Patriot Viper 4 16GB DDR4-3400 C16 Dual Channel Kit Review

Why you can trust Tom's Hardware Our expert reviewers spend hours testing and comparing products and services so you can choose the best for you. Find out more about how we test.

Overclocking, Optimization, Performance And Evaluation

One of the reasons I kept G.Skill's 8GB kit in today's 16GB comparison chart is that it has reached the highest data rate. It sets the bar for overclocking cheaper and/or higher density kits, since either of those factors can impact overclocking. Patriot's PV416G340C6K is both cheaper and higher density.

Surprisingly, the 16GB Patriot Viper 4 DDR4-3400 kit and the 8GB G.Skill Trident Z 8GB kit produced the same maximum data rate. This could be a limit of the motherboard, or the CPU’s integrated memory controller. It's also possible that I limited G.Skill's overclock by setting the motherboard to 1.33V, which resulted in a 1.35V output. I have previously theorized that the DDR4-4000 memory might require the extra voltage that motherboard manufacturers try to sneak in, in order to reach its rating.

Swipe to scroll horizontally
Lowest Stable Timings at 1.35V (Max) on Gigabyte Z170X-Gaming G1 (BIOS F5i)
Row 0 - Cell 0 DDR4-3200DDR4-2666DDR4-2133
Patriot Viper 4 16GB PV416G340C6K16-16-16-32 (1T)13-13-13-28 (1T)11-11-11-28 (1T)
G.Skill Ripjaws V 16GB F4-3600C17Q-16GVK15-16-16-32 (1T)13-14-14-28 (1T)11-11-11-28 (1T)
Corsair Vengeance LPX CMK16GX4M4B3200C1616-16-16-32 (1T)14-14-14-28 (1T)11-11-11-28 (1T)
Crucial Ballistix Sport BLS4K4G4D240FSANot CapableNot Capable12-12-12-28 (1T)
G.Skill Trident Z 8GB F4-4000C19D-8GTZ15-15-15-30 (1T)13-13-13-28 (1T)10-11-11-28 (1T)

Rated at DDR4-3400 CAS 16, the Viper 4 kit wasn't able to stabilize a lower CAS setting at DDR4-3200. Viper 4 DDR4-3400 was stable with reductions in tRCD, tRP, and tRAS, but these didn't catch up to the Trident Z DDR4-4000's low timings.

This specific motherboard becomes unstable at tRAS settings of less than 28 cycles.

Swipe to scroll horizontally
Benchmark Configuration
Autodesk 3ds Max 2013Version 15.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080
WinRARVersion 5.0: THG-Workload (1.3 GB) to RAR, command line switches "winrar a -r -m3"
Grid 2Steam Version, In-Game Test, High Quality, No AA
Battlefield 4DirectX 11, 100-sec. Fraps "Tashgar", High Quality
SiSoftware SandraVersion 2014.02.20.10, Memory Bandwidth, Cache & Memory Latency Benchmarks

Sandra shows big bandwidth gains when all four memory slots are populated, though real-world benchmarks fail to make that differentiation. That means Patriot's 16GB kit competes most-directly against G.Skill's 8GB kit in this specific benchmark. Patriot's DDR4-3400 falls only slightly behind G.Skll's DDR4-4000 using "Optimized Timings" at lower data rates: Platform instability hindered bandwidth measurement at the DDR4-4000 kit's "rated settings."

Patriot's Viper 4 didn't excel in Sandra Memory Latency either, though this could be due to numerous factors such as module organization/density or motherboard-directed secondary/tertiary timing increases. Since it's not a direct match for any of the preceding kits, we'll have to lean even harder on real-world benchmarks to further our understanding of its capabilities.

Viper 4 performs very well in memory-sensitive Grid 2, taking second place behind Crucial's cheap stuff. High performance for those cheap Ballistix modules could be tied to enhanced timings in the advanced section, which would also explain why it overclocks so poorly.

Everyone stumbles somewhere in Battlefield 4, though the results of each product average out fairly close to the class average.

One-second differences in timed benchmarks are usually an exaggeration, since a few milliseconds can affect rounding. Still, the average score favors the expensive 8GB Trident Z kit.

