EPoX began operations in 1995 as a value brand performance alternative in the marketplace. They have a history of providing a wide range of motherboards based upon core logic chipsets from varying manufacturers. EPoX has prided themselves on their research and development capability by offering these products at very competitive price points while maintaining excellent performance standards. More information about the entire line of EPoX products can be found here.

The EPoX EP-9U1697-GLi motherboard is based on the recently released ULi M1697 chipset. EPoX has done an excellent job in utilizing this chipset to create a product whose overall performance competed with boards costing twice as much. This board seamlessly supports NVIDIA SLI technology with the ULi PowerExpress Engine Enabling driver but is not SLI certified by NVIDIA. EPoX will not be supporting or certifying SLI operations on this board and will not include the ULi PowerExpress driver on their Driver/Utilities disk. Of course, now that NVIDIA has completed their acquisition of ULi, we understand this driver will no longer be available for usage.

Our initial impression of the EPoX EP-9U1697-GLi upon opening the box was one of interest. The layout of the board was very clean, although we disagree with the placement of the 24-pin ATX power connector, and Epox utilized high quality components in the design of the board. We were mildly disappointed with the lack of an HD audio codec and Gigabit Ethernet, but we tried to understand the absence of these features based upon a retail price of around US $85 or lower.

During our testing and general usage of the EPoX EP-9U1697-GLi, we found the board's stability to be superb and it delivered very competitive results in the latest synthetic and game benchmarks. However, we were plagued with a single issue that EPoX is working to solve. Our board would randomly lock up after exiting Microsoft Windows XP and this issue would require us to power down the machine before continuing operation.

EPoX utilizes the ULi M1697 HyperTransport PCI Express single chip on the EP-9U1697 GLi. The M1697 fully supports one PCI Express x16 lane or two PCI Express x8 lanes for graphics cards and three PCI Express x1 or one PCI Express x4 expansion slot. The M1697 also offers transfers of up to 16-bit HT downstream and 16-bit HT upstream links at 2.0 GT/s. The M1697 provides dual channel Ultra-DMA IDE support, four SATA ports featuring NCQ, 3Gb/s, eSATA, Hot Plug, RAID 0, 1, 0+1, 5, and JBOD operations. High Definition Audio is natively supported along with eight USB 2.0 ports and 10/100Mb/s Ethernet utilizing a properly supported PHY. PCI Express Gigabit Ethernet can be utilized as discrete solution with a properly supported PHY. Further information about the entire ULi product line can be located here.

The EPoX EP-9U1697 GLi offers a complement of options available including two PCI Express x16 connections (offering x8 SLI operation with PowerExpress driver); one PCI Express x1 connection, and three 32-bit PCI 2.3 connections. The board also offers AC97 audio via the Realtek ALC-655 codec instead of HD audio, PCI 10/100 Mb/s Ethernet via the Realtek RTL8201CL PHY, eight USB 2.0 ports (utilizing two USB 2.0 headers), four SATA 3Gb/s connectors, and two ATA133 Ultra-DMA IDE connectors.

Let's find out how EPoX's offering fares against the competition.



