Asus A7V266-E RAID Motherboard
A7V266-E RAID Motherboard Review
When Asus is mentioned with respect to motherboards, what do you think ?. The words that spring to this reviewer's mind are stability, quality and reliability. Asus have worked hard to gain this enviable reputation, work that has seen them leap to become to #1 manufacturer of PC motherboards world-wide.
The bulk of Asus' revenue is accrued from the OEM channel. You'll typically find an Asus board sitting in a ready-built PC. That's not to say that Asus have not been busy in the retail channel, just one look at their extensive product catalogue informs you of that.
Today we're casting our critical eye over Asus' KT266A offering, the aptly titled A7V266-E. This is Asus' interpretation of how a KT266A should look and perform. Asus, like most other manufacturers, simply replaced their KT266 offering by substituting the more efficient KT266A North bridge in place of the older KT266 North bridge chipset.
Let's have a look at the specification in a little more detail.
On-board IDE RAID controller
Bundle and presentation
Asus have always impressed me with their presentation and the A7V266-E is no different. The board packaging is subtle and not completely over-the-top like some. You receive an excellently written main instruction manual, a quick setup guide for those who simply want to fire the motherboard up with the minimum of reading, a USB 1.1 adapter card, a CD containing the motherboard software and, of course, the ubiquitous ATA cables in the form of two ATA100 and one floppy cable. The software package, although not as impressive as some we've seen, is more than adequate. Everything is stowed away neatly. Once again, the impression of quality is evident here.
Features and Installation
Just a brief glimpse reveals that Asus have loaded the board with some rather nice goodies. We see RAID has been incorporated by the use of the ATA 100 compliant Promise PDC 20265 chipset. We're somewhat surprised that Asus chose not to go with the newer, ATA133 compliant, Promise PDC20276 as seen on other motherboards. However, there is absolutely nothing wrong with the PDC 20265, it provides both RAID 0 (striping) or RAID 1 (Mirroring). As the names suggest, RAID 0 amalgamates 2 IDE hard drives so that they are seen and perform as a single larger drive. Although latency is cut in half, you do sacrifice some security as RAID 0, by it's very nature, has no redundancy. RAID 1 is more commonly used to back up data as one drive simply mirrors what's is on the other. The RAID ports can be used as stand-alone IDE ATA100 ports by the changing the positioning of a single jumper.
On-board sound is having something of a renaissance of late. The very competent C-Media 8738 chip makes an appearance on the Asus A7V266-E. This sound chip capable of 5.1 channel stereo, although not a true hardware solution, is more than adequate for Windows duty. We found it difficult to tell the difference between this and our Sonic Fury, praise indeed for the on-board solution. I'm sure OEM system builders will be more than happy at the inclusion of this little chip, as it almost precludes the need for a true hardware solution.
Layout and Installation.
Starting at the top and working our way down, we firstly see the rotated ZIFF socket. Some users abhor rotated sockets and some like them, I fall into the latter camp as it allows me comfortably attach and remove my heatsink from the motherboard without having to remove it from the case. Simply unplug the memory and away you go. The socket has plenty of room around it ensuring that even large coolers are relatively easy to fit.
Out attention is immediately drawn to the number of jumpers and dip-switches on the board, something we are seeing less of these days. The Asus is almost unique in that it offers both jumper and jumper-free modes of operation. The modes can be toggled by changing the position of the JEN jumper located at the bottom right of the board. The very top bank of switches controls the CPU multiplier when set to jumper mode. One interesting bank of jumpers, also at the very top of the board, controls the CPU type (Athlon XP or Thunderbird, Duron). We've not seen this before, presumably it ensures compatibility with the different processors.
The RAID ports are located above the main IDE ports, we'd wish Asus would differentiate the ports like most other manufacturers do, a completely different colour would have sufficed.
The ATX power connector is located well out of the way of the ZIFF socket so trailing wires shouldn't restrict heatsink airflow. We see three DDR banks that are thoughtfully placed. Unlike on some other boards, you can remove the memory without having to remove the AGP card, there is just enough space for you to do so. We appreciate that little touch. 2 fan headers sit alongside each other just above the AGP port with the DDR voltage jumpers located directly beneath it. Another fan header sits just to the right of the floppy port.
We see the standard PCI slots (5, 32bit, 33Mhz) flanked by a rather oddly placed floppy port. We'd much rather see the floppy port grouped with the other IDE ports. The CNR slot is there purely to serve the needs of the OEM customer. We can see why Asus decided to incorporate a jumper mode, essentially it stops novice users from causing havoc by manipulating sensitive BIOS options. Overall, we're pretty pleased with the layout, the board looks good and everything bar the floppy port is located with thought.
BIOS, stability and overclocking
BIOS duty is handled by Award. Award BIOS' are much like Marmite, you either love them or loathe them. As has been mentioned above, the board can be set to jumper-free mode by changing the position of one jumper. However, even when set to jumper-free, you still have to manually change the DDR voltage by jumpers. This is made even worse by the inaccessibility of the DDR jumpers as they're located directly under the AGP slot.
