What's new
  • Please do not post any links until you have 3 posts as they will automatically be rejected to prevent SPAM. Many words are also blocked due to being used in SPAM Messages. Thanks!

G.Skill PI 2x2GB DDR3 PC3-12800 Review

Status
Not open for further replies.

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
<img src="http://images.hardwarecanucks.com/image/akg/Storage/GSKILL_SSD/gskill_logo-1.png" border="0" alt="" />


<center><b>G.Skill PI 2x2GB PC3-12800 Review</b></center>



<b>Price:</b> <a href="http://www.ncix.com/products/index.php?sku=31478&vpn=F3%2D12800CL7D%2D4GBPI&manufacture=G%2ESKILL">$315 @ NCIX.com</a>
<b>Manufacturer:</b> <a href="http://newgskill.web-bi.net/bbs/G_SKILL/">G.Skill</a>
<b>Manufacturer's Part Number:</b> F3-12800CL7D-4GBPI
<b>Warranty:</b> Lifetime Warranty



<p style="text-align: justify;">With growing availability, lower prices, and more features becoming available not found on previous generation hardware; DDR3 memory is really starting to get talked about. There is more discussion than ever in enthusiast forums and DDR3 terminology has started making its way into general computer hardware stores all over the world. The steps being taken right now are those required in order to get DDR3 into the primary spot in the mainstream memory market. The simple fact that we have been reviewing DDR3 memory for months and months now confirms this gradual shift. As of late though, 2x2GB DDR3 kits seem to be all the rage with compatibility on the various chipsets being what is primarily talked about these days.

G.Skill has recognized this need for high performance 2x2GB kits of DDR3 memory and has sprinkled the pot with a PC3-12800 4GB kit with tight timings of 7-7-7. DDR3 memory frequencies have really started taking off but at much looser timings. The DDR3-1600 and 7-7-7 timing combination really is a steadfast combo that has been common for 2x1GB kits but is a little harder to find in 2GB density sticks. To top off these rather impressive specifications, G.Skill has added in a set of their new PI heat spreaders.

Our last G.Skill review left nothing for the imagination in the performance and overclocking world, breaking records and doing things with bandwidth that have not been duplicated since in the overclocking community. This 2x2GB kit of memory is obviously going to go in a different direction. We are still going to take a quick look at overclocking but we are primarily going to focus on how easy it is to get this kit running stable on the Intel X48 platform. All of the recent forum discussion and e-mails we receive here have been focused on getting 2x2GB kits of memory to simply run their specifications without issue, so we have decided to focus on that with this memory.</p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/index-1.jpg"></center>
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Package & Memory Overview

