Puget Custom Pentium-M Rig: A Silent WC System

Complete|Mobile Systems
Viewing page 6 of 6 pages. Previous 1 2 3 4 5 6

OVERCLOCKING

The extreme efficiency of the cooling system prompted us to see just how much we could stress the system by overclocking it. Unfortunately, the motherboard did not allow the PCI bus speed to be locked or overvolting above the stock 1.308V (undervolting is possible), so our overclocking attempt may have been limited by these factors. In preparation for overclocking, memory timings were loosened to 4-4-4-10 from 2-2-2-8 and the northbridge was overvolted to 1.50V from 1.05V. Because the multiplier could not be adjusted upwards from 15x, all overclocking was done by adjusting the FSB.

Our first attempt at overclocking the FSB to 167 MHz (2,505 MHz processor frequency) was stable enough to run CPUBurn and 3DMark05, but it crashed five hours into a session of Prime95. We then backed off in single MHz increments until all our benchmarks would run properly. Strangely enough, the most unstable application appeared to be 3DMark05; after its first successive run at 167 MHz, it crashed repeatedly at every level above 163 MHz. This level proved to be stable after 48 hours of continuously running Prime95, so we feel confident that this was the highest stable frequency. Multiplied by the 15x CPU multiplier, this translates into 2,445 MHz as the highest stable processor frequency — a 22% overclock.

Puget Custom Computer: Overclocking Test
Activity State
CPU Temperature
AC Power Draw
Idle
31°C
68W
CPUBurn
43°C
81W
3DMark05
41°C
113W
3DMark05 + large file
copy from CD to HDD
41°C
123W

Temperatures did not change significantly in the overclock. CPUBurn only increased the peak temperature by a single degree and every other test we ran produced results identical to those of the non-overclocked CPU.

The effect on AC power draw was similarly small, typically in the range of 2-3W. Running CPUBurn did increase this difference to 5W, but it's important to keep in mind that CPUBurn is an exceptionally CPU-intensive program; most real applications do not match this level of stress on the CPU.

PERFORMANCE BENCHMARKS

While the system was overclocked, we ran PCMark04 and 3DMark05 to obtain benchmark scores. All possible tests were enabled, but all other settings were otherwise left at their default values. The final scores were compared to other scores in Futuremark's benchmark database.


Our score ranked fourth among all Pentium M systems.


This test scores second in Futuremark's database for Pentium M systems with a Radeon X800 XL, but there's only two other contenders.
We could probably have stolen first if we'd overclocked the VGA card too...

SOUND TESTING

In general use, the Puget system was inaudible except when specifically listening for it. We had to put our ear within a foot of the case to clearly distinguish the noise it makes from the ambient noise. The noise comes mainly from the pump motor, and is characterized by a low frequency hum similar to hard drive resonance or DC transformer hum. Puget did not disappoint: In most environments, this system should be silent. The SPL measurement taken here is actually somewhat questionable; it is about the ambient noise level of the lab. It's possible (but unlikely) that the SPL is even lower than 18 dBA at 1m.

The one acoustic flaw in the system is the DVD burner. Puget selected the most expensive DVD burner money can buy for the system, a SATA interface Plextor drive, perhaps hoping that its high-end status would translate into a quieter design. Unfortunately, the Plextor drive cannot overcome the inherently noisy nature of high-speed optical drives; while the drive noise is reasonably smooth, it is far from quiet at full speed.

Puget Custom Computer: Noise Level
Activity State
SPL (dBA/1m)
Idle
18
Hard Drive Seek
18
Full Speed CD Copy
38

CONCLUSIONS

When Puget first contacted Mike, they asked what kind of system SPCR would like to review. Mike's reply was that he'd like to see a system that SPCR audiences would like. A quiet, powerful and unique system would be great, he said. The system that they sent meets all three of these criteria: It's unique, very quiet, and very powerful.

This is probably not the kind of system that most people would buy. The $3700 price tag is more than most people will want to pay for a quiet system. The price includes more than $800 worth of custom parts and labor that are not available except by special request. The custom-built water cooling system accounts for the lion's share of this price, but there's also smaller details such as the soft rubber feet and the custom-mounted hard drive enclosure that contribute to the price.

Taken on its own, the Puget system is a textbook example of overengineering. Watercooling a Pentium M may indeed give the best CPU temperatures possible, but it's far more than necessary. Soft rubber feet can be beneficial in a system with high-vibration drives that sits on a hardwood floor, but they don't really belong on a system of this weight. Building a fanless system that is stable and doesn't overheat is an impressive feat, but it's possible to build a system that is just as quiet using air cooling. Other examples of somewhat redundant modifications include the use of a hard drive enclosure on a notebook drive, and installing Acoustipack in a system that has next to no vibration.

As a demonstration of what Puget can do, this custom system succeeds marvelously. Very few companies that sell whole systems are willing to do the kind of modification that this Puget system required. A self-contained, passively cooled water cooling system is not the kind of thing you can request of just any system builder. With the proper selection of parts, just one of Puget's modifications could make the difference between an audible and an inaudible system.

Puget has demonstrated that they can build a system that is not just quiet but silent under most circumstances. No other system integrator that we know of can make this boast. One reason that the Puget system is so quiet is that the basic components of the system are quiet. The fanless power supply, notebook drive, and low-heat Pentium M would all allow for a quiet system regardless of what other modifications were made. Hopefully, this means that Puget is capable of building quiet systems without taking the extreme approach that they used for our test machine.

Much thanks to Puget Custom Computers for building us this customized review sample.

* * *

Discuss this article in the Silent PC Review Forums.



Previous 1 2 3 4 5 6

Complete|Mobile Systems - Article Index
Help support this site, buy from one of our affiliate retailers!
Search: