AM2 Motherboards-Part 4: ATI Crossfire Xpress 3200

Overclocking

ECS KA3 MVP Extreme Overclocking Testbed
Processor:	AM2 4800+ (X2 2.4GHz 1MB L2 Cache per core)
CPU Voltage:	1.5V (default 1.4V)
Cooling:	AMD Stock Heatpipe FX62 Cooler
Power Supply:	OCZ Power Stream 520W
Memory:	Corsair Twin2x2048-PC2-8500C5 (2x1GB) (Micron Memory Chips)
Hard Drive	Hitachi 250GB 7200RPM SATA2 16MB Cache
Maximum OC: (Standard Ratio)	250x12 (5x HT, 3-3-3-13) 3000MHz (+25%)
Maximum FSB: (Lower Ratio)	263 x 9, 10, or 11 (4x HT, 3-3-3-13) (2862MHz, 2 DIMMs in DC mode) (+31.5% Bus Overclock)

Even after three BIOS revisions and a last minute fix with a beta BIOS, the ECS is still the worst overclocker among the nine AM2 motherboards we have tested - and not by a small margin. Every other AM2 board managed to overclock beyond 300, while the ECS reached 263 at best with a huge amount of effort.

The worst part of the ECS experience was not even the low overclock numbers with this board. It was the pain of getting to any overclock with the severe cold boot issues that were encountered above a 230 clock setting. Above 230, you set the OC speed in the BIOS and the board fails on boot. You then have to shut it down. Upon reboot the board will sit for about a minute or longer and appear to be hung, then suddenly it will spring to life and boot to BIOS at the set speed. You are then presented with an OC fail screen that states the board has failed the overclock. That's probably not true. If you then go into the BIOS again, leave the settings where you have it, and exit, the board will fail to boot again. Turn it off again, restart, and the KA3 normally boots correctly at the speed above 230 after this second shutdown. This is the worst cold boot issue we have seen in a very long time, and frankly it is amazing you can still get another 33 MHz on clock frequency after you encounter the cold boot wall.

In the end all this effort and pain still yields you the worst overclock among AM2 boards. Frankly, it isn't worth the effort when there are cheaper boards in this roundup which don't have these problems. The pain we experienced with the ECS KA3 MVP proves once again the value of true comparative testing. As we have said many times before, an article about a single board without comparison is an advertisement. A review compares the features and performance of two or more boards.

Memory Stress Testing
Optimum tRAS

The ECS KA3 MVP behaved like the MSI in tRAS testing. This is no real surprise since the MSI is based on the same ATI chipset. All testing used a tRAS value of 13 for best bandwidth.

Memory Stress Testing

The ECS has limited memory voltage controls compared to other AM2 boards aimed at the enthusiast. Memory voltage could only be adjusted to 2.2V which was barely adequate for the Corsair memory used in our benchmark tests.

ECS KA3 MVP Extreme DDR2-800 Timings - 2 DIMMs (2/4 slots populated - 1 Dual-Channel Bank)
Clock Speed:	200MHz
Timing Mode:	800MHz - Default
CAS Latency:	3
RAS to CAS Delay:	3
RAS Precharge:	3
RAS Cycle Time:	13
Command Rate:	2T
Voltage:	2.2V

Despite that voltage limitation, with two DIMMs installed, testing was completely stable at 3-3-3-13 2T timings at DDR2-800. However, the ECS board locked up with any attempt to set 1T Command Rate at DDR2-800. The highest speed that could run with complete stability at 1T was DDR2-533. AM2 does not officially support 1T Command Rate at DDR2-800, so it should come as no surprise that KA3 could not handle the 1T settings.

ECS KA3 MVP Extreme Epox DDR2-800 Timings - 4 DIMMs (4/4 slots populated - 2 Dual-Channel Banks)
Clock Speed:	200MHz
Timing Mode:	800MHz - Default
CAS Latency:	4
RAS to CAS Delay:	4
RAS Precharge:	3
RAS Cycle Time:	13
Command Rate:	2T
Voltage:	2.2V

Installing four DIMMs stresses the memory subsystem further, and it is also where the handicap of a maximum 2.2V DIMM voltage rears its ugly head. We had to drop the timings to a 4-4-3 at 2T to get 4 DIMMs to work at 2.2V. These are the slowest timings required for any of the eight tested AM2 boards running 4 DIMMs. Without the ability to provide a bit more voltage it is not clear whether the issue is the board design or just running out of voltage.