We’re back with another high speed G.Skill RAM kit review. This one is rated DDR3-2600 at a relatively tight CAS latency of 10 (full timings 10-12-12-31) and standard 1.65 V rating and model # F3-2600C10D-8GTXD. Can it overclock as well as their DDR3-2666 kit? Well, that’s what we’re here to find out.
Specifications & Product Tour
The specifications are pretty self-explanatory. Like all G.Skill RAM, this set comes with a lifetime warranty.
|3rd Generation Intel Core Processors|
Dual Channel kit
DDR3-2600 MHz (PC3-20800)
54 mm / 2.13 inch
Intel XMP (Extreme Memory Profile) Ready
Removable Top Fin Click here for details
Now it’s time for the fun photo part of our review. Like most high-end G.Skill dual channel kits, the TridentX 2600C10D’s come in a nondescript box with an included fan.
While the sticks don’t get hot, there is an included fan if you choose to use it. If you have anything resembling adequate airflow in your case, you will not need this, but it’s here if you want it.
Here we have the main event, the great looking TridentX heatspreader design. I’ve been a fan since it came out. While it’s mostly unnecessary for heat dissipation, they definitely look good.
These are double-sided modules. Here you can also see the end screw for the top fin. The fins are removable by taking one of those screws out and simply sliding the fin off. This is useful if you need your high speed RAM in a low profile configuration or if you want to use a memory LN2 pot for extreme cooling.
You know me and gratuitous photos, I can’t get enough!
Last one, I promise.
Good looking sticks for sure! Let’s install them and start playing around.
Our test system consists of an Intel i7 3770K CPU, which thankfully has a strong IMC for testing higher frequencies and a Maximus V Extreme. Today we’ll be looking at comparisons from G.Skill itself, Patriot and ADATA.
|CPU||Intel i7 3770K|
|Motherboard||Maximus V Extreme|
|RAM||G.Skill TridentX DDR3-2666 / 11-13-13-35 Patriot Viper 3 DDR3-2133 / 11-11-11-27 ADATA Gaming Series DDR3-2400 / 10-12-12-31|
|OS||Windows 7 Professional x64|
Here’s a photo with the sticks installed on our test board.
Installed and ready to go, time to test out stability.
Stability at Rated Speed
The first torture we put sticks through is the most stringent testing a normal consumer can get. The UltraX R.S.T. Pro3 PCIe card is used not only by crazy reviewers that enjoy torturing RAM, but by several companies for quality control. Companies typically test their sticks for three passes and then let them go out the door. We’re overclockers and normal stability isn’t an option; so we run it for five passes. After that, we can pronounce them forever stable at their rated speed. This TridentX kit passed with flying colors.
Of course, we do try testing that any layperson can run and give ye olde HyperPi a good run through. No problems here.
Absolutely rock solid stable, as expected. No G.Skill kit (zero, zilch) to date that has passed our test bed has failed stability testing, and I’ve tested a lot of RAM over the years.
Measuring RAM performance is one of the more mundane tests we do here. There just isn’t a ton of difference between RAM kits. Being Overclockers, RAM is very important and can absolutely mean the difference between beating someone in a benchmark and losing to them, but real-world testing doesn’t really show a huge difference between kits.
Where you absolutely can see differences is in synthetic testing with programs like AIDA 64. These benchmarks turned out to be quite interesting. First off, a note about latency – When measuring latency, lower is better of course. That’s difficult to graph when considering three other metrics where higher is better. Thankfully, all our memory test graphs show relative performance as measured relative to the kit being tested. So, we just reverse the division for latency. That results in higher latency figures showing worse performance, thus higher in our latency graph is better. For instance, the TridentX kit being reviewed today had a latency of 31.0 ns. The Patriot kit had a latency of 38.4 ns. As shown in the graph, the TridentX kit’s performance is 18.7% better than the Patriot kit. As for the rest, they’re pretty self explanatory. Memory copy is a wash other than the ADATA kit, which tells me that was probably a testing anomaly. In copy, write and read, you can see the speed trade-off for the tighter timings seems to very nearly cancel each other out vs. the higher-speed, looser DDR3-2666 kit.
Video Conversion and Rendering
Real-world testing is where things get boring. There is very little difference between any of the kits here.
These are the programs that benchmarkers care about. The only one where RAM truly makes a difference in multiple seconds is SuperPi 32M, and that’s the one RAM people need to pay attention to. Here you can see that it prefers the tighter latencies of this kit to the faster speeds of the DDR3-2666 kit. WPrime, as always, is a crapshoot with RAM. As long as you have strong RAM (DDR3-2133 or above), you’re fine in WPrime.
The last part of our reviews is far and away the best part when it comes to RAM – Overclocking! For starters, I wanted to see if the timings could be tightened at the default voltage and indeed they could. They went from their rated 10-12-12-31 to 10-11-11-28, SuperPi 32M stable. Remember, they’re already operating at 1T rather than their rated 2T, so this just sweetens the deal.
It wouldn’t let me tighten the CAS latency (CL) though, so I bumped voltage to 1.75 V, leaving everything the same but dropping CL down to 9. Sure enough, passed SuperPi 32M without issue.
Ok, so they’re flexible on timings. How about speed? Why yes, thank you very much, they can do speed too. At their rated timings (and 1T) and voltage (1.65 V), they passed SuperPi 32M at DDR3-2700, no problem.
Throw a bit more voltage at them (1.75 V) and they’ll pass SuperPi 32M at DDR3-2800 at their rated timings. Impressive.
Sure, but how about max frequency? There’s nothing at all to complain about here. At slightly loosened timings (like those on the DDR3-2666 kit, at 11-13-13-35), this kit made it only two tenths of a MHz short of a CPUz validation of DDR3-2940.
Final Thoughts & Conclusion
Well then, after seeing those overclocking results, I think this kit speaks for itself. Like all high-end RAM, you’ll pay a decent premium for it. The G.Skill F3-2600C10D-8GTXD DDR3-2600 kit is currently for sale at Newegg for $164.99 shipped. Honestly, that’s not a bad price for what you get.
/Tangent/ If you follow the DRAM market, you’ve probably seen that prices are on the rise and that’s borne out with the pricing of this kit. Going forward, you can expect this to be a typical price for a high-end, 8 GB kit. They were actually a bit higher than the competition, but that’s just because G.Skill had already adjusted to the new market pricing. When I saw that, I emailed G.Skill and they explained indeed it is higher and the market is just headed that way. I believe the kits priced lower were just old stock, as currently all competitive kits are sold out (except this unfathomably dirt cheap 2666 Team Xtreem kit) and only G.Skill and Corsair remain on Newegg at the DDR3-2600/DDR3-2666 level; and G.Skill is priced better than Corsair. Actually, now that I’m looking, go buy these G.Skill DDR3-2666 sticks while they’re still $149.99, $15 cheaper than this kit. Once those are sold out, they’ll probably be priced higher. /Tangent/
Back to this particular kit. It’s absolutely rock solid stable at its rated speed and timings, it clocks to the moon, it tolerates tighter timings and it’s priced where it should be in the market. There is nothing at all to complain about and lots to love about this kit.