RADIICAL COMPUTING

The review page contains all of the previously reviewed Radiical products along with the testing conducted on products we stock for sale. This area will become the archive for reviews from the test results page. We will maintain regular contact with this page and move material to it as the test results page gets overcrowded.

I became interested in under voltage and default overclocking some time ago. I could not come to terms with incessantly raising the voltage to get additional performance with a corresponding higher risk of component failure. The thought occurred to me to try to achieve higher over clocks using less than or default voltages. The results were of course failure using the traditional forms of water block design. The blocks simply were not up to the task. Enter the X-Block here, hand designed to meet this need.

I have taken an old Abit KR7A fitted an AMD Palomino 1600xp and leaving everything in the bios at default settings including the ram commenced overclocking only the FSB. Rather than go into a complex discussion and presentation I thought that the results and just the results might be of interest here. For the purists, yes, the overclock is stable and the computer is constantly running Folding@Home with zero error messages. The temp results are at an ambient of a chilly 18 degrees Celsius. The temperatures recorded by MBM5 are fully loaded temps with Folding running at 100% CPU usage. If you wish to subtract 3 degrees Celsius from the MBM5 figures you will arrive at the Idle temps for this overclock. Compunurse readings at the core are 23.4 degrees Celsius.

Please Note that this is not the maximum overclock arrived at with this block and this CPU.

I believe that a whole new world of sustainable overclocking at close to default voltages is becoming available to overclockers. The tradition of raising voltages and FSB to very high levels will still be there but with the opportunity to revisit low voltage overclocking with the new water blocks available there seems to be less and less need to do so.

Just some quick data regarding the new water block design. Very early results at present given below in the CPUID graphic. Tim joint maturity is not yet there so this result will improve over time.

A lot of interest has been shown by e-mail in the Radiical™ Bix shroud. To assist people who wish to purchase this shroud I have included below a picture of the Bix in a Chieftec case setting along with the MBM5 temps and detail's of performance.

The picture above shows the Bix with shroud in a chieftec casing. Details of the current overclock for this machine are P4 1.8 Overclocked to 2.5mhz stable.

Please be aware that the Bix was not designed for a shroud to be fitted and that this product does require some adjustment when fitting. All attempts have been made to ensure trouble free installation,however care should be exercised so that the radiator is not damaged.

Testing of this water block has been lengthy due to the many varied adjustments to the prototype device. The balance between presenting a block which is achievable from a manufacturing point of view and cost of production has been challenging. The durability of the design has followed a six months testing process during which time "dirty water" and high ambient temperature testing has occurred. It would be of no value to the purchaser to find that the waterblock corroded after a period of some months or to find that leakage had destroyed their costly graphics card. We are confident in the design elements having passed our tests with flying colors.

The Radiical X-Block© is different in many respects to the conventional "channel" style of waterblock. We do not claim it is a "super" block but the X-Block© performs in the upper range of water block design. This can be stated with confidence as it has compared favorably with "Cyclone 5" and other waterblocks in our testing.

Stability; All tests conducted with Folding@Home running at 100% CPU usage for eight hours.

Stability; All tests conducted with Folding@home running at 100% CPU usage for eight hours.

Conclusions:
The review conducted here should not be considered as definitive and although I have attempted to be as objective as possible please remember that this review is written by the company that manufactures the X-Block.

There are limitations to the degree of accuracy of data that can be reasonably obtained from a review of this type. To obtain scientific results of greater accuracy would require far more sophisticated technology and an improved methodology than that used here. Having said that the review methodology has merit in that it exceeds procedures used in current reviews and starts the development of standards of practice for reviewers of this type of hardware. I look forward to their impressions of the blocks reviewed.

It is anticipated that a further review will occur in the future to evaluate overclocking ability of the blocks tested at 2 volts and over. However, I hold some reservations regarding the Silverprop® Evo-SP performance given the high temperatures already seen at low voltages, however, there are bad eggs in every basket and perhaps this was a Friday unit. Perhaps Silverprop may see fit to send another unit for testing. - Top

I have received a lot of e-mail from people who have read the three-block comparison carried out a little while ago. The majority of the e-mails ask to see results of each block tested on the same motherboard under exactly the same conditions, the same CPU used in each test, the same Ram… you get the picture. Okay here we go for an "in your face" comparison.

The water blocks all performed well however there are differences that can be observed in performance. For the full detail read on.

Case Specifications: Mid tower case (Generic) no sides fitted
CPU: AMD 2100XP "B" Core AIUHB 0302 (Previously tested to reach 2573Mhz)
Motherboard: Epox 8k9AI
Ram: Corsair CMX256A-3200C2
Hard drive: Seagate ATA 100 40gigs capacity.
Operating System: Windows 2000 Pro with service pack 3
Video Card: Nvidia GF4 Ti4200
Water Pump: Eheim 1250
Fans: Evercool aluminium 120mm rated at 84.75cfm
Tubing: Tygon R-3603 ½" ID (3/32" wall)
Radiator: SS3 derivative
Temperature Measurement: Compunurse probes on the surface of the water block and next to the core. Additional cross-referencing was carried out using MBM5.
Ambient temperature: Measured 1.5 meters from the test units by an alcohol based wall-mounted thermometer.

Procedure
1 Each block was tested from default of 1.60 volts. The front side bus (FSB) was then raised until the computer would not boot into windows (BSOD). The voltage was increased and the FSB raised again until BSOD. Final voltage to be used was 2.00volts.

2 The over clocked values reached by a majority of the blocks were then used as test points. When a block loaded into Windows but subsequently "Froze" or "Locked up" it was recorded as a Zero or "failure to complete the test". Measurement of operating temperatures of the blocks was measured and recorded.

3 Maximum over clock was then arrived at by the same procedure of raising the FSB and increasing the voltage until 2 volts was reached then the temperature of each block was recorded.

4 To avoid contamination of the operating system (OS) files the operating system partition was deleted and a new image of the OS files was installed prior to testing a new block. The testing of the blocks was conducted in this way so that any block being tested suffered no impediment from previously corrupt or damaged OS files.

Figure One illustrates the temperatures observed for the three blocks at default voltage of 1.60 volts. (133 X 13)

Figure Two illustrates the performance temperatures of each block as they are over clocked from 1.85 volts to 2 volts. Please note that a zero indicates a failure to complete loading Windows

Of the three blocks tested the LRWW and Radiical X- Block were within the temperature range of 0.7degrees (as measured by the Compunurse) of each other and could not be separated. The maximum over clock for LRWW and RadiicalX- Block was within 1 MHz of the other. This difference is not considered significant.

The EVO-SP performed well at low to moderate voltages but did not respond well to upper the voltages or extreme over clocking. The EVO-SP did not complete the extreme over clocking tests. The picture below of the EVO_SP shows the base plate footprint and the residual Artic Silver 3. It is clear that the contact with the CPU is irregular and unstable. In the picture the darker line to the right of the residue on the base plate is a polished area of the copper base. I believe this indicates severe torsion stressors were placed on the water block base by the attachment mechanism. Couple this with the previous findings of torsion twist in the base and the poor results at high voltages can be explained.

The benefits observed from thin-based designs are that more stable over clocks can be attained and, more importantly, sustained at lower voltages with a reduced risk factor of component failure. It is clear from discussion with many over lockers that they do not "run" their CPU blocks at 2 volts to gain an extra 20Mhz and would rather reach a stable over clock at lower voltages if this were possible. The practice of highly over clocked machines running constantly at high voltages should be regarded as an urban myth.

The over clocking of CPUs to attain greater performance is a "hit and miss" practice. I have seen, under testing, CPUs with the same batch numbers perform entirely differently at the same values. The validity of this review is based upon the data being derived from the use of exactly the same components for each test, the only variable being the blocks undergoing testing. It should be kept in mind that this review in no way guarantees the same performance of any of the blocks with other CPU's.

WATER BLOCK REVIEW:
A Comparative Analysis of the Little River White Water Block, Silverprop® EVO-SP and Radiical ™ X-Block Series1

I suppose anyone new to water-cooling a central processor (CPU) would have to ask the question "why?". I know from my own experience that whenever I mention that I water cool computers I get looks ranging from frank disbelief to incredulity. One lady I met looked blankly at me and said, "Do you mean you put your computer in the bath or something. Why would you want to do that ?"