<b>Package & Memory Overview:</b>
<p style="text-align: justify;">We are greeted with yet another design from G.Skill both in graphics and package size. There is no race car on the front of this package and it is very simple yet at the same time there is an aura of excitement wrapping itself around this light blue package and I for one cannot wait to get inside.</p><center>
</center><p style="text-align: justify;">The very slim but rather large package contains nothing more than a gradient around the lower half and the G.Skill & PI logos annotated with "Where Speed Is!". The backside of the book size package is as simple as the front with a sticker outlining specifications and UPS code for retailers. The package is a bit bigger than most memory packages but that is mostly due to the larger modules inside that the package has to accomodate.</p><center>
</center><p style="text-align: justify;">The front flap also opens like a book providing visual access to the actual modules and our first look at the new PI heat spreaders. We can see that the memory is enclosed in a plastic clam shell to protect it and that the package is merely decorative. The inside flap contains the familiar thermal image of with and without heat spreaders. We hate to be skeptical but it looks more like a 'graphical representation' than an actual thermal image. I don't want to say G.Skill is lying but they show an image of two modules that appear to have exact heat signatures, not to mention the fact that the images simply looked created and not actual images. Either way, we have known for years that heat spreaders help distribute the temperature more evenly across the module as the image depicts.</p><center>
</center><p style="text-align: justify;">The plastic clam shell G.Skill uses appears to be similar to what other manufacturers are using and what G.Skill has provided in the past for protection during transport. This plastic clam shell works well and provides ample protection for the modules. It seals tightly at all four corners and can act as a carrying case or storage case by itself after opening the package.</p><center>
</center><p style="text-align: justify;">The contents of the F3-12800CL7D-4GBPI package are about the same as what we received with the F3-12800CL7D-2GBHZs. There really is no need for an accessory package and G.Skill doesn't provide one. They simply include a G.Skill sticker and a small leaflet with instructions as well as support contact information. The support contact info is really nice to have for those that don't know how to search on Google, and believe it or not, there are plenty out there that need something included in the package with a web site address or e-mail address for support.</p><center>
</center><p style="text-align: justify;">We weren't sure what to expect based on the advertising G.Skill had done with the PI heat spreaders. We were assuming something totally revolutionary was going to come out of the package but we were let down a little bit with that expectation. The PI heat spreaders appear to be just a basic aluminum heat sink with small decorative extensions at the top that help increase cooling surface area. I wouldn't expect the cooling loops at the top to really provide any additional cooling abilities that other solutions don't already offer. So calling the PI heat spreader revolutionary would be rather ambitious of us. Instead, we can call the PI heat spreaders a very smart looking light-weight heat sink that aids in cooling of the memory ICs but doesn't push the memory cooling envelope through innovation as we had hoped.</p><center>
</center><p style="text-align: justify;">Is the heat sink an upgrade from the previous generation of G.Skill memory? I would have to say yes but of course can't really back up those claims with actual evidence due to lack of thermal imaging equipment or other high-tech devices. The additional cooling of the decorative loops probably does aid in the cooling abilities of these new heat sinks, but I would imagine the improvements in thermal material between the ICs and the heat sink is what is really going to benefit these modules from G.Skill modules of past. In the second image above, we are looking at the ends of the modules with a macro lens on the camera. If you look at the outside of the left module, you can see the square edge of the IC and where it contacts the heat sink. You will notice no big thick thermal pad, instead, there is just a small clear enamel-like substance. This is the thermal adhesive and is similar to the likes of Corsair and what they have been using for years now. When we first received our modules, one of the heatsinks was slightly lifted from the ICs but with a little bit of pressure we were able to sit it down and it has stuck well since. Because of this we don't believe the thermal material is cured to the modules but it is a step in the right direction for G.Skill.

We are now going to have a look at the specifications of these modules.
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Specifications

<b>Specifications:</b>
<p style="text-align: justify;">As mentioned, we changed up the order of the review in order to accommodate a short discussion about the new PI heatspreader from G.Skill that is present on these modules. We preface by mentioning that we have handled dozens and dozens of memory modules over the years and have experience with near all of the heat sinks/heat spreaders that have come, gone, and stuck around. We just looked at the form of the PI heat spreaders but now let’s look at the function.</p><center>
</center><p style="text-align: justify;">Upon receiving the G.Skill memory we quickly photographed the package so that we could get at the modules inside. More importantly, the PI heat spreaders that provide the cooling to the memory modules underneath. <img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/specs-1.png" alt="G.Skill F3-12800CL7D-4GBPI Specifications" border="0" style="float: right; margin: 4px 0px 4px 7px";>It was immediately evident that the heat spreader was, none the less sharp looking, nothing more than an aluminum heatsink with a bit of decoration up top that does perform some purpose. At the heart of the cooling where the contact happens, we simply have a flat basic aluminum surface mating to an IC. There is nothing different about this heat sinks mount, or is there? Looking at the mounting adhesive, we see something quite different from G.Skill thermal adhesive in the past. Previous G.Skill modules are still clipped at the top and use a thermal pad much like the original heat spreaders were designed.

