Logo   Banner   TopRight
Foxconn e-bot Small Form Factor PC
Date: Feb 07, 2005
Author: Robert Maloney
Introduction and Specifications

In what seems to be an ever expanding market, we've got yet another look at a new Small Form Factor system on HotHardware.  This time the system comes to us by way of Foxconn.  The same company that's been putting their stamp on motherboards and other OEM hardware since 1991.  Following their recent successful foray into the retail market with their own branded motherboards, Foxconn has a new offering pulling together their motherboard architecture skills with their flair for PC enclosures.  The results, as you will see, appear to be right on target.

Foxconn's initial entry is called the e-bot, which arrived on our doorstop in a bright orange and blue box, similar to the colored packaging of Intel's P4 product line.  Initially, it conjured up odd thoughts as to what, or who, an e-bot was, as there was scant other information on the exterior.  Delving into the contents of the box, we pulled out a shiny silver, black, and orange case which had a faint similarity to a pop-up toaster.  All kidding aside, the Foxconn e-bot seems all at once to take SFF PCs from its humble beginnings to a whole new level, similar to the jump made when PCs shed their beige, boxy casings to the now ubiquitous multi-colored, windowed, and LCD-screened rigs of today.  Powered internally by a custom Foxconn board based on the SiS661FX chipset, it has all the makings of a successful marriage of form and function.


Specifications of the Foxconn e-bot
Not just another SFF PC

•Foxconn 661FXSA

•SiS 661FX Northbridge
•SiS 963 Southbridge

•Intel Socket 478 Pentium 4 (533/800 MHz FSB)
•Intel Socket 478 Celeron with 400 MHz FSB
•Supports Hyper-Threading Technology

•Single-channel DDR 400/333/266
•2x DIMM slots (2GB max)

•One AGP slot supporting 1.5v 4X/8X AGP card
•AGP 3.0 compliant
•On-board high performance 3D VGA controller

•Realtek ALC658 Audio Chipset
•6-channel audio with analog and digital output
•AC '97 v2.3 Specification Compliant

•SiS 900-Based PCI Fast Ethernet controller
•10/100 Mbps operation

•One IDE port supporting Ultra DMA-33/66/100/133
•Supports up to 2 IDE devices

•Onboard SiS 1394 controller
•One front port (4-pin) and one rear port (6-pin)
•IEEE-1394a compliant with up to 400 Mbps bandwidth

Front-panel I/O
•Hidden "pop-up" 7-in-1 card reader
•6 button audio playback control functions
•USB 2.0 ports x 2
•FireWire (4-pin) port
•Microphone port
•Headphone port
•Power and Reset buttons

Rear-panel I/O
•2x USB 2.0 ports
•1x IEEE-1394a FireWire 6-pin connector
•1x LAN connector (RJ-45)
•3x Audio jacks (user configurable)
•1x 15-pin VGA port
•PS/2 mouse port
•PS/2 keyboard port
•External Power Supply Connector

Cooling System
•Triple Copper Heat Pipe System
•Single 80mm Fan for CPU & System Cooling

Optical Drive (included)
•DVD-ROM, CD-R/RW Slimline Auto Tray Drive
•IDE / ATAPI interface
•Speed: 8x / 24x / 10x / 24x

Drive Bays
•1x 3 1/2" Easy Access Hard Drive Bay
•1x 5 1/4" (DVD/CD/RW drive installed)

Expansion Slots
•1x AGP 8x
•1x PCI

Dimensions (L x W x H, in.)
•11.75" x 7" x 10.75"

The Bundle:
The rest of the contents of the box mostly revolved around three main areas: installation, power, and cooling.  There weren't as many items included as one might have expected, and the bundle surely was not as comprehensive as what we saw with the Epox eX5-320S a few months back.

Two manuals cover the hardware side pretty well.  The thicker of the two manuals is a comprehensive guide that covers all aspects of installing and using the system, including BIOS entries and how to use the accompanying software.  The other manual is a quick installation guide, for those who want to just jump right in and get to work.  All the necessary drivers are included on the driver CD, although DirectX 9.0b was one of them, with DX9.0c being readily available for some time now we would have expected the newer version to be included on the CD.  Foxconn has also thrown in a licensed copy of Nero Burning ROM, which will allow for immediate use of the included slim-DVD/CD-RW drive.