Lets face it liquid cooling is not a new thing. Twenty years ago it was the common method of cooling the "super computers" of the day. The emergence of water-cooling and its evolution into today's desk top computer market began with a group of, erstwhile, experimenters trying to get that little bit more power out of their then top of the range CPU's. Little did they imagine where that process would lead.

The heart of the water-cooling system is arguably the heat exchanger or water block, which transfers heat away from the CPU into the fluid medium. It must function for long hours in the confines of the computer case taking heat from the surrounding air and the CPU. It is characteristically made of a metal, which can absorb and transfer heat rapidly. Two common metals are used namely aluminium, for its cost and ease of machining, and copper. Copper has emerged as the "premier" metal although its cost is higher and it is more difficult to machine.

Today we are reviewing three copper water blocks; Cathar's Little River White Water Block, Silverprop's® EVO-SP and Radiical's™X-Block Series1. The first part of the review will discuss the design features of the three blocks, their appearance and their installation procedure. The second part of the review will test each of the blocks under controlled conditions and discuss their performance.

The Little River White Water Block has been made famous by Stew Forster (AKA) "Cathar", a member of Overclockers Australia (OCAU). For some time this particular water block has lead the field in performance. The unique design of micro fins and an injector nozzle combined with a three-barbed flow system add up to a formidable cooling solution.

The installation procedure of the LRWW block uses four mounting holes commonly seen on today's motherboards. Mounting is carried out by inserting the four supplied steel bolts with the acrylic washers placed to insulate the board. Then the compression springs and wing nuts are fitted, For those unfamiliar with the three-barb system the center barb is water "IN" and the outer barbs are water "OUT", returning the heated water to the radiator for cooling. An adapter plate can be purchased to attach the block to Intel based CPUs . A manual is supplied to assist with fitting of this block . For the purposes of this review I have dispensed with the springs and "hard mounted" the water block during testing.

Our second contender on the pictorial catwalk is the latest offering from Silverprop®, well known for his Cyclone series of water blocks. Although the Cyclone series followed a more traditional line of channeled block design the newest versions have also moved into the micro channel and nozzle system. Jeremy has chosen the name "EVO" for his new range of blocks. The EVO-SP is the "top of the range" version of this series. The version suitable for AMD CPUs is tested.

Installation of the EVO-SP uses two three holed Ziff retaining clips much like many of the air-cooling units commonly available. Each retaining clip is held in place with a plastic thumbnut. The retaining clips are inserted into the Ziff tabs and then the water block so that a downward pressure is exerted by the thumbnuts forcing the water block base to contact the CPU. No retaining springs are necessary. A manual is supplied to assist with fitting of this block . For the purposes of this review this method of securing the block to the CPU will be referred to as "hard mounting".

The third contender is our own water block offering from Radiical ™ named the X-Block Series 1. This water block is perhaps, the newest company named block entering the commercial field and it is with some trepidation that the "new kid on the block" comes up against the more seasoned performers. The X-Block Series 1, like the EVO-SP and LRWW blocks, incorporates new innovations one of which is the micro or fine channel base plate design.

Installation of the X-Block Series 1 uses the four mounting holes in the motherboard in the same configuration as the LRWW block using steel bolts, acrylic washers and wing nuts. Where installation differs is that the X-Block Series 1 is not supplied with springs and therefore is "hard mounted" against the CPU face. Some skill is required to accomplish this but the manual supplied greatly simplifies the process.

Product Comments

Before testing begins its appropriate to mention something about the appearance and presentation of each block. It is noticeable that the three blocks show markedly different approaches to solving the problem of heat exchange. Perhaps the most significant difference is their appearance . Two blocks , the LRWW and the X-Block Series 1, have a distinctly industrial "no frills" appearance whereas the EVO-SP from Silverprop® wins hands down on presentation. Now, before you dismiss this observation as merely time wasting, just consider how much importance looks determine the products you buy. We all aspire to get the best-looking products that we can from stereos to cars to houses and so on. The problem with appearance of course is that no matter how good a water block looks its function is to exchange heat. It is imperative that functional performance can be demonstrated irrespective of the outer appearance.