It is impressive to see a newer more effective thermal adhesive being used with the PI heat spreaders but what is more impressive are the specifications to the right. G.Skill has put together a very nice looking kit here with 7-7-7-18 timings putting the tRAS of 18 among the tightest of any 2x2GB kit we have seen at this frequency. G.Skill has confirmed compatibility with Intel chipsets going back as far as Intel P35 right up to the newest P45 as well as NVIDIA 780i/790i chipsets. We will be running this kit on the Asus P5E3-Premium running the Intel x48 chipset.

One thing worth mentioning here that should be listed on the G.Skill web site, but isn't, is the fact that these modules come with XMP profiles for DDR3-1600 7-7-7. This makes for fantastically easy overclocking of an Intel XMP compatible motherboard to get these modules running the specified timings and frequencies. Our previous kits of XMP memory has been very positive so hopefully we will be as fortunate with this kit and the XMP profiles. It is still kind of a shock that G.Skill makes no mention of the XMP compatibility on their web site with this kit.

Before we get into whether or not the G.Skill PC3-12800 2x2GB kit of memory performs stable at the XMP profile, we have to see how they physically fit. Normally this section is a non-issue we breeze over but with the decorative yet functional loops at the top of the heat spreaders, it also provides additional height from standard modules. Let's see if this affects things in the P5E3-Premium, as well as the rest of the hardware we will be using for testing.</p>
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Memory Installation & Test Setup

<b>Memory Installation & Test Setup:</b>
<p style="text-align: justify;">As we have both seen and talked about, the PI heat spreaders on this G.Skill kit are taller than standard heat sinks and will play a role in some setups. We are going to check to see if they play a role in the P5E3-Premium with a Thermalright Ultra-120 CPU cooler. The Ultra-120 is the same size as the newer Ultra-120 Extreme and fits the same motherboards. This heat sink is very popular amongst enthusiasts and is a must test fit for any memory that potentially causes issues.</p><center>
</center><p style="text-align: justify;">Right out of the gate we can see that there is no issue at all with the Ultra-120s wingspan. The primary reason for this is the fact that ASUS has done well to design the first memory slot not to interfere with CPU coolers as big as or smaller than the Ultra-120 from Thermalright.</p><center>
</center><p style="text-align: justify;">The only other fitment issue would be width and these modules despite their elaborate hardware, are no thicker than standard memory modules. They clearly have no problem working in the first slot of the P5E3-Premium and do not interfere with the stock motherboard cooling. These modules also fit in slots side by side without issue for 4x2GB action.</p><center>
</center><p style="text-align: justify;">After installing the memory and getting our system configured to BIOS defaults through a CMOS clear, the XMP profile kicked in like it should and the system immediately booted at 400FSB with a 1:2 ratio so the memory runs at 800MHz or DDR3-1600 with 7-7-7-18 timings. The second SPD tab shown illustrates this XMP profile in action. The P5E3-Premium again recognized that we had XMP memory and without so much as even our input, the system boots at an overclocked setting in order for the memory to run spec. Our overclocking section will determine if the XMP profile is stable or not which is probably more important than whether or not it kicks in.</p><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/setup-1.jpg" style="float: right; margin: 4px 0px 5px 20px"><b>Test Platform:</b>
  • <b>Motherboard:</b> ASUS P5E3-Premium (BIOS 0503)
  • <b>Processor:</b> Intel C2D E8400
  • <b>Processor Cooling:</b> Thermalright Ultra-120
  • <b>Memory:</b> G.Skill PI 2x2GB PC3-12800 7-7-7-18
  • <b>Power Supply:</b> Thermaltake Toughpower 700W
  • <b>Video Card:</b> HIS HD3870X2 512MB
  • <b>Additional Fans:</b> 120mm AD1212MS-A73GL 2050RPM/80.5CFM
  • <b>Hard Drive:</b> 1 x Seagate 7200.9 80GB SATAII 8MB cache
  • <b>OS:</b> Windows Vista Ultimate (with all updates)
<p style="text-align: justify;">The most important piece of hardware to note is the ASUS P5E3-Premium. This is an Intel X48 based motherboard and like we have already discovered, provides XMP support and makes running this memory at specifications truly plug and play. Without an XMP compatible motherboard, the system would likely need to be overclocked but in the case of the NVIDIA platform, that wouldn't be the case because of the un-locked memory ratios.

