First signs of AMD's next-gen Ryzen Threadripper uncovered.
Next-Gen AMD Ryzen Threadripper Might Be Coming Soon : Read more
Next-Gen AMD Ryzen Threadripper Might Be Coming Soon : Read more
My theory is that Threadripper would be a dumping ground for defective IO dies. IO dies are huge and throwing one out because of one bad memory controller would be a huge waste and would be expensive. To differentiate epyc and threadripper AMD might unlock threadrippers. But other than that they are going to be exactly the same. Idk just my theoryThere is one conundrum with a 4th Gen Threadripper release: Package power for the 64-core SKUs.
The new EPYCs already push 280W, which might leave little socket power/thermal headroom for Threadripper to really shine.
I am curious to see how AMD meets this challenge.
Edit: I almost forgot price slotting... the price gap between current Threadripper and new EPYC is actually fairly narrow as far as target markets are concerned.
Isn't Epyc octa channel RAM-wise? And Non Pro Threadripper is Quad-Channel right? How likely would it be for an IO die to be defective in such a way that octa channel doesn't work but quad channel does perfectly? Server RAM has poor timings and is at most 3200Mhz so if anything Threadripper needs the better memory controllers surely?My theory is that Threadripper would be a dumping ground for defective IO dies. IO dies are huge and throwing one out because of one bad memory controller would be a huge waste and would be expensive.
Threadripper 3rd Gen + 1USMUS's Clock TunerTo differentiate epyc and threadripper AMD might unlock threadrippers. But other than that they are going to be exactly the same. Idk just my theory