Home Page

Today's News

Ground ZERO
Latest Rumor
Rumor Archive

 

 
 


mesg. board 
 
 
 

2015 © KickAss Gear

 

 

Athlon Upgrade Kit  
(from May 2001)

Josť Ayala 

This review was done by Josť last Spring, but for reasons beyond his control, the review was never published.  So we thought we'd post it late, rather than never.  Keep in mind that the Upgrade Kit is now made with the Abit KR7-RAID motherboard and PC2100 DDR DRAM.


I recently had the chance to convince a long-time friend of mine to send over one of his custom AMD upgrade kits. I first met John Moffett while working for the All Games Network. Dr. John, (as he is commonly known), runs a small web store that supplies custom Intel and AMD upgrade kits for avid overclockers called KickAss Gear. He was one of the first sponsors AGN ever had, and as such, we used to be in frequent contact.

Now I had been wanting to test out one of AMDís new 1Ghz systems for quite some time now, but I hadnít been able to get AMD to send over some samples. Dr. John quickly came to the rescue, offering one of his custom Thunderbird Athlon upgrade kits for a review. The Athlon upgrade kits officially consist of a 900Mhz Socket A Athlon processor based on AMDís new Thunderbird core and an ABIT KT7A-RAID jumperless motherboard. My review kit consisted of a clock-unlocked 1Ghz Thunderbird processor and an ASUS A7V133 Socket A motherboard, based on VIAís KT133 chipset.

At a base price of $329, the basic Athlon upgrade kit might seem a bit pricey, but with this price you are also guaranteed a processor that will be taken to its highest possible speed. Each kit is fully tested at the highest attained speed and is also burned in for proper and reliable operation. This ensures that you will get a high-performance system as soon as you get it installed on your computer case.

Installation Process:

By now, many of you must have heard of the installation headaches suffered by many early adopters of the Athlon\VIA architecture. Iím happy to report that these installation issues have become a thing of the past, (as has been the case for quite a long time now). The ASUS motherboard went into my Antec mid-tower case without any problems, and it happily accepted my two sticks of 128MB Corsair PC-133 SDRAM, my IBM 34GXP 30GB ATA66 hard drive, Creative Labs GeForce 2 Ultra video card, and Sound Blaster Live! Platinum 5.1 PCI sound card.

The system booted up perfectly and without any fuzz on the first try, a good indication of things to come. Like stated before, the ASUS A7V133 is based on VIAís KT133 chipset, which was released to specifically accommodate AMDís new Duron and Thunderbird processor cores. The board itself also features an integrated Promise Technologiesí Ultra ATA 100 IDE controller with RAID 0 support, support for 1Ghz+ Athlon processors, four IDE connectors, three DIMM slots, 100 and 133Mhz EV6 bus speed support, and jumperless adjustments of bus speeds and clock multipliers. Besides these features, the board is fully PC99 compliant, and also features support for wake-on-ring, wake-on-LAN, suspend to RAM, etc 

It follows a less than ideal 1/4/1 expansion slot set up, featuring 1 AGP Pro slot, 4 PCI slots, and one shared PCI/AMR slot. Personally, I prefer the six PCI/ISA configuration, but then again, the only expansion cards I run are the video card and sound card, so itís not like Iím clamoring for more PCI expansion possibilities. One nice touch included on the AGP Pro slot is that ASUS has covered the extra portion of the AGP Pro slot for users that are installing regular AGP cards. If you were going to install an AGP Pro-compliant video card, then you would have to remove the protective plastic that covers the extra real estate. This will help to avoid users inserting their cards incorrectly.