We talk about it in every review and with a 2x2Gb kit of memory, nothing changes. Intel XMP profiles work, and they work really well. The number one issue with memory in computer hardware forums is the in-ability to get memory to run specified frequencies and timings. With the purchase of memory like the G.Skill PI kit and an XMP compatible motherboard and you can forget all about it. It took a while to get to this point so some leg work was required to get where we are but Intel has definitely perfected the XMP memory profile and fulfilled what their vision was when the idea was first conceived.</p><center>
</center>
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Stability Testing & Overclocking

<b>Stability Testing & Overclocking:</b>
<p style="text-align: justify;">Our stability testing methods for memory have been discussed at length in the past, so we are going to simply rely on the explanation from previous reviews. Here is a run-down of what we consider to be stable.</p><b>Stability Testing Methodology:</b><p style="text-align: justify;">Memory stability, what constitutes stable? What is not considered stable? These questions get hotly debated in enthusiast forums all over the internet like little brush fires on the fringe of an inferno that play havoc with forest fire crews. Everyone has their own opinion about stability, especially when it comes to memory stability. For some, stable means they can do whatever it is on their computer without it crashing, blue-screening, or restarting; whether that means gaming or just surfing the internet. To this user, stable means simply using the computer as they normally would.</p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/setup-3.jpg" alt="" border="0"></center><p style="text-align: justify;">Of course, that is not what we would call stable and we do a lot more thorough testing before labeling a memory frequency/timing stable. One of the toughest types of programs on a system has been found to be distributed computing projects such as [email protected], [email protected], World Community Grid, and more. Running 24/7 crunching for one of these great causes is a sure way to find holes in a system if there is truly some instability, unfortunately it takes a considerable amount of time to use them for stability testing so we use the list of programs below to all but guarantee the system to be 24/7 distributed computing stable:</p><p style="text-align: justify;">Obviously this type of stability testing is for a 24/7 setup as mentioned but it is also very thorough testing a number of different programs that can find instability. We will also show lower levels of stability from time to time but will explicitly outline that we are only showing a specified level of stability. Keeping the above requirements in mind, let's find out if the G.Skill 4GBPI kit is going to give us a hassle at the XMP defaults.



<b>Specification Stability Testing:</b>
<p style="text-align: justify;">Coming into receiving this kit, the most important thing that seemed to be needed by readers was whether or not this memory was stable at the specified frequencies and timings. As well as which motherboards the memory is compatible with. 2x2GB kits of memory seem to really be pushing DDR3 memory controllers at frequencies that 2x1GB kits have no problem running. We will keep this in mind when testing all of our 2x2GB DDR3 memory kits. Here is the screen shot of this G.Skill PI PC3-12800 2x2GB kit running the XMP profile out of the box without so much as a hiccup.</p>
Click for full size screenshot...
<center>
</center><p style="text-align: justify;">As mentioned, there were no issues during testing these modules running their XMP profiles and performance was quite good. The XMP profiles would only select a Performance Level or tRD of 7 but the motherboard should easily be capable of PL6 at 400FSB and would like to see this added to the XMP profile. The screen shot above I actually set PL6 manually and had no issues throughout testing. Let's now talk about our battle with our ASUS P5E3-Premium and clocking these G.Skill sticks.</p>