The BIOS itself is fairly intuitive and reasonably easy to navigate. FSB adjustment can be made in 1Mhz increments from 100 - 227Mhz respectively. Asus also show the relative running speed of the PCI bus. Unfortunately FSB's much over 150 are largely academic due to the lack of 1/5 and 1/6 PCI dividers respectively. Vcore can be raised from 1.75v to the standard overclocker's limit of 1.85v. One interesting thing to note is that Asus seem to over-supply actual voltage. When set to 1.85v, we found that between 1.88v and 1.89v was being supplied to the CPU. Although preferable for overclockers, it certainly puts an emphasis on good cooling. Here's a quick look at the CPU adjustments available.
We could successfully run our XP2100 (13x133) on the board without any problems, an upgraded BIOS may be needed for future processors. You are given the option of running the system RAM at either 100 or 133FSB.
Let's have a look at the DRAM timings available
DRAM can either be left at SPD (automatically detecting the speed of the RAM) or be manually set. When set to manual, you can select the CAS, RAS, RAS to CAS and Precharge timings, the ones shown are the most aggressive available. As you can see the performance-enhancing 1T command is present too. We had to manually set the DDR voltage to 2.8v via jumpers to run at these settings. Using slightly stricter settings resulted in failed boot-ups. Our Samsung DDR2700 has been well over 200FSB @ CAS2 in other boards so we can't attribute the boot up failures to the RAM.
One seemingly annoying feature is the inability to manually turn off the RAID ports when booting. The RAID controller takes a while searching the ports whilst booting, time that would have been saved with a simple option in BIOS. Another observation we made was that the board takes a while to power up once the power switch is depressed, longer than any other board we've had in the labs. The usual power and system setup options are present too.
The board also has a fail-safe booting mechanism. If, for any reason, you cannot load Windows and have to switch the machine off, it will then next restart by automatically clearing CMOS by itself, thereby ensuring that BIOS defaults are loaded. This feature is quite handy if you push the board too far.
Part of Asus' reputation is built on producing extremely stable motherboards. That heritage seems to have carried through to the A7V266-E. We ran the board continuously for well over 16 hours with Prime95 and SETI both taxing the system simultaneously. We weren't surprised to see the board soldiering on even after extended abuse. Top marks to Asus for once again producing an impeccably behaving motherboard.
We mentioned that the Asus seems to over-supply voltage at any given setting. This boded well for our overclocking efforts. Using our XP2100 (13x 133), cooled with a modified dual-fan Taisol 760, and 1.85v Vcore, we were able to push the CPU to 1885Mhz with excellent stability. 1900Mhz was 95% stable too, only failing our Prime95 torture test after 25 minutes. Here's a quick look at our overclocking result.
We tried to find the limit of the motherboard by using a factory-unlocked 1.4Ghz Thunderbird. By reducing the multiplier to 8x, we were able to hit a stable FSB of 164Mhz.
Here's a quick rundown of our test system and benchmarks used
All benchmarks were conducted at 1024x768x32 85Hz with vertical sync disabled. Benchmarks were run 3 times consecutively and an average score was taken. Both systems were configured for maximum performance. Another aspect to bear in mind is the relative running speeds of both systems when set at CPU defaults. When set to 1733Mhz (13x133) in BIOS, the Asus A7V266-E overclocked the FSB such that the actual running speed as measured by WCPUID was 1745.98Mhz. The MSI also overclocked the FSB a little, producing a true running speed of 1742.35Mhz. We don't like to see this happen, but in the cut-throat world of motherboard manufacturing, any increase in performance, however gained, is more than welcome.
We're perhaps being a shade unfair in comparing it with the newer MSI KT3, one based on the KT333 chipset. As a brief recap, the KT333 simply differs from the KT266A by incorporating support, albeit asynchronously, for DDR2700 memory. It also sports an upgraded South bridge which is now ATA133 compliant. It's the same as the KT266A in all other respects.
Our trusty synthetic friend, SiSoft Sandra, leads the way. We'll pass on the CPU benchmark this time as it's simply Sandra's representation of the of the CPU running speed. The memory throughput is a far more interesting comparison.
We can see even the Asus at 1733Mhz producing memory scores above the reference KT266A benchmark. Our strict timings and use of the performance-enhancing 1T command certainly pay dividends here. However, the MSI, running with asynchronous memory, puts a little distance between itself and the Asus. The Asus' scores at 1885Mhz (145FSB) are just what where we expect them to be. Strong performances all-round here.
We'll next turn our attention to Pifast. Pifast calculates the constant Pi to X million decimal places using the fastest method possible. We've been able to empirically prove that Pifast thrives not only on pure CPU speed but also on memory throughput. Shifting large amounts of data from system memory to processor is the name of the game here, Let's see how they compare when racing to calculate Pi to 10 million decimal places.
We're somewhat under-whelmed by the performance of the Asus with the stock XP. Although the memory benchmarks hinted at a difference, we are still surprised to see it almost 3 seconds behind the MSI (albeit running asynchronously). We re-ran the test a number of times and found the results to be consistent. The Asus at 1885Mhz takes an expected lead in this test. It appears as if the better memory optimisations, coupled with running memory asynchronously, really help the MSI here. Still. 92.3 seconds is nothing to scoff at.
