he accuracy of motherboard controller-chips and associated software monitoring utilities have long been found to be highly inaccurate. BIOS readings are often thought to be slightly more accurate, although this is not necessarily true. If there is a fairly accurate monitoring device then it's the CPU internal thermal diodes and other types of internal diodes which are probably most accurate. The reason for this is twofold, first the simple fact they are responsible for fail-safe, or immediate shut-down to preserve the life of the CPU, and second features such as thermal throttling and Intel Speed Step technology are based on these devices. Were it not for a high level of accuracy, Intel and/or AMD could loose millions replacing damaged processors when they begin popping like Orval Redenbacher microwave popcorn. The cost would be even greater in consumer confidence and if a simple diode stands between these scenarios, then they are worth the production cost and you can be fairly sure of their accuracy. The thumbnails below exemplify BIOS screenshots indicating "PC Health" status (voltages). On the right DFI LanParty Nf4 UT BIOS page and on the left Asus P5WD2-P BIOS respectively.
Our next set of thumbnails exemplify voltages as measured through Smart guardian and Asus PC-Probe II / AI-Booster while running S&M v.1.80
found at Overclockers.ru
. All screenshots were taken at 100% LOAD running default then overclocked as indicated below.Opetron-148 / 2.2GHz | Opteron-148 / 2.7GHz | Presler-930 / 3.0GHz | Presler-930 / 3.9GHz Testing:
As the photo above indicates, all voltage measurements found in the charts below were taken by inserting the Exectech MiniTech
digital multi-meter probes into the reverse end of the motherboard power connectors. All measurements were recorded "hot" running S&M v.1.80
. Our first chart reveals the Opteron / DFI system, the second (lower) chart shows the Intel Presler 930 / Asus P5WD2-P system.Noise readings:
Using a SPL meter measuring 50cm from each system our readings were as follows. Inside the AMD system the case temp was 40°C – the dBA meter registered ~7dBA over ambient when the unit was running at full load with the 2x80mm fans at maximum RPM.
Inside the Intel system the case temp was 35° C – the dBA meter could not pick up the XP-650W as it was inaudible over the overall system noise.
Mushkin has done very well with their XP-650 which is a high-quality unit and should be at least as popular as past efforts from OCZ. Perhaps this unit will be the USA Tagan and that is a great compliment. Mushkin continues to innovate yet doesn't do so frivolously and has always been committed to the philosophy: high performance can still be found at a reasonable price.
Held all rails at or slightly above during full LOAD and overclocking
Silent operation at LOAD
Well constructed quality components
Modular cables were difficult to remove
Must void warranty to adjust pots
Combined rail switch would increase amperes for non-SLI systems
While I had my concerns about the modular design beyond those covered in this article, modular PSU’s seemed destined as mid-line budget models for many companies. The XP-650 may be one of just a handful of high-performance PSU's which also incorporate a modular cable system. This PSU performed very well given the fact each test system was watercooled and overclocked for half the tests. The Intel Presler rig in particular used a 12 ~ 24V voltage box on the Alphacool AP1510 pump, this ran continuously on the 18V position. The XSPC CrossFlow raditor included 3 x Zinruilian 12VDC/0.40A 120mm fans running flat out. The highest compliment I can pay to any component is when it replaces another in my system. The XP-650 will replace the OCZ PowerStream in my AMD rig. No price was given at the time of writing but I would expect a decent value from Mushkin whom delivers affordable quality. I would like to thank John and Lee at Mushkin for this "first look."
Questions/Comments: forum thread