**Nvidia GeForce GTX 1080 Ti Thread**

[Dave2150]

There is absolutely no reason for NVidia to release a 1080 Ti ATM.

Failed GP102 chips can be used in the Pascal Titan.

There is absolutely no competition from AMD at the high end.

A further cut down GP102 chip released as a 1080 Ti would be pretty useless performance wise and would be struggling to beat a 1080. Worse still it would be very expensive compared to the 1080.

I also think the next Titan based on the full GP102 chip is about 12 months away unless Vega is a lot faster than expected.

Are you aware you repeated what I said, albeit in many more words?

 

[nashathedog]

If the current Titans are failed quadro, forget the 1080ti. Aint coming.
Titan was the full chip and 780 the failed version. By the time Nov 2013 hit, Titan Black was the full chip, 780ti the failed, and 780ghz was some failed Blacks. Also 780ghz was as fast as the full fledge Titan.

The 780ti and Titan Black used the same full chip (GK110). The first Titan was a cut down version of that same chip (GK110), the 780 was a cut down even further version of that chip (GK110).

780ghz was simply Gigabytes version, wasnt a special gpu really

I think the GHZ chips where B1's where as the regular 780's where A1's, Not 100% positive on that though. I know my 780 Classified was an A1 and a mates GHZ edition was a B1 that clocked better.

 

[Panos]

This is not quite right.

The original Titan was a failed chip, not the full. The GTX 780 was a more failed chip, and also aged very badly, just as a side note.

The 780 Ti was then the full chip, and the Titan Black was a full chip again but with more RAM and DP enabled.

Then the Titan XM was the full chip, and the 980 Ti a failed chip. However due to the clock relationship they were the same speed in practice because the 980 Ti would clock higher (unless you watercooled the Titan XM). Also at this point the Titan didn't have DP enabled, so 'Titan' now meant 'very expensive for no reason'.

Now the Titan XP also doesn't have DP enabled, and is a failed chip. The full chip being 3840 cores.

Please read post #87

 

[nashathedog]

Please read post #87

That the problem with this forum set up, If you're not used to it you reply to someones post and then find many others have already done so as you scroll down further.

Would be nice if they updated the forum software, its pretty dated.

If it was done right it'd certainly help us avoid repeating each others posts

 

[Razor Time]

That the problem with this forum set up,

Would be nice if they updated the forum software, its pretty dated.

 

[subbytna]

If there is to be a 1080Ti in the pipeline, what is it going up against from AMD? Surely they won't have anything that will even get close to a Ti performance?

Depending on cost, a decent upgrade for me would be the 1080Ti, but then I'd have to get a new monitor too...decisions decisions

 

[subbytna]

probably a spoof but saw it and posted it up anyway

 

[Deleted member 651465]

Tip: Take out the code after the v= part of the YouTube link and put it in

 

[Gerard]

If there is to be a 1080Ti in the pipeline, what is it going up against from AMD? Surely they won't have anything that will even get close to a Ti performance?

Depending on cost, a decent upgrade for me would be the 1080Ti, but then I'd have to get a new monitor too...decisions decisions

Doesn't need to be anything from amd, nvidia know that there's people that are waiting for the ti to come out as Titan has just hit. So I'd imagine around October time it's more than likely one will appear.

 

[LeMson]

I'd say November, just before xmas, I'm itching to buy a 1080 zotac amp extreme, but then I'm expecting the ti out in a few months, and then reading about the 1180GTX and HBM2 prolly coming next spring I'm waiting again :X

 

[V F]

I'd say November, just before xmas, I'm itching to buy a 1080 zotac amp extreme, but then I'm expecting the ti out in a few months, and then reading about the 1180GTX and HBM2 prolly coming next spring I'm waiting again :X

Yes. I keep getting impatient wanting to replace my 5870's but I have to keep telling myself, PATIENCE!

I'd rather go for a Ti though.

 

[subbytna]

Tip: Take out the code after the v= part of the YouTube link and put it in

 

[LeMson]

Yes. I keep getting impatient wanting to replace my 5870's but I have to keep telling myself, PATIENCE!

I'd rather go for a Ti though.

Yes buy now or a ti soon, but if a HBM2 1180GTX does come out some time next spring/summer, it would be such a leap forward, like 4 times the memory speed I dunno, money isn't an issue for me, I just prefer bang for buck and my 970gtx is still doing me ok bar some dodgy performance issues recently.

 

[Rroff]

I think the GHZ chips where B1's where as the regular 780's where A1's, Not 100% positive on that though. I know my 780 Classified was an A1 and a mates GHZ edition was a B1 that clocked better.

Yeah the GHZ were all B1 cores - some of the later Classified and 1-2 other models used B1 cores later. I'm pretty sure they used the B1 Kepler as a testing ground for power/clock improvements that would then go into Maxwell. Not really doing them justice to say they weren't "special" as most came out the box boosting 17-19% faster than your average A1 780 and still had around 10% overclocking potential atleast on top of that.

Always found it frustrating when reviewers use 780 numbers and no indication if its an early A1 that is barely doing more than 950MHz a later A1 that could be doing around 1100MHz or a B1 chip that could be running as high as 1200MHz.

 

[setter]

I had a pair of gigabyte wf 780's, (non ghz). Each one was a different revision, and one had a shorter pcb than the other.

 

[Greebo]

Yes buy now or a ti soon, but if a HBM2 1180GTX does come out some time next spring/summer, it would be such a leap forward, like 4 times the memory speed I dunno, money isn't an issue for me, I just prefer bang for buck and my 970gtx is still doing me ok bar some dodgy performance issues recently.

Only meant to be double the bandwidth and Nvidia is only going for stacks of 4.

Main benefit will be half the power consumption.

Although if Nvidia goes for 16Gb it will still use more power than 8gb gddr5x

Not sure how much will be gained from the memory bandwidth performance wise. 10% on pascal doesn't equate to 10% more performance.

All depends how much of a beast the new chip is

 

[Kaapstad]

Only meant to be double the bandwidth and Nvidia is only going for stacks of 4.

Main benefit will be half the power consumption.

Although if Nvidia goes for 16Gb it will still use more power than 8gb gddr5x

Not sure how much will be gained from the memory bandwidth performance wise. 10% on pascal doesn't equate to 10% more performance.

All depends how much of a beast the new chip is

You need clockspeed as well as bandwidth.

I suspect HBM2 will have a hard time trying to beat GDDR5X based cards with a 384bit bus and memory clockspeeds that will be even faster than the current Pascal cards.

 

[subbytna]

You need clockspeed as well as bandwidth.

I suspect HBM2 will have a hard time trying to beat GDDR5X based cards with a 384bit bus and memory clockspeeds that will be even faster than the current Pascal cards.

Isn't HMB2 the next big thing in memory?

Perhaps I need to go read up on what's in the pipeline technology wise.

 

[Greebo]

Isn't HMB2 the next big thing in memory?

Perhaps I need to go read up on what's in the pipeline technology wise.

It is but mainly on power. It's not the massive leap in performance that people are thinking. I've seen it said hbm2 is 10 times faster than gddr5x. It isn't.

 

[Gerard]

I'd be more interested in seeing hbm 2 utilised the way hbm1 was with fiji to make cards drastically smaller. Just about every card these days has some degree of sagging to it, and over time i can't imagine that's good for the pcb with it going through constant thermal cycles, nano and fury x were too short for sagging. Be nice to see a top end geforce around that size.