Overclocking the Pi-Black was a easy task that resulted in what I would call average results for today’s low voltage Tri Channel memory that’s designed for the Core i7 platform.
I’ve been running DDR3 on my Core 2 Quad systems for some time now and I’ve been very pleased with the overall results of its performance. Now, with the introduction of the new Intel Core i7 platform with its Triple Channel memory controller along with its QPI (Quick Path Interconnect) built into the CPU itself instead of the memory controller being intergraded into the Northbridge, it’s going to bring overall memory performance to new performance heights for the competition to try and meet.
Today, we’re going to be running a 6 gig kit of G. Skill’s new high performance low voltage DDR3-1600 Tri-Channel Pi-Black memory through its paces on my newly acquired Intel Core i7 to see just how well it performs. Needless to say, I’m extremely excited to get into this review as it will be my first time running the Core i7 system as well as my first time running a G. Skill product. So, enough of the chit chat and let’s get on with the review.
“G. SKILL”, established in 1989 by enthusiasts, is a leading memory Module Manufacturer based in Taipei, Taiwan. With traditional strengths, we have built excellent reputation by meeting market demand and fostering business competitiveness both locally and internationally with our dedications to:
We are here to provide superior memory products and satisfactory services in order to keep pace with our customers’ growing needs, and help our customers in adding value to their products. And we pledge we will continue to do so and enable both sides to obtain significant competitive advantages in the market segments
Investing in human resource is just one of the reasons why G. SKILL is able to provide such high level of efficient and cost effective services.
G. SKILL is managed as a family, provides a dynamic, challenging and harmonious working environment for all employees. With the open-minded management, each individual¡¯s potential talent can be fully developed.
With this clear goal of providing satisfactory services for customers, our R&D department is constantly developing the fastest and best performing products; our marketing and sales departments ensure all the information needed is available and presented in an clear and understandable format.
At G. SKILL, top priority is the quality standard. All of its products are put through a series of rigorous tests and strict quality control procedures. In addition to commissioning qualified IC testing houses to test its products, all products are hand test 100% twice, in factory and office, to ensure the highest product yield and quality.
G. SKILL strives for the highest and dvanced quality from the initial design, through manufacturing solder-paste printing, and surface mounting, to on-line visual inspection, system compatibility testing, packaging and finally to shipping.
G. SKILL is able to quickly respond to the changing market conditions and fullfill customers’ needs by fully utilizing flexible operation process.
G. SKILL provides the widest breadth of product available in the memory industry and will continue to develop products on the cutting edge of technology.
Our highly skilled team of research, development, technical and sales / service professionals make G. SKILL the choice for businesses who recognize the importance of placing customers’ need first.
THE OLD WAY
Pulled from the Corsair site
Intel’s newest architecture, known as Nehalem (nee-HAY-lem), is an interesting and mysterious beast. There are some of the most significant changes in years inside this bad boy, so let’s get straight to the point.
The biggest and most important change for us to talk about today is the memory controller. The memory controller, much like its name implies, is the interface between the memory and the processor. Typically located on the “North Bridge” chip on the motherboard, the memory controller communicated to the processor through the “Front Side Bus” or FSB.
THE OLD WAY – A SYSTEM WITH A FRONT-SIDE BUS
In this image, the FSB is the connection between the processor(s) and the memory controller. Notice the memory controller is the interface between the CPU and the memory, as well as the interface between the CPU and the I/O controller, which would control your hard drives, USB ports, etc.
In the past, overclocking was usually achieved by tweaking this FSB. For example, on a Core 2 Duo system where the processor’s multiplier is locked, we might have seen the FSB at 800 or 1066. Typically this was twice what the actual host frequency was, so a processor would have a multiplier based on the actual host clock.
Example: The Intel Core 2 Extreme QX9770 runs at 3.2 GHz, and claims a “1600 MHz” FSB. This is a “Quad-pumped” FSB and the host clock is truly running at 400 MHz, still extremely fast. So the multiplier of this processor is 8.
Host clock * Multiplier = Clock Speed
400 MHz * 8 = 3200 MHz, or 3.2 GHz.
THE NEW WAY
But “Nehalem” changes everything. Much like AMD did with their initial launch of the Athlon64 years ago, Intel has finally moved the memory controller onto the processor itself.
Here’s how the new way works.
THE NEW WAY – QUICKPATH INTERCONNECT
The FSB has been removed completely – the memory controller and processor now have very little delay. This allows much lower latency and provides a significant increase in computing performance.
PICTURES & IMPRESSIONS
G. Skill ships the Pi-Black in a typical clear plastic clam shell retail package that fully shows exactly what you are buying with your hard earned dollars.
In my opinion, the Pi-Black is some very nice looking ram. It comes with a heavy gauge aluminum heat spreader that has looped extensions across the top to increase the surface area for better cooling.
Like most other brands of memory out there, G. Skill places a sticker on the back side of each module of their memory kit that includes all the basic specifications of the RAM.