Basic Features: EPoX EP-9U1697 GLi

Specification	Epox EP-9U1697 GLi
CPU Interface	939-Pin Socket supporting AMD Athlon 64 / 64FX / 64X2
Chipset	ULi M1697 - Single Chip Solution
Bus Speeds	200MHz ~ 500MHz in 1MHz increments
CPU Clock Multiplier	Auto, 5x ~ 12x (4000+) in 1x increments, max multiplier dependent upon CPU utilized
Memory Speeds	Auto, 100 MHz, 133MHz, 160MHz, 200MHz, 218 MHz, 240MHz
PCI Bus Speeds	Async with HT frequencies, Disabled
PCI Express Bus Speeds	Async with HT frequencies, Disabled
HT Multipliers	Auto, 200MHz, 400MHZ, 600MHz, 800MHz, 1000MHz
HT Link Speed	Auto, 8-bit, 16-bit, 4-bit, 2-bit
Core Voltage	Auto, -0.200V to +0.250V (in 0.025V increments), (max voltage dependent upon CPU)
DRAM Voltage	Auto, +0.00V to +0.70V, (in 0.10V increments)
Chipset Voltage	Auto, +0.00V to +0.30V, (in 0.10V increments)
Memory Slots	(4) x DIMM, max. 4GB, DDR 400/333, non-ECC, un-buffered memory, Dual Channel Operation supported
Expansion Slots	(2) x PCI-E x16 (first slot operates in 1x16 mode, both slots in 2x8 mode) (1) x PCI-E x1 (3) x PCI 2.3
Onboard SATA	ULi M1697: (4) x SATA II
Onboard IDE	ULi M1697: (2) x UltraDMA 133/100/66/33
SATA/IDE RAID	ULi M1697: (4) x SATA II RAID 0, RAID 1, RAID 0+1, RAID 5
Onboard USB 2.0	(8) USB2.0 ports (four ports, two headers for four more ports)
Onboard LAN	Realtek RTL8201CL PCI 10/100Mb/s Ethernet LAN - PHY
Onboard Audio	Realtek ALC-655, 5.1 channel capable AC97 2.3 Audio Codec
Power Connectors	24-pin ATX 4-pin ATX 12V
Back Panel I/O Ports	1 x PS/2 Keyboard 1 x PS/2 Mouse 1 x Parallel (ECP/EPP) 1 x Serial (COM1) 1 x Audio I/O Panel 1 x RJ45 LAN 1 x Coaxial S/PDIF Out 4 x USB 2.0
Other Features	Thunder Probe - Windows based Monitoring Program Magic Flash - Windows based BIOS Flash Utility Magic Screen - Windows based Bootup Screen Design Program CP80P - Post Port Debug LED EZ Button - On-board Power and Reset Button
BIOS	Award 1.0 (2/06/06)

The EPoX EP-9U1697 GLi is a value-based performance board targeted towards the AMD enthusiast user. The board ships with an accessory package that includes the standard assortment of IDE/SATA cables and power connectors. Epox also includes an extensive driver CD along with desktop utilities. Of special note is Epox's Thunder Probe windows based utility that allows for full temperature monitoring and control of the CPU and system fans. Epox also includes the Thunder Flash set of utilities for automatic retrieval and updates to the latest BIOS, BIOS recovery, and boot-up screen customization. More information about the Thunder Flash system can be located here.

This is the Main BIOS Setup Utility section, which displays the change configuration categories available on the board.

This is the Advanced Chipset Features section that contains the subcategories for the DRAM, PCI-E, Power, and HTT configuration sections.

The Power BIOS configuration section allows you to set individual parameters manually for voltage, multipliers, and chipset frequency settings. The range of features available for change is extensive for an entry-level performance board.

The CPU Clock ratio can be adjusted from the Power BIOS section as well as the Memory Clock ratio. The BIOS allows for an Auto setting that will set the memory clock ratio based upon the SPD settings or you can adjust the memory timings manually. We typically found in our testing that the system would adjust the memory clock ratio based on the conservative side if left on Auto.

The Memory Clock settings are extensive on this board with the Auto setting providing an adjusted ratio dependent upon the SPD settings and bus speeds. We discovered in our testing that the Auto selection would typically adhere to the SPD setting no matter the memory clock ratio or HTT settings. We found it best to set the memory manually based on your system settings and memory capability.



EPoX EP-9U1697 GLi: Features

EPoX designed a decently laid out board with all major connections easily reached except for the floppy drive connector. The EPoX layout provides excellent clearance for cards and components, and was easy to install in a mid-size ATX case. Although the board features a 3-phase voltage regulator power design, it provided excellent stability and allowed for an impressive level of overclocking for an entry-level board.

