AMD Vega 10, Vega 11, Vega 12 and Vega 20 Rumors and Discussion
http://wccftech.com/amd-preparing-l...raphics-cards-gpu-passes-final-certification/
Uuuh, Vega 11 will have HBM2 as well according to the rumour. Was not expecting that at all.
The rumoured naming seems a bit weird tough, Vega 32 and Vega 28. Am I the only one thinking that Vega naming is a bit lackluster even if clear?
Uuuh, Vega 11 will have HBM2 as well according to the rumour. Was not expecting that at all.
The rumoured naming seems a bit weird tough, Vega 32 and Vega 28. Am I the only one thinking that Vega naming is a bit lackluster even if clear?
Rootax
Veteran
Could AMD use Samsung "cheap" HBM for that market ?
https://arstechnica.com/gadgets/2016/08/hbm3-details-price-bandwidth/
"Meanwhile, Samsung has been working on making HBM cheaper by removing the buffer die, and reducing the number of TSVs and interposers. While these changes will have an impact on the overall bandwidth, Samsung has increased the individual pin speed from 2Gbps to 3Gbps, offsetting the reductions somewhat. HBM2 offers around 256GB/s bandwidth, while low cost HBM will feature approximately 200GB/s of bandwidth. Pricing is expected to be far less than that of HBM2, with Samsung targeting mass market products."
https://arstechnica.com/gadgets/2016/08/hbm3-details-price-bandwidth/
"Meanwhile, Samsung has been working on making HBM cheaper by removing the buffer die, and reducing the number of TSVs and interposers. While these changes will have an impact on the overall bandwidth, Samsung has increased the individual pin speed from 2Gbps to 3Gbps, offsetting the reductions somewhat. HBM2 offers around 256GB/s bandwidth, while low cost HBM will feature approximately 200GB/s of bandwidth. Pricing is expected to be far less than that of HBM2, with Samsung targeting mass market products."
I do quite like the differentiation in nomenclature. A bit of space between the numbers and to those who know a bit about the products, you can even tell from the name how much they differ. Not like between 70 and 80 - or 80 and 95, which are completely meaningless. I just hope they keep to this scheme, if it is a scheme after all.
If it is a scheme, they will then need to have Navi naming on next architecture. I think that might end up muddling up the Radeon brand (have you noticed that with Vega we never write Radeon now?) , hurting identity in the long term.
Might Radeon be used only for the lower GPU's and APU's, as I guess of recent times they are seen as a poor relation to the GeForce and GTX brands. Establish Vega as the new high end, similar to what Intel did when they dropped Pentium and introduced Core after the NetBurst/ Pentium D troubles, to be retained as a familiar but lower level brand?
So Navi GPUs would be branded as Vega as well? They could have started going that path when they introduced Fury though.
D
Deleted member 13524
Guest
Vega 32 and 28 simply means 32 and 28 NCUs respectively. Just like Raven Ridge's embedded GPUs are apparently called Vega 8/10/11.
Those are round numbers that are indicative of performance. I like this (name of architecture + 2-digit performance number) better than 3 numbers + suffix, where the first number didn't even dictate the generation and features (I'm looking at you, Pitcairn-based 300 series).
Vega 28/32 coming with HBM2 should put to rest all the "OMG HBM is making AMD burn money" shenanigans, IMO.
Even if they do keep Polaris 10 around and use the new SKUs for "premium mid-end" in laptops.
Those are round numbers that are indicative of performance. I like this (name of architecture + 2-digit performance number) better than 3 numbers + suffix, where the first number didn't even dictate the generation and features (I'm looking at you, Pitcairn-based 300 series).
Vega 28/32 coming with HBM2 should put to rest all the "OMG HBM is making AMD burn money" shenanigans, IMO.
Even if they do keep Polaris 10 around and use the new SKUs for "premium mid-end" in laptops.
I understand that but, I don't think it will work well when evaluating performance between generations, as least for non techie people. How would someone like that know how to distinguish between Vega and Fury, if Fury had the same nomenclature (Fury 64 vs Vega 64). The numbers are exactly the same but performance is very different.
And then you have on Raven Ridge called Vega 11, while they actually have a chip with that codename. Not exactly clear then.
Maybe, I'm just thinking out loud so to speak. Maybe we'll be seeing Navi products branded as RX Navi 64, RX Vega2 64 or even RX Vega 128 next year?
Bondrewd
Veteran
Their current naming scheme looks like "RX (gaming) Vega (uArch name) 64 (nCU number). Assuming Navi will keep it, it'll be something like "RX Navi 64".
Then they should get mor Programmer for primitive shader and make a break for crossfire!
Bondrewd
Veteran
Who knows? They'll find a way (maybe).
D
Deleted member 13524
Guest
The same way people found out the GTX 280 was more powerful than the 9800 GTX (both launched within 2 months), or how the R9 290X was more powerful than the HD7970: the older models are priced progressively lower while they're phased out from the markets.
Last edited by a moderator:
gamervivek
Regular
Perhaps they can add tmu and rop numbers there as well, vega 36-24-36 cards will turn a few heads.
Those are not great examples. In both cases the nomenclature completely changed, which can trigger more investigation from the customer. A situation like Vega 64 vs Navi 64 is way more susceptible to confusion (especially this one where both names have 4 letters and share 2 of them). In this case a customer might end up thinking he is getting a bargain by finding the older GPU at its discounted price and thinking its the new.