On this issue of function lets just look at a couple of early, but nevertheless important, criticism about the fitting of the Silverprop® EVO-SP block. Although the idea of attaching the block using the Ziff holder clips, like a normal air-cooled unit, is not unique in design it seems fraught with inherent dangers to me. The idea of a novice water cooler fitting this unit and not noticing that the recessed edge of the water block base has to be clear of the Ziff holder is a major danger. In commerce it is "buyer beware" but in trying to fit this block I had to refit the block several times before I was satisfied that the unit was fully contacting the CPU and not "resting" on the Ziff socket. When applying the clips to the Tygon® tubing the black plastic thumbnut came into contact with the outlet tubing. I found this to be very frustrating as it disturbed the TIM joint on each rotation of the thumbnut.

For anyone interested in fitting tubing such as Tygon® or food grade tubing with a greater than 1/16" wall size be aware that motion contact with the CPU Tim joint will occur and as a result performance of this block can be seriously degraded. Those issues aside the fitting instructions included by Silverprop® are sufficient for most people to avoid damaging their CPU.

Comments about the internal configuration and design of the insides of the three blocks will not occur here as the Silverprop® EVO-SP and Radiical ™ X-Block Series 1 are both subject to pending patent applications.

The test units are three "open" cases. Case covers were left off so as to reduce internal heat load contamination. An 80mm fan was placed above the area of the mofsets on each case to cool the mosfets and other motherboard electronics as illustrated in the picture below.

Case Specifications
CPU: AMD 2100XP "B" Core AIUHB 0306
Motherboard: Epox 8k9AI
Ram: Samsung PC 2700 DDR 256 Megs
Hard drive: Seagate ATA 100 40gigs capacity.
Operating System: Windows 2000 Pro with service pack 3
Video Card: Nvidia GF4 MMX
Water Pump: Eheim® 1250
Fans: Evercool aluminium 120mm rated at 84.75cfm
Tubing: Tygon R-3603 ½" ID (3/32" wall)
Radiator: SS3 derivatives
Temperature Measurement: Compunurse probes on the surface of the water block and next to the core. Additional cross-referencing was carried out using MBM5.
Ambient temperature: Measured 1.5 meters from the test units by an alcohol based wall mounted thermometer.

Alrighty then I guess you realize we are after replication here. So far so good. Lets get onto the test procedure and see how we go with that.

1 Load testing will be carried out using Folding@Home after a warm up period of one hour once baseline default temperature measurement is completed. Unless otherwise stated all temperatures will be recorded with the machines under load.
2 All blocks tested are hard mounted without the use of springs.
3 Measurements will be taken of temperature base lines at normal ratings for the CPU's of 1733 MHz at a default voltage of 1.60 volts. Each machine will then run Folding@Home at 100% CPU usage for the remainder of test. Temperature measurements will be taken to simulate a "loaded" situation. Ambient temperature will be reported on all tests for which results are presented
4 Each CPU will then be raised to the highest over clock possible in incremental fashion whilst measuring the operating temperature until failure to boot to Windows occurs. When an average attainable over clock is found that each water block could attain the temperatures will again be measured in an "idle" and "loaded" test.

Test 1 Temperatures at Baseline default voltage: 1.60volts
Ambient: 23 degrees Celsius
Default: 1733Mhz
Temp Idle: 28C
Temp Loaded: 29C

Test 2 Temperatures at Over Clock One Voltage: 1.65volts (166 x 13)
Over Clock: 2167.16Mhz
Temp Idle: 30C
Temp Loaded: 32C

Test 3 Temperatures at Over Clock Two Voltage: 1.95 volts (180 x 13)
Over Clock: 2353.28Mhz
Temp Idle: 35C
Temp Loaded: 40C

Comments:
Good low voltage performance ranging up to 1.95 volts with some throttling back at about 1.70 volts. As you would expect still the performer to beat.

