NVIDIA Next-Generation GPU Rumored to Be Called “Ampere”

Megalith

Megalith

24-bit/48kHz
Staff member
Joined
Aug 20, 2006
Messages
13,000
According to Heise.de, NVIDIA is preparing a next-generation GPU known as “Ampere,” which they are planning to unveil at GTC 2018. There are currently no details available, but rumors are that NVIDIA will be jumping straight from Pascal to Ampere, at least on the GeForce front. The site alleges that the Ampere family of graphics cards will succeed the GeForce 10 series cards, which are based on the Pascal GPU architecture.

Now considering this is just a rumor and no other details are mentioned, it’s advised to take this information with a grain of salt. We know for a fact that the Volta GPU is the official replacement for the Pascal GPU, according to the roadmaps. There is no mention of an Ampere GPU until now, and the roadmap past Volta is yet to be updated. Volta has already made an appearance at GTC 2017 earlier this year and is available on the market since September 2017.
 
Two things. One, the fact that they are moving to Ampere so soon tells me they're concerned about the Raja-Intel partnership and are taking it seriously, so much so that they want to put as much daylight between them and their competition is possible. They aren't holding back anymore.

Second thing, WATTA is hilarious. And if Nvidia makes a flop please let it be WATTA.

The memes would write themselves "WATTA the hell happened?"
 
Nolan7689 said:
Nuh uh, Fermi was totally named after a power plant and Tesla is named for a Tesla Coil and Pascal for the pressure Nivida puts on AMD...right right??

I don’t know enough about Maxwell or Kepler to continue.
Click to expand...

You do know there was a Nikolai Tesla who invented that coil right?
 
Zion Halcyon said:
You do know there was a Nikolai Tesla who invented that coil right?
Click to expand...
giphy.gif
 
Zion Halcyon said:
Two things. One, the fact that they are moving to Ampere so soon tells me they're concerned about the Raja-Intel partnership and are taking it seriously, so much so that they want to put as much daylight between them and their competition is possible. They aren't holding back anymore.
Click to expand...

Soon?

It will be OVER two years since the introduction of the 1000 series GPUs.
 
Zion Halcyon said:
Two things. One, the fact that they are moving to Ampere so soon tells me they're concerned about the Raja-Intel partnership and are taking it seriously, so much so that they want to put as much daylight between them and their competition is possible. They aren't holding back anymore.
Click to expand...

WTH, this have to be one of the most uneducated conspiracy theory of the whole year.... all far from a *fact*.. you know the difference between unveil and release, right?. it mean completely different things.
 
They could name it Lucky Rainbow Dragon GPU for all the name matters - it'll still send AMD/RTG shuffling down the hall looking for toilet paper.

Can't wait.
 
My bet is ampere is just an internal code name for either consumer volta or mobile volta.
 
lets assume the story is partially correct

think tick/tock for Nvidia -for Nvidia to change releases with consumer lagging 1 generation behind the HPC

so when consumer gets Volta -...that ampere goes to HPC and when consumer gets Ampere HPC gets Watt


I do believe there was a story awhile back that said Nvidia was planning to do that


also remember that AMD wants to release navi next year - so think changes every -1-2 years now - avg 1.5yrs
 
Last edited:
Zion Halcyon said:
One, the fact that they are moving to Ampere so soon tells me they're concerned about the Raja-Intel partnership and are taking it seriously,
Click to expand...

The heck are you on? Raja couldn't do nothing about nVidia when he was at AMD, and now you think failed Raja starting completely from scratch with Intel makes them catch up within the year to the juggernaut that is nVidia who hasn't had any serious competition for half a decade+?. LMAO.

Sorry, I like optimism but this is just ludicrous.
 
It would only make sense to separate consumer GPUs and compute cards - their requirements and use cases are different enough that you can't cost-effectively make one chip that excels at them both. We've already seen this with Pascal and Volta - the highest-end chips are ultra-expensive specialty chips that don't have the same layout as their gaming or workstation counterparts. I'm sure that producing the same architecture for all use cases makes less and less sense as deep learning tech progresses.

And it's not like they don't have the cash now to fund two different design departments.
 
TimberVD said:
The heck are you on? Raja couldn't do nothing about nVidia when he was at AMD, and now you think failed Raja starting completely from scratch with Intel makes them catch up within the year to the juggernaut that is nVidia who hasn't had any serious competition for half a decade+?. LMAO.

Sorry, I like optimism but this is just ludicrous.
Click to expand...
It just like these people all doom and gloom for Intel when Ryzen came out and how Intel is in a panic. Yet Intel still own 80+ of the market almost a year later..
 
How shit a name would "Watt" be for a GPU? I mean, come on.

"Which GPU should I buy?"
"Watt."
"I said, which GPU should I buy?"
"Watt!"
"I SAID, WHICH GPU SHOULD I... Oh, forget it!"

And they better hope it's power-efficient or the "MegaWatt" or "Killer Watt" jokes will be everywhere.
 
Zion Halcyon said:
Two things. One, the fact that they are moving to Ampere so soon tells me they're concerned about the Raja-Intel partnership and are taking it seriously.
Click to expand...

Whew, had a good laugh on that one. If Intel’s goal is the high end dGPU market, it will be 1-2 years before they get there.

Nvidia is probably ready to make more money...
 
Remember when they announced the "GTX 1080", the huge outcry on this forum about how stupid it was since all the lesser peasants in the world were going to be confused and think it was only for 1080p? Clearly these naming conventions are meant only for us intellectuals. Jen-Hsun is clearly giving us a little winkwinknudgenudge... well at least me, I mean it's pretty obvious if you've been paying attention.

PS I'll see you at the country club next weekend Jen!
 
Wonder how it will handle DX12/Vulkan vs Pascal, that's the obvious weakness that needs to be addressed.
 
After V100 jumped 30% in die size, (muh tensa corez) perhaps the profitability wasn't as good, so maybe they want to keep tiny die margins and hence Ampere. Bit like what AMD tried with Polaris and Nvidia has been doing for years since 290x.
 
Not terribly surprising, Tensor cores aren't marketed towards gaming anyways. And when you strip out the Tensor cores, you got marginally faster Pascal CUDA cores.

So the question is, how fast is Ampere? If they're claiming next generation, I assume it's an architectural change rather than refinement...

Also, can it play Crysis?
 
They should put some history on the nVidia boxes of why it was named so and so. Love the name Volta. I can't say the same about Ampere though.
 
  • Like
Reactions: N4CR
like this
Looks like Razor1 was wrong about Volta being the destroyer of AMD and all the graphics power it had, looks like Nvidia disagreed. I wonder if Volta was only ever planned for the professional side and never for the gaming side. Be interesting to see what Ampere brings to the table.
 
Gideon said:
Looks like Razor1 was wrong about Volta being the destroyer of AMD and all the graphics power it had, looks like Nvidia disagreed. I wonder if Volta was only ever planned for the professional side and never for the gaming side. Be interesting to see what Ampere brings to the table.
Click to expand...
Hynix spilled the beans of a 384bit gddr6 gpu due in q1 ages ago. Razor1 is right far more likely than you are. (insert jumping to conclusions mat here)
 
You must log in or register to reply here.
Back
Top