Curious what E5 config will consume less power than Xeon-D or am I miss understanding and you mean less power than the L56xx dual setup being talked about ?
Pretty sure that'd be the dual setup under discussion. A Xeon D-1537
X10SDV-7TP4F seems to run around 37 watts idle (two RDIMM's and a platinum PSU) and an E5-1620v4 on an X10SRL-F with two RDIMM's and a Seasonic G-650 PSU on the bench here is idling right around 38 watts as well, though that latter is part of an assembled system that has some fans running, and lacks an onboard 10G and HBA, so the comparison isn't totally fair, and obviously the Xeon D is actually winning that once you adjust for the onboard HBA.
Naturally, the 1537 will have a much more aggressive thermal limit with its 35W TDP, whereas the E5-1620v4 is a 140W part. Speed-wise, the 1537 is a 1.7-2.3GHz per core part, whereas the 1620's a 3.5-3.8GHz per core part. Overall, both end up in roughly the same territory in terms of multi-core performance, but single core performance is a massively different story. The Xeon D is a power sipper in comparison.
A dual L5640 would crush either of them, both in multi-core performance and effectiveness as a space-heater. Of course, you can move on up with the E5 platform into the 26xx's, at which point a single socket board with an E5-26* will win out, at the cost of single-core performance. The only real way to win on single-core is to go E5-16xxv4 or something like that where the core count is traded for higher clock.
I think it is fair to say that it'd be nice if the X10SDV-7TP4F had been built with a 1541 instead of a 1537, but they seem to be trying to steer towards the higher core count D-1587. However, at that price point, and with the greater TDP, and the limited board selection, it doesn't seem like there's a huge market for that.