Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Nvidia Volta.
- Thread starter AidenShaw
- Start date
- Sort by reaction score
- Status
http://fudzilla.com/news/graphics/43962-new-geforce-will-be-incremental
It basically means that Consumer GPUs will resemble GP100 architecture, not GV100. Which I have told you months before.
It basically means that Consumer GPUs will resemble GP100 architecture, not GV100. Which I have told you months before.
is this website the breitbart of hardware rumors?
Former Apple man finds life tough outside the reality distortion field
Former Apple man finds life tough outside the reality distortion field
Eh gad all new releases are incremental. I can't remember anything truly revolutionary every since compute units were first announced.
It depends on the perspective.
If you come from GP100 - it will be incremental. If you come from GP102 - it will be huge jump.
Yes but we have to use the first example you make - compare mid range to mid range and high end to high end with each new gen.
There have been some products that broke the pattern and the mold (Radeon Nano being small, powerful and efficient) but in general the duopoly of Nvidia and AMD and their marketing dept make sure updates are incremental with the same propaganda and campaign language every year.
They could ship slightly better products yearly. Undervolters and overclockers have lots of stories about how many cards are shipped with less than ideal default settings. We have seen that some of the desktop chips can run full speed in much small form factors. But they don't ship the products like this because massive coolers and big shrouds are profit making.
But I'm just regurgitating the same complain people have made about CPU and GPU for two decades.
Links?
Oh here's one. https://forums.macrumors.com/threads/2016-nmp.1952250/page-40#post-22681787
March 2016 (not a few years ago). And I don't even see any replies to your post, let alone laughter.
Also, note that the Nvidia paper cites references back to at least 2005. MCM isn't something that ATI just invented.
I remember POWER, but I see that IBM was already using the technology in the 70's.
I wrote: It was before AMD touted Scalability. That post you linked is after AMD touted scalability
.P.S. I was not saying that AMD has invented this, but that reality that GPUs are made from multiple, smaller dies is 100% done deal.
Instead of just saying that I'm wrong, how about posting a link that proves that I'm wrong? (And be sure to include the laugh track to support the rest of your claim.I wrote: It was before AMD touted Scalability. That post you linked is after AMD touted scalability
P.S. I was not saying that AMD has invented this, but that reality that GPUs are made from multiple, smaller dies is 100% done deal.
)
Last edited:
Nvidia’s gaming Volta cards will house the world’s biggest consumer GPUs
"Nvidia likes ‘em big. The first GV100 Volta GPU Nvidia showed off is a mammoth 815mm², but while the gaming Volta chips aren’t going to be quite that large, they will still be the biggest GPUs you’ll have ever seen on a gaming card.
..."
[doublepost=1499565271][/doublepost]Best graphics card... given the bare shelves
"What's the best graphics card on the market right now? Well, it’s the $1,200 Nvidia Titan Xp, obvs. Until we know exactly how those top-end AMD Vega cards perform in-game, the beefy GeForce GPU is the one with all the transistors, all the shaders, all the ROPs and all the FLOPs, which means it's the best. All sorted. We can all go on to live rich, full lives safe in the knowledge the big questions are all taken care of.
..."
"Nvidia likes ‘em big. The first GV100 Volta GPU Nvidia showed off is a mammoth 815mm², but while the gaming Volta chips aren’t going to be quite that large, they will still be the biggest GPUs you’ll have ever seen on a gaming card.
..."
[doublepost=1499565271][/doublepost]Best graphics card... given the bare shelves
"What's the best graphics card on the market right now? Well, it’s the $1,200 Nvidia Titan Xp, obvs. Until we know exactly how those top-end AMD Vega cards perform in-game, the beefy GeForce GPU is the one with all the transistors, all the shaders, all the ROPs and all the FLOPs, which means it's the best. All sorted. We can all go on to live rich, full lives safe in the knowledge the big questions are all taken care of.
..."
Last edited:
Santa Clara’s Nvidia named smartest company in the world
"Nvidia has been named the smartest company on the planet for 2017 by the MIT Technology Review.