The DIMM module slots' color coordination is correct for dual channel setup. The memory modules are easy to install with a full size video card placed in the first PCI Express x16 slot. The ULi IDE port 1 connector is located along the edge of the board directly beneath the memory slots. The layout in this area could have been excellent if the 24-pn ATX connector or floppy drive connector would have been placed to the right of the first IDE port connector.

The ULi IDE port 2 connector is located below the ULi SATA ports. The ULi SATA ports are conveniently located below the M1697 chipset and to the left of the memory slots. The SATA ports feature the new clamp-and-latch design, but are not color-coded.

The chipset fan header is located below the ULi SATA ports. The ULi M1697 chipset is actively cooled with a fan and low rise heat sink that did not interfere with cards installed into a secondary PCI Express x16 slot. The fan was generally quiet during operation, although we believe that a well-designed passive heat sink could have been utilized on this board.

The CP80P post port debug LED along with the power on and reset buttons are located to the left of the IDE and SATA connectors and right of the floppy drive connector. The floppy driver connector is located in an unusual position, and if utilized the cabling could interfere with the SATA and IDE ports. The ULi USB connectors, chassis panel, and system fan header are located on the left edge of the board. The yellow CMOS jumper block is a traditional jumper design located to the left of the BIOS chip and below the battery.

The board comes with (2) physical PCI Express x16 connectors, (1) PCI Express x1 connector, and (3) 32-bit PCI 2.3 connectors. The layout of this design offers a very good balance of expansion slots for a mainstream board.

The first physical PCI Express x16 connector is located below the 24-pin ATX power connector. The configuration jumpers and the PCI Express x1 connector are located next. The additional slot space in this area easily accommodated our dual slot video cards and still allowed the PCI Express x1 connector to be utilized. The second physical PCI Express x16 connector is located next, followed by the three 32-bit PCI 2.3 slots.

We did not have any issues installing our EVGA 7800GTX 512MB or ATI X1900XTX video cards in the second x16 PCI Express slot. Of course, these dual slot cards will physically render the first 32-bit PCI slot useless. We did not have any issues utilizing the first PCI slot with video cards containing single-slot cooling systems.

Returning to the CPU socket area, we find ample room for alternative cooling solutions. We utilized the stock AMD heat sink, but also verified that several aftermarket cooling systems such as the Zalman CNPS9500 would fit in this area during our tests. However, the amount of room in this area for the installation of larger air or water-cooling solutions could be problematic due to the location of the 24-pin ATX connector.

Epox places the 24-pin ATX connector and four-pin 12v auxiliary power connector at the top of the CPU socket area. The power connectors are located in an unusual position and could hamper airflow with cabling that crosses directly over the CPU heat sink/fan; although, we did not have any issues in our case due to the stock heat sink and cabling of our power supply. Epox utilized high quality capacitors and a robust three phase power design that provided excellent stability in both stock and overclocked conditions.

The rear panel contains the standard PS/2 mouse and keyboard ports, parallel port, serial port, LAN port, and 4 USB ports. Located to the right of the parallel and serial ports are the first two USB 2.0 ports. Located next to this series of ports are the next two USB 2.0 ports with the RJ-45 LAN port on top. The audio panel is located next and consists of 3 ports that can be configured for 2, 4, and 6-channel audio connections. At the far left, just under the LPT port, you can also find a coax S/PDIF digital audio out port.



FSB Overclocking Results

Front Side Bus Overclocking Testbed
Processor:	AMD Athlon 64 4000+ (San Diego) AMD Opteron 170 (Toledo)
CPU Voltage: AMD Athlon 64 4000+ AMD Opteron 170	1.500V (1.400V default) 1.450V (1.300V default)
Memory Settings:	2.5-3-3-7 1T - (12x) 2.5-3-3-7 1T - (10x) 2.5-4-4-10 1T - (9x)
Memory Voltage:	2.9V
Chipset Voltage:	2.2V
HT Multiplier:	5x up to 260HTT, 4x up to 325HTT
Memory:	OCZ PC4800 Platinum Edition
Cooling:	Zalman CNPS9500
Power Supply:	OCZ Power Stream 520
Maximum CPU OverClock: (AMD Athlon 64 4000+)	255HTT x 12 (3060MHz) +27% (4000+)
Maximum HTT OverClock: (AMD Athlon 64 4000+)	325HTT x 9 (2925MHz) +62% (4000+)
Maximum CPU OverClock: (AMD Opteron 170)	280HTT x 10 (2800MHz) +40% (O170)
Maximum HTT OverClock: (AMD Opteron 170)	315HTT x 9 (2835MHz) +57% (O170)

