Next in our line of memory reviews is a kit I have actually had at my work desk for a while.  In the land of overclockers, synthetics are everything – if it can get a higher number on a screen, and that number can be pushed, then it is worth it.  Thus in steps comes G.Skill with their high end TridentX range, pushing the boat from DDR3-2400 (both 4GB and 8GB modules) all the way up to DDR3-2800 (4GB modules only).  The kit we are testing today falls right in the middle of all of this, being a 2x4 GB kit of DDR3-2666 11-13-13.  This is an 8 GB kit that retails at $170, and for that money we could easily pick up a 2x8 GB 2400 C10 kit.  Proof will be in the pudding as we put this kit through the testing suite – let us see if it is actually relevant for day-to-day use.

G.Skill 2x4GB DDR3-2666 C11 Kit Overview

Back in our initial memory overview, I gave a couple of formula in order to calculate (at a high level) how a memory kit should act in sequential and random reads.  The formulas were:

Time to read one word: CL * 2000/MHz
Time to read eight words: CL * 2000/MHz + (7*1000/MHz)

Applying this to the G.Skill kit we have in today gives us 8.25ns for a one word read, and 10.88ns for reading eight words.  For comparison, the following memory speeds offer similar scores:

DDR3-2000 C8: 8.00ns/11.50ns
DDR3-2133 C9: 8.44ns/11.72ns
DDR3-2133 C10: 9.38ns/12.66ns
DDR3-2400 C10: 8.33ns/11.25ns
DDR3-2400 C11: 9.17ns/12.08ns
DDR3-2600 C10: 7.69ns/10.38ns
DDR3-2600 C11: 8.46ns/11.15ns
DDR3-2666 C10: 7.50ns/10.13ns
DDR3-2666 C11: 8.25ns/10.88ns
DDR3-2800 C11: 7.86ns/10.36ns
DDR3-2800 C12: 8.57ns/11.07ns

Most of these kits should be available for sale around the same price, but purely by these numbers the 2666 C11 has an advantage over both the 2400 C10 and the 2800 C12.  Moving up to 2666 C10 obviously has the advantages of the lower command rate.

With the G.Skill 2666 C11 kit we are testing today, as it is the highest rated kit we have tested, it clearly gets top marks across almost all of our benchmarks.  Obviously there are some benchmarks that enjoy the MHz (Batman:AA, WinRAR), but others either tend towards a limit (Portal 2 on IGP) due to other factors becoming limiting, or some would prefer more memory as a whole (Thunderbolt copying, more on that later), or others need a nicer balance between MHz and Command Rate (catia in SPECviewperf).

The kit is also great at overclocking.  Like I said, I have had this kit for a while, and part of playing around with the kit was an attempt to push the MHz as far as it would go.  In that respect the kit easily pushed 2800 C11 at stock volts, and 2950 C12 with a bit more juice.  Given that plenty of kits on the market will not move an inch above XMP at stock volts, this is nice to see.

The downsides to this kit are predominantly two-fold.  The cost per GB is astronomical, especially compared to the other kits we have tested.  In our memory overview, the 1866 C9 4x4GB kit performed at almost the same level as this 2666 C11 2x4GB kit for double the capacity and $75 less.  That is a shocking statement – the only way to show that the 2666 C11 kit is a mountain and a half better than the 1866 C9 kit is through synthetics, which rarely have a real-world influence, especially in memory.

As a result, this kit is purely for the extreme overclockers out there – those that need and can feel the difference between 2666 C11 and 2400 C11 by virtue of the number they get on a screen.  It will be some time before a kit of this high specifications comes down to a reasonable price (sub $100) for the rest of the userbase to be interested.

