The question has begun to pop up across the internets, including the the Short-Media
forums: Does upgrading from Arcitic Silver 3 to Arctic Silver 5 provide a significant
change in temperature that justifies the cost? In this review the performances
of the two thermal pastes will be tested and explained at various fan speeds,
voltages, CPU speeds, and CPU loads.

Product Description

From Arcticsilver.com:

Introducing Arctic Silver 5. With its unique high-density filling of
micronized silver and enhanced thermally conductive ceramic particles, Arctic
Silver 5 provides a new level of performance and stability. Now available
at Arctic Silver dealers worldwide, Arctic Silver 5 is definitely […] The
New Reference.

Since the introduction of the original Arctic Silver, the name has been found
in hundreds of thousands of overclockers’ sigs. Arctic Silver rarely ever
had any competitors until approximately a year ago when ShinEtsu released their
G-751 paste. Over time, Arctic Silver lost a few Overclockers; then came Arctic
Silver 5.

Why the jump from Arctic Silver 3 to Arctic Silver 5, though? In Japan, where
Arctic Silver has a large market, the number 4 is synonymous with death. Therefore,
they decided to name the new thermal paste Arctic Silver 5 to avoid the poor
marketing.

Procedure

To test the two thermal compounds the system listed below was used in a controlled
environment:

• A fresh application of Arctic Silver 3 was applied onto the CPU following
  directions direct from Arcticsilver.com.
• One day was allowed for the Arctic Silver 3 to “burn in”; it
  ran for 6 hours and then set for 12 hours turned off.
• Arctic Silver 3 was then tested under the following conditions: 1.5vcore
  2200 rpm idle and load; 1.5vcore 4600 rpm idle and load; 1.7vcore 2200 rpm
  idle and load; 1.7vcore 4600 rpm idle and load; 1.9vcore 2200 rpm idle and
  load; and lastly 1.9vcore 4600 rpm idle and load.
• The procedure was repeated for the Arctic Silver 5
• MBM 5 was used to obtain temperatures.

Test System

• Lian Li Pc61
• Abit Nf7-s Bios 18
• AMD 1800+ @ 1.5vcore, 1.7vcore, and 1.9vcore
• SLK900 U (not lapped)
• Vantec Tornado @ 4623 rpm and @ 2220 rpm
• Antec 480W TruePower PSU
• ATi Radeon 9800 Pro 256MB
• Barracuda IV 60 GB Hard Drive
• Western Digital 120 GB Hard Drive
• Lite-On 16xDVD-Rom
• Lite-On 52x24x52 CD-RW
• Vantec Nexus Rheobus

Testing

First, a fresh application of Arctic Silver 3 was applied onto the heatsink
and cpu following directions direct from ArcticSilver.com. The computer ran
for 5 hours, then sat (turned off) for 15 hours. After this cool down, the computer
was turned back on and results were gathered using a Nexus rheobus and prime95.
Each test was run for 30 minutes to allow a more accurate reading.

The cpu and heatsink were cleaned using 70% Alcohol swabs. Arctic Silver 5
was then applied and temperatures were gathered using the same procedure.

Reading the Results

Each test result shows eight gauges in seperate boxes. The first box in the
upper left is the CPU temperature. The one on the first row 2nd column is the
Northbridge temperature. The box in the 2nd row, 1st column is the Vcore for
the CPU; it shows different readings than the voltage set in the BIOS. In the
next column is Vdimm: the voltage on the memory. In the third row, first column
is the RPM of the heatsink fan (CPU fan). The next gauge shows the RPM of the
fan mounted onto the GPU. In the fourth row, first colum it shows the speed
of the CPU in MHz. The last gauge is the cpu usage percentage. 100% means a
full load and 0-5% means the machine is idle.

Description of Results

In the first column, temperatures were taken at stock speeds, voltage, low
RPM, and idle CPU usage. In the picture to the right, the temperatures were
taken from stock speeds, voltages, high RPM, and low CPU usage. In the next
picture, temperatures were taken at stock speeds, voltages, low RPM, and load
CPU usage. Then in the final picture in the row, temperatures were taken at
stock speeds, voltages, high RPM , and load CPU usage.

	1	2	3	4
Arctic Silver 3
Arctic Silver 5

In this next set, the first column is readings from a CPU at 2GHz (from 1.53),
1.7Vcore (from 1.5), low RPM, and idle CPU usage. The next coulmn is a CPU at
2GHz, 1.7Vcore, full RPM, and idle CPU usage. The next shot is of a CPU at 2GHz,
1.7Vcore, low RPM, and load CPU usage. The last column is a CPU at 2GHz, 1.7Vcore,
full RPM, and load CPU usage.

	1	2	3	4
Arctic Silver 3
Arctic Silver 5

In the last set are readings taken from a CPU at 2.3GHz, 1.9Vcore, low RPM,
and idle CPU usage. The screenshot in the 2nd column are from a CPU at 2.3GHz,
1.9Vcore, high RPM, and idle CPU usage. The next shot is of a CPU at 2.3GHz,
1.9Vcore, low RPM, and load CPU usage. The final shot is from a CPU at 2.4GHz,
1.9Vcore, high RPM, and load CPU usage.

	1	2	3	4
Arctic Silver 3
Arctic Silver 5

Conclusion

As a result of applying Arctic Silver 5, temperatures dropped between 1 and
4 degrees Celsius in the various situations tested above. Arctic Silver 5 out-performs
Arctic Silver 3 at load, idle, high RPM, low RPM, high speeds & voltages,
and low speeds & voltages. Arctic Silver 5 is the clear winner. If you don’t
have the money to go out and buy an SLK947, this is an excellent, cheap upgrade
to push your processor a little farther. Arctic Silver 5 wins in every situation.