Print this article

OCZ 1024-3200 Gold Edition XTC


Date
: 12/27/05 – 02:44:56 AM

Author
:

Category
: Memory

Page 1 : Index

Manufacturer:
OCZ Technology

Price:
$171.93 (NewEgg)

Today we gather here at Overclockers Online around another great looking kit of memory from OCZ. This kit is different than anything else OCZ has sent us. This kit is different than anything else any memory maker has sent us. This kit will revolutionize the way computers work in the years to come! Okay, so maybe not, but the idea behind this new offering from OCZ is definitely interesting.

Heatspreaders were introduced some time ago with the theory that they would help distribute heat across the entire length of the DIMM and help eliminate hotspots caused by individual ICs. This in turn would provide greater stability to the DIMMs on a whole and allowing for better clocks or overclocks. OCZ has been slapping heatspreaders on its performance memory for as long as I can remember and they haven't changed all that much from when they first came out. On October 25 of this year, that all changed.

XTC stands for Xtreme Thermal Convection and the logo itself depicts the idea behind the design. By having holes in the heatspreader, air will be able to come in direct contact with the ICs at the same time that the solid material can still distribute the heat being generated by the ICs to the rest of the DIMM like the traditional heatspreader. Today we will be looking at the 1GB Gold Edition Enhanced Latency PC3200 dual channel kit that sports the XTC heatspreaders. I will not only be doing the standard overclocking and benchmarking with this kit but taking an in-depth look at just how effective these new heatspreaders are.

Page 2 : Package & Contents

The first thing we will look at is the familiar OCZ package.

The heatspreaders may be new on the memory but the package the memory comes in is the same as it has for some time. All OCZ High Performance memory come in a clear blister pack with the modules safely secured by the plastic package itself.

The cardboard insert is a generic one that comes with the rest of the OCZs line leaving any specifics about the RAM inside up to the stickers on the memory. Also, like all other memory kits from OCZ, the stickers on the modules state the important specifications as well as complete part numbers for easy identification.

If the front of the insert hasn't changed I wouldn't expect the backside to have and of course it didn't. The two huge points that you can see on this insert are the
Lifetime Warranty
and the fact that OCZ actually provides
Technical Support
via phone or e-mail for its products. This is unparalleled service in the memory market.

The gold trim of these heatspreaders is just as shiny and reflective as the standard heatspreaders that OCZ uses. Getting a photo of these sticks can present some challenges because of this. The most obvious design changes with the new heatsinks are the big Z smack dab in the middle and the honeycomb interior portion.

The innovation of these honeycombs is to allow air to come in direct contact with the ICs, but as you can see in the close up, there is a layer of what I assume is thermal tape between the honeycomb heatspreader and the ICs. I understand the honeycomb needs that thermal tape there to do its job, but in order for fresh air to actually touch the ICs, the thermal tape would have to be honeycombed as well. I guess we will learn more in a little bit when I remove the heatspreader.

What I don't understand the most about all of this is why OCZ still uses the same sticker on the left hand side like their other memory kits. It is basically covering a quarter of the DIMM and defeating the entire point of the honeycombs for that quarter.

Another aspect I noticed about these new heatspreaders is that they do not connect and interlock at the top like other heatspreaders do. This may be due to the design of the honeycomb heatspreaders but it may also be by design in order to allow for more fresh air to access the ICs. This may also provide an escape for hot air that is being trapped in-between ICs under the thermal tape.

The backside is exactly like the front but without the restricting sticker. Again the big Z is located in the middle where there is the gap between the ICs and honeycombs make up the rest of the surface area. It is now time to get destructive and see if we can get those heatspreaders off without destroying the DIMMs.

In all honesty, getting the XTC heatspreaders off was a lot easier then getting any other OCZ heatspreader off that I have had. The thermal tape used really does stick well, but once you get an edge peeled, the rest seemed to come easier than the other thermal tape they use on the standard heatspreaders. You can see the thermal tape is so thin and light that it is actually translucent. This should aid in cooling and heat transfer. My earlier reservations about the effective cooling by way of the honeycombs have been relieved somewhat based on this photo.

As noted by OCZ, the ICs used on the new XTC modules are Winbond UTT BH5 marked with this part number: X4W560842J-50. Interestingly enough, these and the older gold modules are not the only stick of RAM with these ICs. I happened to recognize the part number as being the same on a kit of the CAS 2.5 value series memory that OCZ puts out. The other thing to note is that my batch number is 0531. I could not find anyone else sporting the 0531 batch of these ICs so no tip-off to the overclockability of this kit was gained.

