AMD Vega 10, Vega 11, Vega 12 and Vega 20 Rumors and Discussion

ToTTenTranz said:
So AMD tried to cock block the GTX 1660 Ti with a momentary discount, and wasn't even brave enough to claim it wasn't momentary?

Wow, that's lame.
Click to expand...

Põe lame nisso!
I had read about it this morning and just finished reading Anandtech review where Vega 56 is the best option at the same price. Then read about the limited supply. Really? Are they selling so well that they can afford to do this publicity stunts?
 
Picao84 said:
Põe lame nisso!
I had read about it this morning and just finished reading Anandtech review where Vega 56 is the best option at the same price. Then read about the limited supply. Really? Are they selling so well that they can afford to do this publicity stunts?
Click to expand...
I think it's more likely that they just don't make many Vegas.
 
Picao84 said:
Põe lame nisso!
I had read about it this morning and just finished reading Anandtech review where Vega 56 is the best option at the same price. Then read about the limited supply. Really? Are they selling so well that they can afford to do this publicity stunts?
Click to expand...
One specific model had discounted price for one batch and people are losing their minds
 
Kaotik said:
One specific model had discounted price for one batch and people are losing their minds
Click to expand...
AMD email said:
Hi,
Gamers looking for the best performance in today’s most popular AAA titles, from APEX Legends to The Division 2, need look no further than the AMD Radeon RX Vega 56 graphics card. Offering high-end performance and incredible features with an attractive price tag, the GPU is now available for $279 on Newegg, Amazon and Best Buy.
Click to expand...
Poor Volta marketing strikes back
 
From Lisa Su's Computex 2019 Keynote, at 1h01m40s:


Lisa Su said:
AMD has had a long history of architectures in the GPU space. We love GCN. GCN has been a great architecture for us. And frankly we're going to continue to see GCN as we go with new versions of Vega in many of the high compute workload applications.
Click to expand...


So AMD is apparently happy with Vega's success for compute workloads, to the point of releasing new GPUs in the future, based on the same architecture.

Could these be based on new chips or other GPUs based Vega 20?
AMD working on a 6nm Vega 20 comes to mind, since TSMC offers a direct upgrade path from 7nm to 6nm.
 
I think it's more likely to be new cards based on Vega 20. I mean, if Navi is more power-efficient, offers more performance/clock, and can clock higher, what's the point of developing new Vega-based GPUs?
 
For FP64 workload ? or compute only stuff ? the performance / clock slide for Navi is very vague, for all we know it can comes from the texturing parts... And maybe in compute world, some clients have already tuned their code for the Vega architecture ?
Or, it's BS, but AMD can't say "well, Vega is obsolete now, even in big compute workload, thx bye".
 
Rootax said:
For FP64 workload ? or compute only stuff ? the performance / clock slide for Navi is very vague, for all we know it can comes from the texturing parts... And maybe in compute world, some clients have already tuned their code for the Vega architecture ?
Or, it's BS, but AMD can't say "well, Vega is obsolete now, even in big compute workload, thx bye".
Click to expand...

Vega 20 won´t be obsolete until they make big NAVI chip with FP64/xGMI support which gaming version of NAVI not supporting, I assume...
 
Alexko said:
I mean, if Navi is more power-efficient, offers more performance/clock, and can clock higher, what's the point of developing new Vega-based GPUs?
Click to expand...

Being gaming-centric, Navi /RDNA might be offering lower compute throughput per watt and/or die area than Vega / GCN 5 (at similar process nodes, of course).

AMD seems to be hinting at forking their gaming and compute architectures.
 
ToTTenTranz said:
Being gaming-centric, Navi /RDNA might be offering lower compute throughput per watt and/or die area than Vega / GCN 5 (at similar process nodes, of course).

AMD seems to be hinting at forking their gaming and compute architectures.
Click to expand...

Possibly, but I see no hint of this at this point.
 
Alexko said:
Possibly, but I see no hint of this at this point.
Click to expand...

You mean aside from Lisa Su saying outright we'll still see future GCN solutions for compute workloads but Navi is a gaming-centric solution?
Sounds the fork is quite clear, at least for 2019.
 
ToTTenTranz said:
You mean aside from Lisa Su saying outright we'll still see future GCN solutions for compute workloads but Navi is a gaming-centric solution?
Sounds the fork is quite clear, at least for 2019.
Click to expand...

Unless she's just saying that there'll be new Vega 20 SKUs. Maybe Ian asked her.
 
Alexko said:
I'm suggesting that Lisa Su may be hinting that there'll be new Vega 20 SKUs
Click to expand...

I'm wondering what other Vega 20 SKUs for compute they could release, besides the already existing MI50 (60 CUs - 16GB HBM2) and MI60 (64CUs - 32GB HBM2).
A fully enabled prosumer card with 32GB HBM2 seems like a waste of money. Radeon VII users who undervolted their cards seem to generally go beyond 1900MHz, so maybe there's an opportunity for slightly higher clocked compute cards when using higher binned chips and better voltage configuration.

I suggested a "simple" transition of Vega 20 to 6nm, but that node is apparently taking so long (coming at the same time as 5nm during late 2020) that I wonder if it makes sense to use it on high-margin solutions.
 
Alexko said:
I think it's more likely to be new cards based on Vega 20. I mean, if Navi is more power-efficient, offers more performance/clock, and can clock higher, what's the point of developing new Vega-based GPUs?
Click to expand...

Because, Dr Lisa Su said that Vega is for compute space now and that Navi is specifically designed for Gaming. Vega & Navi do not share the same architecture anymore and that EVERY Navi transistors, goes towards gaming. There no longer is a compromise for gamers in sharing the uarch of scientific GPUs.

No unwasted die space on Navi, for gamers.
 
w0lfram said:
Because, Dr Lisa Su said that Vega is for compute space now and that Navi is specifically designed for Gaming. Vega & Navi do not share the same architecture anymore and that EVERY Navi transistors, goes towards gaming. There no longer is a compromise for gamers in sharing the uarch of scientific GPUs.

No unwasted die space on Navi, for gamers.
Click to expand...

Did she actually say that, or is that just what people inferred from what she said?
 
You must log in or register to reply here.
Back
Top