ASUS chose to use the popular Award BIOS system, which I personally prefer. I have used BIOS systems from Phoenix, and while they feature nice graphical interfaces, they donít quite offer the same amount of information you get from an Award BIOS. Being a jumper-less motherboard, (but still including jumper blocks for those people that feel inclined to use them), you can adjust every single operating parameter from within the Award BIOS shell. This means that you will be able to specify custom front side bus speeds, custom clock multipliers, (in the case of clock-unlocked processors), set the operating speed of your memory independently from the bus speed, adjust peripheral parameters, etc. The BIOS correctly detected the 1Ghz Athlon processor, and being an unlocked CPU, it harbors some nice overclocking potential

Barring the incident with the ATX power supply connector, (whoopsÖ : ), the only other installation problem I encountered was a single IRQ conflict between my Sound Blaster Live! Platinum 5.1 and the Promise ATA-100 controller. I originally had installed the SB Live! in the second PCI slot, but that slot shares an IRQ with the Promise ATA-100 controller. Thankfully, the Award BIOS detected the problem and let me know that a conflict could occur unless I moved the sound card to another slot. Sure enough, during the first WinME boot up, the system locked up. Moving the sound card to the third PCI slot solved the problem, and the BIOS warning disappeared after making the change. Itís nice to have a BIOS that detects problems and warns you of impending trouble.

ASUS A7V133 Motherboard In Detail:  

 

Even though the standard KickAss Gear Athlon upgrade kit ships with an ABIT KT7A-RAID Socket A motherboard, Dr. John sent me an ASUS A7V133 Socket A motherboard, based on VIAís KT133 chipset. As stated above, the board comes in a 1\4\1 slot configuration, featuring 1 AGP Pro slot, 4 PCI slots, and a shared PCI\AMR slot.

As is the case with all of ASUSí motherboards, the included installation and usage manual is one of the best in the industry. In some instances too informative, the userís manual should be the first place any user should look for when searching for jumper parameters, or for troubleshooting tips. In fact, ASUS went as far as building a list of the shared IRQs between the PCI slots and the different components on the motherboard. This way, you will know instantly where not to install your PCI expansion cards, saving people a lot of time in trying to diagnose IRQ conflicts.

Like the old 440BX motherboards that integrated the then new ATA-66 IDE controllers, the A7V133 ships with four IDE connectors. Two of these support the new ATA100 spec, and are also backwards compatible with the ATA66 and ATA33 standards. The other two IDE connectors are standard accept ATA66 and ATA33 devices. ASUS decided to use the Promise ATA-100 controller for ATA100 drive management. As such, you can connect up to eight different IDE devices to this motherboard, two per connector in the standard master-slave configuration. For this system, I connected an IBM 34GXP 30GB ATA66 hard drive as master on the primary IDE connector and an AOpen 1040 Pro 10X DVD-ROM drive as the master on the secondary channel. The Promise ATA-100 controller also allows a user to run a RAID 0 array with similar hard drives. Unfortunately, I did not have identical hard drives in order to test this feature.

Even though the A7V133 motherboard includes a jumperless design and a compliant BIOS, ASUS decided to keep the jumper blocks on the PCB. One of the blocks controls the CPU multiplier setting, while the other block controls the front side bus speed. Like I said, the motherboard is fully jumperless, but the blocks are there for people that prefer to set these options manually. The A7V133 also includes support for up to four USB ports, thanks to the VIA VT82C686B ďSuper I\OĒ southbridge.

The upgrade kit has already spent two weeks in the lab. Component installation and OS installation took me two days. Since that time, the system has been up without interruption for a week and a half. In that time, the system has only been reset three times. Under this heavy load, the system only crashed one time while playing Counter-Strike. At that point, the system was severely low in resources, so the crash didnít really surprise me. After that period, the system was shut down for the first time around three days ago, when I started my benchmarking runs. But this kind of reliability says a lot about ASUSí manufacturing standards. Personally, when Iím building systems for personal use, I always try to build the most stable system possible, even if I have to sacrifice a little in the performance department. The main reason for this is that I cannot afford any kind of downtime due to malfunctions or conflicts. As such, ASUS motherboards are hard to beat, as they are some of the most rock-solid products users can ever hope to buy.

