Alphacool NexXxos BOLD vs Prescott, Intel's Black Sheep

Water Cooling by KeithSuppe @ 2005-03-10

Can Alphacool?s Xtreme Pro Set mated with their NexXxos BOLD water block dowse the Prescott fire? Intel?s offspring have all been well behaved until Prescott, the Black Sheep came along. Sadly the little (90nm) CPU has had to endure so many epithets, such as claims it can double as a space-heater, and it converts standard H20 systems into phase-change, etc... How cruel the industry can be, and in keeping with mixed metaphors today?s test will determine if one of Alphacool?s best can extinguish the 115W flame within?

Intro/Waterblock Specs

Alphacool Xtreme Pro Set + NexXxos BOLD vs. Intel P4 550


Alphacool has become a force to be reckoned with in the water cooling world. Their designs don't seem overly complicated, or made with exotic materials, so why do they perform so well? One reason I'm sure is Alphacool's choice of components included in their sets, and in particular today's Xtreme Pro Set 240V is an excellent water cooling platform. At the heart of this set is the Alphacool 1500 DC pump made specifically for Alphacool by OASE. The Alphacool 1500 is capable of 900LPH to 1500LPH via a DC-voltage regulator with a 12V to 24V range. Below we see the Xtreme Pro Set with the NexXxos XP, water block and the Black Ice Xtreme double radiator; we'll substitute the NexXxos BOLD S775.

Madshrimps (c)


Alphacool offers several versions of the NexXxos water block and the original NexXxos XP still retains the number one spot at Watercoolplanet.de which has performance chart 50 water blocks deep. To date the best performer has been the NexXxos GOLD which has also dominated Euro H20 charts such as CaseUmbau where it's recently bumped to second replaced by the LICS (formerly Cool-Cases) M-I. LICS have just released their M-II as well which may be the block to beat.

Normally the Xtreme Pro Set were using today is based around the NexXxos XP (reviewed here), however; today we'll be sending chills through Prescott's spine with our NexXxos BOLD. Our first photo below exemplifies one of the most highly polished base-plates I've ever seen. The NexXxos BOLD is mounted on the Socket-775 Lucite mount, and Alphacool provide through-mount for both AMD and Intel models.

Madshrimps (c)


Taking a closer look at the NexXxos BOLD disassembled sitting in front of another BOLD assembled in its A64 Lucite mount, you can see the two copper plates which comprise the block. The top-plate (left) is just 4.5mm thick (approx.), while the bottom base-plate (right) which comprises the impingement zone, is just 4mm (approx.) thick! The NexXxos BOLD has to be one of the sveltest designs on the market today.

Madshrimps (c)


The BOLD's top-plate integrates the inlet which forces the incoming water through 20 small holes creating the dispersion pattern onto the impingement zone. Opposite the inlet the outlet, obviously much larger as not to create any back pressure. The rubber gasket is forced against the Lucite top-plate albeit the AMD, Intel or as in this case the Socket 775.

Madshrimps (c)


Next the bottom or base-plate comprises the impingement zone which is given the moniker extruded pin design. Base plate thickness is just 4mm, and the copper material removed around and between the pins leaves just 2mm of copper between CPU and water flowing above.

Madshrimps (c)


My hypothesis for the success behind Alphacool designs is due to its base-plate thickness. I've seen a propensity among better performing European (particularly German) where a minimalist approach is taken. Alphacool's use of materials indicates an astute grasp for thermodynamics and metallurgy. The laws of physics dictate kinetic energy (heat) where lower water-temp naturally attracts heat from the CPU, therefore placing as little material between H20 and heat-source makes perfect engineering sense.

Madshrimps (c)


A closer look emphasizes base-plate thickness which may even be less than 4mm (approx.), however; relying on the accuracy of mm increments on a common ruler is ill advised.

Madshrimps (c)


Similar to the top-plate the base-plate also utilizes a rubber gasket which pressed against the top-plate (when the four retaining screws are tightened down), creates a water-proof seal. As stated earlier, Alphacool's base-plate sports one of the finest finishes on the market.

Madshrimps (c)


It seems NexXxos BOLD has the potential to be a contender, however; it's up against what is considered to be the hottest running desktop CPU on the market, the 90nm Prescott dissipating 115W. The belief that the LGA Socket 775 (or Socket-T) version Prescott runs cooler then the original Socket 478 version is a conundrum wrapped in misconception. While the processors are basically the same thermally, the Socket 478 version only dissipates 103W. Its thermal problems initially arose from an incongruity with the platform it was prematurely released upon. The pin-design of the .09-micron Prescott and the fact power circuitry on Socket 478 motherboards were never intended to support this processor's core, were the primary cause for its early thermal catastrophes. Stories of motherboards actually melting were not exaggerated. However, the reasons why they did were widely misunderstood.

