current speed stats Xeon e5-2699v4 ES

[diep]

Hello!

VBCurtis asked me to show speed of what i have here. A comparision of some kind to a threadripper would be nice.
The program LLR2 v1.1.1 i'm using gwnum 30.4 at all 44 cores.

That would be a 2 socket Xeon e5-2699v4 ES version (QHUP) running at 2.0Ghz watercooled at all 44 cores without hyperthreading (didn't test with).

It is 1% slower than LLR 3.8.23 gwnum 29.8 yet for now keep using LLR2.

Here is the document i constructed. The current status of my box with last 30 lines of the lresults.txt of all the instances running now.

It's doing everywhere a LLR test of course as that's 5% faster than PRP as it seems. See the attached document.

A comparision versus threadripper would be cool. Broadwell and Hashwell seem same core so would expect same timings. Note i have each socket 4 dimms DDR4 ecc reg.

edit: please note that in the comparenow.txt at the start behind each thing is how many threads it runs.

Under windows how i attach llr2 to a socket so that the instance runs faster at the same socket (better than giving it specific cores) is with either:

move to the correct directory for example:

c:\riesel>cd test01
c:\riesel\test01\> start /affinity 3fffff llr2.exe

or:

c:\riesel\test01\> start /affinity fffffc00000 llr2.exe

This is faster than giving it the exact amount of cores it has as in number of threads.