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!

Mushkin Ridgeback PC3-12800 4GB DDR3 Memory Review

Status
Not open for further replies.

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Founded in 1994, Mushkin Enhanced is known in the enthusiast world as being a major player when it comes to producing top quality memory modules. Located at the base of the Rocky Mountains in Denver, Colorado, Mushkin is not only known for producing memory, also power supplies and digital storage devices like SSDs as well. The memory lines produced by Mushkin range from standard (Proline & Essentials) to extreme (Redline and Blackline), with Ridgeback as one of the latest additions to their higher end yet competitively priced lineup.

In March 2010, Mushkin introduced a new memory cooling solution which they called the Ridgeback. They the Ridgeback cooling system as a world-class memory cooling solution that helps to ensure modules work well in intense, high-labour scenarios while looking elegant in your system. Supposedly, the fin array has been carefully engineered to deliver serious heat transfer performance in a low-profile design while providing a powerful new look. Marketing lingo aside, the new modules look great and they are less likely to interfere with aftermarket heatsink installation than their predecessors.

At the $120-$145, the Ridgeback modules are priced to compete with modules such as G.Skill’s Ripjaws, the Corsair Dominators, and Kingston’s own HyperX modules. These memory lines are among the most dominant in the industry at this moment and it appears that Mushkin hopes to take some market share with innovative new heatspreader designs.

After testing a variety of memory in the same class as the Ridgeback we are very excited to see how these new modules stack up. The kit which will be tested in this review is the 4GB Muskin Ridgeback PC3-12800 6-8-6-24 with a part number of 996826. We will take a close look at the memory itself, what is under the hood, the effectiveness of the Ridgeback heatspreaders and how the memory overclocks. Mushkin’s motto is “superior performance with uncompromised quality”, so let’s put this to the test!

Title1.jpg
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Specifications

Specifications

Before jumping right into photos and testing, let’s take a brief look at the specifications for the Mushkin Ridgeback PC3-12800 6-8-6-24. Specs for the particular kit being tested in this review are as follows:

Model: 996826
Type: DDR3
Pins: 240
Voltage: 1.65V
Speed Spec: PC3-12800
Timings: tCL 6, tRCD 8, tRP 6, tRAS 24
Frequency: 1600MHz
Dimensions: 125mm (L) x 40mm (H) x 7mm (W)
Kit Quantity: Dual Kit
Capacity: 4GB (2GB x2)
Module Config: 256x64
Heatsink: Ridgeback
Registered / Unbuffered: Unbuffered
Warranty: Lifetime

Additional features specific to Mushkin Ridgeback PC3-12800 6-8-6-24 include:

Features.png

The Ridgeback lineup of memory consists of six kits of DDR3 and one kit of DDR2 memory. Looking closer at the DDR3 kits there are two 4GB kits, two 6GB kits, one 8GB kit, and one 12GB kit. Timings range from 6-8-6 to 10-10-10 for the higher capacity kits. Complete details on all Ridgeback kits are as follows:

Memline.jpg


Title2.jpg
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
A Closer Look at Mushkin Ridgebacks

A Closer Look at Mushkin Ridgebacks

In this section we will take a closer look at the packaging for the Ridgeback, the memory itself, and some size comparisons to other memory kits. The following page gives a closer look at what is under the spreaders and how well the spreaders contact the memory chips. The Ridgeback comes in a plastic package, common to many types of memory. However, Mushkin managed to make a very eye catching package.

Spack1.jpg
Spack2.jpg

The marketing on the front package uses the Ridgeback logo front and center with a small Mushkin Enhanced logo above. The color scheme is typical Mushkin green and the black background makes a nice contrast. We really liked the packaging for this kit and found it to be visually pleasing. As you can see the front packaging fits perfectly with the memory modules.

Spack6.jpg
Spack7.jpg

The reverse side of the package gives a short paragraph about Mushkin Enhanced at the top and the remainder of the page is dedicated to installation instructions.

Spack5.jpg
Spack4.jpg

Similar to the packaging, the memory itself is eye catching to say the least and Mushkin should sell several kits based on the looks alone. With a number of competitor products similarly priced and spec'd sometimes looks is what matters and Mushkin hits the mark with the Ridgeback.

