Nvidia GP 100 Titan to debut in April, GTX 1080 to launch in June

[Big-G]

All I want is a single/dual GPU card capable of 4K at 100Hz or higher

[MageTank]

All I want is a single/dual GPU card capable of 4K at 100Hz or higher

And here i am willing to settle for a single card that can do 1440p at 120hz reliably, lol. 4k @ 60 has yet to be completely mastered on a single card, so 4k at 100 on a single card might take some time, even if Pascal is "2x the performance per watt". Dual GPU/SLI/Crossfire should be able to do it though. 

[Big-G]

I'm willing to wait

[lkarej]

What did Nvidia do to you?

 

Not a fan of their closed garden approach to their software etc.

Especially not a fan of having my performance being cut in half by bad drivers.

 

>implying AMD doesn't have driver issues

 

They all have issues, but I hear much less about issues with AMD drivers than NVidia's.

 

 

+What MageTank said

 

Obviously they moved in next door and are playing loud music at night. Why else would he feel that way?

[DocSwag]

Not a fan of their closed garden approach to their software etc.

Especially not a fan of having my performance being cut in half by bad drivers.

 

>implying AMD doesn't have driver issues

 

They all have issues, but I hear much less about issues with AMD drivers than NVidia's.

 

 

+What MageTank said

This is probably true. However AMD does also have its fair share of issues.

I also do hate their closed garden approach. AMD GPUs do tend to have better raw performance, Nvidia just has really good optimization.

I don't like Nvidia's "Let's get to the market first and make our method not open source so that our GPUs will have a big advantage over AMDs." All of AMDs methods are open source and run equally well on Nvidia and AMD GPUs. As well, they don't charge giant price premiums on some of their methods(just look at freesync vs gsync).

Personally right now I think that Nvidia and Intel are both kind of playing cut throat right now, and AMD is the one trying to get everyone to be fair.

[patrickjp93]

Not a fan of their closed garden approach to their software etc.

Especially not a fan of having my performance being cut in half by bad drivers.

>implying AMD doesn't have driver issues

They all have issues, but I hear much less about issues with AMD drivers than NVidia's.

+What MageTank said

Having your performance cut in half? I call bs.

[patrickjp93]

This is probably true. However AMD does also have its fair share of issues.

I also do hate their closed garden approach. AMD GPUs do tend to have better raw performance, Nvidia just has really good optimization.

I don't like Nvidia's "Let's get to the market first and make our method not open source so that our GPUs will have a big advantage over AMDs." All of AMDs methods are open source and run equally well on Nvidia and AMD GPUs. As well, they don't charge giant price premiums on some of their methods(just look at freesync vs gsync).

Personally right now I think that Nvidia and Intel are both kind of playing cut throat right now, and AMD is the one trying to get everyone to be fair.

Mantle never went open. AMD still has proprietary drivers across the board (unlike Intel which does have FULLY open-source drivers which contain no binary blobs). AMD is not nearly the open source champion people claim it to be. If anyone gets that crown, it's Intel. OpenMP, CilkPlus, Thread Building Blocks (incoming), OpenACC, graphics drivers, Ethernet drivers, Infiniband drivers, Linux kernel development, BSD kernel development, FreeNAS, and more.

[Metro]

If every report that was unconfirmed was removed from these forums, the LTT forums would be pretty bare.

 

This article needs to have "Unconfirmed" then, as that source is terrible and if you read it, the information is wrong.

[ACatWithThumbs]

I highly doubt the info is legit.
The Titan Cuda core count seems to low I highly doubt GP100/16nm maxes out on 4000 cuda cores.

I would expect that count maybe from GP104 but GP100 should be easily 5000-6000 unless the size of the cores increased drastically or they want to have a very small die.

[patrickjp93]

I highly doubt the info is legit.

The Titan Cuda core count seems to low I highly doubt GP100/16nm maxes out on 4000 cuda cores.

I would expect that count maybe from GP104 but GP100 should be easily 5000-6000 unless the size of the cores increased drastically or they want to have a very small die.

If you thought doubling the complexity of the SPs (moving from 32 to 64-bit does that) was going to come cheap architecturally, you were being a moron. 4000 cores at 2GHz is still equal to 8000 at 1GHz from a theoretical standpoint. Might get a bit toasty, but it's possible.

[paps511]

Just got my 980ti. Oh well.

 

I will GLADLY take that off your hands!

[Bhav]

Whatever am I going to do with my 1525 / 8262 clocked SLI 970s when 1070 comes out and I preorder a pair? I have such golden Vram on them that I want to keep them.

[bigneo]

This is probably true. However AMD does also have its fair share of issues.

I also do hate their closed garden approach. AMD GPUs do tend to have better raw performance, Nvidia just has really good optimization.

