OCZ’s new DDR PC3000 memory is simply gorgeous. Its red PCB and matching heat spreader can turn even the most reserved geek’s head.
So OCZ has nailed the looks, but does this fancy memory they have the personality to back it up? For PC hardware, a great personality manifests itself in stability and performance. It doesn’t hurt if it’s a cheap date, either.
Did our date with OCZ’s DDR PC3000 memory live up to the claims of 366MHz at CAS 2.5? Would we go out with her again? Read on to find out.
The lady in red
OCZ’s DDR PC3000 memory consists of a red PCB and your choice of one of three colored heat sinks. It’s only RAM, folks. There’s only so much you can do.
RAM doesn’t get that hot, but OCZ specs a simple heat spreader on its PC3000 memory just in case.
While I’m not sure how effective the heat spreader is, it doesn’t add any problematic girth to the DIMM. You should still be able to load up your motherboard’s DIMM slots without the heat spreaders blocking access to any of them. The spreaders will, however, diminish the airflow between the DIMMs you have plugged in.
Unfortunately, the heat spreader on our review sample was a little off center, making it impossible to securely clip into the motherboard’s DIMM slot. It seemed like a simple fix at the time: just pop off the heat spreader and nudge things over a little. Unfortunately, the heat spreader is a real pain to remove.
I don’t foresee many mismounted heat spreaders slipping through the cracks at OCZ, but be sure to check before trying to secure the DIMM to your motherboard with a hammer.
OCZ uses hand-picked Hynix chips in its DDR PC3000 memory, but the markings on the DIMMs don’t match anything in Hynix’s part number decoder. A similar part number can be found in this press release describing Hynix’s latest DDR chips, undoubtedly the same chips found on the OCZ PCB.
It’s worth pointing out that OCZ’s PC3000 moniker really isn’t anything official. OCZ claims that it surpasses the PC2700 standard, but there’s no official PC2700 standard for them to surpass yet.
What to look for in the test results
We’re not comparing OCZ’s PC3000 memory to anything other than itself, which makes interpreting the test results a little more complicated than usual. Instead of seeing several different products tested against each other, you’re going to see the same DIMM benchmarked at different front-side bus speeds and latencies.
So, when looking at our test results, keep in mind that the test system’s hardware doesn’t change at all. Instead, the only variables are the front-side bus and memory latencies that we’ve manipulated in the BIOS. Higher front-side bus speeds and lower latencies should yield better performance. OCZ’s memory will simply allow the system to remain stable at those speeds.
Including memory from other manufacturers in our testing would have been a little redundant. RAM modules are a commodity of sorts, and DDR SDRAM chips on different DIMMs should yield identical performance when running at the same front-side bus speed and memory timings. Regardless of whether the DIMM is from Crucial, Corsair, Kingston, or OCZ, performance should be identical provided the memory settings in the BIOS remain the same. The real test of a DIMM’s worth is how fast you can set the front-side bus and how much you can lower the latency while retaining overall system stability.
What we’re testing here is how well a system performs with an astronomical FSB speed and super-aggressive RAM timings, so you can see what buying a stick of fancy RAM can do for you.
Our testing methods
As ever, we did our best to deliver clean benchmark numbers. Tests were run three times, and the results were averaged.
Our test systems were configured like so:
In order to isolate the memory’s overclocking potential, we’re only running the front-side bus out of spec in our testing. An older, unlocked 1.2GHz Thunderbird makes an appearance because it lets us turn down the CPU multiplier to keep the CPU speed as close as possible to 1.2GHz without going over. It’s like The Price is Right, sort of.
We tested the system at the following speeds and settings to see just how much of a performance boost could be had from only front-side bus overclocking.
As you can see, we were only able to run at CAS 2 up to 183MHz. Beyond that, we had to switch to CAS 2.5 to get a stable system. The memory is only rated at CAS 2.5, so that’s not a big deal. More importantly, we had to disable direct memory access (DMA) modes for the system’s IDE controller at bus speeds above 165MHz. Otherwise, the system wouldn’t boot into Windows. DMA enables devicesin this case the hard driveto send data directly to memory without having to go through the system’s processor. This is especially important to keep in mind when looking at some of the disk-intensive benchmarks, which will put an additional strain on the processor when DMA is disabled.
