AOpen’s Core Duo Flagship i975Xa-YDG motherboard

The i975Xa-YDG is AOpen’s most ambitious MoDT motherboard, designed for Intel Core Solo/Duo with ATI Crossfire dual graphics card capability and a host of performance oriented features. It also has a standard socket 478 heatsink retention bracket, which means big, quiet aftermarket performance heatsinks can be used. Is this a perfect motherboard for the green-conscious, anti-noise gamer?

June 20, 2006 by Mike Chin

Regular SPCR visitors know that this is not the first time an AOpen i975Xa-YDG motherboard or an Intel Core Duo T2600 processor have been featured in one of our articles. They were featured in our April Desktop CPU Power Survey, which examined the best power efficiency achievable with then current processors from AMD and Intel. You may recall that the T2600 proved to be very efficient, as expected of this 31W TDP processor. The AOpen i975Xa-YDG board used for that article was a pre-production lab sample; we promised to get hold of a production sample for a proper review of a product you can actually buy.

The Core Duo T2600 was the fastest of the Intel dual-core processors with the Yonah core, but nothing in the computer world holds the title of “fastest” for long — a 200MHz speedier T2700 is now available, and there’s little reason to think it will retain the speed title for long either. The Yonah Core Duo is descended from the Pentium M, and inherits that mobile processor’s excellent power efficiency. The Pentium M was intended originally to be used exclusively in mobile platforms.

Industrial board makers first began using it in non-mobile specialized applications with micro-ATX and mini-ITX boards. Then AOpen and DFI introduced consumer-friendly desktop 479 socket boards a little over a year ago, which signaled the beginning of MoDT — mobile components on desktop platforms — now a slogan used by AOpen, the visible leader in this field. Intel embraced this practice with the Core Duo / Solo, positioning it from the start as a mobile and a desktop part. It is a component of Intel’s Viiv home multimedia centers as well as the Centrino mobile platform. It also powers Apple’s new 17″ and 20″ LCD iMacs; we reviewed one recently and found it to be a model of efficiency and low noise.

AOpen has many MoDT products for Core Duo / Solo on the market: Several motherboards with Intel 945 chipsets in micro-ATX and mini-ITX formats, as well as SFF barebones (for Pentium M), the (in?)famous MiniPC and its descendant, and complete HTPC sets. The AOpen i975Xa-YDG is their most ambitious MoDT motherboard, introduced in March 2006 with a great deal of fanfare.

It’s packed with features; click on photo for larger view.

At AOpen’s global web site today, you will find no fewer than 10 press releases pertaining to the board. The headlines are worth reproducing here to give you an idea of their positioning of this product:

• The unstoppable force from AOpen i975Xa-YDG
• Five leading technologies of AOpen COO 975X
• Digital entertainment motherboard – AOpen i975Xa-YDG
• AOpen i975Xa-YDG, the gamers’ overclocking motherboard
• AOpen took 55% market share of Intel 975 chipset
• Best overclocking motherboard-AOpen i975Xa-YDG
• Why AOpen i975Xa-YDG becomes so popular?
• The most wanted motherboard in Japan – AOpen i975Xa-YDG
• AOpen i975Xa-YDG offers a significant set of features and impressive performance”- AnandTech
• The motherboard you got to know – AOpen i975xa-YDG
• The flagship MoDT (Mobile on DeskTop) AOpen i975Xa-YDG motherboard powered by the Intel Core Duo CPU has been born

It’s evident that the product has already achieved a degree of success in the marketplace. In truth, it is a very unusual motherboard in many ways. The feature highlights below below explains why.

SPECIFICATIONS

The retail box is pretty big to fit all the standard accessories and parts.

Amazingly, the wee remote even features FSB up and down buttons!
Also shown above is the IR receptor; the other end of the cable fits into a header on the board.

The SATA connector is accessible from the outside;
photo above shows supplied data+power cable just about to be plugged in.

The supplied cable is half a meter long, which is useful only if the motherboard
is in a system that sits atop a desk.

HEATSINKS AND COOLING

The best thing about the i975Xa-YDG for SPCR enthusiasts is that AOpen has fitted it with a standard socket 478 desktop heatsink retention bracket (and four-bolt standard plastic backplate), rather than cripple it with a too-small, proprietary heatsink mount system that dictates the use of a small, too noisy fan. This means that most aftermarket socket 478 heatsinks can be fitted without problems, giving users a huge range of capable models to choose from.

