TestingTesting several XSPC components simultaneously can present a logistics nightmare. In an attempt to clearly exemplify performance, variables must be kept to a minimum. Two primary systems will be built to exemplify the XSPC components performance. The first test system is based on a Yonah T2600 / AOpen i975X a-YDG combo, the second a Presler 930 / Asus P5WD2 Premium (955X) combo. Each system will use the XSPC
Xtreme X20 water block, XSPC
Passive Aluminum Reservoir, XSPC
R120-T Crossflow radiator (triple) and the pump used will be an
Alphacool AP1510 centrifugal pump with adjustable DC-voltage converter. Using the adjustable DC-voltage converter we will test flow rate effects on the Xtreme X20 vs. Alphacool's NexXxos XP water block. For a control group we'll utilize the HSF supplied with AOpen's i975X a-YDG motherboard for the Yonah T2600 and for the Retail Presler 930 the Intel supplied stock HSF. AOpen's 1U cooler can be seen below.
Water Block comparison will place XSPC's Xtreme up against the original X20, Alphacool
CAPE KC42-X2, Alphacool
NexXxos XP and the
Viscool V2. The passive reservoir will be compared to Alphacool's
Cape Coolplex Pro 25 External. Water temp tested between reservoir and radiator. Finally the R 120-T Cross Flow radiator will be compared to Alphacool's
NexXxos Pro III with each radiator running three Xinruilian RDH1225S 120mm fans (98CFM @ 12V).
The (un)Usual Suspects As some of our waterblocks had oxidized over time I used Arctic Silver
Arcticlean to clean and prepare the base surface for TIM application. Arctic Silver
Ceramique was used on all waterblocks and stock coolers. The Xtreme X20 was "on the Lamb" during the "line-up" hence it's absence from the photos below.
Intel Test System |
CPU | 1.) Yonah T2600ES 2.) Presler 930 |
Mainboard | 1.) AOpen i975X a-YDG 2.) Asus P5WD2 Premium |
Memory | 1.) Crucial Ballistix PC2-8000 (2x1024MB) 2.) Super Talent PC2-8000 (2x1024MB) |
Graphics | 1.) Leadtek 7950GX2 2.) AOpen Aeolus 7800GTX-DVD256 |
Storage | 1.) Maxtor DiamondMax SATA 300GB 2.) Maxtor DiamondMax SATA 120GB |
Optical | 1.) Plextor PX755SA DVD/CD-RW 2.) NEC DVD/CRW IDE |
Power Supply | 1.) Corsair HX620 2.) PCPower&Cooling Silencer 750 |
Cooling | Stock (AOpen included) H20 - XSPC Xtreme / XSPC Passive reservoir / XSPC R 120-T radiator (3xXinruilian RDH1225S 12cm 98CFM / 12V) / Alphacool Centrifugal / DC-Voltage Converter |
Operating System | Windows XP |
Each block tested possesed unique attributes. The original X20 was a simple but effective cross-channel design. Alphacool's all copper Cape KC42-X2 utilizes a large "stepped" cylindrical cone seen
here. The cone's base makes contact with the CPU and through this cones heat is transferred. As water flows down around the cone's steps, processor heat is carried away through the KC42-X2's outlet. Finally the Viscool V2 features one of the most unique impingement designs as I discovered in my
V2 review. An amalgamation of a micro-channel, extruded pin, with copper ring to force the water stream into the very center of the impingment area the Viscool V2 is another block which would benefit from inreased flow rate and even more so from a 1/2" system.
To reiterate our first test will involve running the Xtreme X20 at different flow-rates while stressing our overclocked Yonah to full load using
S&M. Temperature measurements will rely primarily on each processor's on-die thermal diodes. These may not be as accurate as other independent devices, however,at the risk of sounding obtuse, so long as they are consistant in their inaccuracy temperature differential between components (performance) can still be shown. CPU makers have invested considerable resources ensuring these diodes are more accurate, especially given the fact they are resposnible for features such as On-demand, Thermal Throttling, Speed Step as well as the safety of the CPU where overheating is at stake. As far as monitoring software is concerned, I've searched the Internet thoroughly testing and re-testing just about every temp monitoring software available. At prsent the program
Core Temp seems to be the most accurate, especially for dual core. To simplify matters I've added the total of both temps and divided by half providing an average. Below is an example of Core Temp v. 0.9.0.91, SpeedFan and S&M running 100% LOAD.
Testing began with noise levels. Considering were using three Xinruilian
RDH1225S 12cm fans displacing 93CFM at 41dBA this is not going to be a system built for silence. These fans do have a speed sensor feature ergo a rheostat can resolve the noise problem when not overclocking. For our purposes I allowed the fans to run full out.
Using an Intellisystems AR824 SPL meter, sound levels were measured from a distance of 1m registering at 42dBA. Measurments were taken with room noise at its lowest, late at night with all windows closed and all appliances turned off (TV, radio, etc.).
Onto temp tests...