The last of the photos is that of the BP branded board OCZ uses. BP makes the best PCB for memory and OCZ uses them exclusively throughout its performance modules with this PC3200 XTC kit being no different. The 808 is a 6-layer shielded PCB as far as I know.

I highly discourage anyone from removing their XTC heatspreaders because as you can see from the first photo, the thermal tape will get messed up and will not allow the heatspreader to sit properly when you go to put it back on. I actually finished my review before removing these heatspreaders for this very reason. Just trust us when we say there are Winbond UTT BH5 chips under there.

Page 3 : Specifications

We will start off the specifications with a short quote from OCZs web site about these new gold modules:

These modules really are impressive with those 2-2-2 timings. Here is the list of the rest of specifications that accompany this fine kit of memory:

Here is the breakdown of the special features listed above provided by OCZ:

The biggest drawback with this kit is the memory voltage requirements. Most motherboards should be able to run the memory at 2-2-2-5 as most motherboards can provide 2.8v to RAM. When it comes to overclocking this memory though, as you will see, there is no substitute for voltage.

Page 4 : Installation & Overclocking

To overclock and test the OCZ 3200ELGEGXT kit, I will be using the following components:

As far as the installation goes, there is nothing different about the new XTC heatspreaders that would effect a straight forward installation. There is no extra bulk on the sides and even less height than standard heatspreaders. There should be absolutely zero fitment issues with this kit.

Being BH5 equipped, I will be using the orange slots on my DFI LanParty UT nF4 Ultra-D as the orange slots tend to work with BH5 chips better than the yellow ones.

As expected, there is nothing to getting this kit installed. I also figured since not everyone will have the luxury of having split dual channel slots like I have here with the Ultra-D, I would test fit the kit in the DFI Infinity nF4 Ultra which has tighter memory slots.

Again, no issues and no problems in getting the modules fitted. The one noticeable difference is that there is very little room for air to get at these heatspreaders when in a configuration like this. Immediately the benefit of them is cut in half. Those with a motherboard where the two slots that will be used are close together might want to keep that in mind. Of course the open top of the DIMMS will still allow for the hot air to escape so there still might be a reason for going with the XTC heatspreaders.

With the modules in place and a cooling fan situated to keep them from melting at crazy high voltages, it is time to get the burn-in out of the way and on to some overclocking. To burn-in memory I like to run it for 24 hours at stock voltages and timings in Memtest86 and then move into windows with another 24 hours in Prime95 torture test blend.

No errors and all looks well so let the overclocking begin. I figure if the memory will run at CAS 2 with only 2.8v, why wouldn't it run 1.5 so we start there. To ensure the memory is the limiting factor the HT multiplier is dropped to X3 and the chipset and CPU voltage will be increased as necessary. I am very familiar with the CPU and motherboard being used so I know what will be necessary along the way.

Much to my surprise CAS 1.5-2-2-5 ran right up to 253 for an effective speed of DDR506. It did take a fair amount of voltage to reach this lofty overclock with 3.46v being fed to the memory at this frequency but the fact that it was able to reach here is impressive. I could not squeeze any more out of the modules at CAS 1.5 by adding voltage or loosening the other timings. Here is the chart of how the memory scaled with the voltages required:

I figured with CAS 1.5 running so high that going to 2-2-2-5 timings would yield further overclocking, but that was not the case. Again, no amount of voltage or looser timings would yield a higher, stable overclock than 253MHz @ 2-2-2-5 with 3.46v.

I was intrigued by the overclocking of this memory to run the same frequency and voltage at both CAS 1.5 and CAS 2. The in-ability to further the overclock by switching to a 2T command rate or loosening timings was also quite different from that of most every other module I have worked with. Obviously these chips are designed to run only at 2-2-2-5. Here is the whole chart of overclocking at 2-2-2-5:

As you can see above, stability at each stage of the overclock is tested by a 32M digit SuperPi run, a jog through Futuremarks 3DMark 01SE and a handful of hours in Prime95 running the torture test on blend. At the final overclocks of 253MHz at both CAS 1.5 and CAS 2, Prime95 was run for an extended period of time to ensure a stable overclock. Because of these overclocking results, I decided testing of the performance would commence at both CAS 1.5 and CAS 2 to see if there was any real performance gains to be had by the lower CAS.