CPU socket sits diagonally in 478 HS retention bracket.

It is important to keep in mind that the absence of a heatspreader has several ramifications (exactly like the Pentium M, as originally researched and reported by Ralf Hutter in SPCR’s first review of Pentium M boards for the desktop):

• The top of the CPU sits approximately 0.075-0.078″ lower than a standard P4 or Celeron 478 with a standard casing (with heatspreader). The base of the heatsink, therefore, must be smaller than the curved-corner area defined by the inside perimeter of the plastic heatsink bracket. Otherwise, the bracket itself can prevent the HS base from making contact with the CPU die.
• Lower CPU core height with a P4-478 heatsink will result in reduced pressure. However, the standard P4-478 casing has a recommended clamping force of 75lb/ft ± 15lb/ft, while the Core Duo / Solo datasheet calls for a maximum of 689KPa, or 14.4lb/ft. pressure. So the 0.075″ lower height might be a good thing after all.
• It is best if the HS mounting mechanism can apply tension gradually, rather than all at once, as with the typical clip. Uneven tension, too high a tension, or tilting the heatsink against a CPU die edge during mounting should all be avoided; the edges of the bare die can be chipped, which could mean the death of an expensive processor. This means the screw-down heatsinks such as the Zalman 7000, 7700 and 9500 series are a bit safer to mount than more conventional dual-clip designs.

A low profile all-aluminum heatsink with integrated 70mm fan is provided with the motherboard. It’s small even compared to the stock HSF that came with the first retail packaged P4 over four years ago. It is probably perfectly adequate to cool any of the current Core Duo processors, which are rated for 31W TDP. Standard metal tension clips that require very little pressure are used for simple installation.

Low profile CPU cooler provided with the i975Xa-YDG.

The worst thing about the i975Xa-YDG for SPCR enthusiasts is that AOpen has fitted the Northbridge chip with a small heatsink and matching fan that promises to be annoying. It’s just too small for it to give any cooling without spinning fast enough to make a high pitched tone. This is a change from the lab sample board, which had a similar size heatsink but was not equipped with a fan. The southbridge chip, on the other hand, has no heatsink at all.

NB cooler with wee fan; SB is bare.

An intelligent cooling decision was made regarding the positioning of the two PCIe X16 slots. They are separated by the width of two slots, which is nearly 3 inches. This means if the dual-card Crossfire mode is used, both graphics cards can be fitted with the large heatsink / fans that are critical for good cooling with low noise.

BIOS

The Phoenix Award BIOS (version 1.03 dated April 4, 2006) is surprisingly basic for a board with such high aspirations. This is not to say that it is a limited BIOS, as clock speed adjustments are provided for CPU and PCIe, and voltage adjustments are available for CPU, memory, northbridge chip, and PCIe. The memory timing settings also have their own submenu, which is quite extensive. The range of settings provided for many parameters is large and extremely fine, as show in the table below.

However, the Power Master feature is a
simple Off / Auto switch, and there is no sign of AOpen’s
SilentBIOS or SilentTEK intelligent fan controller. Many previous versions
of these features (in other AOpen boards) were extremely flexible, with a large
number of user options. Perhaps, as with some of AOpen’s other MoDT motherboards,
the attitude is that with the CPU thermal envelope so low, there’s little benefit
to having complex fan and power management schemes.

The function of Power Master is not fully explained in the manual or in the BIOS help menu. The manual states,

“When the processor loading is heavy, Power Master will raise your processor’s clock frequency to meet your requirement. On the contrary, when the processor is in low loading, Power Master will drop the clock frequency to reduce the noise from the processor fan.”

By carefully monitoring system states, it became clear that when Power Master was set on AUTO, the CPU clock speed dropped by 20% at idle, and increased by 8% at full load. This is documented in the testing section.

TESTING SETUP

The test system was comprised of the following components on an open bench top.

• Seasonic S12-430 (late spring 2006 edition w/ sleeved cables) — 78% efficient at 66W load, 80% efficient
  at 91W load, 81% efficient at 152W load
• Intel Core Duo T2600
  processor. Rated at 31W TDP.
• Corsair XMS2-5400
  DDR2 SDRAM 2 x 512 MB
• Stock AOpen CPU heatsink/fan
• ASUS EAX1600XT Silent/TVD/256M
  fanless graphics card
• Samsung MP0402H 2.5″ PATA hard drive

Details about the Core Duo T2600 are provided below.

