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

Somehow I fear, it will not do too well under DX11. For the die size it needs to be on Titan level.

 

error

 

https://www.techpowerup.com/reviews/AMD/Radeon_Vega_GPU_Architecture/2.html

 

Anandtech has that too, but are no more than the slide from some hours ago

 

I am not sure if about the conclusion. Yes, it a big redesign but it also something like a mix between GP100 and GP102. I fear that those NCUs will not be able to show their power in DX11.

 

Seems they got it wrong. According to AMD's videos on YouTube, HBC is meant to replace the "traditional" video memory architecture.

 

Can anyone explain better how the primitive shader is supposed to work (and interact with current APIs) ?

 

Try this:
http://www.hardware.fr/news/14910/architecture-vega-10-amd-leve-voile.html

 

So is there any answer to the question of WHEN with respect to Vega rather than the obvious answers given about WHAT Vega is about. There were rumours that Vega had passed RRA (South Korea) certification, so when does this baby land?

 

Still there is no have sufficient detail to draw any meaningful conclusion on the "HBC" thingy. Anything from page table magic to real hardware caching is possible.

If one assumes it is purely hardware, for 16GB the tag storage needed would be ranging from >2GB (64B granularity) to ~32MB (4KB page granularity, 52-bit physical address, 12-bit state, more if 64B lines in the page can have their own state). It doesn't seem to be maintained on-chip.

How it would interoperate with cache coherency protocols in the host when being part of an APU is another interesting question — would they just skip this level of allegedly hardware caching for system coherent accesses?

 

1H/2017. Given the many similarities with the Polaris Launch (fresh chips, hot off the foundry, timing of PR etc.), it is not unreasonable to assume a similar timeframe for roll-out.

 

While TPU claims it's not the same as HBM, I hold my ground on saying HBC = HBM, AnandTech seems to agree
http://www.anandtech.com/show/11002/the-amd-vega-gpu-architecture-teaser/3

 

It was a little bit expected, than if it will bring some response or confirmation, thoses ones will raise even more questions. ( and speculation ). ( and i dont speak about HBC/HBM, just for a whole )

 

AMD never announced or hinted an any conference, presentations, livestream. It was clearly labeled as an Architecture Preview and that's exactly what was delivered. no need to be disappointed.

 

HPC is the "Total pool of memory" that was called VRAM in the Past says here:
http://techreport.com/review/31224/the-curtain-comes-up-on-amd-vega-architecture

 

its your own failure to understand disappointment from your own judgement of expectations isnt reality.

 

HBC, but yeah. Let's rename a few more things just to confuse the hell out of people.

 

Considering the countdown actually said "Countdown to Vega Architecture Preview", I don't see what's the fuss is about - they delivered what they promised.