Quote (DCSS @ Dec 7 2017 11:47pm)
issue is those still use system memory iirc so they are very limited in performance just like APU's, not likely to compete well with even a gtx 1050 with these bandwidths in mind. Would be lucky to see gt 1030 performance without implementing dedicated vram and that gets costly real fast
it's always been convenient to put it next to the cpu because it gets to benefit from being where the highest heat transfer takes place. Might not be as much of an issue as I hear low clocked vega is absurdly efficient, however another fan and maybe heat pipes might be necessary on the mobo... esp. if you're implementing dedicated vram. Also you won't have an easy time upgrading or changing your gpu since it won't be tied to the chip you will need a whole new mobo... Might also be some IP concerns with allowing board partners to implement vega directly into their motherboards. Probably not since they won't receive the rights to fab but who knows
chipset with graphics and vram similar to laptop shit but or even use mxm shit, and if you arent using the damn chipset gpu and ram the ram could be used as cache drive, but all that is wishful thinking
wouldnt be hard for them at all
emib and if will be the future
personally I hate the fact I am paying for igpu when I dont need it
like the pII and c2 days when you chose motheboard with different chipset/gpu
This post was edited by yupitsmeh on Dec 8 2017 08:30am