Say hello to TR's new storage test rigs

For years, we've been testing storage performance on a pair of P67 motherboards equipped with Sandy Bridge processors. The machines have served us well, but we're in the process of phasing them out in favor of all-new hardware and updated tests. This is the beautiful beast that's been running benchmarks in the lab all week:

It's not alone, either. We have two identical test systems based on the same hardware. At one point last weekend, both machines were running alongside the old ones in a four-way testing frenzy.

Storage is handled by the motherboard, so that's probably the best place to start. Put your hands together for Asus' Z97-PRO, which is a slightly more upscale version of my favorite Haswell board, the Z97-A.

In addition to sporting loads of 6Gbps SATA ports, Intel's Z97 chipset has native support for M.2 and SATA Express devices. The Z97-PRO takes full advantage of those capabilities, and its M.2 implementation works with both SATA- and PCIe-based mini SSDs.

Although the M.2 slot is limited to dual PCIe Gen2 lanes, we've added a pair of four-lane adapter cards to handle faster SSDs. These cards also come from Asus, and they support Gen3 speeds when plugged into the appropriate PCI Express slot. The motherboard certainly has plenty of those to spare.

As far as I can tell, the adapter cards aren't sold separately. The Z97-PRO is available at Newegg for $189.99, though.

The next ingredient is the CPU, which provides both processing and graphics for our test rigs. Intel hooked us up with a couple of Core i5-4690K processors from the Devil's Canyon family. Asking price: $234.99.

These quad-core chips scale up to 3.9GHz with Turbo and even higher if you're willing to overclock. We've disabled dynamic clock scaling to ensure more repeatable results, so we're actually holding the chips back a little.

On the GPU front, the processor's integrated HD Graphics 4600 runs the games for our load-time tests without issue. The stock cooler is perfectly adequate for our purposes, too. Most of our storage tests barely load up the CPU, and the motherboard's excellent fan speed control keeps noise levels in check.

Some of our updated benchmarks use a RAM disk as a high-speed source and destination drive, making memory particularly important for the new machines. Check out the funky XPG V3 modules Adata sent us:

The XPG DIMMs come with swappable mohawks that add a measure of color customization. The tinted inserts are easy to install, and it's a shame there aren't more colors. Red and gold are the only options for these particular modules.

We need a reasonably large RAM disk for our tests, so each system is loaded with dual 8GB DIMMs. The memory is rated for 10-11-11-30 timings at 2133 MT/s and 1.65V. We've scaled back the clock speed, tightened the timings, and lowered the voltage a smidgen. If you want matching modules of your own, Amazon has 'em for $145.99.

Most of our storage benchmarks run on drives configured as secondary storage. We still need system drives for each rig, and Corsair obliged with a couple of Force Series LS units.

These drives pair Phison's eight-channel PS3108-S8 controller with Toshiba's 19-nm MLC NAND. Each one has 240GB of capacity and a $128.99 street price.

The Force Series LS isn't the fastest SSD around, especially compared to Corsair's Neutron Series XT, which is based on newer Phison silicon. But it's fast enough for our purposes—and a big upgrade from the mechanical drives attached to our old systems.

I haven't taken a separate shot of the Corsair AX650 PSUs destined to power our new rigs, in part because we're borrowing those from the existing machines. Our old AX650s are still going strong after several years of service, so there's no need to replace them. While testing winds down on the old systems, the new ones are being powered by the beefier AX850 units behind our motherboard test rigs.

We'll have more details to share about our next-gen storage testing soon. Stay tuned.

Tip: You can use the A/Z keys to walk threads.
View options

This discussion is now closed.