Date: March 6, 2007 - Author:
OCZ is on the top of their game, which is evident just by taking a look at their recent inventory. I will be taking a look at their latest high-end kit today, which utilize the new water-cooling capable heat spreaders.
OCZ has had so many product releases in the past year, it's hard to recount them all. Even in the weeks past, we've seen a new silent CPU cooler, gaming mouse and even a memory cooler. The gaming mouse is one that stands out to me. OCZ is already known as being a company for gamers, so it's only natural that they now extending their line to cater to their best customers.
Over the course of the last year, the memory market has been constantly busy, but not just with new kits released with higher frequencies. There's only so far you can push a module before it becomes boring, so everyone is trying a new type of heatspreader. Corsair is the only other to actually get a new type of spreader out. While there are many others with new designs, such as GeIL, none have actually surfaced for consumer purchase.
OCZ decided to take a different route from the rest, to stray away from the "must use a fan" mentality. We've wanted to see a kit of memory capable of water cooling for a while (at least I did) and that day has finally arrived. As you probably already noticed, the stature of the modules themselves is intense... they are huge. They need to be in order to have enough room for water to run through while still retaining proper efficiency.
Similar to Corsairs Dominator, the Flex XLC has the same "comb" look, used to dissipate heat quickly. Thanks to the smart design, the kit does not require water cooling to retain stock speeds, but I highly recommend using some form of active cooling, because they will get very, very warm.
Instead of the water flowing over top of the chips themselves, they run through the top portion directly above the PCB. Effectively what this should do is help keep the spreader as a whole at a healthy temperature. I believe there are a few reasons why they chose to not coat the chips themselves with water.
To see this happen, the spreader would have to be twice as thick to compensate for the fact that there are chips on each side of the module. In turn, this would force a larger tubing requirement in order to push enough water through efficiently. So while it's not impossible, it's impractical. Flex XLC is thick enough as is. Any thicker and you might not be able to fit two in your machine side by side.
Since the design doesn't allow water over the chips, the spreader incorporates the usual design goal of smart heat dissipation, which we can see by the rippled sides and fins on top.
The Flex also incorporates a custom 8-layer PCB to compliment the dissipation scheme. OCZ states that this PCB contains conductive power and ground planes to guide the heat away from the interconnects. It's hard to determine whether or not this is a beneficial design, but it's not going to counter the goal either. Solder balls underneath the chips already offer a source of heat transfer, so it could be that this PCB aides in guiding the heat up and out quicker than others.
As you would expect, the Flex XLC are delivered in the usual plastic blister pack. It is however far larger than most, since these modules are huge. This is not really the best method for these modules as far as I am concerned, only because that they do not stay in place. The sheer weight and size made them slide around a bit during initial shipment and even while I store them. I blame the fact that the pack stays shut solely by the two indent tabs on the top. The only solution I can figure would be to make the packaging slightly thicker so that it can handle more action.
Even though they are massive, I really do love the design and style of the modules. They are beefy and just scream a quality build. Each one includes an OCZ badge straight in the middle which works with the rest of the design to offer a great looking product overall.
In the following image you can see how the spreader does not cover as much of the PCB as many others do, including the Dominator. I assume this is to aide in the extra benefits that the 8-layer PCB offers by allowing more room for heat to leave.
The fins found on the top are thick and plentiful.
The tubing connections are found on each end, naturally. These accept 1/4" sizing, which leaves me out since all of my WC'ing rigs use 3/8". That's one thing to bear in mine when pondering a purchase if that's your primary goal. Looking at the surrounding area, it appears that there would be enough room for a larger fitting. I feel that deciding to use 1/4" will leave more people in the dust than if they had gone with one larger, but I am hoping to see a revision in the future. Anything is possible if the current versions sell well.
The front looked great, but the back is vastly different. I would have preferred a replicated look on the back, but I can see two reasons why they decided against that. First, they needed a spot to put the warranty sticker, since the front really offered no spot at all. Notice the three screws as well. Since this is a beefier than normal spreader, it's no surprise that they are required.
While I am not a huge fan of the back, I am pleased with the Flex XLC as a whole. I see a few areas where there could be improvement, but considering that this was the first module sent out to "test the waters", I'm pleased with what I see. To prove just how large the heatspreader is, I provide to you a comparison of the Flex to OCZs own XTC and also Corsairs Dominator.
The Flex completely dwarfs the XTC! When I first received the Dominator, I thought it was huge compared to a regular module, but no more! The Flex is one solid offering, I can say that much.
Even though this is one of the highest-end kits on the market, there is no fun in running any kit of ram at stock. None... absolutely NONE. I was curious to see how much overclocking headroom a 575MHz could possibly muster, but after reviewing Corsairs Dominator 9136 last week, I had an idea of what to expect, assuming that both kits use the same chips.
Well, these modules did not clock as high as the Dominators, but came very close. After a night of tweaking, the max stable OC I found was 625MHz with 5-5-5-18 timings. I need to make one thing clear though, as there seems to be a lot of confusion lately regarding what I consider to be a "max" overclock. I don't conclude with an overclock setting unless it passes 1000% in MemTest for Windows as well as a full 3D Mark 06 run to prove that it's fully 3D stable.