About the only fault I could find with the design of the A7V133 is the fact that there is no BIOS reset jumper. Well, there is a BIOS reset jumper, but itís not the kind of jumper we have all come to know and love. Instead, ASUS decided to include only two small contact points, which can only be shorted out using a small metal part to cover both contact points. As if shorting out a jumper werenít tough enough, we know have to use a metal part to short out two contact points in the unlikely case a BIOS reset is needed. And as it turned out in my case, it was really tough to reset the CMOSÖ at one point during the overclocking tests, I accidentally set the FSB to 133Mhz while keeping the clock multiplier at 12, yielding almost 1.6Ghz. As expected, the computer would not even post, and as such, the BIOS failsafe described in the next paragraph could not engage. Shorting out those damned points took me 30 minutes, as I couldnít get solid contact between both points.

Of course, that bad point is accompanied by a host of good points, the most important of which is the fact that the BIOS will go into ďsafe modeĒ after an overclocking-related lock-up or freeze. In this mode, when the computer is restarted, the motherboard enters the BIOS automatically and sets up the processor parameters for a safe reboot, with the Front Side Bus set to 100Mhz and memory also set at 100Mhz. This measure goes a long way towards protecting your shinny new processor from damage due to failed overclocking attempts.

The ASUS A7V133 also features a clean and uncluttered design, which is always something users should look for when shopping for a new motherboard. The VIA 8363A Northbridge also includes itís own cooling fan and heatsink combo in order to keep itís temperature levels low. The only problem layout-wise lies in the fact that the Promise ATA-100 connectors are placed behind the AGP Pro slot, which could cause interference between the ribbon cables and large AGP video cards. Other than that, the A7V133 design is functional and user-friendly. 

AMD 1Ghz Thunderbird Socket A Processor:

 

            The heart of the upgrade kit is AMDís Thunderbird 1Ghz Athlon processor. One of the biggest improvements over the original Athlon core, (besides moving the processors from their original Slot-A package to a convenient and more economical Socket package), was the ability of running its Level 2 cache at full processor speed.

The original Athlon processors would operate their Level 2 cache memory at certain fractions of the processorís rated speed, due to the fact that the memory chips where not integrated into the processor die, but rather on the card that housed the Slot-A processors. With the new Thunderbird core and AMDís successful move to an .18-micron manufacturing process, they were able to integrate the Level 2 cache into the processor die, which now allows the cache to operate at full processor speed. This means that if you overclock your processor, you also gain more performance from the Level 2 cache memory. So even though the Thunderbird only includes 256KB of L2 cache, (lower than the Athlonís 512KB of L2 cache), the Thunderbird core will outperform the Athlon core because itís L2 cache is operating much faster than the original Athlonís.

Of course, the rest of the Athlon features apply to the 1Ghz Thunderbird, including the EV6 system bus, which is able to operate at 100Mhz or 133Mhz Front Side Bus. But, the EV6 bus is able to work both during the rising and falling signals, so you are basically getting 200Mhz or 266Mhz Front Side Bus speeds, depending on the setting used. Even though all current Socket A Thunderbird processors are clock-locked, (meaning you cannot specify a custom clock multiplier), Dr. John did provide us with a clock-unlocked 1Ghz Thunderbird for testing.

Testing Benchmarks And Procedures:

            Here are the components used to build the kit for testing purposes:

  • Motherboard Ė ASUS A7V133 KT133 motherboard using BIOS revision 1004
  • Processor Ė AMD 1Ghz Athlon Thunderbird Socket A, clock unlocked
  • Hard Drive Ė IBM DeskStar 34GXP 30GB 7,200 RPM ATA66 drive
  • CDROM Ė AOpen 1040 Pro 10X IDE DVD-ROM drive
  • Memory Ė 2x128MB CAS2 Corsair PC133 SDRAM modules
  • Video Card Ė Creative Labs 3D Blaster Annihilator 2 Ultra, NVIDIA drivers 6.50
  • Sound Card Ė Creative Labs SB Live! Platinum 5.1 using original CD drivers
  • Modem Ė 3Com ISDN Pro Terminal Adapter for Dual-Channel ISDN