<b>Stability Overclocking:</b><p style="text-align: justify;">We have to preface this section with the disclaimer that our ASUS P5E3-Premium doesn't like us at all. This review has been pushed to the absolute limits in trying to get this combination clocking but we have just run out of time. Normally, memory doesn't put up much a fight with the hardware we use but the recent addition of the ASUS P5E3-Premium left us in un-familiar ground with the motherboard and the results show. Naturally, we were able to squeeze a good bit out of this kit, even at very low voltages with just the motherboard holding us back at this point. Because of these difficulties, we will for-go the usual clocking and show a bit more fragmented version of what we have been able to achieve. Let's start with the first screenshot that we ran within days of getting the kit. At this point, this is still the highest fully stable screenshot we have.</p>
Click for full size screenshot...
<center>
</center><p style="text-align: justify;">At just above stock voltage, we were able to run the memory easily up to the 830MHz mark. This is with the same tight timings provided by the XMP profile. The excitement for this kit was there early but we were just unable to go any higher. Loosening up timings did nothing for progress, neither did BIOS changes. We did find that overclocking improved when moving from the black slots to the orange slots, but the same gains could be found increasing CLK skew with both being limited to around 880MHz or DDR3-1720 at 7-7-7 or any timing set really. Here is a screenshot of dual 32M SPi stability at 1.90v. Keep in mind, this is just dual 32SPi stable, it may be fully stable but it may also need to drop 10~15MHz for full stability.</p>
Click for full size screenshot...
<center>
</center><p style="text-align: justify;">As we can see, there is a lot of potential in this kit of memory, we just weren't able to un-lock it unfortunately with our motherboard. We also did some testing at tighter timings but lost a reasonable amount of frequency when trying to run at 7-7-6 or 7-6-6. Either way, these modules look to provide some decent headroom for those looking to go past 400FSB on the 1:2 ratio.
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Benchmarking Methodology

<b>Benchmarking Methodology:</b><p style="text-align: justify;">With the lack of overclocking results our benchmarking is going to be a bit slim this review. We have in the past compared numerous DDR3 speeds to various DDR2 speeds with combinations of timings and all testing has shown higher bandwidth for DDR3 but only marginal gains in programs and games with few exceptions. <img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/method-1.png" alt="Just a sample graph outlining where the results will be coming from in the up-coming benchmarks." style="float: right; margin: 4px 0px 4px 7px">Instead, we have decided to compare a 2x1GB kit of memory at the exact same timing set as our G.Skill PI 2x2GB kits XMP profile as well as the maximum stable overclock we were capable of at time of the benchmarking at 7-6-6 timings.

As always, the grey bar in the graphs <b>(1)</b> will indicate the rated frequency and timings of the G.Skill Pi 2x2GB kit of memory. These <b>grey</b> results will in fact be the XMP profiles that this system boots with on BIOS defaults. This means the ultra tight secondary timings programmed into this kit will kick in as well as the increases in the FSB. These features should provide very good performance, even at the stock frequency of DDR3-1600.

The middle blue results in the graphs, <b>(2)</b>, will portray the performance of a 2x1GB kit of memory at the exact same settings as the XMP profile of the G.Skill PI kit with one exception. The 2x1GB kit will run at 1T instead of the 2T that our 2x2GB kit is running at. This is because all 2x1GB kits at these timings and frequency should be able to run stock at 1T despite many being SPD programmed for 2T operation. Our 2x2GB kit just can't run at 1T which is again, a motherboard chipset limitation we believe.

The last red result in the graphs <b>(3)</b> is reserved for the 'non-stock' settings. At the time of the benchmarking, our highest overall overclock wasn't that high but we have since improved upon that with some headway made on the P5E3-Premium. As predicted before in the overclocking section, the motherboard has been heavily holding us back as it has been a battle every step of the way. We will be showing benchmarking results from the highest semi stable overclock at 7-7-7 we are capable of this kit and memory at this time. The specifications for this setup can be found in the Overclocking Section as the highest dual 32M Super Pi stable at DDR3-1720.