MP3 encoding is something that I regularly do, anything that saves time in that process is most welcome. To test the efficiency of our protagonists, we set them the task of encoding a custom 481MB WAV into 128KB/s MP3 format using the popular LAME 3.91 encoder with a Razor-Lame front-end. Let's see how they shaped up.
Interesting, we seem to have validated our previously held assertion that LAME was purely CPU sensitive. The scaling of the Asus at 1885Mhz informs us of that. Still. 111 seconds is the joint-fastest time we've seen here at Hexus, no doubt ably aided by the impressive overclocked speed of the Athlon XP.
Let's move onto another area of media encoding that has historically been heavily reliant on memory throughput, namely DVD encoding.
We're using Xmpeg 2.0, a derivative of the popular Flask encoder, coupled with the Div X 3.20 codec. We've found this combination to be the most stable in our stress tests. Three Kings is the DVD of choice, it's mixture of action and dialogue make it an excellent benchmarking test. The DVD is encoded in full-screen format into YUV2 format. The black borders are cropped to save unnecessary encoding time. The lo-motion codec is used with the bit rate set to 1500 KB/s. We calculate the average FPS after 20,000 frames have been encoded. Let's have a look at the results.
A similar story to the Pifast results, one that should come as no surprise by now. The Asus at stock speeds lags slightly behind the MSI KT333 board but sheer CPU speed / memory throughput see it edge comfortably ahead at overclocked speeds. One thing to take away from this test is the impressive performance all-round. We're looking at speeds greater than real-time here.
As part of our stability testing, we ran SETI. SETI is a scientific experiment that uses Internet-connected computers in the Search for Extraterrestrial Intelligence (SETI) by sifting through large amounts of data. We know that many of our readers are interested in SETI. Through preliminary tests we were able to ascertain that SETI is extremely sensitive to memory bandwidth.
We ran the Ocuk SETI benchmark, a rather tough work-unit with an Angle Ratio of 0.417. This one takes a while to complete. One advantage in this benchmark is it's ability to display results to within 1/10000th of a second, we've rounded the results up to the nearest second for the sake of brevity. Will the trend that we've observed so far continue here ?
It appears so, the Asus can't quite catch the MSI at default speeds but naturally edges ahead at overclocked speeds. Either way, the times posted are extremely impressive in their own right.
So far we've seen the Asus put up a commendable performance in our benchmarks, let's now focus on gaming and see if it continues to do so.
We'll start the ball rolling with one of our favourites, namely 3DMark 2001SE. This synthetic benchmark stresses the CPU, memory sub-system and graphics card to the limit. 3DMark, in theory, should show quantifiable changes when moving from one board to another, let's see what it thought of our systems.
Again, we see the Asus trailing the MSI at stock speeds but managing to overtake it when 'clocked to the limit. It seems Asus' lack of bandwidth is negatively impacting on results. We see the 9000 mark barrier breached once more courtesy of an overclocked XP2100. Impressive results from a stock-clocked Geforce 3 Ti500.
Will the situation be mirrored in Serious Sam 2 ?. We're running the Valley of the Jaguar Timedemo included in the publicly available demo. Our visual preferences were 1024x768x32 Normal settings.
We see our recurring theme present here. Impressive results all-round
Let's pay a visit to page 1 of our benchmarker's Bible and try the old favourite that is Quake 3. We'll try the test at 512 Fastest to garner an idea of total throughput and then at 1024 Quality to represent something that mirrors actual playing resolutions. Onto the benchmarks !.
Can you say consistent?, well, our results certainly are. The same story unfolds here. The Asus is close but not close enough.
And there you have it.
We feel that Asus have done a pretty good job with the A7V266-E. It's well presented, has a multitude of features and is a champion in the stability stakes. The layout of the board is generally good with the exception of an oddly placed floppy drive port. We liked the use of the C-Media 8738 sound chip, we also appreciated the quality of the manual and general appearance of the package. The board satisfies it's probable OEM target market pretty well.
However, we at Hexus expect a motherboard to perform well at stock speeds, we look at motherboard manufacturers to provide that something extra, whether it be with new, innovative features or with outstanding performance. The Asus seems to fall short of this little bit extra, it does a lot of things well but fails to set our world on fire.
We used amongst the most aggressive memory timings available, yet the Asus fell consistently behind the MSI KT333 (even when the MSI was run with DDR266) in every benchmark we threw at it. The performance delta was consistent from start to finish. We would also like to see a completely jumper-free mode in BIOS. The fact that certain jumpers still have to be manipulated is a negative point. Although we're pleased to see RAID make an appearance on this board, the pleasure is somewhat tempered by the use of an older Promise RAID controller. We also disliked the inability to be able to turn the RAID controller off, this increased the boot up times by at least 10 seconds.
Perhaps we're being a bit over-critical, but we've seen some truly excellent motherboards come through the door recently. The Asus, whilst not the best, is certainly in the top half of those that we tested. If you're looking for a solid, feature-laden motherboard, you can do considerably worse than the A7V266-E.