Our 4000+ CPU posted excellent results in the overclocking tests. We were able to run the system at a HTT setting of 260 with very good results. However, the board would not pass part of our multitasking test suite at the 260 HTT setting, but it was extremely stable and passed all of benchmark test suites at the 255HTT setting with an HT multiplier of 5x. We were able to run our memory at a 1T command rate up to the HTT setting of 325. The board actually completed our test suites at an HTT setting of 335, but it required a 2T command rate along with a voltage increase to 1.600V.

Our Opteron 170 also posted excellent results with this board. However, this CPU has reached 2.95GHz on 1.450V in past testing. At the 10x280 and 9x315 overclock settings, the system was able to complete all of our benchmark test suites three consecutive times and run Prime95 and SuperPI without issue. Overall, the board would make an excellent platform for the typical overclocker.


Memory Stress Testing

Memory stress tests look at the ability of the Epox EP-9U1697 GLi to operate at the officially supported memory frequencies of DDR-400, at the best performing memory timings that the OCZ PC4800 Platinum Edition will support.

EPoX EP-9U1697 GLi Stable DDR-400 Timings - 2 DIMMs (2/4 slots populated - 1 Dual-Channel Bank)
Clock Speed:	200MHz
CAS Latency:	2
RAS to CAS Delay:	2
RAS Precharge:	2
RAS Cycle Time:	5
Command Rate:	1T
Voltage:	2.7V

The EPoX EP-9U1697 GLi was extremely stable with 2 DDR modules in Dual-Channel mode at the settings of 2-2-2-5 at 2.7V. We will now install our memory into all four available memory slots, which results in more strenuous requirements on the memory subsystem than testing 2 DDR modules on a motherboard.

EPoX EP-9U1697 GLi Stable DDR-400 Timings - 4 DIMMs (4/4 slots populated - 2 Dual-Channel Banks)
Clock Speed:	200MHz (800FSB)
CAS Latency:	2
RAS to CAS Delay:	2
RAS Precharge:	2
RAS Cycle Time:	7
Command Rate:	2T
Voltage:	2.7V

The Epox EP-9U1697 GLi displayed superb stability with 4 DDR modules in Dual-Channel operation at the settings of 2-2-2-7, but it required the command rate to be increased to 2T. Note that this 2T requirement for 4 DIMMs is typical of nearly all 939 boards; only the DFI SLI-DR Expert managed 1T timings with 4 DIMMs, but overclocking beyond 205 MHz HT bus speeds once again required 2T.



Test Setup

The ULi M1697 chipset fully supports all AMD Athlon 64 processors in both stock and overclocked conditions.

Performance Test Configuration
Processor(s):	AMD Athlon 64 4000+ utilized for all tests
RAM:	2 x 512MB OCZ Technology PC4800 Platinum Edition Settings- DDR-400 at (CL2-2-2-5, 1T)
Hard Drive(s):	2 x Maxtor MaXLine III 7L300S0 300GB 7200 RPM SATA (16MB Buffer), 1 x Maxtor MaXLine III 7L300R0 300GB 7200 RPM IDE (16MB Buffer)
System Platform Drivers:	ULi SATA Driver - 1059 ULi PowerExpress Engine Enabling driver - 1006E
Video Cards:	1 x MSI 7800GTX (PCI Express) for all non-SLI tests 2 x MSI 7800GTX (PCI Express) for SLI tests
Video Drivers:	NVIDIA nForce 81.98 WHQL
Cooling:	Zalman CNPS9500
Power Supply:	OCZ Power Stream 520
Operating System(s):	Windows XP Professional SP2
Motherboards:	Abit AT8 Asus A8N-SLI Premium Albatron K8SLI Foxconn NF4SK8AA-8KRS Asus A8R-MVP (ATI RD480/ULi1575) Asus A8N32-SLI Deluxe