Specifications

	G.Skill				GeIL	G.Skill
Kit Speed	1333	1600	1866	2133	2400	2400	2666
Subtimings	9-9-9-24	9-9-9-24	9-10-9-28	9-11-10-28	11-12-12-30	10-12-12-31	11-13-13-35
Price	$75	$80	$95	$130	$150	$145	$170
XMP	No	Yes	Yes	Yes	Yes	Yes	Yes
Size	4 x 4 GB	4 x 4 GB	4 x 4 GB	4 x 4 GB	2 x 8GB	4 x 4 GB	2x4GB
MHz	1333	1600	1867	2134	2401	2401	2667
Voltage	1.500	1.500	1.500	1.650	1.650	1.650	1.650
tCL	9	9	9	9	11	10	11
tRCD	9	9	10	11	12	12	13
tRP	9	9	9	10	12	12	13
tRAS	24	24	28	28	30	31	35
tRC	33	33	37	38	47	43	48
tWR	10	12	14	16	14	16	16
tRRD	4	5	5	6	6	7/6	7
tRFC	107	128	150	171	362	313	214/213
tWTR	5	6	8/7	9/8	8	10/9	10
tRTP	5	6	8/7	9/8	8	10/9	10
tFAW	20	24	24	25	27	26	27
tCWL	-	7	7	7	11	7	7
CR	-	2	2	2	2	2	2

The big change over previous kits is the tRFC, the Row Refresh Cycle time.  tRFC is defined as the minimum wait in cycles required following a refresh to an idle bank before it can be again activated for access.  This should mean that reads from subsequent rows requiring a refresh will be faster compared to the 2400 C10/C11 kits we have tested.

Visual Inspection

This TridentX kit is essentially the same as the previous TridentX 2400 C10 kit we reviewed - TridentX is one notch above RipjawsZ, and spans kits from DDR3-2400 C9 to DDR3-2800 C11.  The main features on the kit (aside from the speeds) are the heatsinks, to which G.Skill have added a detachable fin.  Without the fin, the module is approximately 9mm above the module, and with the fin the total height is 22mm above the PCB.  That is a lot of height for a memory module that in 99% of circumstances would not produce enough temperature to trouble any build.

In order to remove the fin there is a screw at each end of the module, and the fin slides off effortlessly.  The fins fit very well, but upon attempting to reattach a fin I was unable to get it on as securely as it came out of the box, leaving a little wobble in the fin.  There is no cause for alarm if you get a module with a wobbly fin – nothing is wrong, and it will not affect the heat dissipation as much as most users may think.  Most modules output a few watts at best, so dissipation of several watts of energy without a fin is simple enough.

One thing I should point out with the TridentX modules is the fins are actually very sharp – if you are trying to put these modules into a motherboard where the range of movement to apply pressure is small, wear some gloves.  I actually did take some skin off my hand doing exactly this, trying to use my fingers and palm to push the modules down.

As with the previous TridentX 2400C10 kit, putting such a large module in our TRUE Copper scenario was a recipe for disaster:

Even putting the module in the second slot faired no better:

All the testing for this review was done on an ASUS P8Z77-V Premium motherboard with the Intel stock cooler, so module movement is not much an issue in that case, but big air coolers still get a lot of usage (and is more often than not an investment over several updates), so there are things to consider when purchasing memory.

 

Market Positioning

Despite the price and the market for this memory kit, G.Skill still has price competitive challengers. At the time of writing, here was the state of Newegg, starting with the 2600+ 4x4GB kits, then the 2600+ 2x4GB kits, and finally the 4x4 GB kits in the same price range of the kit being tested today.

