Single page Print

Performance, power consumption, and overclocking
Integrated memory controllers have largely leveled the playing field for memory performance, especially when you're using identical DIMMs running at the same speed and timings. How a motherboard maker chooses to tune an integrated memory controller can influence performance somewhat, but keep in mind we're dealing with two very similar Asus BIOSes.

As you'd expect, there's very little difference in memory performance. Throughput is nearly identical between the two boards, and their access latencies differ by no more than a nanosecond.

The Rampage has fewer traces and onboard components than the Sabertooth, but it also has additional features and custom silicon to support them. So, which board consumes less power? To find out, we measured system power consumption, sans monitor and speakers, at the wall outlet using a Watts Up Pro power meter. Readings were taken at idle and under a load consisting of a Cinebench 11.5 render alongside the rthdribl HDR lighting demo. We tested with Windows 7's High Performance and Balanced power plans.

Part of the extra goodness that comes with the Rampage is Asus EPU power-saving software, which must be run in high performance mode to avoid capping the CPU multiplier at 12X. We ran the Rampage with and without the EPU software enabled.

It didn't make much difference, though. In fact, the Rampage drew slightly more power with the EPU software enabled. If the only way to actually realize power savings is to impede performance by slamming the multiplier, you're better off not bothering. The Sabertooth does just fine without fancy software, and it actually draws five watts fewer than the Rampage at idle.

Asus bundles its own automatic overclocking software with mid-range P55 boards, but it hasn't done so with either of these X58 offerings. The reason? Folks buying into the X58 platform know what they're doing. That said, the Sabertooth does come with Intel's latest overclocking utility, which has a measure of auto-tuning built right in.

Instead of letting software do the grunt work for us, we tackled the base clock speed of each board the old-fashioned way: by lowering the CPU and other system multipliers and then slowly turning up the base clock. Along the way, we used Prime95 and the rthdribl HDR lighting demo to test for stability.


210MHz on the Rampage III Gene...

And the very same base clock speed on the Sabertooth

As luck would have it, both boards marched up to a 210MHz base clock without making a fuss. Neither would even post with a 215MHz base clock, however. I fiddled with QPI settings, added voltage here and there, and even relaxed memory timings, to no avail. 210MHz certainly isn't anything to sneeze at, although given the age of our CPU, I suspect there could be additional base clock headroom lurking in both boards.

Our last stop on the testing front is the wonderful world of onboard peripherals. Here's how the boards' most important ports perform:

HD Tach USB 3.0 performance
Read burst
speed (MB/s)
Average read
speed (MB/s)
Average write
speed (MB/s)
CPU utilization
(%)
Rampage III Gene 215.8 175.8 60.0 2
Sabertooth X58 214.1 176.1 59.7 2

I used Super Talent's USB 3.0 RAIDDrive to test SuperSpeed performance, and the Rampage and Sabertooth offered equivalent throughput and CPU utilization. The write speeds above are a little pokey because the RAIDDrive's flash memory is slow to write, not due to any shortcomings in the USB 3.0 interface of either board.

HD Tach USB 2.0 performance
Read burst
speed (MB/s)
Average read
speed (MB/s)
Average write
speed (MB/s)
CPU utilization
(%)
Rampage III Gene 36.2 35.0 24.8 1
Sabertooth X58 36.2 35.0 25.2 1

If your external storage devices are USB 2.0, well, they'll be slower than equivalent SuperSpeed products. Once more, the Rampage and Sabertooth are locked in a dead heat.

HD Tach Serial ATA performance
Read burst
speed (MB/s)
Average read
speed (MB/s)
Average write
speed (MB/s)
Random access time (ms) CPU utilization
(%)
Rampage III Gene 246.2 134.7 135.2 6.9 1
Rampage III Gene (Marvell) 315.3 136.2 84.5 6.9 1
Sabertooth X58 246.5 136.2 135.1 6.8 1
Sabertooth X58 (Marvell) 318.8 136.2 85.3 6.9 1

Oh, look, another set of near identical results. Although there are differences in performance between the ICH10R's 3Gbps SATA controller and the 6Gbps Marvell chip, the two Asus boards again turn in similar scores. In case you're wondering, we used WD's 6Gbps VelociRaptor VR200M as the target drive for this particular test.

NTttcp Ethernet performance
Throughput (Mbps) CPU utilization (%)
Rampage III Gene 930.3 2.6
Sabertooth X58 711.0 2.5

Remember the Sabertooth's PCI-based Gigabit Ethernet controller? It costs you nearly 220Mbps in throughput versus the Rampage's Intel networking controller—a drop of close to 24%. The Rampage uses the Intel GigE controller built into the ICH10R south bridge, and it offers comparable throughput to what we've observed with PCIe-based controllers from Marvell and Realtek. Tapping the south bridge controller saves Asus a PCI Express lane on the Rampage, and the same approach should've been used with the Sabertooth.

RightMark Audio Analyzer audio quality
Frequency response Noise level Dynamic range THD THD + Noise IMD + Noise Stereo Crosstalk IMD at 10kHz Overall score
Rampage III Gene 1 4 4 4 3 1 1 1 2
Sabertooth X58 5 4 4 5 3 4 5 4 4

Ouch! X-Fi or not, the Rampage board's Via codec does not perform well in our RightMark Audio Analyzer loopback test. A higher score is better here, with six being best. Obviously, the string of ones from the Rampage isn't good. We made sure Creative's software layer wasn't doing any thing funky to the signal by disabling the EQ, EAX effects, and pretty much everything else. RMAA continued to report loads of inter-channel leakage, though.

Oddly, music playback sounds fine on the Rampage. "Fine" being average, or barely satisfactory for anyone with a good ear and decent headphones or speakers. I initially suspected that RMAA might be picking up some issues with the Rampage's line input, which the loopback test uses to record the waveform it pipes through the front-channel output, but we saw similar results using a Xonar DG at the receiving end of the loopback test.