AMD’s 785G Chipset Boards: 780G Evolved

785G is an evolution of 780G that brings some improvements and support for the AM3 socket with DDR3 memory. The onboard graphics chip, dubbed HD 4200, supports DirectX 10.1, HDMI 1.3, UVD 2.0, and Sideport memory which provided a nice performance boost on the more expensive 790GX-based boards.

August 10, 2009 by Lawrence Lee

AMD’s 785G chipset is the successor to 780G, possibly the most important integrated
graphics chipset in their history. The accompanying HD 3200 graphics chip was
the first onboard solution to support hardware acceleration for modern high-definition
video codecs and its 3D performance was far superior to anything that preceded
it. With the release of 780G, motherboards with integrated graphics were no
longer just cheap entry level products. The 780G chipset aged gracefully and now
is the defacto choice for value — 780G boards are heavy
on features and light on price compared to the competition — especially
Intel.

785G is an evolution of 780G, bringing minor improvements and support for the
AM3 socket with DDR3 memory. The onboard graphics chip, dubbed HD 4200, has
a few advantages over HD 3200 including support for DirectX 10.1 and HDMI 1.3
(but not multi-channel LPCM) and allows for the use of Sideport memory, which
provides a nice performance boost on the more expensive 790GX-based
boards. 3D performance should be similar though, as the number of stream processors
and the core frequency has not changed. The UVD (Unified Video Decoder) has
also been updated to version 2.0, allowing for decoding of multiple streams
(i.e. for picture-in-picture functionality while playing Blu-ray) and some extra
post-processing effects.

The Asus and MSI boards we are looking at today use the 785G chipset paired
with the new SB710 Southbridge chip. 785G bridges the gap between 780G and 790GX,
and SB710 is similar, a hybrid that takes the limited RAID support of SB700
(0, 1, 10, JBOD) and combines it with SB750’s ACC (Advanced Clock Calibration)
feature, which aids in overclocking. The vast majority of 780G boards are paired
with SB700 to the dismay of some enthusiasts.

On paper, the two boards share almost the exact same feature set, with the
main differences being the number of expansion slots provided, a result of each
board’s respective form factor.

PHYSICAL DETAILS – ASUS M4A785TD-V EVO

BIOS – ASUS M4A785TD-V EVO

The BIOS allows for plenty of voltage and frequency modification
to satisfy overclockers, and even includes the ability to increase Sideport
memory voltage. Those looking to squeeze better energy efficiency and lower
temperatures out of their CPUs will be somewhat disappointed as the minimum
CPU voltage is -0.30V (1.025V in the case of our X3 720 processor). The fan
control options allow you to set the CPU fan’s starting voltage, and the temperature
range. There are no settings for the other two fan headers.

Editor’s Note: The MSI board came from AMD with a BIOS version assured to be the latest and greatest. The Asus board, on the other hand, came from Asus, with a BIOS dated some 3 weeks earlier than the version AMD was offering for the board on its media FTP site. The more recent BIOS version at AMD was not recognized by our board to be compatible with it, and there were no other BIOS options on the Asus web sites.

PHYSICAL DETAILS – MSI 785GM-E65

TEST METHODOLOGY

Test Setup:

• AMD
  Phenom II X3 720 Black Edition processor – 2.8GHz, 45nm, 95W
• Asus
  M4A785TD-V EVO – AMD 785G chipset, ATX motherboard (BIOS 0211)
• MSI
  785GM-E65 – AMD 785G chipset, mATX motherboard (BIOS V2.2B2)
• Xigmatek HDT-SD964
  CPU cooler
• Corsair
  XMS3 memory 2x2GB, DDR3-1600 @ 1333MHz, 9-9-9-24
• Seagate
  Momentus 5400.6 notebook hard drive – 500 GB, 5400RPM, 8MB
  cache (Asus M4A785TD-V EVO)
• Hitachi
  5K500 notebook hard drive – 500GB, 5400RPM, 8MB cache (MSI
  785GM-E65)
• Asus
  BC-1205PT Blu-ray drive – SATA
• Seasonic
  SS-400ET ATX power supply
• Microsoft
  Windows Vista SP1 operating system – Home Premium, 32-bit
• ATI
  Catalyst 9.6 graphics driver

Measurement and Analysis Tools

• CPU-Z
  to monitor CPU frequency and voltage.
• CPUBurn
  K7
  processor stress software.
• Prime95
  processor stress software.
• FurMark
  stability test to stress the integrated GPU.
• Cyberlink
  PowerDVD to play video.
• SpeedFan
  to monitor temperature and fan speeds.
• Seasonic
  Power Angel AC power meter, used to measure the power consumption
  of the system.
• Custom-built, four-channel variable DC power supply, used to regulate
  the CPU fan speed.

Our main test procedure is designed to determine the overall system power consumption
at various states (measured using a Seasonic Power Angel). To stress CPUs we
use either Prime95 (large FFTs setting) or CPUBurn K7 depending on which produces
higher system power consumption. To stress the IGP, we use FurMark, an OpenGL
benchmarking and stability testing utility. Power consumption during playback
of high definition video is also recorded.

Our main video test suite features a variety of 1080p H.264/VC-1 encoded clips.
The clips are played with PowerDVD and a CPU usage graph is created by the Windows
Task Manger for analysis to determine the approximate mean CPU usage. High CPU
usage is indicative of poor video decoding ability on the part of the integrated
graphics subsystem. If the video (and/or audio) skips or freezes, we conclude
the board’s IGP (in conjunction with the processor) is inadequate to decompress
the clip properly.

Cool’n’Quiet was enabled (unless otherwise noted). The following features/services
were disabled during testing to prevent spikes in CPU/HDD usage that are typical
of fresh Vista installations:

• Windows Sidebar
• Indexing
• Superfetch

Video Test Suite

1080p | 24fps | ~10mbps

H.264: Rush Hour 3 Trailer 1 is a H.264 encoded clip inside an Apple Quicktime container.

1080p | 24fps | ~8mbps

WMV-HD: Coral Reef Adventure Trailer is encoded in VC-1 using the WMV3 codec commonly recognized by the “WMV-HD” moniker.

1080p | 24fps | ~19mbps

VC-1: Drag Race is a recording of a scene from network television re-encoded with TMPGEnc using the WVC1 codec, a more demanding VC-1 codec.

1080p | 24fps | ~33mbps

Blu-ray: Disturbia is a short section (chapter 4) of the Blu-ray version of Disturbia, the motion picture, played directly off the Blu-ray disc. It is encoded with H.264/AVC.

TEST RESULTS

Asus M4A785TD-V EVO

The M4A785TD-V, powered by a X3 720 processor was tested at stock speeds and
underclocked to 1.6 GHz with the lowest stable voltage possible (in our case,
1.025V — the BIOS limit). Running at 2.8 GHz, the onboard IGP breezed through
our video playback suite with CPU usage topping out at 40% during our high bitrate
VC-1 clip. Underclocked, VC-1 playback used a few more CPU cycles but the clips
were still rendered with relative ease. H.264 playback when underclocked resulted
in lower CPU usage because we disabled Cool’n’Quiet — at stock settings Cool’n’Quiet
set the clock speed lower than 1.6 GHz.

MSI 785GM-E65

With the same hardware underneath the hood, the 785GM-E65 delivered nearly
identical results during video playback. The MSI board however exhibited much
lower power consumption, especially when idle. It also used less power when
underclocked/undervolted, though MSI’s BIOS allowed us to undervolt further,
0.940V compared to 1.025V on the Asus board.

Comparisons (Stock Speeds)

The 790GX-based Asus M4A78T-E had slightly lower CPU usage than the 785G boards
during our WMV-HD clip, though 7-9% isn’t enough of a difference to consider
the result anything but an anomaly. We weren’t able to detect any difference
between the HD 3300 and HD 4200 graphics in regards to video playback.

