Vega 7nm early benches

Personally, I think it was more the price factor especially with the demand in mining that made the cards way more expensive, and when one's gaming setup has a CPU that cost 500$ but a pair of GPU's that will cost them between 2000 and 3000$ for both, they go **** that, I'll go with a single card which is more than enough.


GPU makers don't mind since it means that single GPU card setup will run out of steam that much sooner, and make that single high end GPU user change for an even higher performing card even faster that may cost even more, so it's all good.
 
shadow001 said:
Personally, I think it was more the price factor especially with the demand in mining that made the cards way more expensive, and when one's gaming setup has a CPU that cost 500$ but a pair of GPU's that will cost them between 2000 and 3000$ for both, they go **** that, I'll go with a single card which is more than enough.


GPU makers don't mind since it means that single GPU card setup will run out of steam that much sooner, and make that single high end GPU user change for an even higher performing card even faster that may cost even more, so it's all good.
Click to expand...

first pard yes ok cost was part of why I did not buy two 1080 ti strix's and waterblocks and that was at msrp of 779
after the miners went NV cards also it was no way I would pay 1200+ for a second card

second part no most people I know that do or did sli/cfx would upgrade as soon as new carts hit every time
I always did

and there haven't been any new cards to upgrade to for 16 months and counting :mad:

other than the :nuts: Titan V
 
bill dennison said:
first pard yes ok cost was part of why I did not buy two 1080 ti strix's and waterblocks and that was at msrp of 779
after the miners went NV cards also it was no way I would pay 1200+ for a second card

second part no most people I know that do or did sli/cfx would upgrade as soon as new carts hit every time
I always did

and there haven't been any new cards to upgrade to for 16 months and counting :mad:

other than the :nuts: Titan V
Click to expand...



The second part is subjective as for when one decides to upgrade......Many years ago when I was a lot more impressionable and was always trying to go for the highest benchmark results possible, and did upgrades where the gain was sometimes as small as 20~30%, I also got that bitter taste of having blown money on hardware that did indeed make for those records to happen, but when compared to what I was running previously the gain was nothing special.....Another 10 frames here or 15 frames there type deal......Got sick of that eventually.


These days , it's double or nothing so basically going from a 30 Fps situation to a 60 Fps at the minimum, so there's no need to even run benchmark programs, Fps counters, stop watches or anything else that exists out there, the difference is immediately noticeable as soon as you fire up a game and put it at the most demanding settings possible where your current setup couldn't do more than 30 Fps to begin with.



It's why this rumor with there may be a gaming version of Vega at 7 nm coming out early next year, and it seems it is 75% faster because it clocked at 1 Ghz matches the 14 nm version at 1.75 Ghz is a good increase, but it is still a little short of twice as fast in my book......It takes quite a lot to impress me these days, but I guess that's because i've seen so much hardware in nearly 3 decades.



I also hope Nvidia doesn't just release a tweaked Pascal at 12nm, clock it higher and pack it with GDDR6 making the usual 30~40% faster and call it a day while charging a fortune for it.
 
Nunz said:
Bob vodka has explained many times that the whole “DX12 will be easy mGPU” is false.
Click to expand...

Never said it would be easy. Just said it would be seamless. To clarify though, I mean that in the way of.. if the game supports it, it should just work for the user. No profiles etc..

Argoon1981 said:
To support MGPU in Vulkan and Dx12 the developers NEED to implement that themselfs, is not a switch that you toggle and you are in MGPU heaven.

Afaik Nvidia still doesn't support GPU PhysX, if you use a AMD GPU for main render GPU. I would be very surprised if that has ever changed, that decision was not because of true technical difficulties, it was just Nvidia not supporting being relegated to PhysX acceleration and that has never changed.
Click to expand...

1st Part- Yes, I understand that. See the above response to Nunz. :-)

2nd Part- Figured it might be. Just hadn't seen anything on it since the original articles swept the web. lol
 
That said, if the 7nm Vega is released and has that 75% performance increase from one chip, it's basically the performance of both my 14nm Vega's in crossfire, so if i add a second of them still in crossfire, we're talking more than enough speed to keep a 4K monitor that refreshes at 120Hz very busy as the cards would be kicking out 120+ Fps at that 4k resolution which is insane, if it existed of course so it isn't slow that much is sure.




Ball is on AMD's court and I think they should if Navi is only a 2020 product....For once AMD would have the lead over Nvidia in that they'd release a high end gaming card using the latest 7nm fabrication process, while Nvidia may use the 12nm process already leveraged for the titan V.
 
In any case, we're on the last rounds of current lithography technology, so all this discussion about who's faster or uses less power or comes out sooner and cheaper is about to become a moot point within the next 5 years..... The next big step that cuts transistor size in half over the 7 mn process requires manufacturing equipment that isn't even on the drawing board right now, never mind the physics challenges of current leakage ( quantum tunnelling), as certain parts of transistors are nothing but a few atoms wide.



Some will say " there's new materials on the horizon ", we'll I've been hearing that for years now and nothing has materialized in high volume production yet so they're still laboratory experiments for the most part.
 