The cooling solution came in a separate box, and we will cover that on the next page with our installation notes.  That left us with a small baggie with assorted screws, twist ties, and keys for locking down the PC. 
We also found a two piece cable with a transformer on it.  Upon closer review, it was rated at 200W, which made this unit the actual power supply.  Sure enough, a peek inside the chassis showed no oversized box of coils, transistors, and fans, thus freeing up interior space as well as keeping internal heat build-up to a minimum.


A Closer Look at the e-bot


The e-bot struts its stuff
A system chock full of goodies



The Foxconn e-bot's design definitely followed a different plan that what we've seen with most other manufacturer's small form factor systems.  Instead of a typically boxy chassis, the e-bot was more vertically inclined, and looked much more like a mini-tower without the column of drive bays that one would expect.  The front is originally only occupied by a LCD screen with various buttons for power, restart, and audio playback.  Similar to the Epox eX5-320S, the Foxconn e-bot can play CDs and MP3s without booting into Windows.  While the eX5-320S booted into a mini-OS that allowed for a bit more advanced navigation (including reading MP3s from the hard drive) the Foxconn e-bot streamlines the process a bit further.  No additional display was needed to view the tracks as all information is displayed on the LCD screen.  Audio tracks can be paused or stopped, skipped forward or backward, and the volume can be raised or lowered using the conveniently placed large buttons directly beneath this screen.  On one hand, it's cool to be able to access music without fully booting into Windows, but since the e-bot can only access CDs or MP3s from a CD-R it has limited appeal.

Hidden at the top and bottom of the e-bot were various ports for reading media cards and connecting USB devices or audio devices. Beneath the slimline DVD/CD-RW drive which comes pre-installed in the system, a cover is pulled down to allow access to two USB 2.0 ports, a FireWire port, and jacks for headphones and a microphone.  The 7-in-1 card reader up top was accessed by pushing down slightly on the edge of the case and letting go.  All popular card types are supported including memory sticks, Secure Digital, SmartMedia, and Compact Flash cards.  Pushing down a second time locks the reader back into place.  Just past the card reader is a large handle, making it a snap to carry the e-bot wherever needed, such as ever-popular LAN-parties.  For added security, the e-bot's side panel can also be locked to keep idle hands away from and out of the inside of the chassis.


Once the side door was unlocked, it only required a push on a back lever to get inside.  The inside of the e-bot was mostly open and clear of wires, the bulk of which were all located towards the bottom of the system.  Both the pre-installed DVD/CD-RW drive and hard drive will be located here with the IDE ports placed nearby.  The hard drive is actually installed within a metal cage, and is accessed separately from other devices.  Unhooking the bottom plate allows for accessing the cage, and the IDE cable and MOLEX power connector are fed down here as well.  Once again, it's a very innovative move by Foxconn that provides an out-of-the-way location for the hard drive as well as keeping wires from needlessly traveling around the case.  Our only concern would be if the e-bot took a hard hit along the bottom of the case, such as the unit getting dropped suddenly from a table.  The cage is locked into the inner structure, however, rather than placed right along the bottom, which may provide a little more protection.



Once the hard drive was installed, we we're able to move onto installing the CPU and memory.  Without a power supply unit taking up space, there's plenty of room to get to the sockets making installation a snap.  Two DIMM slots were available and can be populated separately since the SiS661FX supports single channel memory.  With the CPU and DDR installed and locked down, we turned our attention to the cooling device.  Similar in some ways to the ICE technology that Shuttle uses, a heatsink is locked down over the Pentium 4 CPU, and three copper heat pipes carry the heat away from the heatsink to a radiator type structure that is mounted on the rear of the chassis.  The heatpipes weren't machined as clean and neatly as we have seen with others, but once installed you'll probably never look at them again.  A single 80mm fan is then placed around the radiator and blows air across the fins thus cooling them off.  At startup, this fan gets quite noisy.  Once it kicks down to normal speeds it's actually not too bad, although you will notice it rev up from time to time.  Overall, it's definitely not the quietest SFF system we've ever tested.  Finally, once everything is installed, we're ready to boot up and install Windows.  If you see a penguin start dancing across the mini-LCD screen then you know you're on the right track.


