Techgage logo

Mushkin 2GB EM6400 DDR2-800

Date: July 28, 2006
Author(s): Rob Williams

With C2D and AM2 out, many are looking for the right DDR2 kit to choose. If you are on a budget, then you will want to check out the Mushkin EM6400, which retail for around $200. Though they are surrounded with a ‘value’ cloak, they deliver surprising overclock-ability.


Today on the ram testing workbench, we have a PC2-6400 2GB value kit from Mushkin. I took a look at the OCZ SOE kit a few months ago, which are rated for the same speed and timings. Their price range is also the same, so I have a feeling that performance will be similar. However, the SOE’s were horrible overclockers, so I am hoping I can at least push these EM’s a little bit further.

Before we jump into the specifications and all the juicy stuff, let’s take a look at what we get.

Closer Look

The memory arrived in a nice clean professional looking plastic blister pack. The recent Mushkin redesign has become evident in their packaging, that’s for sure. One thing I have complained about in the past though, is how they hot glue the sides together. This makes it harder to open than it should be, and you can potentially break the package. That sucks if you plan to re-use it again.

The modules are equipped with green heatspreaders, and are some of the best I’ve seen. Mushkin hasn’t changed their heatspreaders for a while, but they don’t need to… these are great. The top is open to allow for good airflow. If you have a fan pointing at these, the air will actually hit the chips.

The sticker includes all of the information you would expect, such as timings, model number and frequency. Overall, these are some sweet looking sticks. Let’s hope the performance is equaled!

Specs and Overclocking

Because these are value modules, they include value specifications. We have DDR2-800 speeds, with 5-5-5-12 timings. Yes, they are very loose for this speed, however it’s not uncommon by any means. I am unsure of the chips used, but for the sake of them being value, I will assume they are some Elpida variant.

Overclocking these modules proved to be a huge pain at first… they were going nowhere. I first tried 4-4-4 and even at 2.4v, they were not stable at -all-. After a little bit of tweaking, I quickly found out that the main problem was the tRCD. Once I loosened it back to 5, I was able to go with a 4-5-4-12 timing set. After it’s all said in done, here are my stable overclocks:

I have to say, these modules impressed me. I had the mind set that these value were going to go nowhere, but upping the stock by 50MHz and tightening the timings is great. Before we get into benchmarks, please review our benchmarking methodology.

Benchmarking Methodology

Throughout all of our benchmarks regardless of what we are reviewing, testing is done in a clean and stand-alone version of Windows XP Professional with SP2. Prior to testing, these conditions are met:

If you are interested in using the same benchmarks as us, feel free to visit the developers website:

The testing rig used for today’s benchmarking is as follows:

Sandra, Everest

Throughout all of the graphs, the * represents results from the recently reviewed OCZ SOE PC2-6400 2GB kit. It’s also a value kit, so I used it at stock settings to compare directly to the EM6400. In this instance, we can see the EM6400 cleaning house. I did not expect to break 7000 with these modules, but the graph speaks for itself. Between the 280FSB and 270FSB setting, we can see the extra MHz are no match for the extra FSB.

We use specific Unbuffered settings when it comes to Sandra, that seems to differ from how most people do it. You can check out the specific settings we use here.

Now here’s an interesting spin on things. In the Buffered tests, we seen that the 280FSB was superior to the final 270FSB setting. Completely vice versa here, with an ever greater difference than in the first graph. We even broke through 4,500 Float, which I am still impressed to see.

Everest 3.0

Everest 3.0 is only a month old, but the memory results remain the same between version 2.8 and this new one. As we expected, 280FSB took the crown, almost breaking through 8,700. Here, the EM6400 and OCZ SOE are neck and neck at stock settings.

Speaking of the OCZ SOE… it’s interesting to note the vast difference between the latencies at stock. 3ns is not a small difference! Well, it is, and it isn’t. Confused? Great! At our top frequency, we burst through the 70ns mark, which again, was unexpected.

PC Mark, Super Pi, Conclusion

3D Mark 01 is considered outdated for a good reason. It is. One great thing it’s still good for though, is seeing how your CPU overclock scales. I like to pay more attention to PC Mark though, since it’s benchmark is completely memory specific.

It’s unknown to me why the first two ’01 results are far apart, as that version of 3D Mark is extremely CPU dependent. Though 280FSB proved the best for PC Mark, it fell short of 270FSB’s result for 3D Mark. However, PC Mark is a more reliable benchmark here, because the test targets the memory specifically, whereas 3D Mark obviously does not.

Super Pi

Super Pi, like 3D Mark 01 is good for seeing how your CPU overclock scales. The primary benefit is that it does just that… but quickly, especially if you only run the 1 Million method. Memory plays a big part in the results though. The tighter the timings, the faster the roundtrip for the data between your CPU and memory.

33s with 280 is a great result here. In comparison, I have never broke through 30.5s with even $500 kits of memory, even with ‘suicide’ runs. Another reason to avoid the 820 D, really :-)


When I first received the kit, I was hugely skeptical I was going to go anywhere with it. It’s value… what can you really expect? It turns out, I ended up far more impressed than I imagined. It was really like a roller coaster ride earlier in the day. I was hugely upset because the memory was going -nowhere-, but once I realized it was just that troublesome tRCD, things really took off. The memory proved completely stable through a 200% MemTest for Windows run at 450MHz, 4-5-4 and also 32M Super Pi stable. For modules that showed no signs of overclocking, that’s impressive.

Compared to the OCZ SOE PC2-6400 kit, this one wins without a doubt. Sadly, I had no luck with overclocking with that kit, whatsoever. This Mushkin kit though, not only overclocks quite well, it also costs near $50 less than what I see the SOE going for. Looking around at some e-tailers, it’s somewhat hard to find the EM6400 kit in stock. However, after looking at some shoppes, it seems the absolute cheapest place to purchase the kit is directly from Mushkin themselves for $204.

So, is the kit worth that $204? If you are on a tight budget and want ‘better than value’ memory, this is a good choice. Just bare in mind that the tRCD is going nowhere, but there is definitely some potential, especially if you don’t mind moving up to 2.2v. If you plan to do some heavy overclocking, then no these are not worth it. It’s obvious you want a kit with much more freedom, but you will have to haul more out of your wallet for them.

These modules are priced pretty good at $200US, and are right in line with the competition. However, you would not need to pay more than 25% more for a much more overclockable kit. If you are looking for a value kit with potential, I can recommended these ones. I am awarding the EM6400 2GB kit an 8 out of 10. Thanks to Mushkin for handing me a value kit that I was actually impressed by :-)

Discuss in our forums!

If you have a comment you wish to make on this review, feel free to head on into our forums! There is no need to register in order to reply to such threads.

Copyright © 2005-2021 Techgage Networks Inc. - All Rights Reserved.