gaming rig in the sig. Building a new 6700K mini-ITX system for room-scale VR in the living room. Any reason not to pick up an EVGA 970 now then step up to a 1080/1080ti once they launch? Never used the step up program before. Was contemplating just sucking it up and going with a 980ti, but I'd rather not "step down" if the 1080 ends up costing $550 rather than $650 and eating the difference.

This is a good idea. Do this.

 

[xthetenth]

Going with a cheap EVGA 970 seems like a solid solution, you likely won't be able to step up to a 1080 Ti because it's likely a long time before they release though.

 

[Det0x]

The only explanation I have is AMD fans are trying to save face by not having GP104 steamroll Polaris 10 and AMD having nothing until Vega. OTOH, 980Ti/Titan X users are in denial that their card is about to get embarrassed in all metrics, the same way 970/980 made 780Ti look like outdated/overpriced and power hungry tech Day 1 of their release.

If my memory is *correct:

Samsung's 14nm LPE is ~10% more dense then TSMC's 16nm.
Samsung's 14nm LPP is ~16(?) more dense then 14nm LPE

232mm * 1.1 * 1.16 = 296

If both chips were made at TSMC, they would pretty much be equal in size. (296 vs 294)

I just think Polaris 10 will beat GP104 in all new AAA directX 12 games, since they are pretty much all optimized for consoles (amd hw)

* = believe silverforce linked some information regarding samsung 14nm vs tsmc 16nm a few days back.

 

[Arachnotronic]

If my memory is *correct:

Samsung's 14nm LPE is ~10% more dense then TSMC's 16nm.
Samsung's 14nm LPP is ~16(?) more dense then 14nm LPE

232mm * 1.1 * 1.16 = 296

If both chips were made at TSMC, they would pretty much be equal in size. (296 vs 294)

I just think Polaris 10 will beat GP104 in all new AAA directX 12 games, since they are pretty much all optimized for consoles (amd hw)

* = believe silverforce linked some information regarding samsung 14nm vs tsmc 16nm a few days back.

There is no density improvement from 14 LPE -> 14 LPP, the gate/metal pitches were not changed. The difference is entirely in the transistor performance.

 

[SexyK]

You also have to wait in the queue for your turn. Think how many others thinking the same are in the queue as well. Then the early low stock issues, especially with a brand new node and the teething that comes with it.

Another good point. However, I'm assuming that the 970 will be sufficient for most of the room-scale VR experiences that are out there right now/launching soon. Most of these have pretty basic visuals. So, notwithstanding the need to push 90 fps on these VR headsets, my theory is that the 970 should be "good enough" while I wait for my turn in the queue. Of course, this is contingent on the 970 meeting my expectations in the first place, which may be tough coming from the Titan X. I guess if it doesn't measure up, I could just return it straight away and wait it out.

This is a good idea. Do this.

Thank you for the positive reinforcement.

Going with a cheap EVGA 970 seems like a solid solution, you likely won't be able to step up to a 1080 Ti because it's likely a long time before they release though.

Agreed - just using that notation for whatever the "flagship" launch in June/July is in case they mess with the naming scheme again.

As a side note, this is really the first time I've been bitten by the down-side of custom water cooling. It would be nice to be able to throw the Titan X in the VR build and put a placeholder in my traditional gaming build, but it's just not practical. Actually, it would be a massive hassle.

 

[Det0x]

There is no density improvement from 14 LPE -> 14 LPP, the gate/metal pitches were not changed. The difference is entirely in the transistor performance.

Could not find the previous posted link, but i found something else.

Samsung’s 14 nm LPE FinFET Transistors

What should we expect for their 14 nm Low Power Plus (LPP) process? Samsung’s recent press releases suggest that the LPP process will feature a 15% increase in the transistor switching speed and a 15% power consumption decrease. These are being achieved by increasing the transistor’s fin height and enhanced strain engineering. For me, I am anticipating a bit of a process shrink as well to bring the transistor and 6T SRAM cell sizes closer to that of Intel.

For now, we will bide our time until the Exynos 8 SoC or Snapdragon 820 SoC are released in the next generation of Apple and Samsung Smartphones.

About TechInsights: Our comprehensive technical body of knowledge explains the use of patented inventions in hardware and software, proving patent value for technology, legal and patent professionals.

Maybe its me thats misunderstanding whats being said

 

[Arachnotronic]

Could not find the previous posted link, but i found something else.

Samsung’s 14 nm LPE FinFET Transistors



Maybe its me thats misunderstanding whats being said

This person's "prediction" was just wrong, obviously hasn't paid attention to what 14LPP actually is supposed to be.

 

[Det0x]

This person's "prediction" was just wrong, obviously hasn't paid attention to what 14LPP actually is supposed to be.

Ill leave it up to your own interpretation then, but my point were and still are:

Lets say GK104 = 980ti + 20-30% (Aftermarket 980ti + 0-10% which would be a positive surprise)

vs

Polaris 10 = 290x + 30-40% (low figure?)

What will be fastest in the new dx12 games ? I mean just look at the recent history with dx12 games, Nvidia (Maxwell) ain't looking so hot..

Their only saving grace would be if they managed to copy the "old and outdated" GCN AMD architecture which they have set out to do with Pascal.