Here you can get a better idea of how the heat spreader is designed
The G. Skill Pi-Black installed in all its glory. That’s some mighty fine looking memory. Now, on to see if it performs as well as it looks.
Testing & Methodology
To test the G. Skill Pi Black TC DDR3 1600 6GB kit we did a fresh load of Vista 64 on our Core i7 920 test system and ran our normal battery of tests on it. We ran the tests a total of three times and the average of the three test runs is reported here. All patches and updates were applied prior to testing and no background tasks that might interfere with testing were running.
I first ran the complete battery of tests with the complete system set to default settings and only changing the memory speeds in the system BIOS to see how well the memory performs for people that don’t really get into over clocking their CPU.
On the second run through the battery of tests I overclocked the system and, by using a combination of FSB and memory multipliers, set the complete system as fast as possible while setting the memory as close as possible to the tested memory speeds.
Here’s what and where we started the overclocked portion of the test. We simply raised or lowed the FSB until we found the speed we needed.
The motherboard we’re using is the basic version of the Asus P6T (non-deluxe) which has proven to be extremely stable throughout testing.
In this screen shot you can see that the provided SPD setting are a bit on the loose side. Through our testing of this RAM, we found it to be much more versatile in terms of getting much tighter timings out of it while still maintaining stability.
In the following screen shots you’ll see the memory timings that I was able to complete all of the overclocked benchmarks with. During the stock portion of the test, all the memory timings were set by the SPD.
|Case Type||Mountain Mods U2 UFO|
|CPU||Intel Core i7 920 @ 3.6 @ 1.30v|
|Mother Board||ASUS P6T (SLI and CrossFire on Demand)|
|Ram||G.Skill Pi Black DDR3-1600 3 x 2 gig|
|Hard Drives||Seagate 250 Gig SATA 2, WD 80 Gig SATA2|
|Optical||Lite-On DVD R/W SATA|
|GPU||Sapphire Toxic HD4870|
|Case Fans||8 x 120mm fans total|
|Testing PSU||SilverStone Strider SST-ST70F|
|Just an every day little rig.|
Synthetic Benchmarks & Games
SANDRA XII Professional
Everest Ultimate Edition v.4.50.1330
SuperPi Mod 1.5XS
WinRar v.3.71 Compression Benchmark
SuperPi is the “Gold Standard” for many when it comes to judging system performance. We ran 1 and 2 MB calculations, again running each test three times and reporting the average of the results.
Within each group of CPU clock speeds, the difference between the different memory speeds is small, but it does show how well the memory scales to the increase in memory frequencies.
I tossed in a set of dual channel Kingston DDR3-1333 just for a quick reference of how much better the G. Skill performs. As a matter of fact, the Pi-Black is faster at DDR3-1066 than the Kingston is at DDR3-1333.
On the 2 meg runs the G. Skill Pi-Black scales almost identically to the 1 meg runs, so the memory delivers very consistent results.
WinRar v. 3.71
This module in WinRar generates random data, which contains specially introduced redundancy, increasing the load to both the processor and memory. Data is the passed through RAR compression and decompression algorithms, and the output of the decompression algorithm is compared to the source data. If any differences are found, WinRAR then reports “Errors found – Yes” in the command window. WinRAR displays a size of processed data and compression speed, current and resulting, in kilobytes per second.
Here again in WinRar we see virtually perfect scaling of performance as the memory speed increases, as well as the Pi-Black easily beating out the Kingston.
During the WinRar 4x runs, the scaling results were virtually the same as the 1x results with the exception of the DDR3-1727 max overclocked run where the results were less than what I was getting at DDR3-1600. I ran this test countless number of times and regardless of what I tried, the results remained the same. I guess we’ll have to strike this one up to one of those strange little things that can’t be explained, but I’ll be doing some further testing with memory at a later date.
“EVEREST Ultimate Edition is an industry leading system diagnostics and benchmarking solution for enthusiasts PC users, based on the award-winning EVEREST Technology. During system optimizations and tweaking it provides essential system and overclock information, advanced hardware monitoring and diagnostics capabilities to check the effects of the applied settings. CPU, FPU and memory benchmarks are available to measure the actual system performance and compare it to previous states or other systems. Furthermore, complete software, operating system and security information makes EVEREST Ultimate Edition a comprehensive system diagnostics tool that offers a total of 100 pages of information about your PC.”
The way we have the results formatted in Everest is in GB’s.
This benchmark measures the maximum achievable memory copy speed. The code behind this benchmark method is written in Assembly and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing the appropriate x86, MMX, 3DNow!, SSE, SSE2 or SSE4.1 instruction set extension. The benchmark copies a 8 MB sized, 1 MB aligned data buffer into another 8 MB sized, 1 MB aligned data buffer through the CPU. Memory is copied in forward direction, continuously without breaks.
In order to avoid concurrent threads competing over system memory bandwidth, Memory Copy benchmark utilizes only one processor core and one thread.