Spack9.jpg
Spack11.jpg

The specification sticker on the memory gives all the relevant info on which kit you have. Speed, timings, voltage, part number, and capacity are all clearly indicated.

Spack10.jpg
Spack12.jpg

Taking a closer look at the heatspreaders on the Ridgeback, we see a thicker design than usual at 7mm. The ridges are also quite thick and durable so users won't have to worry about them getting damaged or bent. To get some reference on the size of the memory we have a stick of Ridgeback next to a stick of G.Skill Ripjaws and Trident. As you can see the Ridgeback is mid-sized; larger than Ripjaws but smaller than Trident.

Spack13.jpg
Spack16.jpg

Being a mid-sized memory kit, the Ridgebacks may have some issues with CPU heatsink clearance but this depends on the motherboard layout and the CPU heatsink size. Since you use the far slots on an Intel motherboard for a 4GB kit there should be no problem in this case but if you add a second kit clearance issues may arise with some boards. We recommend you measure first to be safe. Just note that unlike some modules, the slimmed-down heatspreader design will cause less installation issues than other enthusiast memory products.

Spack17.jpg
Spack18.jpg

The following section gives an overview of the spreader design and shows what we have under the spreaders.
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Under the Heat Spreaders

Under the Heat Spreaders

The heat spreader design consists of 3 parts: the main heatsink, the side panel and the thermal pads. The screws are countersunk so they don't take from the look of the spreaders and make for easy exposure of the chips. Most memory is held by clips or adhesive thermal pads so users can't remove them without risking damaging the memory and voiding their warranty. Mushkin made the Ridgeback heat spreaders easy to remove, which is nice if you have CPU heatsink clearance issues. Please note that removing spreaders does still void warranty so is not recommended.

Srbdiagram.jpg
Sspread1.jpg

So end users don't have to void their warranty to see what is underneath, we did the work for you. Simply unscrew the three screws and the side panel just pops right off. The thermal pads are not very sticky so provide little to no resistance in removing the side panel.

Sspread2.jpg
Sspread3.jpg

It is as simple as that; the bare stick comes out and the kit is as easy to put back together as it is to take apart. The ICs used in this kit are XDZ PSC chips commonly found in 4GB kits with similar specifications. The full chip code is XDZ025A3G-A, which means PSC ICs, batch XDZ, produced in week 025. XDT and XEJ are also common and the chips can vary depending on when your kit was produced.

Sspread4.jpg
Sspread5.jpg

Although the heat spreaders on the Ridgebacks may be easy to remove and look great, after inspecting the modules closely the contact between the two is questionable. Looking at the marks on the chips from thermal pad contact, it appears that the heat spreaders may not be applying much pressure on the memory. It seems that the screws only apply sufficient pressure to the top quarter of the module while the bottom does not have much pressure. To test this theory we attempted to pull a memory module out of the spreaders without unscrewing the screws and as we suspected it pulled out with minimal effort. This is concerning since it means only a portion of the heat-producing IC will be actively cooled by the heatsinks.

Sspread6.jpg
Sspread7.jpg

Next up we take a look at our test bench and overclocking methodology.
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Test Setup and Overclocking Methodology

Test Setup and Overclocking Methodology


Test Setup

testsetup.png

Our test setup consists of a socket 1156 Intel i7 platform with a EVGA P55 FTW motherboard and an Intel i7 860 CPU. Here are a few shots of the setup:

Stestbench1.jpg
Stestbench2.jpg


Testing Methodology

The following section shows the maximum stable overclock achieved on the Mushkin Enhanced Ridgeback PC3-12800 6-8-6-24 1.65v memory using a variety of voltages and timings. For testing methodology three main stability tests will be used and along with a variety of benchmarks. The first of the two main stability tests will be Linpack (LinX version 0.6.3) with memory usage set to 3,072MB and 25 loops run. In the enthusiast world, Linpack is a benchmark designed to measure performance on Intel CPUs in GFlops. However, it's also a very useful tool for checking the stability of a CPU and memory. LinX picks up memory errors very quickly and if you are able to complete a 25 loop test with the specifications above your system is likely stable or very close to it.