 

[utahraptor]

Going with a cheap EVGA 970 seems like a solid solution, you likely won't be able to step up to a 1080 Ti because it's likely a long time before they release though.

Just before the division came out I got their fastest "silent" one for a really good deal and it came with the Division. It's clocked very high, but one of he fans can rattle from time to time negating the benefit of the quiet technology.

 

[monstercameron]

Another good point. However, I'm assuming that the 970 will be sufficient for most of the room-scale VR experiences that are out there right now/launching soon. Most of these have pretty basic visuals. So, notwithstanding the need to push 90 fps on these VR headsets, my theory is that the 970 should be "good enough" while I wait for my turn in the queue. Of course, this is contingent on the 970 meeting my expectations in the first place, which may be tough coming from the Titan X. I guess if it doesn't measure up, I could just return it straight away and wait it out.



Thank you for the positive reinforcement.



Agreed - just using that notation for whatever the "flagship" launch in June/July is in case they mess with the naming scheme again.

As a side note, this is really the first time I've been bitten by the down-side of custom water cooling. It would be nice to be able to throw the Titan X in the VR build and put a placeholder in my traditional gaming build, but it's just not practical. Actually, it would be a massive hassle.

I would recommend a 970 for any kind of room scale vr -vive? Especially for close to the next gen releases.

 

[raghu78]

This person's "prediction" was just wrong, obviously hasn't paid attention to what 14LPP actually is supposed to be.

You have nothing to prove your statement. Until there is a chipworks or techinsights teardown of Snapdragon 820 with photos of the 14LPP transistor structure which reveals metrics like contacted gate pitch, M1 metal pitch you have no way to say if there is a slight shrink from 14LPE to 14LPP. So just wait and lets see.

http://www.anandtech.com/show/9686/the-apple-iphone-6s-and-iphone-6s-plus-review/3

 

[Silverforce11]

GF also has IBM's 14nm FF as well, not just Samsung. This was recently made public but they had it for awhile.

 

[raghu78]

GF also has IBM's 14nm FF as well, not just Samsung. This was recently made public but they had it for awhile.

IBM's 14nm is SOI FINFET and a high performance CPU focussed process node. I doubt AMD will use that node. For GPUs most definitely its 14LPP. For Zen I don't know what AMD chose - TSMC 16FF+ / Samsung 14LPP / IBM 14nm SOI FINFET.

I don't think 14nm SOI FINFET is anywhere close to volume production while the other two nodes are. So its likely that Zen is 16FF+ or 14LPP.

 

[PPB]

That 14nm SOI FINFET would be ideal for AMD as their performance oriented node if they had the economics to go for it.

Nevertheless 14nm LPP shouldnt be dimissed and its 3rd gen FF succesor either, as we can see by now that 14/16nm process will be long lived ones.

Sent from my XT1040 using Tapatalk

 

[Arachnotronic]

You have nothing to prove your statement. Until there is a chipworks or techinsights teardown of Snapdragon 820 with photos of the 14LPP transistor structure which reveals metrics like contacted gate pitch, M1 metal pitch you have no way to say if there is a slight shrink from 14LPE to 14LPP. So just wait and lets see.

http://www.anandtech.com/show/9686/the-apple-iphone-6s-and-iphone-6s-plus-review/3

"Common design rules for fast migration."

If you need me to spell it out for you: this means that the gate/metal pitches did not change.

Both 16FF+ and 14LPP were about making the silicon fins taller in order to improve drive strength relative to 16FF and 14LPE, respectively.

 

[raghu78]

"Common design rules for fast migration."

If you need me to spell it out for you: this means that the gate/metal pitches did not change.

Both 16FF+ and 14LPP were about making the silicon fins taller in order to improve drive strength relative to 16FF and 14LPE, respectively.

Common design rules does not mean no area advantages. TSMC 16FFC provides area reduction even though it shares the same design rules as TSMC 16FF+.

http://community.cadence.com/cadenc...instream-designers-and-internet-of-things-iot

"16FFC is a “compact” version of the 16nm FinFET+ (16FF+) process technology that is now in risk production at TSMC. It claims advantages in power, performance, and area compared to the existing 16FF+ process, along with easy migration from 16FF+. It can be used for ultra low-power IoT applications such as wearables, mobile, and consumer."

"To migrate to 16FFC, standard cell designers will need to re-characterize their libraries for power, performance, and area optimization. TSMC will offer new, optimized SRAM compilers. I/Os will not be changed from 16FF+. For analog and interface IP, TSMC recommends that users run SPICE simulation to check margins and re-characterize as needed.

Hou said that 16FFC will go into risk production in 2016. By the end of 2015, TSMC will provide foundation IP with fully characterized corners. All interface and analog IP will be finished by the second quarter of 2016. Design rules remain the same as in 16FF+."

 

[Arachnotronic]

Common design rules does not mean no area advantages. TSMC 16FFC provides area reduction even though it shares the same design rules as TSMC 16FF+.

Design rules = metal/gate pitches. Yes, 16FFC sacrifices performance for cost/area reduction, but that doesn't mean that the minimum metal/gate pitches got any tighter (they didn't).

 

[guskline]

I read the EVGA step-up rules and , quite frankly, I'm not sure buying a GTX970 now and "stepping up" to a 1080 will qualify. I would contact EVGA and ask if it would. Get in writing (official e-mail).