Asus Rampage Gene II Review - An Extreme 2 On A Diet?

Motherboards/Intel S1366 by leeghoofd @ 2009-07-20

We got the MicroATX Asus Rampage Gene II on the test bench. Does this PCB shrink mean it doesn´t have the full package on board? Is it really worth cashing out for its bigger brother, the Rampage Extreme II? Some questions that many users would like to see answered in this not so usual motherboard review. Most of you have seen a zillion of photos/reviews already of this piece of art. Madshrimps will take a light approach on its features, but dedicates a few more pages to focus on performance and overclocking!

  • prev
  • next

Essential Tips and Tricks

Essential Tips

Due to the fact I had more than month to dig into this motherboard (Thx Milan) I also had the time to test several BIOS versions. Final bios I used for the Gene II was the 0809 one. Some issues we reported to Asus :

Madshrimps (c)


OC from Memory Level up : Plz do NOT use this feature as it heavily overvolted my low voltage OCZ rams. To be more specific: choosing eg 1600 ram speed: it automatically set 1.8Vdimm, it even gets worse from there on ; at 2000mhz, Vdimm fed a whopping 2.0 volts to my precious dimms. The Corsair rams didn't suffer form this overvolting (mostly likely another XMP profile). I hope the bios engineers either remove this feature or correct it ASAP for those dimms that aren't correctly programmed. I'm pretty sure not all users are aware of this and do not verify all the volts. Think it's better to be safe than sorry !

My board seemed at first to behave very erratic with the Hyper rams. I suffered from daily BSOD's. Sometimes just coming out of the blue (no pun in intended). This even when having the rig at idle, just surfing or when chatting on MSN. I swapped the OCZ Blades for the Corsair Dominators PC12800C8 kit. Yet the blue screening remained pretty persistent. Sometimes I was being able to test the whole night, yet the next morning the first boot being welcomed by a BSOD, seemed very strange. I experienced such behavior on the Rampage Formula SE and the P5Q-E due to faulty Loadline Calibration. On the Rampage SE it required a hardware change (newer PCB revision), with the P5Q-E it was solved via a newer bios version. On the Gene II I disabled LLC and haven't had one bluescreen since. If you experience some wicked BSOD phenomena, this might be worth a shot. I have found other users where it was the solution, others needed to enable LLC to get the board stable. Go figure...

QPI voltage fluctuates big time, I have no idea if this can be solved via a bios update yes or no. For 2000mhz ram speed operations I needed 1.5QPI to remain stable (DMM readout went over 1.53 when under load), on other boards I could use a lower QPI voltage for the same ram clocks. Though it is still far under the 1.65 preset values of some XMP profiles.

QPI data rate is best left on AUTO or the lowest value. I can only post with my sample board when the Uncore of the CPU is exact twice the ram speed. If I go one notch higher on the uncore setting, the Gene II doesn't post when the QPI data rate is not on AUTO or set at the lowest value.

Last but not least the Back to back CAS# delay at Auto is too tight for high Bclocks and high ram speeds with certain hardware configurations. We passed it through to Asus, hopefully to allow better stability out of the box they will adhere to our findings.

Here's a graph with the most common ram used on I7 : PC12800C8

Madshrimps (c)


Do you see the massive influence this setting has on bandwidth ? Really shocking isn't it ? Though some users can run 1600mhz rams with the AUTO setting in the bios, I couldn't complete the Hyperpi 32M 8 threads without freezing. Setting B2B to 6 got the rig smooth sailing, even with 12Gb installed.

Essential Tricks

To get over 222Bclock I had to up IOH voltage over 1.3 and set PCI-E frequency at 115 on my sample. 226 Bclock was the highest I got 2 hour small prime stable and the GENE II was able to complete a 32mb Pi run. Booting higher into windows was no problem, but I couldn't get any stability as the rig froze up each time I stressed it. Also 3D tests were a no go as my GTX 285 doesn't like high PCI-E frequencies at all.

For stable 200Bclock and 1866, 2000mhz ram action, the Back to back cas# delay setting is crucial. When left on AUTO, the bios reads it out as "0". Though testing confirmed that in reality the value is around setting 4. Massman digged deeper on this matter. Cross checking my findings and data with the MSI Eclipse board (These are the only 2 boards to our knowledge that have the effective B2B setting in the bios) Check out his article here : Stabilizing your memory OC on I7

Felix, the programmer of the MemSet tool, will include this setting in the upcoming version. According the Intel tech sheets, B2B seems to be adjustable in Windows.

I tested the 3 ram kits at their original speeds and tested the settings thoroughly with Hyperpi 32mb (8 Threads), HCI Memtest up to 500%, Everest Ultimate and finally it had to complete one last Superpi 32mb run. Loosening the B2B seriously influences ram bandwith (especially Read and Copy bandwith). Latency however was not affected. Compare it with the TRD or Performance Level from the S775 days. At 200 Blcock I tested the following ram configurations:

6gb PC16000 C7, C8 and C9
6gb PC12800 C7 and C8


Observations : A higher CAS level didn't influence the B2B setting at all. Meaning loosening RAM timings didn't allow me to lower the B2B setting one increment. Even though I could pass e.g. the Superpi 32mb test, it still would freeze on me with Hyperpi. The same for voltages, skews, etc; If the B2B is too tight it will not work, no matter what. Below are the which-I-find most stable settings for the Gene II ( they passed all the above tests with flying colors ). For sure tweakers can tighten them by messing with the Skews and voltages depending on what they want to achieve.

Madshrimps (c)


For PC12800 ops use setting between 6 and 8. 1866Mhz users that run into problems might wanna give 8-10 a shot.

For PC16000 ops use setting 10 and when instability remains, try 11 or even 12 back for the back to back cas delay setting.

If you look closely at the chart you will see that a well tweaked 1600mhzC7 with B2B 6 can almost match a more expensive 2000MhzC7 rig with B2B 10. The write performance though remains unmatched and in favor of the 2000mhz rams.

The IOH (some refer to it as the NB) can get very hot. With my Thermalright Air cooler it got close to 65°C under load in a closed case. When testing the Corsair H50 unit, the IOH temp was over 90°C, this due to lack of airflow. Put an active fan over the heatsink to keep it cool, surely when water-cooling your CPU. I always use an Antec Spotcool, that's positioned to blow air over the rams and the IOH.. It's very easy to fit and has a selectable 3 speed 8cm fan.

On the back of the motherboard, near the CPU socket, there are mosfets too. These tend to get very hot under load. So If you are routing cables behind the mobo, make sure they remain clear of this area. Also if you bench old skool (meaning the mobo on a foam or box, make sure the mosfets get some air or they could melt into the surface below.

Installing 6 Dimms is doable, but with my setup I needed to replace the TRUE's plastic fan shroud by the clip system. Think the following picture with only 3dimms tells the tale.

Madshrimps (c)

  • prev
  • next

No comments available.

 

reply