$600: Corsair Dominator DDR3-2800 11-14-14 4x4GB Kit
$500: Corsair Dominator DDR3-2666 12-14-14 4x4GB Kit
$360: Corsair Dominator DDR3-2666 10-12-12 4x4GB Kit
$360: Corsair Dominator DDR3-2600 10-12-12 4x4GB Kit

$600: G.Skill TridentX DDR3-2800 11-13-13 2x4GB Kit
$270: Avexir Core DDR3-2800 12-14-14 2x4GB Kit
$180: Kingston HyperX Predator DDR3-2666 11-13-13 2x4GB Kit
$170: G.Skill TridentX DDR3-2666 11-13-13 2x4GB Kit
$168: Team DDR3-2600 10-12-12 2x4GB Kit
$165: G.Skill TridentX DDR3-2600 10-12-12 2x4GB Kit
$160: Corsair Dominator DDR3-2666 11-13-13 2x4GB Kit

$180: Team Vulcan DDR3-2400 10-11-11 4x4GB Kit
$170: Corsair Vengeance DDR3-2133 9-11-10 4x4GB Kit
$170: Team DDR3-2400 11-11-11 4x4 GB Kit
$155: G.Skill TridentX DDR3-2400 10-11-11 2x8 GB Kit
$150: Kingston HyperX DDR3-2400 11-13-13 4x4GB Kit
$150: Corsair DDR3-2400 10-12-12 4x4GB Kit

As we can see, the more you ramp up the MHz numbers, the price of the kit rises exponentially at this end of the market.  There are lots of crazy numbers in that list, for minute increases in daily performance.  In the 2x4 GB space at that price range, the Corsair Dominator package at $160 seems to offer the same speed but at a cheaper price (overclocking not compared).  Going down to the 4x4 GB range and we see a set of 2400 C10/C11 modules offering almost the same performance and double the amount of memory at the same price.

A quick look at all the 2x4 GB kits available on Newegg gives the following table:

	2x4 GB Memory Kits
	1600	1866	2000	2133	2400	2600	2666	2800
CL 7	$73 Corsair $70 G.Skill $70 Mushkin $65 Mushkin
CL 8	$60 Crucial $55 Corsair $55 G.Skill $55 Crucial $55 Mushkin $53 G.Skill $53 Crucial $52 Mushkin $52 Corsair	$75 G.Skill
CL 9	$86 Corsair $48 G.Skill** $47 Kingston $47 Corsair** $46 Avexir $45 G.Skill** $45 Kingston $45 Avexir** $44 Patriot** $43 Kingston** $42 Corsair $42 ADATA $42 Kingston $42 Mushkin** $41 Corsair $41 Kingston $41 Patriot and below	$90 Corsair $75 Corsair $60 Mushkin $60 Avexir $55 G.Skill $55 Kingston $55 Crucial $55 Mushkin $53 Crucial $53 Kingston $52 Corsair $52 PNY $50 Corsair $50 G.Skill $50 Kingston $50 Patriot $50 Crucial $50 Team $49 Corsair $49 G.Skill $48 GeIL $47 Patriot $46 Kingston $44 Corsair	$75 Corsair $68 Mushkin $65 Avexir $60 Corsair $60 Mushkin	$130 Corsair $120 Corsair $110 Corsair $70 Corsair $70 Avexir $65 Mushkin $64 G.Skill $62 Team $61 G.Skill $60 G.Skill $60 Mushkin
CL 10				$85 PNY $60 Mushkin $56 Team	$100 Avexir $85 Mushkin $82 Team $77 Corsair $75 G.Skill $72 G.Skill $70 G.Skill	$168 Team $165 G.Skill
CL 11	$45 Samsung	$49 Kingston*		$58 Corsair $57 Patriot $56 Corsair $55 Team $50 G.Skill $50 Kingston $50 Patriot $50 Team $50 GeIL	$103 G.Skill $95 Team $77 Patriot $75 GeIL $60 Kingston		$180 Kingston $170 G.Skill $160 Corsair $160 Avexir	$600 G.Skill
CL 12								$270 Avexir
* Plug and Play ** Low Voltage

