theroc44 posted on Apr 30, 2016 12:26:16 PM - Report post
Bunch of people online are saying broadwell-e wont live up to the hype, so i did some research and heres what i came up with. So has-e is 5% slower in instructions per cycle (ipc) than brd-e(based on tick-tock) but the brd-e especially the higher models are running at a 15% lower base clock due to power saving design of the chips. So its a wash there.Another thing the previous released broadwells had something called crystalwell igp,which gave it 128mbs more imbedded dram and that imbedded dram could be used as L4 cache.The extreme edition as we know doesnt have IGP so you lose that right off the bat.So looking at overclocking the original haswell and broadwell releases showed that for example the 4770k and the 5775c OC'ed on average 20-30%. If you extrapolate that to HEDT versions it parallels. Then the broadwell-e will likely match the 40-50% OC range that we see on haswell-e. So if its running at a lower clock its not really that much faster. Lastly the multi-threads(where broadwell-e shines), the haswell k's had 2mbs of L3 per core. The broadwells had 1.5mb per core. In the HEDT version the trend should continue, the broadwell-e is gonna be 6,8,10 core. Then the brd-e and the has-e will both have 20megs of L3 cache and obviously the 10 core 20 threads will win in multi threaded apps. But for gaming you dont need 20 threads. SO all in all unless your heavy multi-tasking, then the broadwell-e isnt really worth the upgrade. It doesnt seem likely that it will overclock better and the small jump in performance 5% isnt great or worth the upgrade IMO. Not to mention it should still be locked at 40pcie lanes as well just like the higher 5930/5960 haswells. Hope this helps you guys/girls stuck on the fence about Haswell-e now or wait for broadwell-e.
Side-note- The skylake extremes should also be coming , those might be worth the change. I also have to give credit to gamers nexus for alot of this info.