The supplied HSF has multicolor LEDs.

Power measurements were performed in three states to establish the full range of
power consumption:

• Enhanced Intel Speed Step (EIST) enabled idle.
• Under maximum CPU load using CPUBurn.
• Under maximum system load load using CPUBurn to stress the CPU and
  RTHDRIBL
  to stress the graphics card.

Two power measurements were taken in each state:

• DC power at the 2x12V (AUX12V) connector on the motherboard. This involves a high precision LTS 25-NP Current Sensor, courtesy of Intel, plugged directly into the 2x12V connector on
  the motherboard, so that all power through this connection passed through
  the power meter. The line voltage (nominally +12V) and the current are measured
  with multimeters, and multiplied together to get the total power running through
  the connection. Because the CPU only draws power through the 2x12V connection
  and nothing else does, this measurement tells us the amount of power consumed by the
  CPU and the voltage regulators on the
  motherboard.
• Total AC power consumed by the system, measured with an Extech
  Power Analyzer / Data Logger 380803 power meter.

Other Test Tools:

• High accuracy Extech MM560 True RMS multimeter.
• Two other multimeters of good precision.
• Fan speeds, temperatures and processor voltage was monitored using SpeedFan
  4.28 and with CPU-Z.

Ambient conditions were 22°C
and 20 dBA throughout testing. All stress tests were run for ~30 minutes before measurements were recorded.

TEST RESULTS

The BIOS was left at default settings for the most part. As mentioned previously, the range of adjustments with SPCR-relevant controls (such as fan control) was minimal.

1. Standard Settings

The processor details at idle and maximum load are shown below. As EIST (Enhanced Intel Speedstep Technology) is enabled, the clock speed and Vcore vary dynamically with load. The effect of the Power Master settings are also reported. Note that the system AC power includes the power demand of the Asus 1600XT graphics card.

AC Power

The fans did not change in speed, regardless of load. The CPU fan was reportedly
~1700 RPM, while the small NB heatsink fan was at ~3400 RPM. At idle, the CPU
temperature was reportedly below 20°C and well below ambient. This requires
many grains of salt to believe, but it’s hardly relevant anyway. Cooling at
load not exceptional but it was perfectly adequate. As is so often the case,
the location of the “board” sensor was not documented anywhere; later,
it became clear that this is most likely the NB chip.

The noise of the two fans was a bit louder and more annoying than the 25 dBA@1m
reading would suggest. The worse offender was the 40 or 50mm chipset fan, which
has a rough character. The
audio recording in MP3 format on the next page gives a pretty good idea of what it
sounds like.

2. Power Master

The effects of the Power Master settings are reported below. The total power of the system dropped a bit in idle, and it rose a bit at full load, as expected due to the 20% underclock at idle and 8% overclock at maximum load.

Power Master had a small effect on temperature, mainly of the CPU. After half an hour of CPUBurn, the processor temperature reached 54°C, six degrees higher than at stock speed,
while the NB temperature stayed the same as before.

3. Overclocking

The default overclocking range provided in the BIOS is 16.6%, achievable by increasing the front side bus from 166 MHz up to 199 MHz. When the JP4 and JP5 jumpers are disabled , the FSB defaults to 200MHz minimum and 320MHz maximum. Since the board is positioned as high performance and overclocking-friendly, we gave it a go with the stock jumper settings.

We started by increasing the bus speed in one fell swoop to 199 MHz, since it’s a relatively modest increase. The 13X multiplier of the Core Duo T2600 put the CPU clock to 2,589 MHz, an increase of 430 MHz. All other setting were left the same. The board was stable running CPUBurn for about 15 minutes, but as soon as RTHDRIBL was added, the system crashed and restarted.