BIOS and Overclocking


The BIOS of the Foxconn e-bot
So many options, so little time



The BIOS that controls the Foxconn e-bot was a customized variant of the Phoenix-Award BIOS that we're typically used to seeing.  The first main difference might be the inclusion of Foxconn specific features.  Here we found four specific settings, SuperBoot, SuperBIOS-protect, SuperRecovery, and SuperSpeed.  The first three settings control the way the system boots up - whether it's streamlining the process with SuperBoot or protecting and recovering BIOS settings with the other two settings.  SuperSpeed is directly related to overclocking by allowing direct input of desired front side bus speeds and changing the CPU to RAM divider when possible.

The remaining sections contained the typical fare for enabling or disabling various components, such as the on-board audio or LAN.  Within the Advanced Chipset Features were three other areas that delved into specific subsystems.  The DRAM timings could by detected BY S PD, using the ratings contained within the ICs, or set manually to more optimized timings, sometimes at the risk of system stability.  We actually ran into a few such problems when enabling performance mode and setting our GEIL DDR manually to 2-2-2-5 timings.  We tried testing both ways after this - with performance mode enabled and DRAM BY SPD, and then performance mode disabled but timings manually set to 2-2-2-5.  The latter offered the better results, which is what we used in our graphs.

Overclocking Tools
A little lacking, but will it be enough?


As we mentioned earlier, the front side bus could be entered in directly from the BIOS features screen, within the SuperSpeed block.  We could choose values between 200 MHz (stock speed) up to 350 MHz, and the CPU:DRAM automatically updated to whatever setting would keep the DRAM speed as close to 400MHz as possible.  This setting could be manually overridden, offering 1:1, 2:1, 3:2, and 6:5 ratios.  This seemed like a great start; a high range of FSB speeds to enter with many CPU:DRAM ratios to choose from.  What we found was sorely lacking, however, were any options to raise any voltages whatsoever in the Frequency/Voltage Control section of the BIOS.  In fact, the only two choices in this section were to "auto detect" the DIMM/PCI clock and enable or disable Spread Spectrum to minimize EMI.  Without the ability to raise the voltages at all, we weren't expecting to get to far with overclocking the Foxconn e-bot.

SANDRA CPU Benchmark

PCMARK04 Benchmark

While we fully expected to get cut short very quickly, we kept adding 5 Mhz at a time while updating the memory divider until we maxed out at 245MHz for the front side bus.  We were able to get a minor overclock on the memory using a 6:5 ratio, equaling 204MHz (408MHz effective) for the GEIL PC3500 DDR.  This was a higher overclock than we were able to achieve with any of our previous SFF PCs, and came without any changes in the CPU voltage which was very surprising.  We even ran a program called SuperStep that came with the e-bot to check on voltages, and found that it was still running close to the expected 1.5V.

As seen in the accompanying screenshots, we were now running our 2.4GHz Pentium 4 as high as 2.94GHz, without any major modifications and did not run into any problems with our benchmark routines like SiSoft's SANDRA 2005 or Futuremark's PCMARK04.  Each of these benchmarks gave us scores that were well above the original scores achieved at the stock speed.  Our new score in SANDRA was well beyond a 3GHz Pentium 4 CPU and was more comparable to the performance seen with an AthlonXP 3200+.  Perhaps jumping the gun a little bit, the PCMARK04 results we got while overclocked were 22% better for CPU performance and 12% better for the memory.  We were really impressed by how the Foxconn e-bot was able to get these kinds of gains with very little time or work invested by the user.  Of course, mileage may vary from user to user, so these results should not necessarily be expected with each setup.