Exclusive: AMD Senior Vice President Jim Anderson Resigns – Moving On To Become CEO [Updated]
Click to expand...
Edit: AMD has confirmed our story with the following update:

After the close of business today there were two AMD announcements of note that I wanted to bring to your attention and provide further detail around:
1.We are manufacturing all our 7nm products at TSMC, and
2.We are promoting Saeid Moshkelani, a 30+ year AMD semiconductor veteran, to SVP & GM Client Compute given the departure of Jim Anderson to Lattice Semiconductor.

7nm Update

For the past several years we consistently executed our multi-generational, leadership product and architectural roadmap across our CPU and GPU portfolio. Our upcoming 7nm product portfolio represents the next major milestone for AMD computing and graphics leadership, including our second generation “Zen 2” CPU core and our new “Navi” GPU architecture. We have already taped out multiple 7nm products at TSMC, including our first 7nm GPU that is planned to launch later this year, and our first 7nm server CPU that we have sampled to strategic customers and plan to launch in 2019.
Click to expand...

https://wccftech.com/exclusive-amd-...jim-anderson-resigns-moving-on-to-become-ceo/
 
?


https://hexus.net/tech/news/industry/121508-amd-accelerates-7nm-process-adoption/




The AMD Chief Technical Officer (CTO) claimed that, though 7nm has been a big challenge it decided to go all-in, and thanks to its partners is now able to bring ahead the launch date for the technology.
Click to expand...


"We knew 7nm would be a big challenge, so we made the bet, we shifted our resources onto the new node," Papermaster said in an exclusive interview with CRN. "We didn't just dip our toe in the water. We went all in."
AMD had originally planned to release the first 7nm Vega GPU in 2019 but, thanks to "immense focus" has been able to move the date forward to later this year. Thus we will see the next-generation Radeon Vega Instinct GPU, demonstrated at Computex with 32GB HBM2, launched before the year is out.
Click to expand...
 
Megaman said:
?


https://hexus.net/tech/news/industry/121508-amd-accelerates-7nm-process-adoption/
Click to expand...

vega 20


For the past several years we consistently executed our multi-generational, leadership product and architectural roadmap across our CPU and GPU portfolio. Our upcoming 7nm product portfolio represents the next major milestone for AMD computing and graphics leadership, including our second generation “Zen 2” CPU core and our new “Navi” GPU architecture. We have already taped out multiple 7nm products at TSMC, including our first 7nm GPU that is planned to launch later this year, and our first 7nm server CPU that we have sampled to strategic customers and plan to launch in 2019.
Click to expand...

https://wccftech.com/exclusive-amd-...jim-anderson-resigns-moving-on-to-become-ceo/

I care about Zen 2 most at this point and sounds like it maybe taped out Navi also so puts them in Q1 if all goes well
 
Pity GF could not pull of a decent 7nm, would have given AMD good second source of 7nm chips. I think AMD would have been GF's biggest customer for 7nm and if its not to scratch for AMD... its probably pragmatic to quit digging and stick with 12/14 until they get there **** together
 
Says the same here:


https://www.anandtech.com/show/13277/globalfoundries-stops-all-7nm-development




And so there are only 2 left and even then it's only the 7nm process with no idea if even smaller is possible after this point ( Samsung and TSMC )…..Global foundry owners are tired of sinking billions and not making a return on the investment for the last 10 years, and any research in 5nm or 3nm is completely halted too....



This could mean that semiconductor development that requires a new and smaller process will hit a brick wall in the next 5 years....No more competition between Nvidia, AMD, intel as they can't make them smaller at the fabs themselves....:eek:
 
shadow001 said:
Says the same here:


https://www.anandtech.com/show/13277/globalfoundries-stops-all-7nm-development




And so there are only 2 left and even then it's only the 7nm process with no idea if even smaller is possible after this point ( Samsung and TSMC )…..Global foundry owners are tired of sinking billions and not making a return on the investment for the last 10 years, and any research in 5nm or 3nm is completely halted too....



This could mean that semiconductor development that requires a new and smaller process will hit a brick wall in the next 5 years....No more competition between Nvidia, AMD, intel as they can't make them smaller at the fabs themselves....:eek:
Click to expand...

Multi-chip-module tech may help with that. Instead trying to make giant low yield 800mm monolithic cores that run at 2GHz, you build tiny high yield modules and stitch them together with high speed interconnects.
 
Global Foundries has been sucking for years. Their 14nm had problems too and they had to license from Samsung. GF just isn't a first tier fab.

A lot of electronics use older nodes because they're mature and reliable and cost less, it's only stuff like GPUs/processors that really need the best process. So, maybe they can focus on producing that sort of thing. But it seems like they basically cannot hang with TSMC or Samsung anymore.
 
GloFo sucks. I’ve said it ever since they moved to them. They’ve been nothing but terrible, haven’t kept up with the competition, and they hurt AMD in the long run.
 
Problem is if TSMC becomes sole source they will up prices big time. We might find the 2000 gen Geforce prices as just a start...

Hope Samsung offers some competition there but it doesnt look likely...
 
You must log in or register to reply here.
Back
Top