Extreme Kit Testing

Radiical Extreme Kit versus the AMD Phenom 2 Thuban 6 Core 1055T

The Extreme Kit

E KIT Contents

Introduction

Perhaps the biggest problem with being an early adopter of new technology are the teething problems that are likely to occur. In our experience with the 1055T the first problem to emerge after building our system was the lack of recognition of the new CPU. This particular problem is reflected in the number of bios upgrades being released by the major motherboard suppliers to accommodate the new 6 core CPU structure. For most people the first build task will be a bios upgrade to first identify then support the new CPU in older motherboards. In our case even though we used a Gigabyte 890GPA-UDH3 Version 1 motherboard the release bios did not identify our CPU and a bios upgrade was required. We used the beta bios F7c for our overclocking.

The other pressing teething problem we found in our initial testing of the 1055T CPU was the ram memory speed used. We initially tested with Gigabytes recommended GSkill F3-17066CL9D-4GBTDS but met an unidentified issue with crashes to the desktop at anything above a CPU frequency of 250. This problem was rectified using some Corsair CMX4GX3M2a1600CP sticks and once round and over the 250 hill we replaced the Corsair with the GSkill to run up to 4Ghz. Some time later we managed to identify an issue with this bios where the turbo boost would intermittently turn itself on giving us an impractical boost to 4.5Ghz. An overclock which could not be sustained at the voltage and ram settings used.

Our original goal was to produce a mid to high range 24/7 overclock of 1Ghz above default and with the assistance of our new Extreme Kit sustain this at standard voltage. This was found to be difficult due to a voltage droop teething problem with the bios that was quite noticeable. We ended up adding to the bios setting to stabilize the voltages at as close as we could to the default. This solved many instability problems in particular multiplier testing of Bad Company 2 (see testing below) so I would imagine that a more mature bios would address this issue.

Okay so final teething problem temperature measurement. AMD have released a very useful program called AMD Overdive (AOD). This is a compound program of stress testing, monitoring and overclocking just to mention a few functions. It is at the time of writing probably the authoritative temperature measuring software. The result coming from this program demonstrate that the AMD hexa-core CPU is one cool customer. The general thoughts about the temperatures coming from this program are that there should be an offset of 10 degrees added to the temperatures being reported by AOD. I will leave that to others to debate. Our position will be to measure water temperatures for the purpose of obtaining test results for the new Extreme water cooling kit.

Testing

We decide to run multi testing Prime Blend and AOD Stress at the same time to test to establish stability. Testing with Prime and AOD was uneventful and even with the voltage droop mentioned earlier stability was confirmed. However, we could not maintain the same degree of multiplayer game testing stability in the game Battlefiled Bad Company 2 unless we added on the additional voltage described earlier. The results presented below are with the added voltage.

Testing at 3808Mhz

Coolant temperature testing

Okay so on with the additional testing for the E kit. To extend the software results presented above we carried out water in/air in delta results at 3808Mhz to compare with our existing data on the E kit.

Testing was carried out with an ambient temperature of 19.20 degrees air in.

Coolant idle temperatures at 3808MHz were 21.20 Degrees Celsius

Coolant load temperatures were carried out with Prime 95 Blend and AOD Stress test running at the same overclock.

Coolant load temperatures at 3808MHz were 27.10 Degrees Celsius

Interestingly the AOD voltage for the CPU was reported as 1.40 volts (1.488Volts in CPU-Z) which is probably a function of rounding down of the voltage. The full load temperature was reported at 26.5 degrees a figure 0.6 degrees less than the returning water temperature.

Testing at 3920Mhz

Testing at 3920Mhz

The same Prime blend and AOD stress testing was used.

Only marginal changes were seen during this testing and are more probably related to slight variations in ambient and heat soak effects. Overall temperature performance for this CPU is quite amazing given that the delta variation is minimal.

We did not bother testing this kit at over 4Ghz with the bios limitations we had.