Chaintech VNF4 Ultra Review - PAGE 8

HDTach IDE Testing

Not much of a difference at all in any single drive IDE implementation. Processor usage between the four boards is in the margin of error for testing. The spread for the burst rate is just merely 0.7MB/s.

HDTach SATA Testing

The MCP on the nForce 4 has a bit more juice than the VIA 8237R southbridge from yesteryear. The 8% difference in throughput is pretty large and for those that are planning on RAID configurations, I suspect the gap will grow even wider.

USB2 Throughput

USB2 throughput on the VNF4 is very good coming in at second place just behind the Soltek K8T890 board. Processor usage remained very low also.

LAN Throughput

LAN Througput on the nForce 4 series is very good. The VNF4 is roughly average in comparison to other nForce 4 board implementations in terms of throughput and it is definitely better than the general add in chip solution found on the K8T890 boards.

When comparing processor usage relative to throughput, the nForce 4 definitely has the edge again. While the Marvell solution on the Gigabyte board has very respectable throughput, the processor usage is through the roof. For testing purposes here we are still using the 6.39 driver set which we noted had a problem with ActiveArmor and connection offloading. Last week we were finally able to get a hold of a driver build that has a working implementation of ActiveArmor. Processor usage on the VNF4 dropped down into the 26-28% range while keeping up with the throughput, a very impressive feat. While this is not quite the 20% that was quoted, this is still a significant drop. Special thanks to NVIDIA for their continued patience in working with us through this one. The driver set with the fixed ActiveArmor is not publically available yet but we have been assured that this will get rolled into the next driver update.