Test Bed

Test Bed
Processor	i7-3770K @ 4.4 GHz 4 Cores / 8 Threads
Motherboard	ASUS P8Z77-V Premium
Memory	G.Skill 1333 MHz 9-9-9-24 1.5V 4x4GB Kit G.Skill 1600 MHz 9-9-9-24 1.5V 4x4GB Kit G.Skill 1866 MHz 9-10-9-28 1.5V 4x4GB Kit GeIL 2400 MHz 11-12-12-30 1.65V 2x8GB Kit G.Skill 2133 MHz 9-11-10-28 1.65V 4x4GB Kit G.Skill 2400 MHz 10-12-12-31 1.65V 4x4GB Kit G.Skill 2666 MHz 11-13-13-35 1.65V 2x4GB Kit
CPU Cooler	Intel Stock Cooler
Graphics Cards	Intel HD4000 ECS GTX580
Power Supply	Rosewill SilentNight 500W Platinum
Storage	OCZ Vertex3 240GB
SATA 6Gbps to USB 3.0	Thermaltake BlacX 5G Docking Station
Thunderbolt Device	Lacie Little Big Disk 240GB
Test Bench	Coolermaster Test Bed
Operating System	Windows 7 x64 Ultimate

Many thanks to...

We must thank the following companies for kindly donating hardware for our test bed:

OCZ for donating the USB testing SSD
ASUS for donating the IO testing kit
ECS for donating NVIDIA GPUs
Rosewill for donating the Power Supply

ASUS MemTweakIt

With our overview of the ASUS Republic of Gamers range of products, one piece of software caught my eye while I was testing.  The ASUS MemTweakIt allows for almost complete control of the memory subtimings while in the OS, such that users can optimize their settings for memory reads, memory writes, or for pushing the boundaries.  The upshot of this software in our context is that it takes all the sub-timings and settings and condenses them into a score.  As the memory kits we test contain XMP profiles, these profiles determine a large majority of the sub-timings on the kit and how aggressive a memory manufacturer is.  We should see this represented in our MemTweakIt score.

As we do not know the formula by which ASUS calculates this value, it has to be taken with a pinch of salt.  It could be weighted in favor of one of the settings versus the other.  Normally I would not put such an non-descript benchmark as part of our testing suite, but the MemTweakIt software does give us one descriptor – it gives us a theoretical rate of improvement across the range of kits we test, and allows us to order them in the way they should perform.  With this being said, the results for our kits are as follows:

Compared to our previous kits tested, we were under no illusion that a 2666C11 kit would take top spot.  But what this also means is that MemTweakIt does not differentiate between one memory stick per channel or two – it should however differentiate between single and dual channel orientations.  Note that according to the scores, we have ~20% increase in MemTweakIt score over a 4x4GB 1333 C9 kit for a whopping 127% increase in price (as well as half the capacity).

Metro2033

Metro2033 is a DX11 benchmark that challenges every system that tries to run it at any high-end settings.  Developed by 4A Games and released in March 2010, we use the inbuilt DirectX 11 Frontline benchmark to test the hardware at 1920x1080 with full graphical settings.  Results are given as the average frame rate from 4 runs.

Our Metro2033 IGP results seem to mirror the expectations shown from the MemTweakIt results, albeit by a factor of a 1/2 (MemTweakIt predicts 20% gain over 1333 C9, we only see a 10% gain).

Civilization V

Civilization V is a strategy video game that utilizes a significant number of the latest GPU features and software advances.  Using the in-game benchmark, we run Civilization V at 1920x1080 with full graphical settings, similar to Ryan in his GPU testing functionality.  Results reported by the benchmark are the total number of frames in sixty seconds, which we normalize to frames per second.

Despite the expected difference the 2666 C11 kit should give according to MemTweakIt and Metro2033, we actually fall behind the 2400C10 kit by a very small margin.  Overall however we are only getting a ~6.7% gain in frame rates over 1333 C9 for that 127% increase in price.

Dirt 3

Dirt 3 is a rallying video game and the third in the Dirt series of the Colin McRae Rally series, developed and published by Codemasters.  Using the in game benchmark, Dirt 3 is run at 1920x1080 with Ultra Low graphical settings.  Results are reported as the average frame rate across four runs.