The M4A785TD-V is the least energy efficient AM3 board we’ve tested, using
12W more idle, 5-14W more during video playback, and a 2-3W more during full
load compared to its MSI counterpart. The M4A78T-E fared poorly in idle power
like the M4A785TD-V, but used 4-6W less than the 785G boards during CPU load
only. Most of the other results fell in-between.

The 785GM-E65 uses a 4+1 phase array in its VRMs, while the M4A785TD-V has
a 8+2 phase design. That would seem to indicate that the MSI board is more suited
for low-load systems while the Asus is more efficient when paired with a higher
power CPU, though this was not the case when we pushed our 95W X3 720 to the
brink on both boards. The 785GM-E65 also has “active phase switching”
which adjusts the phases used by the board’s VRMs depending on CPU load. The
M4A785TD-V, like most Asus boards has a similar feature called EPU though all
it seems to do is make minor adjustments to various operating voltages which
results in only slight energy savings.

Comparison (Underclocked, Undervolted)

The M4A785TD-V used the most power when undervolted to the lowest stable voltage
with our processor running at 1.6 GHz, though it we weren’t able to lower it
below 1.025V due to the BIOS. In any regard, it drew 14W more when idle and
10-12W more during video playback and CPU/GPU load compared to the MSI 785GM-E65.
The 790GX-based M4A78T-E used only slightly more than the MSI board, 5W at most
during most of our test states, but during full system load, it used 10W more,
2W less than the M4A785TD-V.

Fan Control

To test the board’s fan control, we connected the CPU fan to a manual fan speed
controller so we could slow it down to induce higher CPU temperatures during
load. A 92mm Scythe Kama PWM (2500 RPM) fan was connected to the CPU fan header
and a 80mm Scythe Kama Flow (1500 RPM) fan was connected to the System fan header.

The Asus M4A785TD-V’s fan control followed precisely the settings we entered
in the BIOS. The fan ramp up was gradual, starting at 1220 RPM at 30°C core
temperature and topping out at just over 2500 RPM at 51°C. By tweaking these
settings you can make the fan control as aggressive or as passive as you like.
On the downside, the CPU fan’s minimum speed is a bit too high, and the secondary
fan spun 50% speed throughout, not reacting to changes in temperature.

The MSI board had a much lower minimum CPU fan speed — it spun at only 200
RPM until the core temperature reached 42°C (3°C lower than the value
entered in the BIOS), and ramped up very quickly, reaching its maximum speed
a mere 6°C later. You can select when the CPU fan cranks up, but you can’t
dictate the speed at which it does so. The secondary fan spun at 630 RPM when
set to 50% in the BIOS (75% and 100% are the other options). Like the Asus board,
it does not respond to increased heat.

SpeedFan Support

For Windows users, SpeedFan is our application of choice for fan control. It
can be configured to raise or lower multiple fan speeds to designated limits
when any specified temperature threshold is breached.

SpeedFan reported fan speeds from all three headers as well as the “CPU”
and “MB” temperatures indicated in Asus’ PC Probe utility. SpeedFan
can control the CPU_FAN header (only if a PWM fan is connected) and the CHA_FAN
header. To enable fan control, select the “IT8712F-J” chip in the
Advanced tab of the Configuration menu and set all the PWM modes to “Software
Controlled.”

Unfortunately SpeedFan does not yet support the sensors on the MSI 785GM-E65
— it could only report the hard drive and CPU core temperatures. SpeedFan
also does not support fan speed readings or fan control of any kind. To make
matters worse, the system monitoring and tweaking utilities offered on the support
CD would not install.

Cooling

Lower cost boards ship with simple heatsinks on the Northbridge and Southbridge
while those targeted at enthusiasts typically have large heatpipe coolers and
heatsinks on the voltage regulation modules near the CPU socket. A well-cooled
motherboard can deliver better power efficiency and stability.

