CPU Miner Memory Optimizations

So I am running dual xeon 2630v4 CPU’s (40 cores hyper-threaded with 64gig 2100mhz ECC ram) with Tromps solver averaging only 27 sol/sec and only about 4-6 gigs of my 64 gigs of ram are being used. Is there a way to optimize the miner so each core gets its own dedicated memory? I thought the reason why people said GPU’s would not be that much faster with thousands of cores was because they would all end up sharing the same ram on the video card - yet here I have 40 cores pretty much doing the same thing on my CPU - can anything be done to use all the available ram or am I completely off base and lookin’ a fool right now?