Dirt 3 results seem a little at odds with what we have expected – moving from 2133 C9 to 2400 C10 caused a decrease in frame rates, whereas 2133 C9 to 2666 C11 caused an increase.  Nonetheless the biggest gains over 1333 C9 seem to be with 1600 C9 and 1866 C9.  Any memory kit above that is a miniscule raise at best.

Portal 2

A stalwart of the Source engine, Portal 2 is the big hit of 2011 following on from the original award-winning Portal.  In our testing suite, Portal 2 performance should be indicative of CS:GO performance to a certain extent.  Here we test Portal 2 at 1920x1080 with High/Very High graphical settings.

With Portal 2 on IGP we seem to hit a ceiling very early on in our testing, around the 2400C11/2133C9 range.  After this the frame rate levels out, suggesting that something else in the system is the limiting factor.

Batman Arkham Asylum

Made in 2009, Batman:AA uses the Unreal Engine 3 to create what was called “the Most Critically Acclaimed Superhero Game Ever”, awarded in the Guinness World Record books with an average score of 91.67 from reviewers.  The game boasts several awards including a BAFTA.  Here we use the in-game benchmark while at the lowest specification settings without PhysX at 1920x1080.  Results are reported to the nearest FPS, and as such we take 4 runs and take the average value of the final three, as the first result is sometimes +33% more than normal.

Batman: AA is still the main beneficiary of faster memory speed in our IGP testing, and the jump from 2400 C10 to 2666 C11 does seem to make a difference (even if it is only 0.75 fps out of 58.25).

Overall IGP

Taking all our IGP results gives us the following graph:

The only kit that gets the performance increase predicted by MemTweakIt is Batman:AA.  There is no doubt that out of the kits tested that the 2666 C11 2x4 GB kit performs the best, but it is hard to argue for the kit that costs twice per GB than the other kits except in an extreme overclocker scenario.

Conversion – Xilisoft 7

Another classic example of memory bandwidth and speed is during video conversion.  Data is passed between the memory and the CPU for processing – ideally faster memory here helps as well as memory that can deal with consecutive reads.  Multiple threads on the CPU will also provide an additional stress, as each will ask for different data from the system. If we combine this with the capability of using GPUs that Xilisoft Video Converter 7 allows, we can attempt to really tax the memory.  Our test consists of converting 26GB of various sized videos (1080p to VGA), totalling 45+ hours of video time, to MP2 format using our test bed.

In both of our conversion tests there seems to be a hard limit at which the CPU becomes more of the limiting factor – as the CPU cannot process data quickly enough, any copy times or latency to the memory are hidden by the lack of CPU power.  As a result, we do not see any speed up beyond 1600 C9/1866 C9.

Folding on GPU

Memory usage is all algorithm dependent – if the calculation has a lot of small loops that do not require additional reads memory, then memory is unimportant.  If the calculation requires data from other sources in those calculations, then memory can either be stressed randomly or sequentially.  Using Ryan’s Folding benchmark as a platform, we are testing how much memory affects the serial calculation part of a standard F@H work unit.

As before, the serial copying to the GPU, or CPU specific code, is not affected by memory speed when GPU Folding.

WinRAR x64 4.20

When compressing or converting files from one format to another, the file itself is often held in memory then passed through the CPU to be processed, then written back.  If the file is larger than the available memory, then there is also loading time between the storage and the memory to consider.  WinRAR is a variable multi-threaded benchmark, whereby the files it converts and compresses determines how much multi-threading takes place.  When in multithreaded mode, the rate of cache misses can increase, leading to a less-than optimal scaling.  Having fast memory can help with this.

Our WinRAR test remains the standout test in terms of what memory can improve.  WinRAR deals a lot with memory accesses, so having that faster memory can help with compressing data.  As we see, the 2666C11D has an advantage over the 2400C10Q by 22 seconds, or around 5%.  While WinRAR may not follow exactly the proportions that the MemTweakIt software predicts, it has shown the order in which these kits should perform.

