Question Why did Intel abandon further desktop development of the Broadwell-DT i7-5775c/i5-5675c?

80251

Distinguished
Jan 5, 2015
98
2
18,635
Intel must've known the Broadwell DT i7-5775c could outperform most all other CPUs of the time (not always of course), yet they just completely abandoned all further development of eDRAM LLC for their desktop CPUs. I seem to remember people selling these Broadwell-DT parts at greater than MSRP on ebay.

I don't know if AMD's 3D-vcache is comparable but it is a large, fast, LLC and it's already proven itself for the last 3 generations. Was Intel's decision to abandon further development of eDRAM equipped CPUs a business/marketing decision or an engineering one? Considering AMD's proven success w/their 3D-vcache it seems like Intel's decision was a stupid one.
 
Intel must've known the Broadwell DT i7-5775c could outperform most all other CPUs of the time (not always of course), yet they just completely abandoned all further development of eDRAM LLC for their desktop CPUs. I seem to remember people selling these Broadwell-DT parts at greater than MSRP on ebay.

I don't know if AMD's 3D-vcache is comparable but it is a large, fast, LLC and it's already proven itself for the last 3 generations. Was Intel's decision to abandon further development of eDRAM equipped CPUs a business/marketing decision or an engineering one? Considering AMD's proven success w/their 3D-vcache it seems like Intel's decision was a stupid one.
I wouldn't say it was stupid, those tiles were very expensive to manufacture and they we're already kings of the market, so they would only be competing with themselves. They could have started looking to re-introduce the cache tile concept starting with coffee lake and onward, but it still didn't make sense since they were very competitive without it. Then we had the thermal and power limit issues that they started to run into with the 10th and 11th gen, and 12th gen was also very competitive without it. It's really only after 12th gen that they needed to worry about something like that, and AMD beat them to the market. AMD didn't introduce their 3D V-Cache chips until several generation after they had their tile process down, making it a bit easier to be able to add that extra cache in a way that made sense. The bigger deal was figuring out the power and thermal limits, and how to stack it. It still wasn't a cheap add on. Intel will likely start looking to add cache die's in their future cpu's since they've also moved to a tile based approach, we may have to wait a little while for them to become available though. It wouldn't surprise me if we don't see them make an appearance until after Arrow lake. In general I wouldn't say it was for any engineering reason that they didn't keep rolling with those extra cache tiles, and it was purely cost and market reasons. There was no reason for it to exist, until there was.