And the following benchmarking software and games were used

  3D Mark 2001 Ė Mad Onion

  Content Creation Winstone 2001 version 1.0.1 Ė ZDBOP

  Quake 3: Arena version 1.17 Ė id Software

  Dethkarz version 1.0 Ė Melbourne House

  Unreal Tournament version 4.36 Ė Epic Games

In the case of game benchmarking, all tests were run five times, rebooting the machine each time for each benchmark run. Later, the five results were added and then averaged. In the case of standard benchmarks, the results were run once after a clean system reboot with only Explorer.exe and Systray running in the background. OS used was a clean installation of Microsoft Windows Millennium and the version 4.29a VIA 4-in-1 driver release. Even though overclocking tests were done, for the purpose of the gaming and standard benchmarks, the processor was left on its default 1Ghz speed, (10x clock multiplier and 100Mhz Front Side Bus, memory operating at 133Mhz).

Graphics Subsystem Performance: 3D Mark 2001

            Like I stated on my first impressions article, Iím really not fond of Mad Onionís 3D Mark benchmarking software. While it does an excellent job of taxing computer systems with the latest and greatest graphical effects, I donít find its performance figures useful for real-world comparisons. Sure, it is nice to compare it against a similarly configured machine, but what do I care if my next-door neighbor with the same computer gets 300 more 3D Marks than me? The individual frames per second results on the different test games though, are much more helpful. But, since you guys are hooked on 3D Mark 2001 and wanted scores from it, hereís a screenshot of my result:

 

3,274 3D Marks might seem like a really low number, especially when you consider the fact that people consistently get 7,000 to 8,000 3D Marks with 3D Mark 2000. But the main thing to remember here is that 3D Mark 2001 is heavily dependant on your video cardís support for the new Direct X8 features. So my GeForce 2 Ultra 64MB DDR SDRAM video card was simply not up to the task that the upcoming GeForce 3 is bound to fill. Still, that score is much higher than the 2,345 3D Marks I got from my Pentium III 750Mhz system. And the average frame rates I got from the different benchmark games were very respectable, hovering around 35-40 fps on the high-detail demos.

Content Creation Winstone 2001 Scores:     

CC Winstone 2k1

AMD Athlon 1Ghz System

Pentium III 750Mhz System

 

 33.2

28.4

          As was to be expected, the 1Ghz Athlon system beat out my reference Pentium III 750Mhz system by a good margin. Still, there is no question that the 750Mhz did reach a pretty respectable score. But then again, when you take into consideration the low cost of a 1Ghz AMD Thunderbird processor, that number becomes fairly attractive.

Quake 3 Arena: OpenGL Performance

            Under Quake 3 Arena, the 1Ghz Thunderbird system was tested using the two demo files id software included with the game, Demo1 and Demo2:

 

Q3A Demo1

AMD Athlon 1Ghz System

Pentium 750Mhz System

640x480x32

118.5

94.2

800x600x32

116.2

91.7

1024x768x32

103.7

82.4

Here we can see that the KickAss Gear Athlon kit takes a sizeable advantage over my reference PIII 750Mhz system. Ideally, I should have compared the Athlon system against a 1Ghz PIII system, but I was not able to secure a PIII 1Ghz system. Since the KickAss Gear unit has to go back next week, either I compared it to my PIII 750Mhz system or I only used the Athlon numbers. Of course, this article will be updated as soon as I can secure a reference PIII 1Ghz system.

Q3A Demo2

AMD Athlon 1Ghz System

Pentium III 750Mhz System

640x480x32

115.4

90.4

800x600x32

114.8

88.1

1024x768x32

105.3

82.8

These demos also serve to illustrate that the NVIDIA GeForce 2 Ultra GTS chip requires at least a 900Mhz or above processor in order to take full advantage of itís power and high-speed DDR SDRAM memory. As we can see from both demo scores, no one will be able to complain from maintaining frame rates above the 100 mark. These tests were all run in 32-bit color, but if you take the resolution to 16-bit you will be able to reach 140 frames per second on the Athlon system, while breaking the 120 frames per second mark on the PIII 750Mhz system.