Greysky's x264 HD 5.0.1

The x264 HD test, now version 5.0.1, tests the time to encode a 1080p video file into a high quality x264 video file.  This test is standard across a range of reviews from AnandTech and other websites allowing for easy comparison.  The benchmark is capable of running all cores to the maximum.  Results here are reported as the average across four attempts for both the first and second passes.

For our x264 test, the 2666 C11 does come out top, albeit by the smallest of margins.  Moving from 1333 C9 we are still only getting a 6% increase in frame rates – similar scores are achieved with 2133 C9.

Synthetic testing has a way of elevating what may be a minor difference between hardware into a larger-than-life comparison, despite the effect on the usage of the system being near minimal.  There are several benchmarks which straddle the line between synthetic and real world (such as Cinebench and SPECviewperf) which we include here, plus a couple which users at home can use to compare their memory settings.

SPECviewperf

The mix of real-world and synthetic benchmarks does not get more complex than SPECviewperf – a benchmarking tool designed to test various capabilities in several modern 3D renders.  Each of these rendering programs come with their own coding practices, and as such can either be memory bound, CPU bound or GPU bound.  In our testing, we use the standard benchmark and report the results for comparison.

SPECviewperf gracefully provides us with a mixed bag of results.  For the majority of situations where the 2666C11 kit is on top (Lightwave, Maya, TCVis, snx), the kit is only just above the nearest challengers, meaning in each case the 2133 C9 kit still makes sense.  From the point of view of the other renderers (especially catia), the 2133 C9 represents a happy medium over all the kits provided.

Cinebench x64

A long time favourite of synthetic benchmarkers the world over is the use of Cinebench, software designed to test the real-world application of rendering software via the CPU or GPU.  In this circumstance we test the CPU single core and multi-core performance, as well as the GPU performance using a single GTX 580 at x16 PCIe 2.0 bandwidth.  Any serial factors have to be processed through the CPU, and as such any memory access will either slow or speed up the benchmark.

While nothing significant happens on the CPU side when using 2666 C11 memory, the IGP OpenGL portion of Cinebench gives our highest result, albeit with a less than 4% improvement over 1333 C9.

USB 3.0 + Thunderbolt Backup

Our backup testing takes a typical set of user files – specifically just under 8000 files across 7.5 GB, some large files but mostly small.  For USB 3.0 testing, these files are copied from our SSD onto an OCZ Vertex3 which is connected via a SATA 6 Gbps to USB 3.0 device using the UASP protocol that the ASMedia controller on our test bed affords as well as the chipset driven Intel USB 3.0 under ASUS' Turbo mode.  The copy test is conducted using DiskBench, a copying tool with accurate copy timing.

Similar to our USB 3.0 Backup test, Thunderbolt testing carries the same files directly through to our LittleBig Disk which contains two 120 GB Intel SSDs in RAID-0. 

For all the USB 3.0 copy tests, we see that the movement from 1333 C9 to 1600 C9 gives a distinct gain, but above this not much gain is to be had – only small differences (e.g. USB 3.0 with Intel + Turbo).  However, our Thunderbolt test causes a big spike with our G.Skill 2666 C11 kit.  This is because of how our test is set up:

- Create 4 GB RAM Disk, populate with data (to eliminate read speed limitations of SSDs)
- Use DiskBench to copy from RAMDisk to Thunderbolt Device

The way Thunderbolt works is to copy the data from the source into the memory, then onto the device.  Due to the 2666 C11 kit only being 8 GB total, there is not enough space for the RAMDisk and a copy of the RAMDisk’s contents to be stored in memory.  This means the system has to copy as much as it can to begin with, wipe the copy from the memory, then transfer the rest into memory for moving onto the device.  This adds latency.

