2.
Great review - excellent effort put into details and in-depth study rather than just "run it and write about it" as some other websites do. For one thing, at least xbitlabs is not too arrogant to think that they do not need to check through the manual in case of any special feature/requirement that individual products may require - some reviews would state that they didn't know how to do this or that, or didn't know what this or that was for, when the answers would have been easily found had they read the friggin' manual. :P
The only shortcoming of the review is perhaps the lack of subsystem tests - SATA/IDE, USB, FireWire and especially network (throughput/CPU utilisation) and audio (e.g. using RightMark audio analyser). The last test is particularly relevant because the IN9 uses a codec from Realtek - which has picked up a reputation for faulty EAX implementation and lower audio quality (THD/crossover/SNR) compared to the 'other' major HD audio codec manufacturer, i.e. Analog Devices. Of course some would dismiss this and say that any user serious about audio quality should be using a discrete audio card, but then not every user can afford to do that. Besides, that's defeating the point of having onboard HD audio - might as well just stick to AC'97.
The hissing sound may be due to the power circuitry running out of spec or at the limit - common in switch-mode power supplies. Digital PWM certainly promises greater stability due to the very accurate and fast feedback for tight Vcore control. It also promises greater life expectancy due to the fact that it does not require electrolytic capacitors for bulk smoothing - just small ceramic SMD ones (which do not dry up).
However, this latter feature may be a potential weakness of digital PWMs when raising Vcore above nominal, especially during overclocking when the CPU frequency, and hence power draw, also increases well above nominal. At least one review has noted that some implementations of digital PWMs appear to do only so-so when it comes to handling highly out-of-design-spec loads as the SMD capacitors and/or Pulse coils heat up significantly in such scenarios.
In light of that, if xbitlabs had an infrared thermometer or thermocouple, perhaps they could study the temperature gradient of the SMD capacitors, Pulse coil and even the Volterra regulators (which would mean without the heatsink) when increasing Vcore and/or current draw. The infrared thermometer would probably be more useful as it can measure the tiny SMD caps more easily than with a thermocouple. xbitlabs could also rig up a quick sound detector... use a mini condenser microphone to try to locate the source of the hiss.
With all this data, and xbitlabs' typical thoroughness, xbitlabs could help the community examine the characteristics of digital PWMs under various scenarios... after all, if it was so perfect, there must be some reason why some major manufacturers are still reluctant to implement it over the current analog feedback PWMs (existing contracts with component suppliers aside).
Just my 2 cents! :)
[
Posted by: Stele

|
Date: 05/02/07 07:33:15 AM]