The MSI 785GM-E65 was easily the winner in our thermal testing. Its chipsets
measured 6-7°C than Asus’ board which makes sense considering the board’s
lower power consumption. The heatsinks on both boards were fairly cool by relative
standards. Cooling isn’t an issue on either.

3D Performance

To get a rough estimate of how well each board’s integrated graphics plays
games, we ran 3DMark05/06. As synthetic benchmarks they have limited value,
but they give a rough idea of how well an IGP performs.

The HD 3300 IGP on the 790GX chipset is still a faster chip, posting a 17%
and 22% advantage in 3DMark05/06 respectively over HD 4200. Taking into account
our previous results with the 780G chipset and X2 4850e processor, after extrapolation/scaling,
we estimate HD 4200 has cut the difference in half, sitting right between HD
3200 and HD 3300.

FINAL THOUGHTS

The 785G chipset is an evolutionary step, bringing a few feature
improvements over 780G. It provides a capable and affordable alternative
to 790GX for users looking to setup a true AM3 system, DDR3 memory and all.
790 boards offer faster 3D performance (in the order of 20% if you consider
3DMark a good indicator), more bandwidth for CrossFire configurations (8x/8x
vs. 16x/4x) and RAID-5 capability from the SB750 Southbridge chip. If none of
these advantages appeal to you, a 785G board might be superior as HD
4200 includes an updated version of UVD, and it is more affordable.

The MSI 785GM-E65’s main draw is its impressive energy efficiency. At idle
it used 12W (about 20%) less than the Asus board, and was thrifty during video
playback as well. The heatsinks on the board components run fairly cool, a result
of its high efficiency. The BIOS has plenty of voltage/frequency options and
allowed us to undervolt much more than the M4A785TD-V. The only aspect of the
785GM-E65 we disliked was the fan control system. Once the target temperature
is reached, the fan ramps up incredibly fast and the system fan can only be
set to three different but static speeds. It is not yet supported by SpeedFan.

The Asus M4A785TD-V EVO offers more expansion slots
and the option of using CrossFire. Our sample’s power consumption is decidedly
poor at low levels, though it matches the 785GM-E65’s efficiency at
full load. Like the MSI board, its components are well-cooled and shouldn’t
be problem if paired with a high performance 125W+ processor, for which it seems
more suited. The BIOS is enthusiast-oriented but has an CPU undervolting limit
of -0.3V. Fan control was better than on the 785GM-E65. The settings in the
BIOS allow you to adjust its aggressiveness to suite your tolerance, and both
fans can be controlled using SpeedFan which is a huge bonus. On the other hand,
the CPU fan’s minimum speed is too high, and if you use BIOS control, the auxiliary
fan does not react to changes in temperature.

Both of these boards offer a slew of features including HDMI, DVI, S/PDIF,
eSATA and FireWire, RAID 0/1/10, a 100% solid-state capacitor design, and a
very capable onboard graphics chip. The M4A785TD-V and 785GM-E65 retail for
about $30 less than their 790GX counterparts. The new chipset won’t blow anyone’s
socks off, but it does give price-conscious users a reason to choose the newer
AM3/DDR3 platform rather than settling on an AM2+/DDR2 solution simply because
of the price. It is also a compelling alternative to an Intel based platform.
The only Intel motherboard we can find that matches the features of these two
boards is the Gigabyte GA-EG45M-UD2H which retails for $130. The bottom line on Intel vs. AMD for affordable HTPC: The MSI 785GM-E65
is $40 cheaper and has faster onboard graphics.

Our thanks to Asus
and Micro-Star International
for the motherboard samples.

* * *

Articles of Related Interest
Intel DG41MJ: Affordable 775 Mini-ITX
DFI LanParty JR 790GX-M3H5 mATX
Motherboard
Zotac IONITX-A: An ION / dual-core Atom
Mini-ITX Board
Asus M4A78-HTPC/RC: A Complete HTPC
Solution?
Zotac GeForce 9300-ITX WiFi: Good
Things Come in Small Packages
Asus M4A78T-E AM3 motherboard

* * *

Discuss this article in the
SPCR forums.