The second stability will be a minimum of three hours of HCI memtest using all memory available. This will pick up any memory error that LinX may have missed and where time allowed we ran Memtest overnight. On top of that, the third stability test will be 3 runs of 3DMark Vantage. This tests the 3D stability of the overclock as well as CPU, BCLK and memory. Once an overclock passes these tests but fails anything further, this is the point deemed as “stable” for the purposes of this review.

The EVGA P55 FTW (657) BIOS used for testing is E657_A51, released on December 29, 2009. This BIOS worked well for our testing and we had no issue with compatibility. Our i7 860 test CPU has been run well over 1000Mhz with other memory kits so we can confirm that the CPU's memory controller is not limiting our overclock with the Ridgeback.

A number of combinations of memory timings will be tested at stock voltage of 1.65v. 1.65v set in the P55 FTW BIOS is exactly 1.65v as measured by a digital multimeter. We also kept VTT at a safe maximum of 1.30v throughout all testing. Higher VTT will allow for slightly higher memory clocks but can also be dangerous for your CPU.

Stay tuned for overclocking results!
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Overclocking Results

Overclocking Results

Now that we have had a good look at the memory and the chips it uses we can move on to how the Mushkin Enhanced Ridgeback 6-8-6-24 kit overclocks! It is always nice to know which ICs memory uses when overclocking as you can have an idea how it will react different timings and voltages.

The chips in our test kit are PSC chips, XDZ025A3G-A to be more specific. We know from past experience that these Powerchips will be highly dependent on tRCD for how they clock and that we will see the best results with CAS being two clocks lower than tRCD. The spec speed of 6-8-6 supports this theory.

All tests were ran at default voltage of 1.65v using a number of timing variations; eight in total. We started with TRCD below spec at 6-7-6 and ended at 9-11-9. In all cases tRP was left at the same clock as CAS since it did not provide additional clocks at a higher setting. We also stuck to 1.30v for VTT for consistency and to prevent damage to the CPU. More VTT and vDIMM will likely produce higher clocks but should be used for short benching sessions only. The tests in this review represent 24/7 stable settings for the memory for your daily overclock.

The chart below is a summary of our overclocking results and below that we will go through the overclocking process and provide screenshots of the results. As always, your results may vary and overclock at your own risk.

ocresults.png

First off we want to say this kit was a pleasure to test. We saw no compatibility issues at all with the EVGA P55 FTW board and the memory reacted to timings and voltages as expected. Starting below spec timings at 6-7-6 we managed an overclock of 729Mhz, 71Mhz below spec speed. However when we moved up to spec timings of 6-8-6 we saw close to a 150Mhz gain, with stability all the way to 887Mhz! Mushkin really left some headroom here and from our tests it appears they bin the Ridgeback relatively conservatively.

Increasing CAS and tRP by 1 really does not do much as we only saw 2 extra Mhz; however when we moved tRCD to nine we were greeted by another 100Mhz+. So as you can see, this kit is all about the tRCD and it is best to keep CAS and tRP two clocks below tRCD.


Additional headroom when increasing from tRCD = 9 to tRCD = 10 was less than at lower settings but we received good increases when moving CAS and tRP up by one. For example 6-8-6 vs. 7-8-7 was an additional 2Mhz while 8-10-8 vs. 9-10-9 was an additional 31Mhz. Moving up to the top end speed of the memory we managed to hit 1,173Mhz at 9-11-9, which is great for kit in this price range. Up around these speeds however we were getting close to our CPU’s BCLK stability limit so it is very possible 10-11-10 or even 9-11-9 would crack 1,200Mhz on a CPU with a better IMC or with additional VTT.


All in all we thought the Mushkin Ridgeback 6-8-6 kit overclocked great for the price range and was saw speeds close to the top dog Elpida Hyper kits. We liked that there was plenty of headroom above spec timings and speeds and that gaining over 100Mhz was as easy as bumping tRCD up one clock. Now exactly how much performance do you lose by bumping tRCD up to 9, 10, or even 11, and what is the ideal speed and timings for the Ridgeback? Continue on to the next section to find out.