I know this must be getting a little old looking at every single benchmark result where the Pi-Black is scaling perfectly as well as it beating out the Kingston by a fair amount. When I was running through all the tests, I found myself wishing that the Kingston would win on at least one of the runs.
This benchmark measures the typical delay when the CPU reads data from system memory. Memory latency time means the penalty measured from the issuing of the read command until the data arrives to the integer registers of the CPU. The code behind this benchmark method is written in Assembly, and uses 1 MB alignment, 1024-byte stride size. Memory is accessed in forward direction.
Memory Latency benchmark test uses only the basic x86 instructions and utilizes only one processor core and one thread.
This benchmark measures the maximum achievable memory read bandwidth. The code behind this benchmark method is written in Assembly and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing the appropriate x86, MMX, 3DNow!, SSE, SSE2 or SSE4.1 instruction set extension. The benchmark reads a 16 MB sized, 1 MB aligned data buffer from system memory into the CPU. Memory is read in forward direction, continuously without breaks.
In order to avoid concurrent threads competing over system memory bandwidth, Memory Read benchmark utilizes only one processor core and one thread.
This benchmark measures the maximum achievable memory write bandwidth. The code behind this benchmark method is written in Assembly and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing the appropriate x86, MMX, 3DNow!, SSE or SSE2 instruction set extension. The benchmark writes a 16 MB sized, 1 MB aligned data buffer from the CPU into the system memory. Memory is written in forward direction, continuously without breaks.
In order to avoid concurrent threads competing over system memory bandwidth, Memory Write benchmark utilizes only one processor core and one thread.
Here the memory write test we ran into another one of those unexplainable situations where the results between two different frequencies were the same and the RAM wasn’t scaling well regardless of what I tried. Still, overall, the Pi-Black put out some excellent performance numbers.
SiSoft Sandra 2009 SP1
“SiSoftware Sandra (the System ANalyser, Diagnostic and Reporting Assistant) is an information & diagnostic utility. It should provide most of the information (including undocumented) you need to know about your hardware, software and other devices whether hardware or software. It works along the lines of other Windows utilities, however it tries to go beyond them and show you more of what’s really going on. Giving the user the ability to draw comparisons at both a high and low-level. You can get information about the CPU, chipset, video adapter, ports, printers, sound card, memory, network, Windows internals, AGP, PCI, PCI-X, PCIe (PCI Express), database, USB, USB2, 1394/Firewire, etc.”
Benchmark the memory bandwidth of your computer. Shows how your memory sub-systems compare to other computers in terms of bandwidth.
The benchmark is based on the well-known STREAM memory benchmark.
Multi-Processor (SMP) support for up to 32/64 CPUs & SMT.
Arithmetic Operations Benchmark (assignment, scaling, addition, triad)
Floating-point Operations Benchmark (assignment, scaling, addition, triad)
Test needs at least 50% or more of free physical memory free.
As they say back in the old country, a picture is worth a thousand words. As toy can see in the screen shot of the over clocked DDR3-1600 benchmark. There’s nothing currently on the market that keeps pace with the Intel Corei7 with its onboard QPI memory controller. Having some awesome memory installed helps as well.
I have absolutely nothing bad to say about the G. Skill Pi-Black DDR3-1600 Tri-Channel 6 GB memory kit. Overall, throughout the entire test, the Pi-Black was one of the most well behaved and stable sets of memory that I’ve ever had the pleasure of working with. From the beginning, it was simply plug and play and it didn’t matter if I used the XMP settings, SPD settings, or dug deep into the system bios and did my own thing. And the best thing of all is that it did all this without going over the dreaded 1.65 volt limit that Intel recommends for the Core i7 memory.
The RAM also ran extremely cool to the touch throughout all phases of the review in my Mountain Mods chassis. This is mainly due to excellent air flow through the case, as well as the large surface area provided by the large heat spreaders. The heat spreaders themselves are something that won’t be a hit with everyone because of their design. Everyone has their own likes and dislikes, as they say, but for me, I like the looks as well as the performance.
Overclocking the Pi-Black was also a fairly easy task that resulted in what I would call average results for today’s low voltage Tri Channel memory that is designed for the Core i7 platform.
We are trying out a new addition to our scoring system to provide additional feedback beyond a flat score. Please note that the final score isn’t an aggregate average of the new rating system.
- Performance 9.5
- Value 8.5
- Quality 10
- Warranty 10
- Features 9.5
- Innovation 9.5
+ Extremely stable.
+ Easy over clocker with great timings.
+ Great looks with black aluminum heat spreader.
+ Superb performance.
- Just a bit on the expensive side for some folks.
For building a outstanding Tri Channel DDR3-1600 that does everything asked of it and a little more, all the while running cool and stable, we award the G. Skill Pi-Black DDR3-1600 a:
Final Score: 9.5 out of 10 and the Bjorn3D Golden Bear Award.