This kit -can- be clocked higher, but on my particular system it caused instabilities. I don't see a point of running a kit so insanely high that it causes your computer to spontaneously reboot itself. Sure high numbers are fun, but it's not practical.
I can say one thing though, you -will- want active cooling of some sort, because these spreaders get hot very quickly at 2.3v and above. Since these modules are meant to use 2.4v, it's good to have that in your game plan. While they may run fine and dandy without a fan, I don't feel particularly good knowing that the ram inside my computer is too hot to touch. I should also point out that I am talking about an 80mm or 120mm fan solution, as OCZs own XTC cooler in addition to Corsairs Airflow fan are too short to fit over the modules properly. You can sit these on top of the modules, but simply moving your computer will make it fall off.
That said, although DDR2-1250 was my personal max overclock, your situation might be different if you are not scared of applying higher than average voltage. The eVGA 680i tops out at 2.5v, which is a tad low when compared to other boards that hit 3.0v. After a lot of experimentation, I found that the memory scaled well with voltage, so extra dabs should help improve your overclock. OCZs in-house enthusiast, Tony Leach, has hit DDR2-1300 as a 100% stable overclock, but I don't recall the voltage he needed to do so. DFI's latest Crossfire board, the ICFX3200-T2R/G has insane capabilities, including (literally) intense voltage possibilities, so that might be the ideal board for this ram.
That said, here are the five stable settings I used (CL-tRCD-tRP-tRAS-tRC):
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:
The testing rig used for today's benchmarking is as follows:
Most of the results listed deal with how all of our settings scaled, although there will be direct comparisons to other kits of ram as well.
On Core based systems, it's always been rare to see a score in Sandra above 6,000MB/s, but that's easily surpassed with stock speeds like this.
Obviously, thanks to the sheer clock speed, our 625MHz setting took the cake with 6651/6651. Yes, it was one of those times where both the Int and Float happened to be identical. You can see by comparing the middle and bottom results that frequency doesn't play a large role in overall bandwidth. Even though the middle overclock had a full 133MHz above the bottom, it's bandwidth was a mere 50MB/s better.
Next on the list is Everest 3.5, with it's read/write and latency tests. We broke through the 8K mark in our read tests and came quite close to 6K in our write. These benchmarks are heavily CPU bound as well though, so the higher your CPU frequency, the better effects your memory overclocks will have.
The latency results were a big surprise to me, as our top overclock obliterated the stock speed with a full 9ns gain, despite using identical timings. In fact, stock latencies were equalled to DDR2-800 4-4-4 settings, oddly enough.
Sciencemark showed similar results to what we saw with our Sandra tests with obvious benefits with our top overclock.
Lastly we have Super Pi, using the 8 million setting. This benchmark is heavily CPU bound like a few others, but faster memory is beneficial. More so with tighter timings than higher frequencies, however.
Finally, let's wrap up the review with some comparisons and our final thoughts.
Here, we compare our overclocks of the memory being reviewed alongside other recently evaluated sticks. These graphs include benchmarks with each kit of ram at DDR2-1000 4-4-4-12-13 2.1v along with each kits own top overclock. While the DDR2-1000 results should not vary much, the top end overclocks will, given that each kit will top out differently.
For reference, here are the top overclocks for each kit of ram included:
Additional kits will be included as time passes. I had other kits on hand, such as Kingstons HyperX PC2-8000, but they did not give realistic results. I assume they are not fully compatible with the 680i motherboard.
In our first comparison chart, only the top overclocks of each kit were used. No surprises here. Corsairs Dominator inched over the Flex, thanks to the slightly higher CPU frequency.
Once again, aided by higher frequencies, the Flex falls behind the Dominator, by 1.1ns.
Equipped with our 2812MHz CPU and 625MHz memory, we crunched 8 million Pi digits in 3:37.
In the end, the 9200 Flex is one impressive kit of memory. Although it didn't keep up to the previously reviewed Dominator with the same voltage, it was very close. I mentioned in the overclocking section that there is more headroom available to those who are willing to use additional voltage. The eVGA 680i is one of the best motherboards I've used in recent years, but the lack of "extreme" vdimm capabilities is one of the only things that holds it back. Using upwards of 2.7v should deliver some amazing results.
If there are any immediate downsides to this memory, it's that it costs more than Corsairs 9136 kits at every e-tailer I checked, by about $20 - $30. However, this is the premium you are paying for their exclusive Flex XLC spreader. If you are a water cooling enthusiast, paying $30 for good blocks is usually something to get excited over. So in that respect, they are priced right. The benefit here though, is if you -do- choose to hook them up to water, you could essentially be one step closer to a completely silent PC. Although I didn't have the opportunity to connect them to my water cooling, others who have seem to be quite pleased with both temperature and overclocking performance.
I am awarding the OCZ Flex an 8 out of 10 with our Editors Choice award. Although pricey, it's the only kit on the market to include support for watercooling out of the box, not to mention at the extremely fast 575MHz frequencies. If you are looking to go big without killing a warranty, these are modules you should be considering. I just hope in the future to see revisions made to support larger tubing, as sticking to 1/4" might exclude a fair amount of the audience.
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 © 2005-2006 Techgage. All Rights Reserved. | Proudly hosted by Neosurge.