With the second quarter of 2017 now at our doorstep, it is about time to start counting down the days to AMD's launch of the RX Vega. We've already seen some details through various leaks and demonstrations over the last few months and this week, we learned that the RX Vega would be available in 4GB and 8GB HBM2 versions, which goes a long way towards explaining why seven different Vega device ID's were found in recent Linux GPU drivers.
According to AMD's Scott Herkelman, the RX Vega launch is ‘just around the corner' and while the original Fury X was limited to 4GB of HBM, Vega will be able to bump that up to 8GB of HBM2, though a 4GB option will still exist.
At a recent AMD Tech Day (via Videocardz) Herkelman explained that “Vega will use HBM2 that has different capacity stacks. You will see from our board partners different configurations, whether that’s 4GB or 8GB”.
So we can expect to see different configurations from AiBs once Vega launches. 4GB versions will have a single memory stack with 8GB versions will have two. However, if HBM2 can be produced in 2GB stacks, then we may see 4GB GPUs with two full stacks of memory as well.
Aside from learning that Vega is ‘just around the corner' we also learned that Vega would in fact be available in gaming notebooks later down the line. So AMD will be competing with Nvidia in the high-end desktop and laptop GPU space.
KitGuru Says: Hearing that Vega is ‘just around the corner' is encouraging. Hopefully that means we will see these new high-end GPUs shortly after the RX 500-series launch, which we already know will be a Polaris refresh.
Im with GTX1080 ti still.
Well, it’s gotta be a LOT cheaper than the Ti even if it competes with it performance-wise due to the vRam deficit (and that’s coming from a Fury owner). Another generation where nVidia is one-upping them on vRam (in the high-end at least). I run the LG 38UC99 so thankfully 8GB of HBM is probably just enough for me over the next 2/3 years.
Come on, release it, cuz I need a Acer XR382CQK.
Another generation? The R9 390 and 290 both had miles more vRAM than the 970, which was meant to be the competition…
The idea behind the new cache system is that it will perform as well as it would with more VRAM. With less they can price it cheaper (potentially) too, or make more profit.
Tbf my Fury does alright at 3840×1600, although I’ve had to run at lower settings than I’d like (so it’s probably helped).
Which is why I said high-end.not middle (although middle is probably where the money is at)
Nice! “So AMD will be competing with AMD in the high-end desktop and laptop GPU space.”
Its like saying Vega sucks already
you can’t release a product without extensive quality testing and refinement/optimization. that’s like selling raw Milk before it’s pasteurized and tested for a number of parameters.
Just around the corner tells us nothing new. It was just around the corner last year according to half the internet journalists. That said a lot of them missed the class on integrity at journalist school and probably come up with half the news they post while sat on the throne taking a dump. Here’s thinking of you WCCFtech. The site that produces enough manure to keep 40 acre’s of working farm land nourished for a full year..
ROFLMAO!!! @ last two sentences!