Testing Setup and SANDRA comparisons


How we configured our test systems: When configuring the test systems for this review, we first entered the system BIOS and set each board to their "Optimized" or "High-Performance Defaults".  The hard drive was then formatted, and Windows XP Professional with Service Pack 2 was installed. When the installation was complete, we hit the Windows Update site and downloaded all of the available updates, with the exception of the ones related to Windows Messenger. Then we installed all of the necessary drivers, and removed Windows Messenger from the system altogether. Auto-Updating and System Restore were also disabled, and we setup a 768MB permanent page file on the same partition as the Windows installation. Lastly, we set Windows XP's Visual Effects to "best performance", installed all of our benchmarking software, defragged the hard drives and ran all of the tests.

Test System Specifications
Can you say "Intel Inside!"

Foxconn e-bot

SiS 661FX
Intel Pentium 4 2.4GHz
2x256MB GEIL PC3500 DDR

NVIDIA GeForce FX 5900XT
On-Board 10/100 Ethernet
On-Board Audio

WD "Special Edition" 40GB
7,200 RPM IDE Hard Drive

Windows XP Pro SP2

NVIDIA Forceware v66.93
DirectX 9.0c

Epox eX5-320S "Mini-Me"

Intel i865G
Intel Pentium 4 2.4GHz
2x256MB GEIL PC3500 DDR

NVIDIA GeForce FX 5900XT
On-Board 10/100 Ethernet
On-Board Audio

WD "Special Edition" 40GB
7,200 RPM IDE Hard Drive

Windows XP Pro SP2
NVIDIA Forceware v66.93
DirectX 9.0c

Shuttle SB61G2

Intel i865G
Intel Pentium 4 2.4GHz

NVIDIA GeForce FX 5900XT

On-Board 10/100 Ethernet
On-Board Audio

WD "Special Edition" 40GB
7,200 RPM IDE Hard Drive

Windows XP Pro SP2
NVIDIA Forceware v66.93
DirectX 9.0c
Preliminary Benchmarks With SiSoft SANDRA 2005
Synthetic Testing of the e-bot

We began our testing with SiSoftware's SANDRA, the System ANalyzer, Diagnostic and Reporting Assistant. SANDRA consists of a set of information and diagnostic utilities that can provide a host of useful information about your hardware and operating system. We ran three of the built-in sub-system tests that partially comprise the SANDRA 2004 suite (CPU, Multimedia, and Memory). All of these tests were run with the Foxconn e-bot powered by an Intel Pentium 4 2.4GHz CPU with 512MB of Geil DDR.

SANDRA CPU Benchmark
Pentium 4 @ 2.4GHz
512MB DDR (CL2)

SANDRA Memory Benchmark
Pentium 4 @
512MB DDR (CL2)

SANDRA Multimedia Benchmark
Pentium 4 @
512MB DDR (CL2)

We had a tough time finding exact comparisons within Sandra's database, so instead we will just give a summary of its relative performance throughout the benchmarking routines.  In the CPU Arithmetic test, the Foxconn e-bot's performance came closest to that of a Pentium 4 3GHz CPU, yet falling way shy of that of an AthlonXP.  The score we achiveved was much higher than a 2.4GHz P4, but using a slower bus speed.  The CPU Multimedia test flip-flopped a bit, where the benchmark score was generally closer to AMD's offerings and far behind the 3GHz P4.  Again, however, we were able to handily beat the P4 2.4GHz score from the database.  Finally, in the memory benchmark, we saw what will probably be the e-bot's biggest handicap; the single channel memory.  The memory bandwidth in this test was nearly a third slower than what was found with i865 or i915 boards, which is exactly what we will be using for comparisons in the upcoming benchmarks.


PCMark04 Benchmarks

Futuremark PCMark04
More Synthetic CPU and Memory Benchmarks

For our next round of benchmarks, we ran the CPU and Memory performance modules built-into Futuremark's PCMark04.  For those interested in more than just the graphs, we've got a couple of quotes from Futuremark that explain exactly what these tests do and how they work...

 "The CPU test suite is a collection of tests that are run to isolate the performance of the CPU. There are nine tests in all. Two pairs of tests are run multithreaded - each test in the pair is run in its own thread.  The remaining five tests are run single threaded. These tests include such functions as file encryption, decryption, compression and decompression, grammar check, audio conversion, WMV and DivX video compression."