Unreal Tournament 4.36: Direct3D Performance

         Even though Unreal Tournament as a game is getting old, there is no question that it still remains one of the most demanding Direct3D applications currently available. As such, it only seemed fair to include performance numbers from Reverendís Thunder demo:

UT: Thunder

AMD Athlon 1Ghz System

Pentium III 750Mhz System

640x480x32

 89.45

75.76

800x600x32

 86.32

74.16

1024x768x32

 84.79

71.92

Once again, the numbers speak for themselves. The 1Ghz KickAss Gear upgrade kit is pretty much able to reach and sustain 90 frames per second in Unreal Tournament, while my PIII 750Mhz system tops out at 75 frames per second. Theoretically, the human eye is not able to perceive any difference on anything over 60 frames per second, but you can feel the difference when controlling the on-screen action, and the higher the frame rate, the smoother the controls will be.

Overclocking Test Results:

Being an unlocked processor, I was anxious to see what was the limit for this puppy. So using the jumperless overclocking features found on the ASUS A7V133, I started out slowly to see what was the absolute limit for this particular processor.

Going with the clock multiplier route first, I increased the clock to 10.5 on a 100Mhz Front Side Bus, which yielded a 1,050Mhz speed. This isnít exactly a huge jump, so I moved up to an 11x multiplier, which yielded a better result of 1.1Ghz. After that, I moved up to 11.5, which resulted in a 1,150Mhz speed. Unfortunately, using this multiplier resulted in the computer not detecting my hard drive. Resetting the system only resulted in a hard lock-up. So going with a multiplier-only overclock, the most stable speed achieved was 1.1Ghz. Not a huge jump, but it did have a good increase in the Q3A benchmark numbers:

Q3A Demo1 OC

AMD 1.1Ghz Overclocked

640x480x32

128.3

800x600x32

126.7

1024x768x32

116.4

 

Q3A Demo2 OC

AMD 1.1Ghz Overclocked

640x480x32

124.6

800x600x32

123.9

1024x768x32

118.7

With this overclock, the CPU would max out with a 58 degree Celsius temperature under load, which is pretty close to the maximum operational temperature for the 1Ghz Thunderbird of 60 degrees Celsius. Still, Iím confident that with better cooling solution, (the processor came with a standard-issue AMD heatsink and fan combination), I would have been able to operate the processor reliably at the 1.2Ghz level.

Going with a combined multiplier and front side bus overclock, the system fared a bit better. Using a 7.5x multiplier with the KT133ís support for the 133Mhz EV6 Front Side Bus, you can reach 1Ghz and take advantage of the higher performance obtained from operating at a 266Mhz Front Side Bus and the memory at 133Mhz. Using this method, the highest front side bus achieved was 145Mhz, which yielded a speed of 1,089Mhz. I fell a bit short of replicating the 1.1Ghz results obtained by overclocking the multiplier only, as the system became unstable and crashed repeatedly when operating at a 147Mhz FSB speed.

In both cases, running 3D Mark 2001ís demo mode for five straight hours tested the maximum limits attainable. If the demo failed for any reason, the system was reset, and if it failed again, the system was reverted to the setting before the crash.

Conclusion:

Frankly, for $329, you really canít ask for anything better than what  KickAss Gear offers with their Athlon upgrade kits. Even though the official kits feature ABITís KT7A-RAID motherboard, there isnít a single user that could complain about the stability of ABITís offering. You are also assured a system that is both stable, (thanks to the burn-in process done at the store), and running at the highest possible clock speed, (which means the processors arrive clock-unlocked for maximum overclocking potential).

Offering blistering performance and good value, itís really hard not to recommend the KickAss Gear Athlon upgrade kit.

 

© Copyright, May 6th, 2001