We tested our MSI 7800GTX video cards using NVIDIA 81.98 WHQL drivers to provide recent performance results. Resolution in all benchmarks is 1280x1024x32 unless SLI is enabled. Resolution in SLI benchmarks is 1600x1200x32 with 4XAA and 8xAF where applicable. 3DMark and Aquamark3 benchmarks use a "Standard Score" setup at the 1024x768 video resolution for both SLI and non-SLI testing.

Note that while newer Beta drivers are now available from NVIDIA's nZone site for the 7900 series cards, the 81.98 remain the most recent official release. Testing began prior to the launch of the G71 cards, but the ULi PowerExpress Engine drivers appears to work with all current Beta drivers as well.



General Performance & Encoding

The EPoX EP-9U1697 GLi is very competitive in the synthetic benchmarks with scores consistently near the top. Of course, the largest spread in PCMark05 is only 3.5%, which means that in real-world use you wouldn't be able to tell the difference between any of these boards without running benchmarks. The combination of the ULi M1697 chipset and Epox's BIOS capabilities has resulted in a very cost-friendly, performance-oriented solution. Our encoding tests will soon change to the DivX 6.1 codec and additional multimedia tasks.



Memory Performance

We recently switched to version 2.50 of Everest, so these scores are not comparable to previous tests with version 2.20. The memory latency test shows a slight advantage over the nForce4 boards, but the read and write performances of the ULi and ATI based solutions are impressive and accounts for their performance advantages in memory intensive applications.


Overclocking Performance

The overclocking performance graphs have been added to the standard benchmark test suite and should allow for a better comparison on the overclocking capabilities of tested boards. For more details on the specific overclocking abilities of this board, please refer to the Overclocking and Memory Stress Test section in the Basic Features section.

The EPoX EP-9U1697 GLi is a very competitive overclocking platform for a value based performance board. Hopefully, the BIOS that we tested and components on the board are the same as production level boards. Epox has assured us that this is the case.



Gaming Performance

The overall gaming performance of the EPoX EP-9U1697 GLi is outstanding in all titles. In actual game play, the board exhibited excellent manners and we did not witness anomalies or issues in our extended on-line gaming sessions. At stock settings, this board competes very well against boards costing twice as much.



SLI Performance

We installed our SLI bridge connector and the ULi PowerExpress Engine Enabling driver with the NVIDIA 81.98 WHQL driver set for the SLI testing. Upon installation of the ULi PowerExpress driver and a system reboot, the NVIDIA 81.98 driver properly recognized our additional MSI 7800GTX video card for SLI operation. EPoX has stated they will not be including an SLI bridge connector or the ULi PowerExpress driver in their retail package. Also, the board will not be certified for SLI operations.

The performance of the Epox board in our SLI gaming benchmarks is very competitive and, at times, better than our NVIDIA nForce4 boards. In fact, we are very surprised at the level of performance in these benchmarks compared to the NVIDIA nForce4 solutions and we understand why NVIDIA chose to pursue ULi.

The performance pattern continues in the synthetic benchmarks with the Epox board scoring a sweep in the 3DMark tests. Overall, we were very impressed with the SLI performance of the EPoX EP-9U1697 GLi and did not find any performance or graphic issues in our test suites. We also played numerous games without issue and believe that the ULi PowerExpress Engine Enabling driver is a very stable solution.



Disk Controller Performance

With the variety of disk drive benchmarks available, we needed a means of comparing the true performance of the wide selection of controllers. The logical choice was Anand's storage benchmark first described in Q2 2004 Desktop Hard Drive Comparison: WD Raptor vs. the World. The iPeak test was designed to measure "pure" hard disk performance, and in this case, we kept the hard drive as consistent as possible while varying the hard drive controller. The idea is to measure the performance of a hard drive controller with a consistent hard drive.