The test systems’ Windows desktops were set at 1024×768 in 32-bit color at a 75Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests. The motherboard’s BIOS was tweaked with the most aggressive settings possible, without sacrificing system stability.
We used the following versions of our test applications:
- SiSoft Sandra Standard 2001.3.7.50
- ZD Media Business Winstone 2001 1.0.2
- ZD Media Content Creation Winstone 2001 1.0.2
- MadOnion 3DMark 2001 Build 200
- Quake III Arena 1.30 Final Release
- Serious Sam v1.02
All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.
First up we have Sandra’s memory bandwidth tests.
In the ALU test, the performance scales predictably.
Sandra’s FPU scores differ a little from what we saw with the ALU results. Here, CAS latency becomes more important than bus speed, and our 183MHz CAS 2 setup scores higher than the 195MHz CAS 2.5 configuration. The battle between CAS latency and memory bus speed will be one to watch through the rest of the benchmarks.
Business Winstone 2001
Typically, you wouldn’t be doing extreme front-side bus overclocking to improve your word processor performance, but Business Winstone does produce some interesting results.
At bus speeds above 165MHz, where DMA had to be disabled, Business Winstone’s performance tanks spectacularly.
Content Creation Winstone 2001
On to Content Creation Winstone, whose suite of tests is more intensive, in some ways, than its Business sibling.
The lack of DMA really hurts Content Creation Winstone’s scores, but that’s no surprise. Multimedia content creation moves a lot of information back and forth from the disk.
Quake III Arena
Quake III Arena starts off our game benchmarking. Let’s take a look at the results.
Thankfully, Quake III Arena’s frame rate doesn’t suffer when DMA is disabled. Notice the importance of CAS latency: our 183MHz bus at CAS 2 scores noticeably higher than our 195MHz bus at CAS 2.5.
Serious Sam is up next. Will CAS latency continue to play an important role?
Sort of. Here the CAS latency and MHz differences seem to balance each other out, with the two fastest scores tying. DMA continues to be a non-issue in gaming, at least in terms of average frame rates.
3DMark 2001 rounds out our testing. Let’s see how the performance stacks up.
Like the Serious Sam results, 3DMark 2001 shows almost identical performance for 183MHz at CAS 2 and 195MHz at CAS 2.5. Again, the lack of DMA isn’t a hindrance for the faster front-side bus speeds.
OCZ’s DDR PC3000 memory looks great, but more importantly, it performs exceptionally well. Though only rated for 183MHz at CAS 2.5, I was able to run the test system’s bus at 183MHz with CAS 2, and at 195MHz with CAS 2.5, with no stability problems. You may or may not be into the look of OCZ’s DDR PC3000 memory, but you can’t diminish that kind of a performance.
Since no one else makes “PC3000” memory, we have to look at the next best thing for a price comparison: PC2700. On Pricewatch, generic PC2700 memory runs $70 for a 256MB stick. OCZ’s PC2700 runs $78. At $88, OCZ’s DDR PC3000 memory doesn’t see a huge markup. Consider for a moment that Corsair’s lauded XMS2700 CAS 2 DIMMs go for $132 on Pricewatch. Our sample from OCZ ran at “PC3000” speeds and CAS 2 without a hitch, and for less than 70% of the cost. That makes it a pretty good deal in my book.
Because we ran our system so far out of spec for this review, I’d classify our testing as extreme overclocking. That said, OCZ’s memory will work in faster platforms, where reaching 183MHz won’t be such a huge jump. SiS’ 645 chipset already calls for a 333MHz DDR memory bus, and VIA’s KT333 is also on the way. These chipsets require PC2700 memory to run at their stock speeds without any overclocking. Currently, OCZ’s DDR PC300 memory is the only memory rated by its manufacturer to handle front-side bus overclocking on these DDR333-capable platforms. So while you can get your fix of extreme overclocking on older platforms, OCZ’s DDR PC3000 also gives you a shot at some more moderate overclocking with future products. DDR333-capable motherboards shouldn’t require disabling DMA modes to get the most out of these DIMMs, either.
If you’re looking for RAM that scales to incredible front-side bus speeds, look no further than OCZ’s DDR PC3000 memory. With its performance matching its great aesthetics, and a price that you’ll be hard pressed to beat, this one goes to 11.