While all three systems were using the same setup, we did see some slight difference in the PCMark04 benchmark.  The Epox machine garnered the top spot overall, with a slight difference between it and the Shuttle and then again between the Shuttle and the Foxconn e-bot.  We usually find that these slight differences can be caused by reported CPU timings, but in this case the Foxconn e-bot had the 2.4GHz P4 listed at 2.41GHz, which was actually higher than the Shuttle SB61G2 running at the typical stock speed.  All in all, the degree of separation between the three systems will be mostly negligible in benchmarking results.

"The Memory test suite is a collection of tests that isolate the performance of the memory subsystem. The memory subsystem consists of various devices on the PC. This includes the main memory, the CPU internal cache (known as the L1 cache) and the external cache (known as the L2 cache). As it is difficult to find applications that only stress the memory, we explicitly developed a set of tests geared for this purpose. The tests are written in C++ and assembly. They include: Reading data blocks from memory, Writing data blocks to memory performing copy operations on data blocks, random access to data items and latency testing."

The order of the boards didn't change in the memory benchmark, although the delta between the Foxconn e-bot and the other two SFF PCs was much more pronounced.  This is solely due to the single channel DDR configuration that the SiS661FX is bulit around.  The single channel bandwidth simply can't keep up with the systems using dual channel memory, typically found on motherboards such as the i865G.  We will see in our later benchmarks how this will affect overall performance.


Windows Media Encoder 9 and 3DMark03

Windows Media Encoder 9
Digital Video Encoding

We continued our testing with another application from Futuremark, 3DMark03, and with a video encoding benchmark, Windows Media Encoder 9.  In the WME9 test, we took a 416MB Digital Video file and encoded to a WMV9 format used for streaming video.  Times were recorded in minutes and seconds, with lower times indicating better performance.

This benchmark was mostly a wash, as all three systems put up nearly identical times.  What was most interesting to note, however, was that even though the Foxconn e-bot has consistently shown lower scores in the synthetic CPU and Memory benchmarks, it technically came in first in this "real-world" test.

DirectX Gaming Performance - Sort Of

It's not an actual game, but 3DMark03's built-in CPU test is a "gaming related" DirectX metric that's useful for comparing relative performance among similarly equipped systems.  This test consists of two different 3D scenes that are generated with a software renderer, which is dependant on the host CPU's performance.  This means that the calculations normally reserved for your 3D accelerator, are instead sent to the central processor.  The number of frames generated per second in each test are used to determine the final score.

Using only the CPU benchmark as the basis for comparison, we've got the Epox eX5-320S back on top, followed by the Shuttle SB61G2 and finally the Foxconn e-bot.  Not too surprising, since this is exactly the same order of systems that we saw with the PCMark04 CPU benchmark.  The dropoff was much larger than we expected, and mulitple attempts at running 3DMark always produced the same results. 


Business and Content Creation Winstones

Business &Content Creation Winstones
Real-World Application Performance

Synthetic benchmarks only tell part of the performance story, so we took the Epox eX5-320S to task in some "real world" scenarios as well.  For our first set of real world tests, we did some benchmarking with Ziff Davis' Business Winstone 2004 suite, followed by the more demanding Content Creation Winstone 2004 suite

       The PC Magazine Business Winstone 2004 test utilizes the following applications in its benchmark:

  • Microsoft Access 2002
  • Microsoft Excel 2002
  • Microsoft FrontPage 2002
  • Microsoft Outlook 2002
  • Microsoft PowerPoint 2002
  • Microsoft Project 2002
  • Microsoft Word 2002
  • Norton Antivirus Professional Edition 2003
  • WinZip 8.1

The PC Magazine Content Creation Winstone 2004 test utilizes the following applications in its benchmark:

  • Adobe Photoshop 7.0.1
  • Adobe Premiere 6.50
  • Macromedia Director MX 9.0
  • Macromedia Dreamweaver MX 6.1
  • Microsoft Windows Media Encoder 9 Version
  • NewTek's LightWave 3D 7.5b
  • Steinberg WaveLab 4.0f

It was here that we really began to see what a difference the extra memory bandwidth provided by a dual-channel memory controller would have on overall system performance.  The two Springdale-based PCs from Shuttle and Epox were producing on roughly the same level.  Their scores were nearly identical for both Winstones.  The Foxconn e-bot was consistently behind their pace, dropping just over 4% back in the Business Winstone and over 6% in Content Creation.