We played back Anand's raw files that recorded I/O operations when running a real world benchmark - the entire Winstone 2004 suite. Intel's iPEAK utility was then used to play back the trace file of all IO operations that took place during a single run of Business Winstone 2004 and MCC Winstone 2004. To try to isolate performance differences to the controllers that we were testing, we used the Maxtor MaXLine III 7L300S0 300GB 7200 RPM SATA drive in all tests . The drive was formatted before each test run and a composite average of 5 tests on each controller interface was tabulated in order to ensure consistency in the benchmark.

iPeak gives a mean service time in milliseconds; in other words, the average time that each drive took to fulfill each IO operation. In order to make the data more understandable, we report the scores as an average number of IO operations per second so that higher scores translate into better performance. This number is meaningless as far as hard disk performance is concerned, as it is just the number of IO operations completed in a second. However, the scores are useful for comparing "pure" performance of the storage controllers in this case.

The performance patterns hold steady across both Multimedia Content IO and Business IO, with the ULi based disk controllers providing the fastest IO operations followed by the on-board NVIDIA nForce4 SATA controllers. Of particular note is the excellent performance generated by the ULi IDE controller logic while the SATA performance is incredible compared to the nForce4 chipset.



Firewire and USB Performance

After looking at many options for Firewire and USB testing, we finally determined that an external USB 2.0, Firewire 400, and Firewire 800 hard disk would be a sensible way to look at USB and Firewire throughput.

Our first efforts at testing with an IDE or SATA drive as the "server" yielded very inconsistent results, since Windows XP sets up cache schemes to improve performance. Finally, we decided to try a RAM disk as our "server", since memory removed almost all overhead from the serving end. We also managed to turn off disk caching on the USB and Firewire side by setting up the drives for "quick disconnect" and our results were then consistent over many test runs.

We used 1GB of fast 2-2-2-5 system memory set up as a 450MB RAM disk and 550MB of system memory. Our standard file is the SPECviewPerf install file, which measures 432,533,504 bytes (412.4961MB). After copying this file to our RAM disk, we measured the time for writing from the RAM disk to our external USB 2.0, Firewire 400, or Firewire 800 drive using our Windows based timing program. The copy times in seconds were then converted into Megabits per second (Mb) to provide a convenient means of comparing throughput. Higher Rates, therefore, mean better performance in this particular test.

Possibly the most striking finding in our Firewire and USB throughput tests is the continued performance of an external hard drive connected to Firewire 800. Our benchmarks show Firewire 800 is up to 46% faster than a drive connected to the more common Firewire 400, and about 29% faster than USB 2.0.

The Epox board does not offer a Firewire option. The USB 2.0 performance is consistent with other ULi based controllers and continues to lag behind the NVIDIA nForce4 chipset solutions in throughput.



Ethernet Performance

The current motherboard test suite includes LAN performance measurements. All of these boards utilize PCI Express controllers with the only difference being the supplier of the core logic.

The Windows 2000 Driver Development Kit (DDK) includes a useful LAN testing utility called NTttcp. We used the NTttcp tool to test Ethernet throughput and the CPU utilization of the various Ethernet Controllers used on the Intel motherboards.

We set up one machine as the server; in this test, an Intel system with an Intel CSA Gigabit LAN connection. Intel CSA has a reputation for providing fast throughput and this seemed a reasonable choice to serve our Gigabit LAN clients.


At the server side, we used the following Command Line as suggested by the VIA whitepaper on LAN testing:

Ntttcpr -m 4 ,0,‹server IP› -a 4 -l 256000 -n 30000

On the client side (the motherboard under test), we used the following Command Line:

Ntttcps -m 4 ,0,‹client IP› -a 4 -l 256000 -n 30000

At the conclusion of the test, we captured the throughput and CPU utilization figures from the client screen.