Patriot's 2-DIMM 16GB kit falls behind G.Skill's 4-DIMM 16GB kit in WinRAR, which isn't a very fair comparison if you're shopping specifically for 2-DIMM kits. The good news for Patriot is that it beats the other 2-DIMM kit, though that one is only 8GB.

What all this means is that if you want high data rates in a mid-priced kit, you'll want the Patriot Viper 4 DDR4-3400 set. And, if you're not so concerned with data rates, you can save money on the Ballistix kit.

G.Skills Trident Z DDR4-4000 has fallen in price since its review, but this 8GB memory kit is still more expensive than any of the 16GB kits we've tested on this platform. Comparing the capacity of each kit to the previous value chart makes Patriot's DDR4-3400 look like an even better overclocking value, even as Crucial's cheap kit looks like a superb value for non-overclockers.

MORE: Best MemoryMORE: All Memory Articles
MORE: Memory in the Forums

Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware, covering Cases, Cooling, Memory and Motherboards. Follow him onTwitter.

Follow Tom's Hardware on Twitter, Facebook and Google+.

Thomas Soderstrom
Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware US. He tests and reviews cases, cooling, memory and motherboards.
  • Sabishii Hito
    Samsung 8Gbit B-die at work. Too bad Patriot isn't consistent with the ICs they use for this kit, some have Samsung 4Gbit E-die (16 chips per stick, double-sided) and thus can't overclock as high.
    Reply
  • svan71
    8.1 ? why?
    Reply
  • vaughn2k
    I still would like to see what this means in real world situation.
    Reply
  • nitrium
    I still would like to see what this means in real world situation.
    Exactly. I commented similarly in the last DDR4 review. I have a feeling DDR3 would be pretty much just as good for Skylake, but we'll see. RAM is still expensive - I have 16GB of DDR3 1866 but am running it on an antique i5 760. If I upgrade to Skylake I would like to know if I should just get a DDR3 1151 mobo and keep my RAM or go for a DDR4 one with all new RAM. Apparently an article on just this is coming next week.
    Reply
  • Crashman
    17598193 said:
    I still would like to see what this means in real world situation.
    Exactly. I commented similarly in the last DDR4 review. I have a feeling DDR3 would be pretty much just as good for Skylake, but we'll see. RAM is still expensive - I have 16GB of DDR3 1866 but am running it on an antique i5 760. If I upgrade to Skylake I would like to know if I should just get a DDR3 1151 mobo and keep my RAM or go for a DDR4 one with all new RAM. Apparently an article on just this is coming next week.
    Already published last week with high-end DDR3-2800 and DDR4-3000 :)
    http://www.tomshardware.com/reviews/asrock-fatal1ty-z170-gaming-k4-d3-ddr4-vs-ddr3,4431.html

    Basically, if you want to upgrade your memory capacity, you should consider replacing it with DDR4. If you think you'll be happy with your capacity for a long time, it's OK to stick with DDR3.

    Reply
  • educalifa
    I still would like to see what this means in real world situation.
    Exactly. I commented similarly in the last DDR4 review. I have a feeling DDR3 would be pretty much just as good for Skylake, but we'll see. RAM is still expensive - I have 16GB of DDR3 1866 but am running it on an antique i5 760. If I upgrade to Skylake I would like to know if I should just get a DDR3 1151 mobo and keep my RAM or go for a DDR4 one with all new RAM. Apparently an article on just this is coming next week.

    I just went from this EXACTLY same setup to a 6700K and DDR4. Now, I obviously can't comment on how much of an impact the different memory type had vs CPU/platform as I changed them at the same time. But despite the leaps and bounds you'd expect from an upgrade of parts (kept GPU/PSU/Case/etc) I do feel DDR4 is the way of the future and, depending on your needs, this is already a good time to making the jump.