The problem seemed to be a minor droop in the CPU voltage under full load in the overclocked state. The 12V feed from the PSU to the system was 13.2V at idle, and it dropped to about 12.9V under full load. This is definitely on the high side, substantially over the +5% tolerance generally allowed for the 12V line (which means a high of 12.6V). Still, the high input voltage seemed to cause no problem for the motherboard. The CPU Vcore was reported by CPU-Z as 1.26V when the CPUBurn was first turned on. After RTHDRIBL was also turned on, the Vcore sagged to ~1.22V, at which point the system crashed. So using the BIOS controls, the Vcore setting was raised in incremental steps until the system remained stable running both CPUBurn(x2) and RTHDRIBL together for at least half an hour. This was achieved with a Vcore setting of 1.28V, which saw sags under load of 1.25V. Unfortunately, as soon as the Vcore was set manually, the dynamic voltage adjustment of EIST appeared to stop working, and the very low power draw of the CPU at idle was lost.

CPU power draw increased as did overall system power demand, as expected, and temperatures went up a fair amount, but still remained within safe limits. With no changes in fan speed, the noise remained unchanged.

There seems little doubt that the board and the CPU can be overclocked to higher speeds, but we had little interest in exploring this capability any further. A more enthusiastic OC site already addressed the issue anyway: AnandTech’s review gushed that…

“…the overclocking aspect of this board and the Intel Core Duo required so little effort to reach 50%+ overclock levels that we are still surprised by how easy it was. The stability of the board was superb in our benchmark, gaming sessions, and general application testing whether the board was at stock or overclocked settings. The AOpen i975Xa-YDG is certainly the most stable board we have tested in recent memory…”

The above Anandtech review also lauds the Core Duo’s gaming performance, an opinion strongly supported by X-bit Labs, who stated in their Core Duo Overclocking article that…

“The performance of the overclocked Core Duo processor working at 3.0GHz is admirable. It is outpaces top dual-core processors from AMD and Intel, such as Athlon 64 FX-60 and Pentium Extreme Edition 965. And the performance advantage is more than tremendous in most tasks.”

TEST RESULTS (continued)

4. Other Heatsinks

Since it is possible to use high performance aftermarket CPU heatsinks, we had to try a couple of our favorites. After all, the standard 478 retention bracket is one of the best aspects of this board for silencers.

A much-abused Scythe Ninja that’s seen
a lot of duty around the lab was installed. Care had to be taken with the clips,
which apply a lot of pressure. If you are going to do this, we’d recommend installing
the heatsink with really good access all around the CPU socket area. You’ll
need to be careful not to let the HS tilt on the CPU die while engaging the
clips. A 120mm Antec TriCool fan was brought into service and set on low. It
was the nearest fan at hand. It was simply set to rest on the stock NB cooler,
with its fan unplugged. With the amount of airflow the 120mm fan can generate
even on low, this seemed safe enough to try, as the NB would benefit from some
of the TriCool’s airflow. The fan was set to blow through the Ninja; it also
provided cooling for the passive Asus graphics card.

A Scythe Ninja is ridiculous overkill for a 31W TDP Core Duo T2600, but it could be done, so it was.

A modified Zalman 9500 heatsink was also called into play. It’s one where the stock fan, deemed a trifle too loud for us, was swapped out with a Nexus 92mm fan. Even with the screws employed for mounting, getting the 9500 on the CPU safely wasn’t that easy. Again, we’d recommend installing the heatsink with really good access all around the CPU socket area. As with the Scythe Ninja setup, the NB fan was simply left unplugged. The Nexus 92 fan on the Zalman 9500 was simply plugged into the CPU fan header on the motherboard.

Zalman 9500 modded with Nexus 92 fan in similar setup as above.

The results, as shown in the table below, were predictable.

Idle

CPUBurn OC’d*

Idle

The idle CPU temperature reported with either heatsink/fan is difficult to believe, but it’s duly reported for the record. The temperature of the CPU at load was amazing with the Ninja, and only superb with the Zalman. The big fan used with the Scythe also kept the NB temperature way down as well, under all conditions. The much smaller, slow-spinning Nexus 92 used in the Zalman didn’t provide as much airflow for the NB, and with the NB fan turned off, the temperature actually exceeded that of the stock cooler. Still, the difference was modest, and there was not even a whisper of instability throughout.

The acoustics also improved dramatically. With the Antec 120 TriCool on low,
the general character of the sound improved to a smooth, very quiet whoosh.
Now the Samsung notebook hard drive could be heard occasionally, along with
the quiet 120mm fan in the Seasonic S12-430 power supply — very nice, overall.
With the Zalman 9500 / Nexus 92, the fan is so quiet that it’s basically inaudible
from a meter. Whatever noise the sound level meter was picking up came not from
the HSF but from the power supply fan and the residual ambient level of the
room.

