Jump to content

Future nvidia gpus

Atlant
By Atlant
in Graphics Cards
 Share
Go to solution Solved by MageTank,

OP is referencing this: http://www.anandtech.com/show/7900/nvidia-updates-gpu-roadmap-unveils-pascal-architecture-for-2016

 

PascalBoard_575px.jpg

 

 

 

But the rabbit hole goes deeper. To pull off the kind of transfer rates NVIDIA wants to accomplish, the traditional PCI/PCIe style edge connector is no good; if nothing else the lengths that can be supported by such a fast bus are too short. So NVLink will be ditching the slot in favor of what NVIDIA is labeling a mezzanine connector, the type of connector typically used to sandwich multiple PCBs together (think GTX 295). We haven’t seen the connector yet, but it goes without saying that this requires a major change in motherboard designs for the boards that will support NVLink. The upside of this however is that with this change and the use of a true point-to-point bus, what NVIDIA is proposing is for all practical purposes a socketed GPU, just with the memory and power delivery circuitry on the GPU instead of on the motherboard.
NVIDIA’s Pascal test vehicle is one such example of what a card would look like. We cannot see the connector itself, but the basic idea is that it will lay down on a motherboard parallel to the board (instead of perpendicular like PCIe slots), with each Pascal card connected to the board through the NVLink mezzanine connector. Besides reducing trace lengths, this has the added benefit of allowing such GPUs to be cooled with CPU-style cooling methods (we’re talking about servers here, not desktops) in a space efficient manner. How many NVLink mezzanine connectors available would of course depend on how many the motherboard design calls for, which in turn will depend on how much space is available.

However, as i discussed with a friend of mine on skype, it is unlikely this will be the case once they actually roll around. I still think traditional PCIe slots will exist, and that NVLink will be adapted in another fashion.  After all, that article is over 2 years old now, and we still have not reached the full limitations of PCIE 3.0 yet. 

Posted

i heard that the new gpus (pascal or volta) will not be perpendicular to the mobo and that instead they will be parallel to it(for better bandwith), can any1 confirm that with a source?

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

I don't exactly get what you mean.

 

Are you meaning it's going to be similarly to how an M.2 SSD is generally mounted?

Sig under construction.

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

is that even possible?

Thats that. If you need to get in touch chances are you can find someone that knows me that can get in touch.

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

Where'd you hear that?

|  Athlon 750K   |   EVGA GTX 750 Ti FTW  |  Acer Aspire E15 - i5 6200u, 940M  |

|  Sennheiser HD 518   |   FiiO E10K DAC and Amp  |

Moto G (1st Gen)

 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

wut

"The of and to a in is I that it for you was with on as have but be they"

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

What would be the point? We can't utilise PCIe 3.0 16x right now, there's no need for more bandwidth, unless the cards are much, much, much faster than current ones. 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

I don't really get it either? Especially with pcie 4.0 coming. That will have retarded bandwidth. Maybe if they can't figure out what comes after that nvidia will have to step up on the gpu side and do something interesting but I don't really see that happening. 

Ryzen 3700x -Evga RTX 2080 Super- Msi x570 Gaming Edge - G.Skill Ripjaws 3600Mhz RAM - EVGA SuperNova G3 750W -500gb 970 Evo - 250Gb Samsung 850 Evo - 250Gb Samsung 840 Evo  - 4Tb WD Blue- NZXT h500 - ROG Swift PG348Q

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

I don't see how that would improve performance nor do I see how it's physically possible.

"It pays to keep an open mind, but not so open your brain falls out." - Carl Sagan.

"I can explain it to you, but I can't understand it for you" - Edward I. Koch

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

i dont think that will happen because of pascal will be PCI E 3.0 or at least the first iteration will be. nvidia would not change the pci e connector because then motherboard manufacturers would have to make new motherboads for the sake of one gpu

My Personal PC 'Apex' https://uk.pcpartpicker.com/user/LiamBetts123/saved/3rTNnQ

Intel Core i9 9900k, ASUS Z390-A, RTX 2080TI, Meshify C, HX 850i, 32GB Gskill Trident Z RGB @ 3200MHZ, 500GB NVME, 500GB SSD & 2 x 4TB Baracudas 

 
 
 
 
 
 
 
 
 
 
 
 
 
1
Link to comment
Share on other sites
Link to post
Share on other sites
Posted

The entire concept behind being parallel to the mobo giving better bandwidth makes no sence 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

i dont think that will happen because of pascal will be PCI E 3.0 or at least the first iteration will be. nvidia would not change the pci e connector because then motherboard manufacturers would have to make new motherboads for the sake of one gpu

Actually I think it could even be PCIE 4.0 by the time it comes out, which should be plenty of bandwidth.

