Mushkin HP2-8500 2GB Memory Kit

Although I've heard that Denver is an awesome place, there are very few things associated with Denver that I like. I hate the Denver Broncos (I'm a Steelers fan), I hate Coors beer, and the Colorado Rockies really let me down: now I have to hear all about the Red Sux from Paul until at least next April.

Now that I've finished my anti-Denver rant, let me introduce you to Mushkin, the enthusiast memory maker based in... you guessed it, Denver, CO. Today HL takes a look at Mushkin's newest kit,the HP2-8500.

Although DDR2 memory is far from dead (thanks in large part to DDR3's premium pricing, and generally cool reception), who would look at a PC2-8500 kit with 5-5-5-18 timings and come away impressed?

Sometimes, the specs that accompany a product are not representative of the performance, and as you'll see, Mushkin's latest HP2-8500 kit is a perfect example of this. 

Mushkin's performance line of memory is divided into four different categories. The Redline Series targets the hardcore enthusiast with 1GB and 2GB kits matching 4-5-4-11 latencies with PC2-8000 clock speeds. The XP Series is one notch below it, although no less impressive, with kits ranging from DDR2-667 at CAS3 to DDR2-1100 at CAS5. The HP Series is marketed as a balance between performance and budget with kits ranging from PC2-4200 to PC2-8500, supporting latencies from CAS3 to CAS5. The EM Series, targeting the budget market, offers clock speeds from PC2-4200 to PC2-8500, with latencies typically found on budget kits (CAS4 to CAS5).  

Bring up the topic of RAM at any enthusiast forum, and you'll almost assuredly be bombarded with chit-chat regarding the ICs (Integrated Circuits) used on different kits, and there may even be talk about the PCB (Printed Circuit Board). At HardwareLogic, our philosophy is a bit different. While some ICs are known to perform better than others, no matter which chips are found under the heatspreader, you're never guaranteed a level of performance beyond that which is printed on the sticker. Moreover, it's the screening process that can really make or break a line of memory, and the specific chips used on a kit of memory are subject to change at any time depending on the market.

One of the more public examples of this occurred in DDR's heyday. Enthusiasts frothed at the mouth over kits sporting Samsung TCCD chips, which had a reputation for high overclocks. When it was discovered that OCZ had switched from TCCD to TCC9 chips in their Platinum Revision 2 line, the community was up in arms. However, users for the most part where experiencing the same level of overclocks because the screening process didn't change, just the specific chips used (which are never advertised in the first place).

My test bench has been made up of a Gigabyte P35C-DS3R motherboard for the last couple of months, mainly because the board supports both DDR2 & 3. However, my little house of cards came crumbling down when I installed the Mushkin HP2-8500 kit in my Gigabyte board. The kit failed at default timings, regardless of voltages or settings. Now, this isn't the first time I've encountered this issue, and Mushkin was great, as they immediately sent out another kit in the event that there was an issue with the first memory kit. I tried again with the new kit, and encountered the same results. After talking with the Mushkin support staff, I found the issue to lie with the Gigabyte board's secondary memory settings simply being too aggressive. 

"Please read the following information if you have issues with the 2nd kit. As I recall, the Gigabyte boards set some subtimings very tight.
Here are the subtiming values we use with ASUS boards:
tWR 5
tRRD 3
Rank Write to READ 11
Write to Precharge 12
Refresh to Act 42
Read to Precharge 5
Note that the Read to Precharge value will most likely read two clocks lower when using memset in Windows. We test in dual channel above spec with RST Pro2 so our test environments are almost identical."

Overclocking Disclaimer

First and foremost, when purchasing a kit of RAM (or any component, for that matter) overclocking results are never guaranteed. Many factors can influence what a kit is capable of, including complimentary components (motherboard, processor, power supply), user experience, and simple luck of the draw. Two identical systems will not necessarily yield identical results, and anything over the rated specification should be treated as a bonus, not an expectation.

HardwareLogic has some of the most stringent memory testing on the web. When it comes to stability, we require a kit pass five cycles with the Ultra-X R.S.T PRO2 Memory Tester (The same testing hardware used by Corsair Memory and several other memory manufacturers to validate kits, however even most memory companies only require a kit to pass three cycles). We then require the test kit pass our entire benchmarking suite without error. If a memory kit can pass this torture test... we say its 100% stable. 

Pushing memory to its very limits is common in the overclocking community, however it's not the best option for everyone. Some people look to balance tighter timings with with higher CPU clock speeds, while some people are limited by their hardware and not always able to achieve the same results many review sites achieve. 

In that frame of mind, I typically place more value on a kit's versatility than its ultimate clock speed. While two PC2-8500 kits may perform almost identically at their rated clock speed, what separates a good kit from a great kit is not only how far you can push the clock speed at acceptable timings, but how tight you can set the main latencies (tCL, tRCD, tRP, and tRAS) at acceptable clock speeds.

While we've touched on the default timings as being less than exciting, I wanted to see how much I could get out of the Mushkin HP2-8500 kit with tighter timings and less voltage.

While I couldn't get the kit to boot with 1.8V, the standard for DDR2 memory, I was able to push the kit all the way to DDR2-909 @ 4-4-4-12-2T with only 1.85V. It is worth noting that 1.85V was the minimum I could use with 100% stability. I'd also like to point out that regardless of voltage, I was not able to run the PC2-8500 with 100% stability at 1T.

The Mushkin HP2-8500 overclocked like a champ, topping out at an impressive DDR2-1173 with tighter than advertised timings of 4-4-4-12.  Those of you running DDR2 in excess of 2.1-2.2V should be using active cooling over your DIMM slots anyway, but I do have to point out that the kit was smoking hot at anything over 2.2V.

* Run at 3:2 to keep the processor above its default 1333FSB

Loosening the latencies to their defaults (5-5-5-18-2T), I started over and was still unable to boot at 1.8V. I increased the voltage to 1.85V and was able to push the kit clear to DDR2-1090, which again, is much better than the specs listed on the label. Continuing on, I was extremely impressed with the performance, as 2.0V took me almost to DDR2-1150 (DDR2-1147).

* Default settings for this kit. 

Our Recommendation

Other Reviews of Note

It's always nice to have more than one opinion on a component before you spend your hard earned money. For one, we may see something others missed, or vice versa. As with all reviews published at HardwareLogic, we'll not only give you our recommendation, but also point out reviews from other great sites around the web.
N/A