Marketing hyperbole has no effect on the laws of physics, and from 130nm to 90nm is almost a 50% reduction in the size of voltage traces (Sudhian).

Additionally D-VID circuitry which allowed fluctuations in Prescott's Vcore from its default 1.375V ~ 1.525V were not engineered into Socket 478 motherboards. When overclockers attempted to set vcore manually in their Socket 478 BIOS (even if it was just to set it at the default vcore level), this led to a plethora of adverse results. Some of these results were melted mobo power circuitry, and insanely high mosfets temps, but they also may have been responsible for original Prescott's high operating temps. None-the-less Intel engineers have worked feverishly to bring things under control, and while they haven't made substantial changes at the Fab level, where motherboard power circuitry specifications are concerned temps have dropped somewhat.

Now onto the testing ->

System/Tests/Conlusion

Test Setup:

Test System
CPU Intel P4 550 Retail-3.4GHz LGA 775
Mainboard Abit AA8 Duramax
Memory Kingston Hyper-XPC2-5400 1024MB kit
Video ATI Radeon X800 XT
Power Supply TTGI USA TT-600K04 Modular
Operating System Windows XP SP2
H20 System NexXxos BOLD S-775 CPU-kuhler, Alphacool 1500 (@ 800LPH) pump, Black Ice X-double radiator, 2x120mm/57CFM Papst, Cooplex 25 reservoir, 6mm-ID


SOFTWARE:

Temps were measured using the TTGI USA Fan-Master SF-609 rheostat, I placed one external thermistor between water block and CPU (remounting several times for the best average). I placed another thermistor within the case; the side-panel was removed due to the radiator size. Finally the third thermistor was placed outside the case to measure ambient (room) temperature.

Additionally I drew data from Abit's µGuru which monitors the Intel P4 550's internal thermal diode, among other devices. Also employed was the Panopsys utility Throttle Watch. This utility measures CPU activity as well as thermal-throttling TM1, and TM2. To reproduce a full-load environment on the CPU I used the fairly new stress tester S&M.

I've included a photo (below) of the test set-up to give an idea of the environment.
Madshrimps (c)


Below we see a close-up of the NexXxos BOLD installed on the Abit AA8 Duramax.
Madshrimps (c)


The wire entering the top of the block is the thermistor from the TTGI Fan Master. Contact between the block base, and Prescott's IHS was very snug, and to avoid air-gaps I could only get the first few millimeters of the thermistor between these surfaces. The Thermal Paste used was Arctic Silver Ceramique which I simply placed a rice-sized grain of in the center of the IHS, and then installed the water block holding the thermistor in place. I prefer this method over spreading which can incorporate air into the paste. Allowing the pressure of installation tightening to compress the paste not only disperses it naturally, it can actually force out any air pockets.

Testing consisted of running S&M, Throttle Watch, CPUID, µGuru and taking a screenshot of the resulting temps. Since S&M (as most other utilities) recognizes the Pentium-4's Hyper Threading feature as two distinct processors labeled "1" and "0", the program then split's the LOAD. While Throttle Watch indicates the processor running at 100% LOAD, S&M will often read processor version "0" as 47% and processor "1" as 54%. S&M also sees the LOAD fluctuate between versions as some screenshots captured either "1" or "0" at 100% or 99%. I've provided two thumbnails below, clicking on the left thumbnail shows Throttle Watch reading a 100% CPU-load, while S&M indicates processor "0" running at 49%. Choosing the right thumbnail will also show 100% CPU-load read-out on the Throttle Watch GUI, while S&M indicates processor "0" running at 99%.

Madshrimps (c) Madshrimps (c)


Now onto our reslting test data, I decided to run the CPU at 201FSB, 240FSB, and finally 250FSB, employing S&M I pushed the processor until the temperature reached and maintained a peak temp. My room temp was/is very low at 18°C/64F. Below we have our graph which is an extrapolation of all the data.

Madshrimps (c)




Conclusion:

Alphacool continues to release one successful design after the next.

Their grasp on PC-thermodynamics and water cooling encapsulates more then simple CPU-water coolers. In so far as this reviewer is concerned, Alphacool embraces cooling, and water cooling from the standpoint of the overall system, let's call it thermal management.

So long as they continue to view water cooling from the system perspective they will avoid the tunnel vision that comes with focusing on just CPU water blocks.

I'd like to thank Taner and the staff at Alphacool for their assistance.

My next Alphacool review will feature their NexXxos XP, NexXxos BOLD, and CAPE42-XC on the DFI LANPARTY with an Athlon 3500 Winchester against Cool-Cases CF-1 Rev.2, CC-Magic Silver, and the venerable LICS M-II in a small 6mm system round-up. Look out for my Polar Flow TT review next which will be a preclude to the 1/2" ID round-up.

Questions/Comments: forum thread
  翻译: