Testing
The Thermalright SI-128 SE was tested on a red-hot Pentium 4 EE based test rig consisting of the components listed below. The ambient room air temperature was maintained at 23°C ±0.5°C. Four instances of CPUBurn were executed at the same time (two to load both physical cores, and two to load the two virtual HyperThreading cores), which resulted in 100% CPU usage. Tests were conducted with four different 120mm fans to see how the cooler performed with varying degrees of airflow.
- 120mm Nexus (1000 rpm, 36.9 CFM, 22.8 dBA)
- 120mm Panaflo FBA12GL1A (1700 rpm, 68.9 CFM, 30.0 dBA)
- 120mm Antec TriCool 3-Speed (1200/1600/2000 rpm, 39/56/79 CFM)
- 120mm Delta SHE (3700 rpm, 152 CFM, 53 dBA)
Test Rig Configuration
- Asus P5N32-SLI Deluxe motherboard
- Pentium 4 Extreme Edition dual core 955 @ 3.46 GHz
- (2) Corsair CM2X512-8000UL DDR2
- NVIDIA 7800 GTX 512 MB video card
- Western Digital WD1200JD S-ATA HDD
- SilverStone Olympia 750W PSU
- Windows XP Pro with SP2
- nVIDIA 91.31 nForce driver
A small Omega thermocouple is attached to the side of the 955 IHS with Arctic Alumina thermal epoxy to provide accurate CPU temperatures. The measurement equipment used during testing included:
- CPU/IHS – Barnant Model 115 digital thermometer (accuracy +/- 0.4º C)
- Ambient air – Barnant Model 115 digital thermometer (accuracy +/- 0.4º C)
- Extech Model 407738 digital sound level meter (accuracy +/- 1.5 dB)
Software Utilities
- Lavalys Everest Ultimate Edition 2006 (hardware monitoring)
- CPUBurn (load the CPU)
For comparison, I’m including the results from five other popular HSFs we recently tested on the LGA775 platform. All HSFs were tested on the same EE 955 CPU under the same conditions.
- Scythe Andy Samurai Master with bundled Scythe 120mm fan
- Enzotech Ultra-X with bundled Delta 120mm fan
- Thermaltake Big Typhoon VX with bundled fan
- Thermalright Ultra-120 with Antec TriCool 3-speed fan
- Thermalright XP-120 with Antec TriCool 3-speed fan
The following data is presented for comparative purposes only. Your actual results may be different depending on the variables unique to your system (CPU, overclock, ambient temperature, case air flow, temperature monitoring, etc).
Amb – Ambient room air temperature
CPU – Temperature reported by Everest utility (internal diode)
Tc – Temperature obtained with calibrated thermocouple attached to 955 IHS
∆T – Fully loaded Tc temperature rise above ambient temperature
dBA – Sound pressure level recorded 3’ away (background ~29 dBA)
Note: My original P5N32-SLI Deluxe motherboard died awhile back and the CPU temperature reported by the replacement board is so far off (~20°C) its not worth mentioning – the thermocouple attached to the 955 IHS is much more reliable.
The Thermalright SI-128 SE CPU cooler did an excellent job of keeping the fully loaded CPU cool. This is the first down-draft style cooler we have seen that offers close to the same level of performance as some of the larger tower style coolers on the market. For example, when compared to the Thermalright Ultra-120 (both using the same 2,000 rpm fan), the full load CPU temperatures were within 0.2°C of each other (well within the experimental margin of error).
When used with the Nexus Real Silent fan, the SI-128 SE is virtually silent, but the minimal airflow didn’t do much for cooling the socket area components. Changing fans to provide more airflow improved performance considerably at the expense of noise. If you can stand the noise, a high output fan will produce outstanding results. One of my personal favorites is the Panaflo L1A, which delivers good cooling with minimal noise. For greater flexibility, a good high-speed fan and speed controller combination might be a good choice (120mm Delta fan AFB1212H-F00, rated for up to 82.7 CFM and 38.5 dBA at 2,500 rpm when operating at 12V).