Test 1 Temperatures at Baseline default voltage: 1.60volts
Ambient: 23 degrees Celsius
Default: 1733Mhz
Temp Idle: 30C
Temp Loaded: 32C
Test 2 Temperatures at Over Clock One Voltage: 1.65volts (166 x 13)
Over Clock: 2167.16Mhz
Temp Idle: 33C
Temp Loaded: 35C
Test 3 Temperatures at Over Clock Two Voltage: 1.95 volts (180 x 13)
Over Clock: 2353.28Mhz
Temp Idle: 44C
Temp Loaded: 50C

I have some reservations about this block. It is without doubt the most troublesome block that I have ever fitted and I don't regard myself as a novice in this area. It took quite some time to get the unit fitted correctly simply because I began believing that the 3/32 wall thickness Tygon tubing would fit well to this block as it had to all the others tested. It did not fit as I have described above disturbing the TIM joint on each turn of the Thumbnuts. I eventually had to remove the Tygon and replace it with Eheim hose. The effect this had on potential performance is unknown. When I had replaced the Tygon tubing I still had difficulty getting anywhere near the default temperature figures of the other two blocks. After repeated refits of the block I checked the AS3 footprint to see if I was getting good contact over the base area the picture below shows the uneven and disturbed footprint with what appears to be a section of the CPU face which is not in contact with the water block base. (My apologies for the out of focus view).

I have no idea how this footprint could occur unless the base plate was flexing under pressure. I have been in contact with and sent Silverprop® a copy of this picture to find out if this has occurred before. Silverprop® have attempted to assist me with the issues here but we have been unable to address this problem. Silverprop® have been kind enough to offer to inspect the block themselves if I return it to them. I will do so and retest the block later on as an addition to this review however for the purposes of this review the findings must stand on their merit or credibility will suffer. The purpose of this review was to observe the out of the box performance of each block.

On the subject of temperatures I have taken a moderate view to this blocks use. I must admit it is able to handle some under clocking but with the contact footprint above I was very reluctant to push it. I have not undertaken a maximum over clock on this unit for this reason.

Test 1 Temperatures at Baseline default voltage: 1.60volts
Ambient: 23 degrees Celsius
Default: 1733Mhz
Temp Idle: 28C
Temp Loaded: 30C
Test 2 Temperatures at Over Clock One Voltage: 1.65volts (166 x 13)
Over Clock: 2167.16Mhz
Temp Idle: 31C
Temp Loaded: 33C
Test 3 Temperatures at Over Clock Two Voltage: 1.95 volts (180 x 13)
Over Clock: 2353.28Mhz
Temp Idle: 36C
Temp Loaded: 41C

The Radiical™ X-Block Series 1 performs up there with Cathar's LRWW block. Like Cathar's LRWW block, the Radiical™ X-Block Series 1 was developed out of a keen interest in producing a solid performing block easily adapted to both AMD and Intel CPU's. Radiical ™ have developed a new design feature by using a two-barb system rather than three barbs.. The major design features are set for new AMD CPU releases based on the use of heat spreaders. The Radiical ™ X -Block Series 1 can also be used on an Intel system with an alternative attachment system which makes it a cost-effective block.

Comparisons of temperatures observed at Default voltage, Over clock One voltage of 1.65volts and Over clock Two voltage of 1.95 volts.

I think it was the TV personality Professor Sumner Miller who used to ask the question "why is it so". After reviewing the three blocks I was dissatisfied with the results of the EVO-SP block and this very question remained unanswered. I returned to re test this block simply because I could not believe that my temperature findings were the best this block could do. I decided that a bit of lapping action was required on the base, yep there goes the warranty and then a very careful refit. Switch on and we are away.

Starting with 1.60volts and 166Mhz I decided to raise the FSB by 1Mhz at a time. The result was quite incredible in comparison to my earlier attempts the temperature remained at a constant 33degrees Celsius all the way through the overclocking until I reached a FSB of 174Mhz when the CPU demanded 1.65 volts. I still have not reached maximum overclock at this voltage but have already reached the sustained test overclock (181Mhz FSB 2353Mhz) of the LRWW block at a voltage 0.15volts lower. In addition there is only one degree difference in temperatures between the two blocks. The interesting part for EVO-SP owners is that this block and this particular CPU work better at low voltages rather than high voltages where the EVO-SP is out gunned. I will continue to test this block in the next few days to establish maximum overclockability at the lower voltage range.