For all of the benchmarks, appropriate lengths are taken to ensure an equal comparison through methodical setup, installation, and testing. The following outlines our testing methodology:

a/ Windows is installed using a full format.
b/ Intel chipset drivers and accessory hardware drivers (audio, network, GPU) are installed followed by a defragment and a reboot.
c/ Programs and games are then installed followed by another defragment.
d/ Windows updates are then completed installing all available updates followed by a defragment.
e/ Benchmarks are each ran three times after a clean reboot for every iteration of the benchmark unless otherwise stated, the results are then averaged.

We have listed the benchmark versions on each graph as results can vary between updates.</p>
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Memory Benchmarks

<b>Memory Benchmarks:</b>

Everest Ultimate v4.50<p style="text-align: justify;"><i>Everest Ultimate is the most useful tool for any and all memory tweaking. Everest will reliably measure memory bandwidth of the system as well as latency. With this information, one can discover what is and is not improving bandwidth when tweaking or even just trying various timing and frequency combinations.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-1.png"></center><p style="text-align: justify;">The bandwidth numbers are really nice cracking 10K in Everest Read with the overclocked G.Skill PI kit. There isn't really much else to say, at these timings and frequencies, these are the types of bandwidth numbers DDR3 should be putting up.</p>


ScienceMark v2.0<p style="text-align: justify;"><i>ScienceMark is an almost ancient benchmark utility at this point in time and hasn't seen an update in a long time. It is, however, still a favorite for accurately calculating bandwidth on even the newest chipsets.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-2.png"></center><p style="text-align: justify;">The ScienceMark bandwidth and Everest copy result is a bit interesting with the 4GB kit out-pacing the 2GB kit at the same timings aside from the 2GB kit running at 1T. The larger density must be playing a role there.</p>


Futuremark PCMark Vantage<p style="text-align: justify;"><i>The latest in a long line of system benchmarking utilities, PCMark Vantage has quite the heritage and following in the overclocking and benchmarking community. For this memory review, we will simply be utilizing the memory benchmark portion of the software.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-3.png" alt="" border="0"></center><p style="text-align: justify;">The PCMark Vantage memories suite are pretty much what we would have predicted at these clocks but we weren't anticipating to see such gains shown to the 2GB kit over the 4GB at the XMP defaults. Switching between 1T and 2T does lower bandwidth slightly but 200 points in PCMark Vantage is way more than we were expecting.</p>


Everest Ultimate v4.50<p style="text-align: justify;"><i>Again, Everest Ultimate to the rescue in measuring memory latency. Like bandwidth, measuring latency can give us some insight into how the whole memory sub-system is performing.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-3.png" alt="" border="0"></center><p style="text-align: justify;">Like the bandwidth numbers, the latency was also lowered slightly when switching from the 4GB kit at 2T to the 2GB kit at 1T. This is 100% expected and confirmed with these results.</p>
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
System Benchmarks

<b>System Benchmarks:</b>

Super Pi Mod v1.5<p style="text-align: justify;"><i>When running the 32M benchmark of SPi, we are calculating Pi to 32 million digits and timing the process. Obviously more CPU power helps in this intense calculation, but the memory sub-system also plays an important role, as does the operating system. SPi 32M has been a favorite amongst benchmarks for these very reasons and is admittedly the favorite benchmark of this reviewer.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-5.png"></center><p style="text-align: justify;">In the benchmark results above, we see the 4GB G.Skill Pi kit fall the slightest bit behind the 2GB kit at the same settings. Again, we are pretty much looking at the difference between 1T and 2T because there is nothing else that is different about the two setups. It appears the having more memory is not always necessarily better. We have also seen that going from 1T to 2T isn't that much of a penalty at all and that 2x2GB kits really do hold up well overall and show little to no downside in the higher density modules...except of course the inability to run at as tight timings as 2x1GB kits which does effect performance a little bit.</p>