    My use is mostly gaming (FPS) and video editing but I couldn't justify the premium of X99 over Skylake. I could have saved some $ by keeping my old DDR3 but honestly, I'd have saved $120 to keep a set of memory I'd had for almost 5 years. Not worth it IMHO.
    Reply
  • f-14
    17598193 said:
    I still would like to see what this means in real world situation.
    Exactly. I commented similarly in the last DDR4 review. I have a feeling DDR3 would be pretty much just as good for Skylake, but we'll see. RAM is still expensive - I have 16GB of DDR3 1866 but am running it on an antique i5 760. If I upgrade to Skylake I would like to know if I should just get a DDR3 1151 mobo and keep my RAM or go for a DDR4 one with all new RAM. Apparently an article on just this is coming next week.
    Already published last week :)
    http://www.tomshardware.com/reviews/asrock-fatal1ty-z170-gaming-k4-d3-ddr4-vs-ddr3,4431.html

    Basically, if you want to upgrade your memory capacity, you should consider replacing it with DDR4. If you think you'll be happy with your capacity for a long time, it's OK to stick with DDR3.

    that was a terrible article where low power green energy ddr3 was pitted against ddr2 on steroids ddr4 which can't even adequately challenge ddr3 3000+
    it's like the conversion from ddr to ddr2 when ddr ended at 400mhz and ddr2 picked up with 266mhz.
    ddr4 is the equivelent of ddr2 with even lower power on a new die shrink allowing higher clocks. it's not even a real upgrade, especially starting out at speeds 800 mhz under ddr3's highest rated speeds which was 3200...seriously ddr4 2133? stop recycling old ram spec on new die shrinks if you can't make the new starting speed atleast faster than ram that's over 8 years old
    this article further reinforces it with only a 100mhz gain over crucial ballistix tactical with tracer leds even at 1.5v running stock 1600mhz at 8-8-8-24. and this ram is 4 years old. what's ddr4's excuse? NONE over priced slower recycled spec of previous generation 2 ddr on a new die shrink. wait for ddr4 to get to the 4,000mhz mark or wind up in the same boat you were with ddr2
    WE ARE ON TO YOUR GAME RAM MANUFACTURERS.

    this ddr4 374cycles row refill rate is horrendous, you can load the ram up, but emptying and reloading eats so many cycles it allows ddr3 1333 refresh's to blow right past it at almost 3 times per charge.

    the sad thing is crucial ballistixs tactical tracer ddr2 1066mhz running 5-5-5-15 with a row refresh of 32 to 24 can still score on the board against ddr4, yes that ddr2 will be slower, but on a pentium or i3 you are barely going to notice the difference.

    i'm sure if you die shrank ddr2 to 14nm and die shrank core2quad you could crank out 4k mhz clock rates all across the board at 12-12-12-32 at row refresh of 128-140
    and stomp sky lake arse into the mud with a 30% increase in round the board/table performance especially with the new encryption standards implemented.

    learn from the ddr2 lesson upgrade your ddr3 right now to the fastest possible ram and get another 5 years out of your current system easy and wait for intel generation 2020 ddr5 with a year to spare for that generation prices to go down
    Reply
  • Very nice! But I recommend the Corsair Vengeance 16GB over this its far more better!

    Thanks
    Reply
  • Crashman
    17610083 said:
    17598193 said:
    I still would like to see what this means in real world situation.
    Exactly. I commented similarly in the last DDR4 review. I have a feeling DDR3 would be pretty much just as good for Skylake, but we'll see. RAM is still expensive - I have 16GB of DDR3 1866 but am running it on an antique i5 760. If I upgrade to Skylake I would like to know if I should just get a DDR3 1151 mobo and keep my RAM or go for a DDR4 one with all new RAM. Apparently an article on just this is coming next week.
    Already published last week with high-end DDR3-2800 and DDR4-3000 :)
    http://www.tomshardware.com/reviews/asrock-fatal1ty-z170-gaming-k4-d3-ddr4-vs-ddr3,4431.html

    Basically, if you want to upgrade your memory capacity, you should consider replacing it with DDR4. If you think you'll be happy with your capacity for a long time, it's OK to stick with DDR3.

    that was a terrible article where low power green energy ddr3 was pitted against ddr2 on steroids ddr4 which can't even adequately challenge ddr3 3000+
    You've shown up in various threads to say that, and at first I thought you were mistaken, so I previously explained why your statement isn't true. Since you already know the truth, this repetition can logically only be a lie. Worse, harming other readers appears to be the intent of your lie.

    Please report me, because I'm not allowed to beat up on readers :) Readers are more-important than writers because they're harder to replace. But then again, you've proven several times that you never read the article. The other option would be that you read the article and lied about its contents, but I'm not giving you that much credit.
    Reply
  • phil brinkle
    JEDEC #5 column of CPU-Z is wrong! 1037 MHz and respective timings have been determined incorrectly. It's a buggy app.
    Reply