NVIDIA GeForce GTX TITAN X 3DMark performance From [ videocardz ] & [ChipHell]

[Memphaestus]

Single GPU card vs. dual. No single GPU card in history has beaten the previous generation's dual-chip flagship.

Fair enough.  I just thought that since Nvidia is always so much better than AMD, that Nvidia would crush the 295x2 with their new uber card.  /S

 

But really I guess it isn't that big of a deal.  I'm just hoping that AMD gives Nvidia a run for their money this time around, as Nvidia has been owning for a little too long right now. 

 

C'mon $999.99, daddy needs a new GFX card!

 

The benchmarks look great and I think I found the card for my system, sorry not gonna wait for the 8gb 980s when a 12gb titan comes around.

Will I finally get to play games smoothly at 11520x2160?  4x 980s get me "good" frame rates and at very high settings, but will 4x titan xs fare better? 

 

Jesus man! I'll take a couple dollars off your hand if you've got some to spare.

[merolis]

Subtracting all the Nvidia vs AMD hate going on.

How does the sli scaling on a card go

1st overclock  100%
2nd overclock 185%
3rd overclock  202%
4th overclock  240%

The 2nd card is at 85% strength.
The 3rd card is at 17%.
The 4rd card is at 38%.

I will give that nvidia has good drivers but that scaling is bullshit.

[mr moose]

Fair enough.  I just thought that since Nvidia is always so much better than AMD, that Nvidia would crush the 295x2 with their new uber card.  /S

 

But really I guess it isn't that big of a deal.  I'm just hoping that AMD gives Nvidia a run for their money this time around, as Nvidia has been owning for a little too long right now. 

 

They will, like all competing companies they frog hop each other,  It's just that AMD have had an unstable decade where nvidia and intel didn't.  if you go back ten years AMD were killing it.  In fact you were a fool if you bought intel for gaming.  GPUs have always bounced between the two.  I think the problem at the moment is all the fanboys who make it sound like it's all one way, to prove this If I say what do you think the hottest and most unstable GPU was, if you say AMD you'd be wrong, they have been unstable but not the hottest, Nvidia 480 was both cooking breakfast and unstable at the same time.  But we only think of AMD when we mention hot power hungry GPUs. 

[Opcode]

But we only think of AMD when we mention hot power hungry GPUs. 

I think most of us immediately think of Nvidia when it comes to them terms due to their track record (so many hot GPUs). Their the only company that anyone can remember that launched a GPU that can literally cook itself.

[Dabombinable]

I think most of us immediately think of Nvidia when it comes to them terms due to their track record (so many hot GPUs). Their the only company that anyone can remember that launched a GPU that can literally cook itself.

*sigh* kind of like their Geforce 256, the dinky heatsink and fans used on them were terrible at actually cooling them, and I've measured my one at 127^oC in use with its heatsink (fan broke years ago). The funny thing is that it still works without locking up unlike later Nvidia cards. Just with really, really bad artifacting.

[patrickjp93]

That's wasn't the point of my post at all. Price to performance ratio is the single biggest factor in this market. The R9 390X performance is expected to fall somewhere around the TITAN X while speculation suggests the card to drop at a MSRP of $599 (seems about right). I don't know about you but I'm not ready to pay double for a GPU that doesn't offer any difference in performance. At the end of the day the value is simply not there. You could home brew your own Bermuda for the same price and that's really saying something.

No, there are a ton of factors which influence purchases. Why do you think AMD, current holder of the performance and performance/price crowns, can't seem to budge Nvidia's marketshare? Reliability, perceived quality, customer service, outspokenness... If price/performance was everything no one would buy Intel CPUs or Nvidia GPUs.

 

Opcode you really need to step back again and chill with the blatant fanboyism.

 

Also, no way in Hell does the 390X drop for just $600 out of the gate. Between HBM, new memory controller development, 35% larger die, and the liquid cooler, no way in Hell it's less than $750 out of the gate. The economics of it just make no sense to launch at $600. The market is hot and itching and needs a strong and unyielding strike. AMD has to seize on that while also making itself money.

 

We also do not yet know the official Titan X pricing. Given it can't actually serve as a poor man's Quadro this time around (no 64-bit SPs on Maxwell, so only 1/16 or 1/32 the performance), I predict it'll land between $800 and $900.

[patrickjp93]

Fair enough.  I just thought that since Nvidia is always so much better than AMD, that Nvidia would crush the 295x2 with their new uber card.  /S

 

But really I guess it isn't that big of a deal.  I'm just hoping that AMD gives Nvidia a run for their money this time around, as Nvidia has been owning for a little too long right now. 

 

 

Jesus man! I'll take a couple dollars off your hand if you've got some to spare.

AMD needs to come up with a mature, sophisticated marketing system and a way to convince people 12 GB of GDDR5 is pointless (best of luck on that second one). The performance needs to be there, the reliability needs to be there, the price needs to be right, and so does the marketing.

[mr moose]

I think most of us immediately think of Nvidia when it comes to them terms due to their track record (so many hot GPUs). Their the only company that anyone can remember that launched a GPU that can literally cook itself.

 

Really? don't know what forums you've been on recently but on hear there are more AMD heat jokes and complaints about driver stability than anything Nvidia related.

[Guest Kloaked]

I think most of us immediately think of Nvidia when it comes to them terms due to their track record (so many hot GPUs). Their the only company that anyone can remember that launched a GPU that can literally cook itself.

 

I don't think most of the forum is old enough to remember anything that long ago.

[Opcode]

Really? don't know what forums you've been on recently but on hear there are more AMD heat jokes and complaints about driver stability than anything Nvidia related.

