This year is Patriot’s 30th anniversary in the memory market, so I’m sure the majority of you have heard of them. When I think of Patriot, RAM is the first product that comes to mind, followed by flash memory products like SSDs, USB drives, and SD cards. As I already mentioned, they’ve been around for a long time, and have made a good name for themselves in the computer industry. Today, we have a kit of their enthusiast level Viper 4 DDR4 RAM to test to our heart’s content.
|Patriot Viper 4 Specifications|
|Unit Weight||105 g|
|Unit Dimensions (LxHxW)||13.3 x 4.26 x 0.71 cm|
|Capacity||16 GB (2 x 8 GB)|
|Memory Speed||2800 MHz|
|Memory Type||Non-ECC Unbuffered|
|Memory Configuration||Dual Channel Kit|
Here’s the SPD tab from CPU-Z showing the tested frequencies, timings, and voltage of the RAM. The XMP 2.0 profile should make booting at those settings pain free by enabling XMP in the BIOS before booting.
Patriot Viper 4
The packaging and clear view of the the contents provides some basic information on the RAM, such as capacity and frequency. Information I’d like to see that’s missing are the four major timings (not just the CAS) and voltage tested for the advertised speed, which are 16-18-18-36 at 1.2 V. It’s interesting that Patriot mentions quad channel on this kit’s packaging even though there are only two sticks of RAM. This kit, or any two stick kit for that matter, cannot run in quad channel. For this RAM, you’ll need another kit and an X99 motherboard for quad channel.
Patriot’s Viper 4 uses a simple black and red heatsink that looks good, but isn’t overbearing. I can also appreciate the black PCBs for aesthetics’ sake, because the standard green PCBs rarely ever match with anything. Patriot also threw in a case badge for those who like to display their hardware brands.
I wanted to remove the heatsinks to check the RAM chips, but those heatsink are quite secure and I was afraid I might rip some memory chips off the PCB if I pulled too hard.
My test setup is pretty basic with the only thing worth noting is the CPU is manually set to run 4.2 GHz with all the power saving features that drop the multiplier disabled. Yes, that 8800GTX is not a typo…
|CPU||Intel i7 6770K @ 4.2 GHz, 1.3 V|
|CPU Cooler||Noctua NH-D14|
|RAM||Patriot Viper 4 2×8 GB DDR4-2800 16-18-18-36, 1.2 V|
|Graphics Card||NVIDIA GeForce 8800GTX|
|Hard Drive||Western Digital Black 640 GB|
|Power Supply||SeaSonic Platinum-1000|
|Operating System||Windows 7 Pro x64 SP1|
AIDA64 Engineer (v5.50.3600)
POV Ray (v3.7)
x264 HD Benchmark (v5.0.1)
Intel XTU (v126.96.36.199)
The XMP profile worked flawlessly to get the RAM to advertised speeds for stock testing. The first overclocking I tried was running MemTest86+ v5.01 at DDR4-3000 15-15-15-35 with 1.36 V. I let the test run for four hours without errors before stopping. I did this primarily to make sure this RAM would be suitable for a permanent spot in my test system for future reviews.
Before I started overclocking, I decided to limit myself to 1.4 V and timings at or below the stock timings of 16-18-18-36. Now, since I knew DDR4-3000 15-15-15-35 1.36 V was stable, I started increasing just the RAM multiplier to see how far I could get. That only got me to 3100 MHz before not being able to boot when trying for more. Then, I decided to set the timings to the default 16-18-18-36 before trying for more speed, which got me through 3200 and 3333 MHz. The next available multiplier after 33.33 was 34.66 which was too much for the default timings. So, I had to try using the 33.33 multiplier and increasing the bclk by a little to test 3366 and 3402 MHz, but to no avail, even when using up to 1.4 V. My next step was going back to DDR4-3333 16-18-18-36 and trying to reduce timings. I was able to get DDR4-3333 16-16-16-36 to pass the XTU 5 minute quick memory test, but it would instantly throw up errors when trying HyperPi 32M at anything but the default timings. So, back to DDR4-3200 16-18-18-36 and testing reduced timings, which worked well down to 16-16-16-36. That’s where I ended up, DDR4-3200 16-16-16-36 1.36 V (1.354 V according to my multimeter).
|Tested Memory Overclock Settings|
Here’s the HyperPi 32M run at DDR4-3200 16-16-16-36:
The tests that show the biggest difference between the stock and overclocked settings are memory specific AIDA64 and MaxxMEM2 tests. Reads and Copies are over 20% faster with Write being at least 10% better. Intel XTU also shows a good boost with a 8.7% better score going from 1230 to 1337. SuperPi 32M and wPrime 1024M have a small benefit of ~2% with the overclocked RAM. The least affected benchmarks are SuperPi 1M, wPrime 32M, Cinebench R15, PoV Ray, and x264 which have negligible benefits of around 1% or less.
This Viper 4 kit is aesthetically pleasing with its understated red and black heatsink combined with the black PCB showing Patriot’s attention to detail. This particular kit ran its XMP profile settings as expected, and also overclocked decently to a stable 3200 MHz while also running tighter (lower) timings of 16-16-16-36 at 1.354 V. I was able to get up to 3333 MHz when using the stock timings, but that was outperformed by 3200 MHz at 16-16-16-36. So, after spending some time with the Viper 4, Patriot’s 30 years of experience shows with their DRAM lineup.
The MSRP of $159.99 seems high according to most similar DDR4 kits out there, but the Viper 4 is being sold for $114.99 at Amazon which is what we would expect. So, as long as the RAM keeps selling around $115, then I have no quibbles with the price and the Viper 4 gets the Approved stamp.
– Matt Green (MattNo5ss)