Benchmark Tests 

Utilizing ATITool v0.26, we decided to see just how far we could take this ECS 8800GTS card. Stock clocks were a bit askew of standard reference clocks (500/800mhz) for 8800GTS cards, with the core clock @ 513mhz, and the memory clock @ 792mhz. We used the Ntune Control Panel adjacent to utilizing ATITool, though not for over clocking. We used it for overriding dynamic fan speed control, opting to lock the ECS cards fan at full speed. Thankfully though it is somewhat louder, it's not obtrusively so this way, certainly not to the extent that our old bench ATI X850xt was.

We spent the better part of an hour playing with the automated clock ramping functions for both core and memory clocks, before we settled on a combination that we found to be utterly stable, 620mhz core and 950mhz memory clocks. These are increases of 23% and 20% respectively, and thus none too shabby. Though the G80 architecture supports a separate clock for it's stream processors, in the case of the 8800GTS, this is 1200mhz, there is no way to adjust this clock rate short of editing card BIOS's, at least for now.

3DMark05 and 3DMark06 are the veritable standards for synthetic benchmarking. Though their usefulness as indicators of real world performance are a bit specious, they still serve a purpose for evaluating GPU and driver performance, and at least inferring what can be gained from an over clock.

As you can see below, 3DMark05 does return some impressive performance for this midrange card. However you will also note performance when over clocking doesn't scale linear to clocks here, with 3DMark's over clocked score only 8% faster than stock. This is likely to do there being little gain to be had when utilizing 2.0 Shaders.