Sig under construction.

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

My mind is now processing faster than these parallel cards

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

Actually I think it could even be PCIE 4.0 by the time it comes out, which should be plenty of bandwidth.

Not even plenty, it would be way too much for a long time xD 

 

We're not even close to PCIe 3.0 16x. Cards that need more bandwidth than that will be insanely fast compared to current ones. 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

Actually I think it could even be PCIE 4.0 by the time it comes out, which should be plenty of bandwidth.

yes exactly and forcing people to change to somthing there not used to i never a good idea and i dont think nvidia would risk there 80 percent market share for the sake of a vertical card.

My Personal PC 'Apex' https://uk.pcpartpicker.com/user/LiamBetts123/saved/3rTNnQ

Intel Core i9 9900k, ASUS Z390-A, RTX 2080TI, Meshify C, HX 850i, 32GB Gskill Trident Z RGB @ 3200MHZ, 500GB NVME, 500GB SSD & 2 x 4TB Baracudas 

 
 
 
 
 
 
 
 
 
 
 
 
 
1
Link to comment
Share on other sites
Link to post
Share on other sites
Posted

Are you talking about NVlink? NVlink should increase the rate at which the CPU and GPU can interact and also increases bandwidth by a great deal so that's probably what you are thinking about. 

[ Rig: CPU: 4930K, GPU: EVGA 780TI SC x2, RAM: 32GB Corsair Vengeance 1600MHz, Mobo: ASUS P9 X79 LE, Storage: 120GB Samsung EVO + 2TB Seagate Barracuda, PSU: Corsair RM1000 ]

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

I don't see how that would improve performance nor do I see how it's physically possible.

That's what I was thinking

Because he had a hard drive.

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

Not even plenty, it would be way too much for a long time xD 

 

We're not even close to PCIe 16x. Cards that need more bandwidth than that will be insanely fast compared to current ones. 

PCIe *3.0* x16. The generation is important - we have saturated PCIe 1.0 x16 for example.

"It pays to keep an open mind, but not so open your brain falls out." - Carl Sagan.

"I can explain it to you, but I can't understand it for you" - Edward I. Koch

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

PCIe *3.0* x16. The generation is important - we have saturated PCIe 1.0 x16 for example.

I swear I typed 3.0. Did on my first post. 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted
  • Author

yes, from anandtech ( they wrote about what they heard on the last conference)

I don't exactly get what you mean.

 

Are you meaning it's going to be similarly to how an M.2 SSD is generally mounted?

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

i heard that the new gpus (pascal or volta) will not be perpendicular to the mobo and that instead they will be parallel to it(for better bandwith), can any1 confirm that with a source?

 

I believe you're referring to HBM, like how AMD has theirs.  The VRAM is stacked on the GPU core or directly next to it to improve bandwidth and save on silicone costs for the PCB, as well as generally less complex electrical traces.

QUOTE ME IN A REPLY SO I CAN SEE THE NOTIFICATION!

When there is no danger of failure there is no pleasure in success.

Link to comment
Share on other sites
Link to post
Share on other sites
Posted
  • Author

they wrote something about being able to mount the cpu coolers the same way on the new gpus, im searching the post but i cant find it

I believe you're referring to HBM, like how AMD has theirs.  The VRAM is stacked on the GPU core or directly next to it to improve bandwidth and save on silicone costs for the PCB, as well as generally less complex electrical traces.

Link to comment
Share on other sites
Link to post
Share on other sites
Posted
  • Author

how does nvlink physically look like? Any1 knows?

Are you talking about NVlink? NVlink should increase the rate at which the CPU and GPU can interact and also increases bandwidth by a great deal so that's probably what you are thinking about. 

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

how does nvlink physically look like? Any1 knows?

Blog post directly from NVIDIA: http://blogs.nvidia.com/blog/2015/04/23/data-in-the-fast-lane-how-nvlink-unleashes-application-performance/

They have a Youtube video embedded in there showing how it works. I did not find any physical slot.

"It pays to keep an open mind, but not so open your brain falls out." - Carl Sagan.

"I can explain it to you, but I can't understand it for you" - Edward I. Koch

Link to comment
Share on other sites
Link to post
Share on other sites
Posted

So it seems that pascal is going to be fast as hell?

The clockspeed wil be OVER 9000!

Link to comment
Share on other sites
Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share
Go to topic listing Graphics Cards

×