TheOverclocker Issue 18 | Page 16

GEIL EVO TWO PC3 19200 QUAD CHANNEL RAM RRP: $145 | Website: http://www.geil.com. Test Machine • • • • • Intel Core i7 2600K ASUS MAXIMUS IV Extreme Plextor M3 256GB SSD ANTEC HCP 1200 PSU Windows 7 64-Bit I t’s always interesting how INTEL platforms always manage to set the new benchmark for the “standard” memory amount in people’s desktops. This is also true for overclockers as well because all of a sudden 16GB sets are all the rage at least for performance sets. As cheap as DRAM has become though, 2400MHz 16GB sets are mostly selling north of the $200 mark and despite this being a reasonable price it’s still a lot of money to sink into RAM in addition to all other costs associated with the X79 platform. Good thing then we have 16 The OverClocker Issue 18 | 2012 sets like the GEIL EVO TWO 8GB 2400MHz kit. 8GB of memory is still more than most people will require at present especially for overclocking. 8GB of memory, a PC3-19200 rating and a $145 price tag, you have a really cost effective set for enthusiasts. For this kind of money, the GEIL EVO TWO is surprisingly impressive. We had an issue with the XMP profile which would automatically set timings that would not POST on our particular system, which would hang the system at “31” on the POST LED. After some fiddling we realized that the secondary settings applied via X.M.P were too aggressive. Oddly enough, manual tuning of these, using the recommended settings allowed the system to boot. The offending settings were DRAM RAS Value Award # to RAS Delay and DRAM READ to PRE Time. The XMP profile configured both of these settings to a value of 4, however the ideal settings was 5 and 6 which fixed the issue. After that it was pretty easy going all the way to the most comfortable limit of our CPU IMC The default timings of 11-11-11-30-2N are nothing to write home about and are quite loose to be honest, however we quickly realized that 10-11-10-281T was doable with a slight adjustment in VDIMM. Setting 1.66V stabilized the system and as result we got some pretty good performance out of the set. Sadly however this was as low as we could go because anything tighter than this rendered the system unable to POST. Dropping the Memory multiplier and selecting 1866MHz divider allowed