WinRAR v7.1<p style="text-align: justify;"><i>The WinRAR benchmark is simply timed until the test reaches 500MBs of data. Compressing data with WinRAR is all about the memory and the results should reflect this. WinRAR is very sensitive to frequency and timings of memory and one of the highest influenced programs out there when it comes to memory performance.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-6.png"></center><p style="text-align: justify;">Like the testing in the 32M digit calculation of SuperPi, WinRAR definitely favors the 1T setting for the 2GB kit more than it does the extra space of the 4GB kit. Once we start overclocking, the time required to reach 500MB compressed just drops like a rock.</p>


DivX Converter v6<p style="text-align: justify;"><i>Moving from some of the more 'synthetic' benchmarks, we find ourselves smack dab in the middle of a VOB to DivX encoding task. We will take a VOB rip of the movie Rounder’s, and convert it into DivX using the default 720P setting of DivX converter v6.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-7.png"></center><p style="text-align: justify;">Just as we finish talking about how there is little to no difference between the two kits, in steps a daily task that shows a big advantage to larger amounts of memory. We can see that the 4GB setup takes almost 4 minutes less time to convert the movie and that is with identical settings over the 2GB setup. Here, we start to see the real benefits of running 4GB kits of memory in our daily systems. A lot of the time, benchmarks don't really measure the difference between a 2GB and 4GB system but here, that clearly isn't the case.</p>


Photoshop CS3<p style="text-align: justify;"><i>Adobe Photoshop CS3 is full x64 compliant and ready and able to use every single CPU cycle our processor has available. Since digital photography is as popular as roller skates were in the 70's, we are going to be timing how long it takes to convert 100 RAW images from a Canon 20D into half size JPG files of maximum quality.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-8.png"></center><p style="text-align: justify;">The Photoshop RAW to JPG conversion test we have shows similar results to the rest of the benches we have looked at today, but keep in mind, this is simply converting RAW to JPG images. In heavy filtering or adjustments, the larger memory pool to work with will really let Photoshop stretch its legs. Working in the program is a lot smoother with 4GB of memory instead of just 2GB. Everything is snappier and there are far less hour glasses when working with many or large files.</p>
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
3D/Gaming Benchmarks

<b>3D/Gaming Benchmarks:</b>

Futuremark 3DMark Vantage<p style="text-align: justify;"><i>Like PCMark, 3DMark is now into the Vantage stage. The most popular 3D benchmark series of all time was updated this year and we have finally gotten around to getting used to 3Dmark Vantage so it has beat out the spot of 3DMark 06 as the default 3DMark benchmark in memory reviews. It should be interesting to see how memory affects this benchmark. For now, we will simply rely on the Performance benchmark defaults for testing.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-9.png"></center><p style="text-align: justify;">It appears that the extra room of the 4GB kit equals out to the 2GB kit running at 1T. Either that or the Performance preset of the Vantage doesn't show a lick of preference to memory size or timings. With such a small difference in results, it is hard to say whether it is just the benchmarks margin for error or if either one of the kits results in better scores. One thing is for certain though...increasing CPU frequency along with memory frequency definitely increases the score, as if that wasn't an obvious conclusion.</p>

Crysis<p style="text-align: justify;"><i>Like all other benchmarks, Crysis is benchmarked with all patches and updates done to it. We use the popular Crysis benchmark tool running the Sphere level time demo at 1680x1050 with detail levels set to High in DX9 and 64-bit. This level has all the elements of the Crysis environment that make the game so great so we chose to use it as the level to run our benchmarks on.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-10.png"></center><p style="text-align: justify;">The results are pretty obvious and it doesn't take much effort to see what really changes in Crysis with 2GB less memory. After our initial results be re-checked all settings and repeated all steps only to come to the same conclusions time and time again. With the 2x2GB G.Skill PI memory, the minimum frame rate was consistently lower by almost 10FPS in our time demo. With other time demos recorded on other levels, we found similar results. Maximum and average frame rates were marginally different, but the minimum frame rate definitely changes when making the jump to 4GB of memory. This sort of caught us off guard but is good to see. Minimum frame rates are about as important as average because the average frame rates are only so good if you keep dipping minimum frame rates down to stuttering levels. Let's see if the same type of pattern shows up for COD4.</p>