Title2.jpg
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Timing Comparisons

Timing Comparisons

We achieved a wide range of maximum stable memory speeds by different timing configurations (from 729Mhz at 6-7-6 to 1,173Mhz at 9-11-9) but which will be optimal for performance? To give a little insight on what kind of difference timings make we decided to run some additional tests using everyone’s favorite benchmark, SuperPI. RAM timings affect the speed of the PI calculation and generally the looser the timings, the slower the PI time. To rule out other influence on PI time such as CPU speed, QPI, and RAM speed we ran all tests using the same CPU speed, memory speed, QPI, and secondary memory timings. Seconradies were set manually for consistency and all that was adjusted were the major timings: CAS. tRCD, and tRP. In all tests below CPU speed was 3198Mhz, QPI 2617Mhz, and memory speed was 727Mhz. We chose this speed because this was the highest we were able to achieve at 6-7-6 and it could be applied to all eights sets of timings. Secondary memory timings used for each of the eight tests were the following:

picomp3.png

First let’s take a look at the differences in PI time for the eight sets of timings tested. Again all settings were kept consistent other than CAS, tRCD, and tRP.

picomp.png

As expected, the tightest timings yielded the fastest PI times and each test we ran gave a slightly slower result. The delta in PI time from the fastest time (6-7-6) to the slowest (9-11-9) was 9.3 seconds, really not that much in a 10 minute calculation. So which timings should you run? Let’s add memory speed to the chart for context and go from there.

picomp2.png

Ideally we want to find a middle ground where we get a high frequency for the memory with the tightest possible timings. Applying this theory to the chart, we want to find a spot where the PI time line starts to slope steeply upwards and memory speed levels out. 7-9-7 appears to be a good place to start as 7-9-7 gets us nearly 1000Mhz and also a good PI time. Looking at the chart line slopes, after 7-9-7 the PI time shoots upward over seven seconds to 714.2 at 9-11-9. Running at 7-9-7 will put less stress on your IMC than the speeds near 1200Mhz with looser timings and will allow for tighter subtimings (such as tWR) which provide additional performance gains.

The rule of thumb for maximum performance from your overclock is CPU Mhz is number one, memory Mhz is number two, QPI three, and memory timings should come fourth. Basically you want to max out your CPU and then find your maximum stable memory clocks at the tightest possible timings while using the highest stable QPI. For example, if your max CPU clock and memory ratio put the ram speed at 950Mhz then your best bet is 7-9-7. Or if your CPU and ratios put memory at 1,100Mhz then you will need to loosen to 9-11-9.
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Memory Benchmarks

Memory Benchmarks

Sandra Memory Bandwidth and Memory Latency


SiSoftware Sandra (the System ANalyser, Diagnostic and Reporting Assistant) is an information & diagnostic utility. The software suite provides most of the information (including undocumented) users like to know about hardware, software, and other devices whether hardware or software. The name “Sandra” is a (girl) name of Greek origin that means "defender", "helper of mankind".

The software version used for these tests is SiSoftware SiSoftware Sandra Professional Business 2009.SP4 and the two benchmarks used are the Memory Bandwidth and Memory Latency. These benchmarks were chosen as they provide a good indication of memory performance and will tell us which settings are most effective. The bandwidth test shows performance of memory sub-systems and the memory latency test shows performance of the memory controller.


Sandrabw.png


Sandralat.png

Using the max clock speeds at 6-8-6, 7-9-7, 8-10-8, and 9-11-9 at 1.65v we see that both memory bandwidth and memory latency improve with speed. This tells us that maxing out your memory at 9-11-9 will likely yield better performance in most applications than using lower TRCD and lower speed. Taking the timing comparisons (previous page) into perspective, it is unlikely that the drop in PI times with looser timings will be as significant as the increase in performance from extra bandwidth and less latency. That said, we recommend running this memory at as high a frequency as your CPU overclock will allow.
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Temperature Analysis

Temperature Analysis

Mushkin specifically designed the Ridgeback heat spreaders to be top notch in terms of looks and heat transfer. At Hardware Canucks we always like to test out any and all manufacturer claims so in this section we will take a look at the Ridgeback kit’s temperatures with and without the heat spreaders installed and with and without active cooling.

