The Shuttle AK35GTR Motherboard Review
Raising the Bar on Quality and Performance

By, Jeff Bouton
February 12, 2002

 

Overclocking Shuttle AK35GTR Motherboards Is it worth it??

Once the AK35GTR was up and running, we were anxious to get this board cranking and see what our Athlon 1.2GHz. Thunderbird could do.  After we got all of the BIOS settings set for optimal performance, it was time to start playing with the multiplier settings.  It was amazing to see that we were able to set the multiplier to 12X without a single complaint.  This pushed our Athlon 1200MHz. processor to 1596MHz.  This was a completely stable setting without any further adjustments.  Next we started pushing the BUS speed up a few MHz. to see if we could squeeze a tad more out the system.  With the BUS speed set at 134 and a slight increase to the CPU voltage our test rig was able to reach an astounding 1608MHz. Not only was this the highest we've ever gone with this particular processor, the board was one of the easiest to over clock.  After all was done, we were able to increase our CPU output by a jaw-dropping 34%!  If there is one thing to be said about the AK35GTR motherboard it's that this board is ready for some serious over clocking.

Next up is SiSoft Sandra to show what this translates to in performance.

HotHardware Test Systems Athlon All the Way!!

AMD Athlon @ 1200MHz.

Shuttle AK35GTR Via KT266A

256MB IBM PC2100 (CAS 2)

GeForce 3 Ti500 (23.11 Drivers)

Hercules Game Theater XP

Seagate Barracuda IV ATA100 7200RPM 20GB HD

Creative 52X CD-ROM

Standard Floppy Drive

Windows XP Professional

DirectX 8.1

Via 4-in-1s v.4.37's

 

Benchmarking with Sisoft Sandra 2002 Pro Starting with the Synthetic...

The first benchmark we'll run is SiSoft Sandra 2002 Professional.  Sandra is a synthetic benchmarking application that gives the user a visual gauge of their systems performance in comparison to a variety of popular configurations.  Below we've listed the results after we ran them at the CPU default 1200MHz. and the overclocked rate of 1608MHz.

CPU@1200MHz.

CPU@1608MHz.

In both the CPU and Multimedia tests, the system ran as expected for a 1.2GHz. Athlon processor.  The real fun was after we ran the same tests while overclocked.  Running at 1.61GHz. the test system rivaled the performance of every processor on the scale.  We were surprised to see that the Athlon Thunderbird was even beating the new Athlon XP processors but soon came to realize that this is probably not an accurate representation of the chips performance.  Clock for clock, the Athlon XP's performance is far superior to that of the Athlon Thunderbird.

Hard Drive - Single

Hard Drive - RAID

At the default BUS-Speed of 133MHz., the system memory performance scored slightly higher than the equivalent reference system.  With the system over clocked at a BUS-Speed of 134 we weren't expecting a huge gain in memory performance and we were right.  With the test system running a single hard drive performance was slightly lower that the reference point, but still quite good.  Once we changed over to a RAID-0 configuration, the scores were well below average, only slightly higher than the single drive score.  We do want to say that this is in all probability not a shortcoming of the system board but rather the combination of hard drives and RAID controller.  We've heard scattered reports of poor RAID-0 performance with Seagate Barracuda IV hard drives.  The latest line of thinking is that the drives are simply too fast for the RAID controllers buffer to handle.  Unable to keep up with the data output, the buffer fills and performance degrades.  The jury is still out on this theory, but it is something to look into.  My only question is if the drives are so fast, why is the single drive score not leaping off the page?  If any more information becomes available on this topic we'll update this and any future reviews.

 

Benchmarks and Comparisons Quake 3 and MadOnion's Video 2000...

Synthetic benchmarks are a great place to start when testing a motherboard, but with the various issues we had to contend with, we feel the most accurate way to display the AK35GTR's performance prowess is with some real world tests.

Quake 3:

The first test we ran is Quake 3 "timedemo demo001" to determine the CPU limit of the AK35GTR  motherboard.  With the test running at 640x480x16 with all textures set to their minimum setting, we are assured that the timedemo doesn't come close to taxing the capability of the video card, giving us a score that is purely based on the CPU capacity.  We ran the test at both the default CPU speed and while over clocked.

The Shuttle AK35GTR had no trouble burning through the Quake 3 test reaching the CPU limit at an excellent 203.4 FPS.  Once we over clocked the system we achieved an increase of 20.3FPS which translate to a gain of just under 10%. 

MPEG-2 Encoding Performance with Video 2000:

Next we'll run MadOnion's Video 2000 MPEG-2 Encoding Performance test to once again demonstrate the CPU performance.  The MPEG-2 test is a good gauge of the systems processing power since the MPEG encoding process is solely a CPU based routine.

Once again the AK35GTR has demonstrated that it is "the little KT266A that could."  Next we'll run some more familiar "real-world" tests that do a great job of stressing a systems potential to its limit.  Of course, we're talking about the Winstone family of benchmarks.

The Winstones