maze3 v cyclone 5 at ocau

[Pinkster]

initial tes results show that they perform pretty similarly... although they are only initial tests i think that repeated tests will not show much variation..
http://forums.overclockers.com.au/sh...threadid=70476

interesting... very interesting... very very interesting... still it goes without saying that a cyclone 5 rapes everything else for pelt use...

[Brad]

/me isn't going to register to view a forum page

[DigitalChaos]

yea.. you should post the results over here.... i really dont want to register just to view a page

[UpAllNight]

no regestering here either

[mkosem]

Warning: The following may shock you. Hold onto your hats, 'cos this may be a rough ride.

This is a brief one-off test conducted to do a quick comparison with the 3mm base Cyclone 5 vs the Danger Den Maze 3.

Introduction/Rant

This needs to be said here and now before you go on further and complain about my high temperatures. I made every effort to eliminate all sources of additional cooling of the in-scoket thermistor that is used to measure CPU temperatures. As such, my temperatures will seem somewhat high to many reading this review. I believe that they are actually fairly accurate. If you think I must be doing something wrong, then please consider that I am able to get most any AthlonXP AGOIA stepping CPU up to 1920-1950MHz stable (Prime95 for 6 hours) in my water-cooling setup. This has been achieved with at least 8 different CPUs. A 1400MHz T'Bird was capable of running at 1790MHz stable in my system. The reported temperatures though were always quite high by most other people's standards. If the temperatures were an indication of the poorness of my cooling setup, such good overclocks would simply not be achievable. The temperatures are as reported by my system. Nothing more, nothing less. Do not use them as a judge of if I'm doing something wrong. I am not, I assure you.

Test Setup

Motherboard: Abit KR7A-RAID

CPU: AthlonXP (unlocked)

Pump: Eheim 1250 with 1/2" brass barbs on inlet/outlet

Waterblocks: Danger Den Maze 3, Silverprop Cyclone 5 w/3mm base

Radiator: The "Big Arse" with baffled shroud using 2 x 120mm 12V 104CFM rated Panaflo H1A fans at full-speed

Thermal Probe: Dick Smith Temptec "Bodgemaster 1000" with supposed 0.1C accuracy - agrees fairly well with 2 other alcohol thermometers with respect to absolute temperatures - accuracy probably in the +/- 0.3C range - not great - but good enough for my limited needs. Measurement of the radiator air intake temperature (ambient), and water return temperature were taken.

Motherboard Thermistor: On board KR7A in-socket thermistor coated with silicone thermal paste and pushed up. When CPU was inserted it had to be pushed down with around 5kg's of force to ensure the CPU sat flat due to the thermistor pushing up hard against the base of the CPU. Great care was taken to eliminate all source of socket area air-flow to reduce the impact this would have on temperature readings. When compared with an Epox 8K3A which measures temperatures using the on-die AthlonXP thermistor, the temperatures reported by the KR7A when setup in this way are actually around 1-2C higher than that reported by the Epox 8K3A, but doesn't fluctuate wildly like the Epox 8K3A thermistor does. I believe the KR7A is giving a fairly good and repeatable method of measuring CPU temperature.

Water Loop: pump->waterblock->radiator->reservoir/pump

Flow Rates

Flow rates were measured by timing how long it took to fill a 5 litre container. lpm = litres per minute, 1US gpm ~= 3.9lpm

Pump alone (with 1/2" inlet barb): 11.5 lpm

Pump + Cyclone 5: 8.5 lpm
Pump + Cyclone 5 + Radiator: 7.5 lpm

Pump + Maze 3: 7.9 lpm
Pump + Maze 3 + Radiator: 7.0 lpm

Test Loads

3 temperatures were gathered for each block under different conditions.

The AthlonXP was set to run at 1900MHz with 2.11v. This correlates to a Radiate calculated 125W of heat load under full CPU load.

The different load tests that were run:

CPU Idle Temperature
Sandra Multimedia Burn-in Loop Temperature
K7Burn "Full Load" CPU Stress Test

Each "test" was run until equilibrium was reached, which was identified by CPU and water temperatures remaining stable or oscillating between two values for at least 30 minutes.

Results

Silverprop Cyclone 5 w/3mm base

Idle: Ambient 22.9C, Water 24.0C, CPU 38.0C
Sandra: Ambient 22.9C, Water 24.7C, CPU 46.0C
K7Burn: Ambient 22.9C, Water 24.9C, CPU 50.5C

DangerDen Maze 3

Idle: Ambient 22.9C, Water 23.9C, CPU 38.0C
Sandra: Ambient 22.9C, Water 24.7C, CPU 46.5C
K7Burn: Ambient 22.9C, Water 24.9C, CPU 51.0C

Discussion

Look to an upcoming review from me soon (2-3 weeks away) at the OCAU front-page. I'll be comparing the Cyclone 5 range and will be using the Maze 3 to keep it all honest. Quite frankly, I'm downright bloody impressed by the performance of the Maze 3. With all this speculation of "rice engineering", it seems the lads at Danger Den actually know a thing or two about water-block design and how to make it look good at the same time. The Cyclone 5 is just able to pip it at the post, but it really is a close race. Far, far closer than I would've imagined just by looking at the Cyclone 5 and Maze 3 side by side.

[xerka]

hah...thats great he called it "rice"...except this rice can back up its looks

[Brad]

just depends if you are in AU or the US. The performance is practically the same