I don't like Nvidia's "Let's get to the market first and make our method not open source so that our GPUs will have a big advantage over AMDs." All of AMDs methods are open source and run equally well on Nvidia and AMD GPUs. As well, they don't charge giant price premiums on some of their methods(just look at freesync vs gsync).

Personally right now I think that Nvidia and Intel are both kind of playing cut throat right now, and AMD is the one trying to get everyone to be fair.

You last point caught my attention. Fair? In business? In this kind of business, I think anyway, the one that is more of an asshole will win. The one that is ahead will get more ahead easier, money makes money yeah? Tho, I really hope that AMD will kick ass this time so we can get cheaper cards that are capable to put smile on our faces.

[DocSwag]

You last point caught my attention. Fair? In business? In this kind of business, I think anyway, the one that is more of an asshole will win. The one that is ahead will get more ahead easier, money makes money yeah? Tho, I really hope that AMD will kick ass this time so we can get cheaper cards that are capable to put smile on our faces.

Yeah I am aware that in business usually you do need to be a little dirty to win, but that doesn't mean I can't favor AMD. As well it would do good for Intel and Nvidia if AMD was competitive(it would probably be better for Intel, just look at how slowly their CPUs have progressed). Every 2 years nvidia and amd have probably gotten around 30-50% more performance out of their Gpus, but on the CPU side it has only been 10-20%. This is why the Fury X sometimes is actually being bottleneck end by a 4960x at 4.2 ghz (anandtech). His part I actually find ironic since the reason AMD high end Gpus are being bottlenecked is because Intel CPUs aren't progressing fast enough and the reason for that is AMD isn't competitive enough.

Basically AMD is bottlenecking their own GPUs lol.

[-BirdiE-]

This article needs to have "Unconfirmed" then, as that source is terrible and if you read it, the information is wrong.

 

I understand that it's not confirmed, but how can you say the info is wrong? Do you have a confirmed source saying otherwise?

[patrickjp93]

Yeah I am aware that in business usually you do need to be a little dirty to win, but that doesn't mean I can't favor AMD. As well it would do good for Intel and Nvidia if AMD was competitive(it would probably be better for Intel, just look at how slowly their CPUs have progressed). Every 2 years nvidia and amd have probably gotten around 30-50% more performance out of their Gpus, but on the CPU side it has only been 10-20%. This is why the Fury X sometimes is actually being bottleneck end by a 4960x at 4.2 ghz (anandtech). His part I actually find ironic since the reason AMD high end Gpus are being bottlenecked is because Intel CPUs aren't progressing fast enough and the reason for that is AMD isn't competitive enough.

Basically AMD is bottlenecking their own GPUs lol.

Intel's CPUs are as much as 50% faster now than they were in the days of Sandy Bridge. Newer software just doesn't keep up with newer instruction extensions and good programming practices to use SIMD effectively for non-obvious workloads.

[Metro]

I understand that it's not confirmed, but how can you say the info is wrong? Do you have a confirmed source saying otherwise?

 

Well the fact that this source is not reliable, it comes from an article made 3 months ago and NVIDIA did not have a working sample at CES, it seems quite obvious that high end Pascal is not coming out this year, I expect mid end Pascal to come out this year

[DocSwag]

Well the fact that this source is not reliable, it comes from an article made 3 months ago and NVIDIA did not have a working sample at CES, it seems quite obvious that high end Pascal is not coming out this year, I expect mid end Pascal to come out this year

How do you know they didn't have a working sample?

Maybe they were just being secretive, trying to catch AMD and the market off guard.

[don_svetlio]

How do you know they didn't have a working sample?

Maybe they were just being secretive, trying to catch AMD and the market off guard.

Cause Nvidia never miss a chance to brag and build hype

[SolarNova]

Chilled water? Just do chilled Ethanol. You can go way colder and have a superior heat transfer gradient.

Im not aiming for sub zero, just sub ambient. Not going into full details, the idea is simply to have a 4-10c water temp (will depend on the Dew point, but it will be able to hit 4c at the lowest), and with the plan I have ready to go, I wont have to worry about condensation, it will also allow for the system to look just like any other high end water cooled rig. (usually sub ambient / sub zero requires a ugly chill box to prevent condensation when used for long durations)

[-BirdiE-]

Well the fact that this source is not reliable, it comes from an article made 3 months ago and NVIDIA did not have a working sample at CES, it seems quite obvious that high end Pascal is not coming out this year, I expect mid end Pascal to come out this year

 

Again.. Nothing confirmed in the statement you've made that would prove this rumor to be false.. You can believe what you'd like, and you may be right... But you can't dismiss another rumor because it doesn't fit with the rumor you believe is more likely.