To ensure that we were consistent with the testing methodology, the same speed, stress test and test duration were used for all four tests. The memory was run at max overclock of 1,177Mhz at 9-11-9 and 1.65v and the temperature was recorded at the 10 minute mark of a LinX test. First we tested the temperature of a stick of Ridgeback with the heat spreader installed. To do so we slipped a temp probe under the spreader so that it was touching the very bottom of an IC. Then we lifted the fan and retested with no airflow. The results are as follows:

Stemp1.jpg
Stemp2.jpg

Next, we taped the temperature probe to be bottom of a bare chip and reran the tests using the same methodology. The results are as follows:

Stemp3.jpg
Stemp4.jpg

As you can see, the heat spreaders allowed the memory to run cooler in both active and passive cooling situations by 2.4 and 6.6 degrees respectively. This is particularly interesting since as we saw, the thermal pads used don’t make all that good contact with the ICs.

It is also important to note that this memory gets rather toasty without any airflow so it is always a good idea to have lots of airflow in your case; especially when it makes close to a 20C difference.

temps.png

So in conclusion, our temperature analysis tells us that the spreaders do help cool the memory to some extent. This kit does produce a fair amount of heat when running at full overclock / load so active cooling is always a good idea. The more airflow, the more the Ridgeback spreaders will dissipate that heat.
 
Last edited by a moderator:

Eldonko

Well-known member
Joined
Apr 5, 2007
Messages
5,416
Location
Calgary, AB
Conclusion

Conclusion

Adding another line of memory to an already saturated mid-range enthusiast market is a tough proposition since any new product really has to stand out. You have to be a step ahead in price, overclockability, or design to grab that elusive market share from competitors. Has Mushkin done this? Yes, but there are a few minor caveats along the way to perfection.

You can tell at first glance that a lot of effort was put into the design of the new Ridgeback heat spreaders. Their design may be similar to products like G.Skill Ripjaws but once you get them in your hands, the quality does seem to be a step beyond the competition. We saw in our temperature analysis that the heat spreaders do in fact gain you a few degrees over a bare module and even more with active cooling. Unfortunately, the screws at the top of the module do not provide sufficient or equal pressure across the module. It would be interesting to see what kind of temperatures these modules could have achieved without this design oversight.

cons2.jpg

One of the major benefits we see in this kit is its ability to overclock and indeed the Ridgebacks produced some excellent bandwidth when running at high frequencies. Even at stock timings of 6-8-6 we were able to hit 887Mhz, 87Mhz over spec speed. It looks like Mushkin conservatively bins the Ridgebacks and that is definitely a good thing for the end user. Loosening up timings to 9-11-9, we were able to push that all the way to 1,173Mhz at stock voltage of 1.65v and 1.3v VTT. This is impressive to say the least.

Mushkin’s lifetime warranty is consistent with most other memory manufacturers with two slight differences. First, this is an American company so customers may find then to be easier to deal in RMA situations. The other difference is that Mushkin requires a proof of purchase in order to RMA while some other manufacturers do not. This can be inconvenient if you buy memory second hand or forget to save the receipt.

All in all, we feel Mushkin hits the mark with the Ridgebacks and should be able to take a bite out of the competition’s market share. The memory looks great, has an innovative design and most importantly overclocks well. However, while this is a great product there is still very little to differentiate it from the competition and as such, the choice of the consumer will likely boil down to one based on price rather than stand-out qualities. If you can find yourself a kit of Ridgeback for a good price though, grab it and you won’t be disappointed.


Pros:

- Lots of overclocking headroom
- Attractive heatspreaders
- Mid-sized for CPU heatsink clearance
- Great looks


Cons:

- Heat spreaders do not fit snugly



 
Last edited by a moderator:
Status
Not open for further replies.

Latest posts

Twitter

Top