Cause it's higher bandwidth. As showcased by the 3070 vs 3070ti, the difference is miniscule, though.

GDDR6 seems to be optimal between 100GB/s to 500GB/s, HBM2 is clearly optimal at 1000GB/s bandwidths and above. The big question is the 250GB/s to 500GB/s area, where HBM2 1-stack or 2-stack lives at.

Hello! I was looking for an upgrade for my 1080 and stumbled upon two different 3080 variants ASUS GeForce RTX 3080 12GB TUF OC and ASUS GeForce RTX 3080 TUF Gaming OC V2 I noticed a difference in memory between the two. One with 12gb of gddr6 and the other with 10gb of gddr6x. That made me wonde...

Some technical info is available from Micron, but unfortunately seems like they haven't added in GDDR6X to their broad "thermal applications" sheet, unless they intend for it to follow the same thermal guidelines as GDDR6 🤷‍♂️ You'll get better info straight from Micron, but I believe it's rated up to 85c for case/package temperature during operation (operating temp listed as 0-95c ...

This is a decent video on high vs low bandwidth memories (e.g. HBM2/GDDR vs DDR), but I didn't hear anywhere a direct comparison of HBM2 vs GDDR6. When asked specifically to compare them at different points, the answer was always something like: "they're basically the same, just give them enough bandwidth".

I am confused if I should choose GDDR6 graphics on DDR5 setup as would it run properly without any incompatibility issues or should I reconsider my decision and go back to DDR4?

In general, GDDR6X uses 15% less power than GDDR6 to transmit each bit. However, since it is much faster, the total power consumption over a time period still exceeds that of GDDR6." If a 3060ti FE GDDR6 and 3060ti FE GDDR6X were both power limited to 100w max power consumption, which would get more FPS? Thank you.

Does the X affect any compatibility with rest of the system?Also I'm planning on switching MSI 3060ti to Asus 3060ti would i need to use ddu?

From all account the 3060 ti GDDR6X is pretty much identical in performance as a 3070. So if you want to wait for it, there is a difference in the base TI vs the GDDR6X version.

The 3080 and up use GDDR6X, which has another speed bump over the more "traditional" GDDR6. The Radeon R9 290X is a 4GB card on a 512 bit bus. Should be 16 GDDR5 memory chips at 256MB each. HBM2 on something like a Radeon Vega would be 4 "chips" or stacked dies at 1024bit each. As for which is better between HBM & GDDR?