I think it's a comparison to Carrizo…
I could swear I heard the guy saying it was being compared to an Intel Core.Yeah, it's pretty obviously being compared to AMD's previous APUs
"HW Accelerated Data Transfer Between"
Between what?
I think it's just the fact that the GPU has an embedded NVMe controller within the new HBCC, so the data transfer doesn't have to run through a PLX controller like the previous Radeon SSG. Latency and peak bandwidth should be a lot lower.What is that HW-accelerated data transfer thingy? On-the-fly compression?
Same thing.Where does it say HBM2? It's 16 gigs of high-bandwidth cache. This presentation is pretty bad, even if the hardware is not.
Ryan Smith said:Meanwhile it’s very interesting to note that with Vega, AMD is calling their on-package HBM stacks “high-bandwidth cache” rather than “VRAM” or similar terms as was the case with Fiji products.
Eerm... between the two items that are immediately below?
Between the 2TB NVMe and the 16GB HBM2.
P100 is last years architecture. Would be more interesting to se how it compares to V100.Vega vs P100 in DeepBench:
Frontier Edition...
I would not bet on the 8-hi stacks. It is not to be ruled out yet that Vega can talk to four stacks - maybe even that Vega sample shown at the Radeon Tech Summit only was the smaller Vega 11 or not the fully equipped version of Vega 10. Smoke screens and mirrors.Yup, 16GB HBM2 / 8-Hi stacks triple confirmed, as well as ~1.5GHz core clock.