Call Of Duty 4<p style="text-align: justify;"><i>As with the Crysis benchmarking, COD4 was also benchmarked using a custom time demo, this time during a muli-player free-for-all on the Crash level. All details levels are set to high with 4X AA at 1680x1050 resolution. The time demo was recorded with more a dozen players and includes indoor/outdoor action as well as some heavy fire fighting throughout.</i></p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/bench-11.png"></center><p style="text-align: justify;">Those are hardly conclusive results, but we think it is safe to say that 4GB kits of memory may start to show their worth in 3D gaming minimum frame rates. Although not as dramatic, the 4GB kit at XMP defaults again shows a visible gain in the minimum frame rate achieved during our benchmark. The difference isn't going to save you $100 in a GPU purchase, but it certainly appears to be there, with this setup in this installation of Vista it does anyway.</p>
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Conclusion

<b>Conclusion:</b><p style="text-align: justify;">Winding down another journey through DDR3-ville is always a good part of the day. Like driving home from a hard day at work on Friday when your favorite song comes on the radio, ending this review just puts a smile on my face that can't easily be wiped off. When the weekend comes, you know all your hard work during the week was worth it despite some trying moments on Monday and Tuesday that make you think the week will never end. That is pretty much what it was like with this kit but was entirely the fault of the motherboard and operator.</p><center><img src="http://images.hardwarecanucks.com/image/3oh6/gskill/F3-12800CL7D-4GBPI/conclusion-1.jpg" alt="" border="0"></center><p style="text-align: justify;">Naturally though, the memory did play a small role in the problems being a 2x2GB kit of memory but any 2x2GB kit of memory was causing the overclocking issues we had with this kit on the motherboard. What we were able to accomplish was pretty darn impressive though and many users who find themselves opening a box from their favorite retailer with G.Skill PI DDR3 memory inside will have a lot of success overclocking. There really wasn't much else to discuss negatively when it comes to this kit other little issues like the fact that the heat spreaders are just a re-hash of an old idea or the availability being limited still in North America.

Overall though, the G.Skill PI 2x2GB PC3-12800 7-7-7 memory kit we looked at today is quite an impressive pair of modules. The XMP profiles are about as fast as you are going to be able to get at these frequencies with extremely aggressive sub-timings. Our Intel X48 based motherboard immediately recognized the XMP profile and from first boot we were running at rated frequencies and timings with this kit. The overclocking potential is there as we saw some pretty impressive results on a less than willing platform, imagine what we could do with a motherboard that wants to do its job? As we usher in a new set of Intel chipsets in the P45, we hope this won't be the last you see of the G.Skill PI PC3-12800 2x2GB kit. Don't forget to keep your eyes on the discussion thread of this review in our forum as these modules put more time in when newer motherboards come through.</p>
<b>Pros:</b>
  • 2x2GB memory kits are as hot as a volcano’s basement...popularity wise
  • XMP profile is bang on and provides excellent performance
  • Despite large heat sinks, the PI heat spreader is very thin and extremely lightweight
  • For the first time ever I think, DDR3 memory has the price in the Pro side on the list

<b>Cons:</b>
  • Motherboard capabilities are going to determine overclocking ability
  • The PI heat spreader despite being nice, does not push the memory cooling envelope
  • Availability on this kit is scarce here in North America and absent in Canada...so far

We like to hear feedback here at Hardware Canucks so feel free to bring up any questions or comments in the Discussion thread for the G.Skill PI PC3-12800 2x2GB kit.
 
Last edited by a moderator:
Status
Not open for further replies.

Latest posts

Twitter

Top