 

What we know:

NVIDIA did not bring and display a working sample of Pascal to CES

Samsung had begin mass producing 4GB HBM2 modules

 

That hardly makes it "quite obvious" that high end Pascal is not coming out this year. You've also made an unsubstantiated assumption that mid pascal will be coming out before high end pascal. There aren't even rumors to suggest that.

[Trixanity]

Instead of disregarding what is known that disproves this article, maybe try to find one shred of evidence that the article isn't made up.

It's quite clear that it's a fabrication. Half of it is question marks and the rest is speculation and shaking of the magic 8-ball. There is literally no source to indicate why a chip would be released in April. No leaked slides or specifications, no shipping manifests, no inside source. Nothing.

What we do know is, as you mention, that Samsung is producing HBM2 but that doesn't mean Nvidia is releasing a Pascal card now. They're not solely producing for Nvidia in any case.

The second is that Nvidia presented a Pascal chip at CES, which was in fact false as they used a Maxwell chip in its place despite stating otherwise. So that points to the fact that the chip is not ready.

Now I don't believe it has been stated that PX2 will use GP100, so that makes things more uncertain. But if the big chip launches first, then I would expect in very small quantities because of the yields.

But it's still very improbable that we'll see anything so soon. And the article still doesn't have any credible evidence.

[DocSwag]

Instead of disregarding what is known that disproves this article, maybe try to find one shred of evidence that the article isn't made up.

It's quite clear that it's a fabrication. Half of it is question marks and the rest is speculation and shaking of the magic 8-ball. There is literally no source to indicate why a chip would be released in April. No leaked slides or specifications, no shipping manifests, no inside source. Nothing.

What we do know is, as you mention, that Samsung is producing HBM2 but that doesn't mean Nvidia is releasing a Pascal card now. They're not solely producing for Nvidia in any case.

The second is that Nvidia presented a Pascal chip at CES, which was in fact false as they used a Maxwell chip in its place despite stating otherwise. So that points to the fact that the chip is not ready.

Now I don't believe it has been stated that PX2 will use GP100, so that makes things more uncertain. But if the big chip launches first, then I would expect in very small quantities because of the yields.

But it's still very improbable that we'll see anything so soon. And the article still doesn't have any credible evidence.

Oh, so all rumors are confirmed?

Many early rumors for upcoming products are extremely questionable, but that doesn't mean they are all wrong.

None of us know whether this is true or not. The only people that know are people at Nvidia. This could be right or wrong. Only time will tell.

[-BirdiE-]

Instead of disregarding what is known that disproves this article, maybe try to find one shred of evidence that the article isn't made up.

It's quite clear that it's a fabrication. Half of it is question marks and the rest is speculation and shaking of the magic 8-ball. There is literally no source to indicate why a chip would be released in April. No leaked slides or specifications, no shipping manifests, no inside source. Nothing.

What we do know is, as you mention, that Samsung is producing HBM2 but that doesn't mean Nvidia is releasing a Pascal card now. They're not solely producing for Nvidia in any case.

The second is that Nvidia presented a Pascal chip at CES, which was in fact false as they used a Maxwell chip in its place despite stating otherwise. So that points to the fact that the chip is not ready.

Now I don't believe it has been stated that PX2 will use GP100, so that makes things more uncertain. But if the big chip launches first, then I would expect in very small quantities because of the yields.

But it's still very improbable that we'll see anything so soon. And the article still doesn't have any credible evidence.

 

DocSwag is completely right. No one is claiming this to be fact. I'm arguing against people saying "this should be taken down because it can't be confirmed". Pretty much no information about Pascal is confirmed

 

As I said before, the LTT forums would be pretty bare if you took down every unconfirmed rumor.

 

And for those saying "this should be taken down because it's blatantly wrong"...  Their claims are no more substantiated than this rumor.

 

Until there are confirmed reports about Pascal I, as well as many others, would like to know what is rumored to be happening with it.

[Trixanity]

I'm not saying it should be taken down but it should be dismissed for what it is: false.

Rumors are fine if they have evidence to support it, even if vague. This does not. In fact, so many other things contradict this article.

Half of the rumor articles involve some source to base the article on, the rest are fabrications aka click bait. This sits firmly in the latter category.

For example let's take the Samsung Galaxy rumors. Some of them cite benchmarks appearing in databases, credible sources with good track records or other evidence. Then there's articles saying "it will have a 6" 8K display, 20 core processor, 16 gigabytes of ram, 2 tb internal storage, finger and iris scanner plus it can shoot laser beams out of the bottom doubling as a self-defense weapon". Point being they make shit up to get clicks.

This article looks very much like an example of this.

Edit: the problem isn't necessarily the click bait articles, it's also the people who believe them. That's how these things spread and how people suddenly start believing in myths.