The M1697 natively supports 10/100Mb/s Ethernet operations, so Epox utilizes the PCI based Realtek RTL8201CL 10/100Mb/s Ethernet PHY for this purpose. ASRock was able to utilize a discreet PCI-E Gigabit solution on their ULi board and we feel that Epox should have followed suit or at least offered a PCI Gigabit Ethernet option on this board model.

Obviously, the performance of the 10/100 Realtek PHY is not competitive with the Gigabit based solutions with an average throughput of 98.9Mb/s. Also rather odd is the CPU usage required for this throughput. Most 100 Mbit connections require very little in the way of CPU time, due to the much lower bandwidth requirements.

All standard Ethernet tests were performed with standard frames and the NVIDIA Active Armor suite disabled unless otherwise noted. Gigabit Ethernet supports Jumbo frames as well and provides a further reduction in CPU overhead.



Audio Performance

We limited audio testing to the Rightmark 3D Sound version 2.2 CPU utilization test and tested with sound enabled to show the performance effects on several games. The Rightmark 3D Sound benchmark measures the overhead or CPU utilization required by a codec or hardware audio chip.

The Realtek ALC-655 AC97 audio codec was tested with the recently released 3.84 driver set. The Realtek DirectSound audio drivers do not support more than 26 hardware buffers and the OpenAL 1.1 drivers do not support more than 23 hardware buffers at this time, so the scores cannot be directly compared to the HDA Mystique 7.1, Realtek ALC-882, and Creative Labs Sound Blaster X-FI audio solutions in the benchmarks. The Realtek OpenAL 1.1 driver increases CPU utilization up to 8% more than the Realtek DirectSound drivers.

The Realtek ALC-655 AC97 audio codec has average CPU utilization rates with reductions of up to 4% in the 3D tests compared to the previous driver release. The HDA Mystique 7.1 Gold has the highest overall utilization rates of the audio solutions tested. The Realtek ALC-655 performance is good, but it does not match the audio quality of the ALC-882 HD audio codec. The Sound Blaster X-FI has the lowest overall rates as expected. Let's find out how these results translate into real world numbers.

The audio performance numbers remain consistent as the Realtek ALC-655 generally finishes near the HDA Mystique 7.1 and SoundBlaster X-FI. Serious Sam II suffers a loss of 43%, Splinter Cell at 1%, Battlefield 2 at 21%, Call of Duty 2 at 2%, and F.E.A.R. at 4%. The output quality of audio with the Realtek ALC-655 is good and continues to improve with each driver release, but in no way compares to the HD audio codecs or most discreet audio solutions. The majority of home/office users should have no issues utilizing the ALC-655, but we would not recommend it as the primary audio solution for a gaming or HTPC system, considering the overall quality of audio. In fact, the audio quality in most applications sounded flat and lacked clarity.

If you are a serious gamer, then a dedicated sound card is still a requirement to ensure consistent frame rate averages across a wide variety of games. We noticed in previous testing of our Battlefield 2 and Half Life 2 benchmarks that the Realtek AC97 audio codecs would cause stuttering in intensive scenes. The 3.84 driver release does not have stuttering in our current benchmarks while improving performance across the board.

The Realtek 3.84 driver installation installs a basic control panel that features a built-in 10-band equalizer along with the standard mixer and speaker controls. We found the control panel to be user friendly and a definite improvement over the standard Windows audio properties application.



Final Words

The EPoX EP-9U1697 GLi offers excellent performance at a bargain price. The performance of the board in the majority of the benchmarks was extremely competitive with the ATI and NVIDIA chipset offerings. The stability of the board was superb at stock settings and very good at overclocked settings. We find it refreshing that a board with this performance is being offered for a retail price around US $85. While the ASRock 939SLI32-eSATA2 offers a greater feature set for the same price, the EPoX board offers significantly better performance in the overclocking area.

With that said, let's move on to our performance opinions regarding this board.