You must be a young padawan that's following the wrong group of users. Those of us who actually lived long enough know Nvidia is at fault more than ATI/AMD in these particular circumstances. Tho even with your addition to the thread Nvidia is the only one to actually release drivers that brick cards faster than the earth orbiting the sun.

[mr moose]

You must be a young padawan then that's following the wrong group of users. Those of us who actually lived long enough know Nvidia is at fault more than ATI/AMD in this particular circumstances. Tho even with your addition to the thread Nvidia is the only one to actually release drivers that brick cards faster than the earth orbits the sun.

 

Do you even follow the discussion or do you just  assume the person your talking is wrong? Maybe I need to spell it out for you:

 

If you have a read of any GPU thread on these forums over the last year all you will see is heat and power consumption being used as a reason not to buy AMD.  This leads all the new guys to thinking that AMD are worse for heat and power consumption.  Only those of use who have been around a while (not "young" or "padawan" following the "wrong group" as you so erroneously put it) know that this is not the sole domain of AMD and that Nvidia (as stated before) are the kings of power consumption and heat.

 

Please try to keep up.

[Dezz]

Does the flame next to the GPUs means that they burnt during the benchmarks?

[Opcode]

Does the flame next to the GPUs means that they burnt during the benchmarks?

I guess it would be an indication of the card running on the hot side.

[Guest Kloaked]

Does the flame next to the GPUs means that they burnt during the benchmarks?

 

 

I guess it would be an indication of the card running on the hot side.

 

I hope you guys are joking.

[OlekKing]

AMD needs to come up with a mature, sophisticated marketing system and a way to convince people 12 GB of GDDR5 is pointless (best of luck on that second one). The performance needs to be there, the reliability needs to be there, the price needs to be right, and so does the marketing.

Lol what are you talking about? When i think of AMD's marketing, maturity and sophistication is all that comes to mind:

 

 

”On Sept 18th and 19th we’ll be watching the #Game24 hashtag and the @AMDRadeon twitter channel for pictures of you in attendance at the events, proudly wearing your favorite red T-shirt. Boldly tweet pictures of yourself wearing your colors while at these events, and we’ll send some love your way. The funnier and more inventive the photos, the better – remember, this is a celebration! Make sure to mention @AMDRadeon so we can find your posts easily.”

 

/s

[Tiber1337]

I'd say between 24 - 30% performance increase over a reference 980 is a massive difference.

 

Why? its double the price if I am generous, get a second 980 for the same money,  and you have at least 80% performance increase

[FabulousOtter]

Why? its double the price if I am generous, get a second 980 for the same money,  and you have at least 80% performance increase

 

In no* application will you ever have quote:'at least' 80% performance increase by adding a second GPU

 

 

*Outside of select synthetic benchmarks

[Tim Drake]

At the incredible high price these will cost, it isnt worth it by far.

Isn't worth it? You realise this is halfway between a workstation and a gaming card?

 

It hits at a great price tbh..

[Victorious Secret]

Isn't worth it? You realise this is halfway between a workstation and a gaming card?

 

It hits at a great price tbh..

 

And if you added 12GB of VRAM to any other card, you'd end up with some freakishly high prices as well. R9 290Xs with 8GB of VRAM hover around 700 USD on average. So yea, adding in another 4GB might put you at ~1000 on average. 

 

VRAM isn't cheap no matter how you slice it. HBM certainly won't be cheaper. If the Titan X shows up at 1500, it'll carry some Titan branding surcharge but if it comes in at 999, it'll be a steal of a deal for what its packing. 

[Tim Drake]

And if you added 12GB of VRAM to any other card, you'd end up with some freakishly high prices as well. R9 290Xs with 8GB of VRAM hover around 700 USD on average. So yea, adding in another 4GB might put you at ~1000 on average. 

 

VRAM isn't cheap no matter how you slice it. HBM certainly won't be cheaper. If the Titan X shows up at 1500, it'll carry some Titan branding surcharge but if it comes in at 999, it'll be a steal of a deal for what its packing. 

Can't tell whether you agree with my point or not..

 

V Ram isn't all that matter in a workstation card btw..

[patrickjp93]

Isn't worth it? You realise this is halfway between a workstation and a gaming card?

It hits at a great price tbh..

1) price not announced yet.

2) Maxwell has no native 64-bit SPs. There is no halfway workstation card this time around. It's still fine for CUDA research I suppose, but you're limited to single precision.

[Tim Drake]

1) price not announced yet.

2) Maxwell has no native 64-bit SPs. There is no halfway workstation card this time around. It's still fine for CUDA research I suppose, but you're limited to single precision.

Huh...

 

*scratches head confused*

[patrickjp93]

Huh...

*scratches head confused*

Nvidia has not yet officially announced the price. We only have rumors at this point.

Many people have X-rayed the GM 204 chips and seen there is no native 64-bit SP. In other words, the best performance you can get is 1/32 by using the mantissa extension algorithm to hook 2 32-bit SPs together.

The GM 200 die is only 53% larger than GM 204. To replace all the existing SPs with 64-bit ones would have required a 60%+ increase before adding a single core, and that would have broken the limit TSMC puts on its maximum die sizes (700mm sq.)

[Tim Drake]

Nvidia has not yet officially announced the price. We only have rumors at this point.

Many people have X-rayed the GM 204 chips and seen there is no native 64-bit SP. In other words, the best performance you can get is 1/32 by using the mantissa extension algorithm to hook 2 32-bit SPs together.

I'm even more confused now..

 

I'm 16, not a tech god

[patrickjp93]

I'm even more confused now..

I'm 16, not a tech god

There is no Double Precision performance, so it can't serve as a workstation card for rendering or precise scientific computing.