I really hope this is true. I 1TB/s GPU with 8GB of memory would make me seriously weak at the knees.
How weird/hard would it be to have two distinct memory pools, 4GB HBM and 4GB GDDR5 (in a 128bit bus, for example)?
Even if both pools would have different bandwidths, nVidia has made at least 2 graphics cards using memory banks with uneven performance: GK104 in 660 Ti and GM104 in GTX 970.
It just seems to me that 4GB will be too much of a limitation for the performance these cards are rumoured to have.
I don't think HBM1 tech itself would prevent using 4Gb density chips for example, but it's better to go safe on new standard I suppose?I never really understood why HBM 1.0 should be limited to 1GB per stack. It seems like something that should be purely a function of density, and independent of the revision.
IMHO, it is only a matter of time when we get these to DirectX, since the cross lane operations provide nice performance gains for many algorithms (reduced GPR usage, reduced LDS usage, less instructions, etc) and at the same time make writing these algorithms much simpler.
They are just quoting/linking fudzilla, as with half of their GPU articles.WCCFTech also claiming that 390X will come with 8GB HBM:
http://wccftech.com/amd-r9-390x-8-gb-hbm/
They are just quoting/linking fudzilla, as with half of their GPU articles.
4GB vs 12GB in this "4K-tests". And R600 was more slower than G80.Looks like a repeat of R600 if this really is an HBM chip.
Liquid water cooler require ~20/30WAnd it uses more power
I really don't think that is right at all.Liquid water cooler require ~20/30W