Abit BP6
Published on 2000-02-03 13:30:25

This review will not seem complete without a little background, so...

Welcome to Celeron History 101!

Does anyone remember the days of the first Celerons?  You know, a computers lifetime ago and then some.  Has it been a whole year?  When the Celeron first hit the market it was nothing but a castrated and cacheless PII.  It barely compared with the likes of the mainstream processors like the K6-2 and the PII.  Never the less it was cheap, and where there is cheap, there is room for tinkering.  When there is room for tinkering, watch out!  The next thing we saw was people soldering and drilling Celeron PCB's, pin taping, and other such things that in general make me shudder.  Then is happened.  Tom's Hardware I believe was the first to post (If I am wrong please let me know and please forgive me) the article about a guy from Japan who made slot 1 Celerons work in SMP mode.  I will not repost the link, but trust me, it was ugly (actually quite elegant) J  Then the PPGA Celerons hit the market.  Wait, what about my slot one board?  A few companies manufactured adaptors (slockets) to enable them to fit into slot one boards.  The first slockets were crude.  They were nothing more than a PCB that adapted the socket to the slot.  No big deal.  Next thing you know MSI is building the granddaddy of all slockets with voltage tweaks and as many bus multipliers as we could handle.  Now, this may not seem all that important, but it was.  With Celerons costing only 50.00 or so each, and slockets running about 15.00, and dual boards in the 150.00 to 180.00 price range, it was enticing.

Enter Abit!  Abit built a board that added 2 socket 370's and enabled us to get rid of the slockets.  Further more they incorporated ATA-66, voltage tweaks, a multitude of FSB settings, and a jumperless design, and did it all at an affordable price.

  • Chipset: Intel 440BX

  • Bus Speeds: 66/72/75/78/80/82/83/84/85/86/87/88/89/90/91/92/93/94/95/96/97/98/ 99/100/104/106/108/110/124/133

  • Multipliers: 2.0x-8.0x (0.5 increments)

  • DIMM Slots: 3

  • PCI Slots: 5 ISA Slot: 2

  • AGP Slot:1 (x2 supported)

  • SB-Link: Yes

  • Form Factor: ATX

  • Fan Headers: 3

Test System

  • Abit BP6  

  • Intel Celeron 366a's @ 550 Mhz

  • Cooling by 2 FA6EXB?s same as Global Win CPM25603-16
  • 128 MB PC100 SDRAM Western Digital Caviar 313000 ATA-66 Hard Drive
  • Memorex CD-402  
  • 3dfx Voodoo 3 
  • Creative Labs Sound Blaster Live!
  • Netgear 10/100 Network Interface Card
  • Windows 2000 Professional RC2

Setup and Softmenu II

This has been the biggest seller for Abit.  Since the days of putting together old QDI TX based jumperless boards, I always understood the value of the jumperless design.  Abit took it a step further with Softmenu II.  If you can't set it in Softmenu II, it just does not need to be set J  The ease of setting the board up with this feature is amazing.  I only have one problem with Softmenu II.  When you are pushing a system to its limits, and you push it too far, it is a pain to go back and have to clear the CMOS to get it to boot again.  I guess that is just something I need to learn to live with.  Other than that, how can you beat it?

Monitoring

The Abit BP6 has temperature sensors provided by Winbond in the form of their W83781D/W83782D/W83783D chipset.  It is accessed through the Winbond hardware doctor in the bios but can be accessed by a number of other hardware monitoring products such as Motherboard Monitor from your desktop.  These sensors do not only provide information but also allow the system to slow itself down in times of high stress (defined by the user).  This has saved many an overclocker from an untimely hardware failure.

Overclocking

As far as overclocking is concerned, NO ONE MAKES A BETTER BOARD THAN ABIT!  You can try to refute that all you want, but we all know it is true.  As much as I am an opponent to hardware/software monopoly/alliances, ABIT has the market on the board for the serious overclocker.  No one has more bus speeds, voltage options, or multipliers, and on no other board can they be set more easily.  Sorry guys, please keep trying.

Benchmarks ($%#^@%#&@)

  

Anyone who knows me knows I am not a big fan of benchmarks.  I am especially not fond of benchmarking motherboards.  Trying to say a motherboard is responsible for performance increases in a particular subsystem requires a lot of research and comparison.  I can not say whether the high end Winstone 99 Disk Marks are better on this board than they are on an Epox KP6-BS.  There are a lot of contributing factors.  Using the exact same hardware is a practical way to get rid of the guesswork, but then I need to test every board on the market to conclude which one is best.  This is a big pain in the butt, however we are working on these tests here at 2CPU.com and plan to release them in the near future.  For now, I will give you some results of some of the tests we have run and let you make up your own mind.        

CPU

(2) Intel Celeron A @ 551 MHz

Video Board

3dfx Interactive, Inc. Voodoo3

Video Mode

1024x768 @ 24bits/pixel

RAM

128 MB

OS

Windows 2000 5.0.2128

 

Area Tested

Value

CPU Integer

3232.759 MIPS

CPU Floating Point

1333.333 MFLOPS

Video(2D)

56.79591 MPixels/s

Direct3D

29.86571 MPixels/s

OpenGL

11.26151 MPixels/s

Memory

977.035 MB/s

Cached Disk

193.5631 MB/s

Uncached Disk

1.798381 MB/s


I would love to post some favorable gaming reviews for the BP6 but currently I just do not have any.  I am really not sure what to make of the results I have got from a technical standpoint yet, but something is not right.  I am unsure if it is Q3ATest code, or the OS, but the results are not consistent with an SMP based machine.  Since the Winstone tests, and RC5-64 tests seem to be consistent with an SMP based machine, I am going to go out on a limb here and say that I think the Q3ATest code is screwed, or I am not doing something right.  The latter is likely the case, but I have not seen anyone else post big framerates from an SMP machine yet either.  I guarantee it is not for lack of people trying. 

RC5-64 decryption is impressive.  3.08 Mkeys/s consistent!  By way of comparison, our single PIII 450 @ 558 gets about 1.58 Mkeys/s.  So the lowly Celeron box almost doubles that performance.  Of course the RC5-64 decryption program is not threaded, so there could be some issues there.  At this point the client simply starts a cruncher and feeds each processor.  We do not know if a multi-threaded client would work better or not.  I did not contact Distributed.net about this yet.   

SMP is finally available for everyone.  I built my first SMP machine a couple years ago, paying 200.00 each for my P200 non-MMX chips and running them in a Tyan Tomcat III (HX chipset) with 256 MB of EDO RAM.  People were envious, but there was not much to do with it.  With Windows 2000 on the horizon, and Linux SMP support, we are well on our way to making good use out of SMP systems on the desktop.  Kudos to Abit for a job well done with this board.  

Jason Mccoy