In the video area, the inclusion of dual PCI Express x16 connectors provides dual card capability with eight PCI Express lanes per graphics connector available during dual card operation. The board utilizes a manual jumper system to switch from single x16 PCI-E operation to dual x8 PCI-E operation. While not as convenient as a digital switch, it does assist in keeping the cost of the board down. The performance of the board under SLI testing matched that of our nForce4 boards and offered full SLI compatibility with the ULi PowerExpress Engine Enabling driver although the board is not SLI certified. We tried this driver with the 81.85, 81.95, 81.98, and 82.12 drivers without any issue in a myriad of benchmarks and games. EPoX will not be supporting or certifying SLI operations on the board. The board also fully supported our ATI X1900XTX and EVGA 7900GTX video cards in limited testing.

In the on-board audio area, the EPoX board offers the Realtek ALC-655 AC97 audio codec. While the ULi M1697 chipset fully supports 7.1 HD audio, EPoX only provided 5.1 AC97 capability, probably due to the implementation cost of the Realtek HD audio codecs. The audio output of this codec in the music, video, and DVD areas is decent for an on-board solution. The audio quality in gaming was okay, but it did not match the output of the Sound Blaster X-FI or even the Realtek ALC-882. If you plan on utilizing this board for gaming, then our recommendation is to purchase an appropriate sound card for consistency in frame rates across a wide range of games and audio quality. However, the Realtek ALC-655 should suffice for the majority of home/office users.

In the storage area, the EPoX board offers the full complement of storage options afforded by the ULi M1697 chipset. The board offers RAID 0, 1, 0+1, 5 capability, NCQ, Hot Plug, and 3Gb/s support along with dual channel ATA133 Ultra DMA capability. The board also offers eight ULi USB 2.0 ports when utilizing the two USB 2.0 headers. The performances of the ULi SATA and IDE controllers were excellent and easily exceeded the nForce4 solutions.

In the performance area, the EPoX EP-9U1697 GLi generated outstanding benchmark scores in the majority of applications, considering the price of the board. The board's performance was consistently competitive with other ATI, ULi, and NVIDIA chipset offerings in the majority of benchmarks and applications. The stability of the board was excellent during testing and general usage.

The EPoX EP-9U1697 GLi is a board designed and marketed for the AMD enthusiast on a budget, yet it excels at equaling the performance of boards costing up to two times more. EPoX is the first manufacturer to market with the ULi M1697 single chipset configuration offering and we are impressed with their efforts.


However, we feel that EPoX made the following errors in the design and execution of the board. The location of the floppy drive connector at the bottom of the board, the location of the 24-pin ATX power connector right above the CPU socket, and the use of active cooling on the ULi M1697 chipset could be improved.

Although we understand the budget nature of the board, we believe that the lack of PCI-E or PCI based Gigabit Ethernet LAN controller is an issue for a board being targeted to an enthusiast market. Also, the lack of an HD audio codec hampers the audio capability of the board and while we understand most serious gamers will opt for a discreet audio solution, we still feel that this option should have been included, considering the native support provided by the ULi M1697. EPoX will be offering the 9U1697 GLi-J board with PCI-E Gigabit via the Marvell 88E8053 controller, but it will only be upgrading the audio to the 8-channel capable ALC-850 AC97 codec. EPoX has not disclosed pricing or availability of this model at the time of our writing.

The BIOS issue that we encountered, which caused random lockups after exiting Windows XP, was disconcerting, although we did not notice any stability or performance issues within Windows or during testing. EPoX has provided us with an updated BIOS for testing. However, it also has issues. At the time of this writing, we are still waiting for an acceptable solution from EPoX.

We feel that it is unfortunate that the ULi M1697 chipset utilized on this board will probably have a short life span in light of the NVIDIA acquisition, but we are glad to see that EPoX has done a wonderful job in bringing this board to market at a low price-point while offering superb performance and stability. If you're willing to live with the few minor shortcomings mentioned above, this is a great board given the budget price.