5. Audio Recordings in MP3 Format

Audio recordings were made of the test system with the three different HSFs. Keep in mind that the recordings paint only part of the acoustic
picture. The microphone is within 3-4″ of the primary noise source, and as a result, they tend to exaggerate differences (with good audio playback gear), much as if you placed your ear to within a few inches of the actual sound source. This is intentional, it’s meant to help you examine the noise the way we do in the lab.

AOpen i975Xa-YDG test system with stock HS fans: 25 dBA@1m
The first eight seconds have both HSF running. The next eight seconds are with the CPU fan stopped. Then it’s with the NB fan stopped. Then we have both fans stopped, and what you hear is the background noise of the Samsung 2.5″ notebook drive. Finally both HS fans are allowed to ramp up again.

AOpen i975Xa-YDG test system w/ Scythe Ninja + Antec TriCool 120: 22 dBA@1m

AOpen i975Xa-YDG test system w/ Zalman 9500 + Nexus 92 fan: 20 dBA@1m

Reference: Nexus 92mm
case fan @ 5V (17 dBA@1m) Reference

All this leaves little doubt that that 478 heatsink retention bracket is totally worthwhile on this board for any silencer. It’s also clear that for very little additional cost, AOpen could do much better than the small all-aluminum CPU and NB HSFs they’re supplying on this board. The bigger challenge is for noise-conscious gaming users of this board: How to cool two hot graphics cards quietly enough so that the acoustic advantages of the cool-running Core Duo CPU and the big 478 heatsink capability are not lost.

6. External SATA Functionality

The hot-swap capability of the External SATA connection was tested with a formatted
WD Caviar SE16 500GB drive and an unformatted WD Raptor 150GB drive. In both
cases, the eSATA featured worked perfectly. As soon as the cable connected the
HDD to the board, the external drive was recognized correctly, allowing it formatted
if necessary, and access was provided immediately to the formatted disc. The
AC power demand rose by 10~11W when the drives were connected. Note that this
is not a standard eSATA connector; it is an adaptation of internal SATA. “Real”
eSATA external drives use a slightly different connector, are self-powered (with
their own AC power supply or adapter) and do not need power from the board.
The question asked earlier is worth repeating: How many users will simply use
a bare hard drive outside the case?

CONCLUSIONS

The AOpen i975Xa-YDG is an interesting product. On the
one hand, it is designed solely for use with Intel Core Solo / Duo processors,
which run exceedingly cool, thus enabling very low power consumption. On the
other hand, it supports ATI’s dual-graphics card Crossfire technology, which,
if employed with high-end cards, can mean power demand approaching 250W (in DC power) for the video
subsystem. In fact, aside from the processor, the i975Xa-YDG allows a plethora
of high power devices to be run off it, which gives it something of a schizoid
personality, in our view. What is the point of employing a CPU that draws under
35W if the rest of the system is allowed to draw >300W? It is an approach
that’s almost the antithesis of AOpen’s own MiniPC or the Core Duo based LCD
panel iMacs from Apple.

Apparently, SPCR’s point of view is not the norm, as this motherboard seems very popular among well-heeled Japanese aficionados. We have to assume they have deep pockets for their hobbies, as the steep ~US$300 price of this board is only one of the many items needed to build a system around it. Core Solo / Duo processors are also not exactly bargains at this time; a T2600 fetches upwards of US$500.

Nevertheless, the AOpen i975Xa-YDG does what it does very well. Although the stock heatsink/fans are not that quiet, it is easy to replace them with much better aftermarket coolers. The sheer range of features is probably enough to attract many buyers, and no doubt there are gamers among the SPCR audience who will eagerly seek out this board. The combination of Core Duo performance and overclocking potential means the AOpen i975Xa-YDG is arguably the most power efficient dual-graphics card capable motherboard for extreme gamers today.

Much thanks to AOpen for the i975Xa-YDG review sample.

* * *

SPCR Articles of Related Interest

AOpen i915Ga-HFS ATX Pentium M Motherboard
17″ Apple iMac: The Official SPCR Review
VIA EPIA EN12000E: Today’s most efficient CPU & mainboard
Desktop CPU Power Survey, April 2006

Asus A8R32-MVP Deluxe w/ATI CrossFire Xpress 3200 Chipset

* * *

Discuss this article in the Silent PC Review Forums.