Page 5 : Performance

I decided to take a bit of a different path when it came to testing this OCZ EL Gold GX XTC dual channel kit. I figured that since the modules will both run the same frequency at CAS 1.5 and CAS 2, why not take this opportunity to see how benchmarks are affected by the CAS difference. I used the standard benchmark suite relying on SiSoft Sandra Pro 2005 SR3 and Lavalys' Everest Ultimate to determine the bandwidth pushing capabilities of the kit. I also did 1M and 32M digit passes with SuperPi Mod 1.4 to depict whether or not a lower CAS had that much effect on the number crunching ability of the memory. For a general computing demonstration of the two sets of timings I ran Futuremarks PCMark 04 and 05 as well as 3DMark 01SE. The Futuremark series should give us a general idea of what we can expect for performance differences in every day tasks. Finally, I used BenchEm All to run three gaming benchmarks at a low resolution so the system would act as the bottleneck to see if any performance gains could be had by dropping down to a CAS of 1.5. The games were Far Cry, Half-Life 2, and Unreal Tournament 04. Let's have a look at the results.

I wasn't sure exactly what to expect as I had never done a direct comparison of CAS timings before, and certainly not between a CAS of 2 and 1.5. As we can see, the difference isn't dramatic at all and in some instances, the bandwidth at CAS 2 exceeds that of CAS 1.5. There does not appear to be a clear cut winner as far as synthetic bandwidth benchmarks are concerned.

In the PCMark results we see, if anything, a slight advantage for the lower CAS of 1.5 but again, the margins are minimal to put it mildly. In 3Dmark 01SE we almost see a performance increase with the lower CAS but it isn't drastic by any stretch of the imagination. At most it is 400 points at the highest clocking of 253. Perhaps this means the gaming will indicate further increases with a CAS of 1.5.

With the nature of the SuperPi benchmark a performance gain by the lower CAS was expected and was achieved but again, it is minimal at best. It shaved off less than a tenth of a second in most instances of the 1M digit run and less than a single second on the 32M run. It really appears that either CAS 1.5 does not make much difference or some limitations of the motherboard are not allowing the lower CAS to fully show its worth.

The gaming benchmarks are conclusive only in pointing out that there really is no real world benefit to running this RAM at CAS 1.5. The results go back and forth but the entire time in all three games, the gap is less than a single frame. We can draw from these results that running a CAS of 1.5 will only be beneficial if you are trying to squeeze out ever last hundredth of a second on a SuperPi run for a forum contest, otherwise you are best left to run this kit at 2-2-2-5. If your motherboard doesn't support CAS 1.5 then don't bother running out to find one that does, you will be wasting your money.

I wanted to put this kit through a battery of thermal tests to see just how effective the XTC heatspreaders are and if they are any more beneficial than the standard heatspreaders we have come to know and love. Unfortunately the inability to effectively remove and replace the heatspreaders has made that impossible. The only true way to test this kit would be to put it up against the older version of this memory that came with the standard heatspreaders but retained the same PCB and ICs. I, of course, do not have a kit of that memory. I can tell you that I did run the memory at 3.46v for extended periods of time and even without the 92mm fan blowing directly on the memory, the kit remained stable. A thermal probe was place under the heatspreaders in the center of the module behind the big Z throughout testing and with active cooling the temperature never exceeded
31C
. Without the active cooling, the temperature climbed up to
42C
on occasion.

Page 6 : Conclusion

The Gold series from OCZ is not targeted to the overclocking crowd. The Gold series was developed for the gamers who want the absolute best performance with the tightest timings at specified frequencies. The fact that OCZ uses Winbond UTT BH5 chips that happen to thrive on high voltages does in fact make this a very decent overclocking kit of memory if , and only if, you have a motherboard and the active cooling to provide over 3v. This means only if you own a DFI motherboard or one of a small handful of others that can accomplish this voltage feat.

With a limit of only 3v one can expect to squeeze about 230MHz out of this kit, perhaps a touch less. If you have only 2.8v at your disposal then a maximum frequency in the 220 range is all you can expect. Myself, I do have access to the extreme voltages that were needed to run this kit at 250 or DDR500+, but I am not sure how comfortable I would be running this machine 24/7 with 3.4v+ running through the modules. The lifespan is surely going to decrease drastically even with active cooling.

For the suicide-shot overclockers that have a CPU with a large multiplier, this is the memory for you because nothing will be able to touch it at 250+ with timings of 2-2-2-5. The gamers out there with higher end CPUs that also have higher multipliers will fall in love with the tight timings at moderate overclocks as well. It is hard to say whether the XTC heatspreaders are any better than the standard ones but the theory behind them is solid and they sure look much better in my opinion. Keeping the temperature of the modules at 40C with no active cooling and high voltages is definitely no luck, these XTC heatspreaders really do their job well.

Advantages

• Beautiful heatspreaders
• UTT BH5 ICs will accept ridiculously high voltages
• Great overclock with high voltages
• Retains 2-2-2-5 timings throughout OC range

  Disadvantages

• High voltages required to maximize overclock

  Overclockers Online would like to thank OCZ for supplying the sample for this review.