Real World Results with Wolfenstein and UT2004

Wolfenstein: Enemy Territory
OpenGL Quake Engine Gaming

To start our in-game testing, we ran through a batch of time demos with the OpenGL game Wolfenstein: Enemy Territory.  Wolfenstein: ET is a free, standalone multiplayer game that is based on the original Return to Castle Wolfenstein, that was released a few years back. It uses a heavily modified version of the Quake 3 engine which makes it a very easy to use benchmarking tool.  We ran the test using the "Fastest" setting at a low resolution of 640X480, using 16-bit color and textures.  Running this test with a higher-end graphics card, at these minimal settings, isolates processor and memory performance, without being limited by the graphics subsystem.

The frame rates we got with Wolfenstein - Enemy Territory clearly point out the advantages gained by the extra bandwidth afforded by the dual-channel setups.  The Foxconn e-bot put up respectable numbers, albeit at 640x460 with every setting turned down, but was still anywhere from 7 to 9 frames per second behind the other two systems.  Our performance delta is now looking at about a 7- 9 percent dropoff from the leaderboard.

Unreal Tournament 2004
DirectX Gaming Performance

Lastly, we did some benchmarking with Epic's Unreal Tournament 2004.  When we tested these systems with UT 2004, we ensured that all of them were being benchmarked with the exact same in-game settings and graphical options, and we dropped the resolution and detail levels to isolate CPU and memory performance.

More of the same occurred in Unreal Tournament 2004, although the Shuttle SB61G2 and Foxconn e-bot were much closer in framerates this time around.  The Epox ex5-320S placed well above the other two, at exactly eight frames faster than the e-bot.   This equated to the largest difference in any of our real-world tests; nearly a 10% difference from top to bottom.


Summary and Conclusion

Benchmark Summary: Due in large part to the single channel memory that the SiS661FX chipset supports, the Foxconn e-bot didn't fare as well against two other Pentium 4 driven SFF PCs such as the Shuttle SB61G2 and Epox eX5-320S.  Both of these systems are based around the i865G chipset, which utilizes dual memory channels, providing much more bandwidth for applications to use.  When testing benchmarks solely based on CPU performance, the results were usually similar, with 3DMark03 being the sole exception.  In all memory intensive testing, including gaming and business related applications, the e-bot generally fell behind the other systems anywhere from 5-10%.


Honestly, the Foxconn e-bot does so many things well, that it's a shame that its performance isn't up at the same levels as other Small Form Factor PCs that are in the same price range.  It's quite clear that Foxconn knows how to build a unique device, and there are design ideas used with the e-bot that will undoubtedly find their way into other manufacturer's plans.  Removing the power supply unit provides needed space within such small confines while also keeping heat buildup to a relative minimum.  The e-bot was also one of the first mini-pcs we've seen to come with a pre-installed slimline drive.  The drive serves as a DVD-ROM as well as CD burner, and is much thinner than the typical drive that a DIY builder might have added.  Again, space is always an issue with SFF PCs, so having a single drive that serves multiple purposes is a definite plus.

CD/MP3 playback, while not necessarily unique, is also a welcome addition although we wonder how often we would use this feature over playing music from a standard CD player.  We much preferred the way Epox handled this idea with the eX5-320S, where we could sample MP3 tracks from the hard drive rather than being limited to using only the CD/DVD ROM drive. 

The Foxconn e-bot is priced slightly higher than we might have originally expected, finding it currently selling at about $350 on most major retailers.  This price is somewhat mitigated when you consider that adding in your own DVD and/or CD burner would cost an additional fifty dollars or so.  For performance freaks, this system might not necessarily be what you're looking for.  For just about everyone else, however, the e-bot is an elegantly designed small form factor system, suitable for everyday computing needs.  We're giving the Foxconn e-bot an 8 on the Hot Hardware Heat Meter...


Get in HotHardware's PC Hardware Forum and
Chat with a ton of Techies, Just like You!


Content Property of HotHardware.com