This is technically how the USB 3.0 test works as well too, but the USB 3.0 test seems to clean up as it copies in order to hide copy-from-source-to-memory latency.

When it comes to memory overclocking, there are several ways to approach the issue.  Typically memory overclocking is rarely required - only those attempting to run benchmarks need worry about pushing the memory to its uppermost limits.  It also depends highly on the memory kits being used - memory is similar to processors in the fact that the ICs are binned to a rated speed.  The higher the bin, the better the speed - however if there is a demand for lower speed memory, then the higher bin parts may be declocked to increase supply of the lower clocked component.  Similarly, for the high end frequency kits, less than 1% of all ICs tested may actually hit the speed of the kit, hence the price for these kits increase exponentially.

With this in mind, there are several ways a user can approach overclocking memory.  The art of overclocking memory can be as complex or as simple as the user would like - typically the dark side of memory overclocking requires deep in-depth knowledge of how memory works at a fundamental level.  For the purposes of this review, we are taking overclocking in three different scenarios:

a) From XMP, adjust Command Rate from 2T to 1T
b) From XMP, increase Memory Speed strap (e.g. 1333 MHz -> 1400 -> 1600)
c) From XMP, decrease main sub-timings (e.g. 10-12-12 to 9-11-11 to 8-10-10)

There is plenty of scope to overclock beyond this, such as adjusting voltages or the voltage of the memory controller.  As long as a user is confident with adjusting these settings, then there is a good chance that the results here will be surpassed.   There is also the fact that individual sticks of memory may perform better than the rest of the kit, or that one of the modules could be a complete dud and hold the rest of the kit back.  For the purpose of this review we are seeing if the memory out of the box, and the performance of the kit as a whole, will work faster at the rated voltage.

In order to ensure that the kit is stable at the new speed, we run the memory test within OCCT for five minutes.  This is a small but thorough test, and we understand that users may wish to stability test for longer to reassure themselves of a longer element of stability.  However for the purposes of throughput, a five minute test will catch immediate errors from the overclocking of the memory.

With this in mind, the kit performed as follows:

F3-2666C11D-8GTXD – 2x4 GB rated at DDR3-2666 11-13-13-35 2T 1.65 volts

Adjusting from 2T to 1T: Passes Linpack
Adjusting from 2666 to 2800: Passes Linpack
Adjusting from 2800 to 2933: No Boot
Adjusting from 11-13-13 to 10-12-12: Fails Linpack

It should be noted that overclockers are getting very good results from these high end G.Skill kits – the fact that I was able to push this kit from 2666 C11 to 2800 C11 with nothing more than changing the memory strap is great.  G.Skill inform me that a lot of these kits will do 10-12-12 with a small voltage increase as well, taking performance a stage higher.  In fact, as part of my hobby of competitive overclocking, I have got this memory to 2950 MHz C11-13-13 by pushing some more voltage into the kit.

I wonder if this means that there may be recourse for manufacturers to release kits with a 1.75 volt profile, especially if it pushes the performance a little.  Then again, it is still debatable if the majority of users will see the performance increase.

Conclusions

Trying to argue the case for a 2x4 GB DDR3-2666 C11 kit for $170 is a monumental task.  4x4 GB kits of 2400 C10 offer double the capacity at a similar price point with no real change in real-life performance.  Arguably neither does 4x4 GB 2133 C9 which is even cheaper, or 2x8 GB 2400 C11 which offers the same amount of modules but double the density.

From our benchmarking results, the 2666 C11 was predicted by MemTweakIt to come out on top, and in all the benchmarks we have that show results, it did, often by the smallest margins.  The biggest difference seems to be with respect to WinRAR, where for our 450+ second test the 2666 C11 kit performed 22 seconds faster than a 2400 C10 kit.  If that 5% change can be justified in the price for what you do, then this kit could be for you.