Central to Nvidia’s success of late has been its development of processors for artificially intelligent systems, the Technology Review noted. As cloud computing and autonomous driving take off, the firm’s processors are in high demand.
..."
"Nvidia has been named the smartest company on the planet for 2017 by the MIT Technology Review.
Central to Nvidia’s success of late has been its development of processors for artificially intelligent systems, the Technology Review noted. As cloud computing and autonomous driving take off, the firm’s processors are in high demand.
..."
Last edited:
How Nvidia’s ‘brains’ are dominating the self-driving race
"SANTA CLARA — As Google, Uber and Tesla fight for control of the self-driving car market, another company better known for its gaming chips than its wheels is positioned to cash-in on the transportation revolution.
Nvidia has quietly become one of the top suppliers of the “brains” that control self-driving cars — the super computers that sit nestled inside the trunk or perhaps tucked behind the glove compartment and interpret visual cues like red lights and pedestrians, telling the vehicle how to respond. The Santa Clara-based chipmaker has racked up deals with big carmakers including Tesla, Audi, Mercedes, Toyota and Volvo and is poised to power a large portion of the burgeoning self-driving car industry, a market that is expected to hit $77 billion by 2035.
..."
Although the story references the Pascal-based Drive PX2, Volta-based automotive supercomputers nicknamed the PX3 are rumoured.
Our "sources" at WCCFTECH are speculating about the Volta-based "Xavier" SOC:
Also see https://en.wikipedia.org/wiki/Drive_PX-series#Xavier_AI_Car_Supercomputer
"SANTA CLARA — As Google, Uber and Tesla fight for control of the self-driving car market, another company better known for its gaming chips than its wheels is positioned to cash-in on the transportation revolution.
Nvidia has quietly become one of the top suppliers of the “brains” that control self-driving cars — the super computers that sit nestled inside the trunk or perhaps tucked behind the glove compartment and interpret visual cues like red lights and pedestrians, telling the vehicle how to respond. The Santa Clara-based chipmaker has racked up deals with big carmakers including Tesla, Audi, Mercedes, Toyota and Volvo and is poised to power a large portion of the burgeoning self-driving car industry, a market that is expected to hit $77 billion by 2035.
..."
Although the story references the Pascal-based Drive PX2, Volta-based automotive supercomputers nicknamed the PX3 are rumoured.
Our "sources" at WCCFTECH are speculating about the Volta-based "Xavier" SOC:
Also see https://en.wikipedia.org/wiki/Drive_PX-series#Xavier_AI_Car_Supercomputer
Last edited:
I have some interesting info from my sources.
GTX 2050 Ti, will not require 6 pin connector, and will give slightly higher performance than GTX 1060 6 GB offered. No word on GDDR memory bandwidth and bus width. Pricing will be around 149-169$, tho.
That is pretty great perspective. Intel i7 7700T plus GTX 2050 Ti can be very efficient 1440p computer.
GTX 2050 Ti, will not require 6 pin connector, and will give slightly higher performance than GTX 1060 6 GB offered. No word on GDDR memory bandwidth and bus width. Pricing will be around 149-169$, tho.
That is pretty great perspective. Intel i7 7700T plus GTX 2050 Ti can be very efficient 1440p computer.
Considering 1050Ti doesn't require power connector, and has slightly higher performance than 960, it is safe to assume that it will again be the case with 2050ti.
Stop pretending like you have like some kind of insider sources please, it only destroys your credibility.
Last edited:
It depends on end product TDP, but I will say - yes.
There were versions of GTX 1050 Ti with lower voltage, but the same core clocks, hence why it was able to fit in 55W thermal envelope, and not required active cooling or 6 pin connector.
Why would what you are writing disprove that I have sources?
... and it doesn't need 350 watts and water cooling!
https://blogs.nvidia.com/blog/2017/07/22/tesla-v100-cvpr-nvail/
Waiting for the GV104 cards....
I would love a single slot gpu. Any news about nvidia supporting macos better? Seems amd has better relationship/support with apple.
- Status