Table of Contents
The introduction of the X79 Sandy Bridge-E platform brought along with it the ability to use more powerful quad channel memory. Today, we examine the Patriot Viper Xtreme Division 4 DDR3-1600 CL8 kit to find out whether quad channel memory truly delivers more bandwidth than its predecessors.
Meet the RAM
- Extreme Performance PC3-12800 (1600 MHz)
- Low Latency (8-9-8-24)
- Voltage: 1.65 V
- XMP 1.3 Ready
- Equipped with an extruded aluminum shield to provide improved cooling
- 100% Tested and Verified
- RoHS Compliant
- Only compatible with Intel X79 chipsets
First let’s take a look at the retail packaging. You can see a tastefully designed box by Patriot showing a picture of what is inside. There is a sticker on the top of the box showing exactly what model sticks are in it. Inside the retail packaging one can see that two sticks have their own clamshell providing adequate protection for the product.
Last are the actual sticks. Like my previous Patriot review (Viper Xtreme DDR3 2000 MHz 3×4 GB), you know I love the subdued look of the Xtreme series custom copper and aluminum heat sink. The Division 4 sticks have the exact same heatsink. These sticks do not have a matching black PCB like the previous version, but in the slot (or out really) it hardly detracts from the look going on here.
Test System and Methods
- i7 3930k (stock)
- Gigabyte X79 UD7 (Overclockers Approved!)
- Patriot 4×4 GB DDR3-1600 (8-9-8-24 @ 1.65v)
- Asus Matrix GTX 580
- SeaSonic 1000W
- Windows 7 64bit SP1
You are not going to see much out of the ordinary in a lot of these tests. Faster speed RAM just doesn’t matter in most of these tests. As you can tell, the 1600 MHz quad channel RAM has kept up with the 2133 MHz RAM across most tests. Cinebench, both R10 and R11.5 both show negligible differences even between dual and triple channel.
You can see notable differences in SuperPi 32M and Wprime 1024M however. As most benchers know, faster and tighter timing RAM knocks the seconds off these long runners and we see that holds true here.
Last, but not least, are the synthetic bandwidth and latency tests. Starting out looking at AIDA64 and MaxxMEM2 you can see that the original versions of these are single threaded and are therefore not showing multi channel RAM in its best light. However if you look at the last MaxxMEM2 (multi) result, you can see scaling up through triple channel results. Hopefully they update this version to really test out what quad channel can do. Sisoft Sandra (2012) is also showing what we would expect across the board, doing better than its triple channel counterpart even with a severe speed deficiency (1600Mhz to 2133Mhz). Goes to show the potential of quad channel memory when its able to be taken advantage of by the software.
I’ll start off by saying this experience was a lot better than my last go at overclocking RAM. The voltage is already pretty high for the platform, but within specifications at 1.65 V, so I decided to leave that the same for testing. As far as tightening down at stock speeds, I was able to to go 8-8-8-21 (SPi 32M stable) but was not able to drop to CL7 no matter what, at that voltage.
For overclocking, I was initially able to reach 848 MHz with stock timings and voltage. Not a bad increase, as we have seen less in the past, but not a lot of headroom out of the box with its factory timings. After raising the timings to a reasonable level (9-10-9-27) I was able to push past 933 (1866) MHz to 952 (1904) MHz Super Pi 32M stable. So loosening things up allowed a fair amount of headroom.
Overall these chips do a bang up job at their stock settings, as they should, have a bit of headroom to overclock on stock timings, and more if you loosen them up. With the SB and SB-E platforms not requiring raising the bclk to overclock, it doesn’t matter to terribly much in the big picture. But if you want to squeeze out a few extra MHz out of your overclock via bclk, or use the 1.25x multiplier, this kit has a bit in the tank to take you there without lowering the speed.
Finally, a set of sticks that does well out of the box (expected), and overclocked! Like 99.99% of all brand new in box items, you expect them to work per the factory specifications as described, and they do. Overclocking on this set of RAM was successful up to 952 MHz which inst bad at all for a 4 x 4 GB kit. We weren’t breaking any records with this kit, but those trying aren’t buying 4 x 4 GB 1600 MHz kits to do so either.
The aesthetics of the kit should not sway anyone from choosing them. Unless you have some off the wall theme, the black aluminum with copper will match mostly any setup. The only concern may be with larger tower-style CPU heatsinks and clearance issues as the RAM heatsinks do stick up a bit above the PCB. I had no trouble though with these with a Coolermaster Hyper 212+ on a G1.Sniper2 when I tested. They were precariously close if not touching though, but all was well.
The MSRP on this kit is $119.99. I haven’t been able to find them at the usual websites (Newegg for example) yet, but a quick trip to Google Shopping and you can find this kit. To wrap this up, you have a good looking quad channel kit for the X79 platform that does its job at stock settings, has some headroom available to overclock, and is priced quite well among its peers making this kit Overclockers approved.