The only other placement for this kit is really in the hands of extreme overclockers.  I am a competitive overclocker – it is one of my hobbies that I find fascinating, even if it is only trying to get a number on a screen higher than the next guy.  But overclocking is like drag racing for computers – there are very few people in the market to purchase these kits compared to the majority of users that need memory.  As a kit for overclockers, it works quite well, hitting 2800 C11 without a bump in voltage, and I was able to boost this to 2950 C11 with a bump in voltage.  No doubt with some more knowledge and tweaking this could break the 3000 MHz barrier.

The immediate future for memory still lies in DDR3 – the next iteration, DDR4, is still several years away.  We are told that on the Intel side of things, Haswell is DDR3, as will be Broadwell, the Haswell replacement.  Reports expect DDR4 to be less than 10% of the market in late 2014 (early adoption in the high end space), but 50%+ across 2015.  DDR4 is expected to have a base speed of 2133 MHz up to 3200 MHz for initial enthusiast applications – though given the rise in enthusiast speeds this could seemingly be pushed to 4266 MHz+ over the course of the development cycle.  DDR4 is also expected to be a single module per channel, paving the way for up-to-quad channel in the mainstream arena.

But for now, if synthetics matter, then the 2x4 GB F3-2666C11D-8GTXD 2666 C11 TridentX kit offers some overclocking headroom.  All other increases (at XMP) are marginal at best from previous kits.  Kits like this will become more commonplace when processors adjust the tested IMC rated speed – the high end Trinity processors are rated at 1866 MHz minimum, so it only needs to raise another 1000 MHz or so.  This should hold true in 2015 when DDR4 is part of the desktop hierarchy.

If you are on the lookout for a 2x4 GB memory kit, as of 10/18, here is what Newegg looked like:

	2x4 GB Memory Kits
	1600	1866	2000	2133	2400	2600	2666	2800
CL 7	$73 Corsair $70 G.Skill $70 Mushkin $65 Mushkin
CL 8	$60 Crucial $55 Corsair $55 G.Skill $55 Crucial $55 Mushkin $53 G.Skill $53 Crucial $52 Mushkin $52 Corsair	$75 G.Skill
CL 9	$86 Corsair $48 G.Skill** $47 Kingston $47 Corsair** $46 Avexir $45 G.Skill** $45 Kingston $45 Avexir** $44 Patriot** $43 Kingston** $42 Corsair $42 ADATA $42 Kingston $42 Mushkin** $41 Corsair $41 Kingston $41 Patriot and below	$90 Corsair $75 Corsair $60 Mushkin $60 Avexir $55 G.Skill $55 Kingston $55 Crucial $55 Mushkin $53 Crucial $53 Kingston $52 Corsair $52 PNY $50 Corsair $50 G.Skill $50 Kingston $50 Patriot $50 Crucial $50 Team $49 Corsair $49 G.Skill $48 GeIL $47 Patriot $46 Kingston $44 Corsair	$75 Corsair $68 Mushkin $65 Avexir $60 Corsair $60 Mushkin	$130 Corsair $120 Corsair $110 Corsair $70 Corsair $70 Avexir $65 Mushkin $64 G.Skill $62 Team $61 G.Skill $60 G.Skill $60 Mushkin
CL 10				$85 PNY $60 Mushkin $56 Team	$100 Avexir $85 Mushkin $82 Team $77 Corsair $75 G.Skill $72 G.Skill $70 G.Skill	$168 Team $165 G.Skill
CL 11	$45 Samsung	$49 Kingston*		$58 Corsair $57 Patriot $56 Corsair $55 Team $50 G.Skill $50 Kingston $50 Patriot $50 Team $50 GeIL	$103 G.Skill $95 Team $77 Patriot $75 GeIL $60 Kingston		$180 Kingston $170 G.Skill $160 Corsair $160 Avexir	$600 G.Skill
CL 12								$270 Avexir
* Plug and Play ** Low Voltage

Please note this table does not take into account warranty or any 'extras' or sub-timings that may come from each kit.