In broad performance-per-dollar terms, we can see that, as usual, the lowest cost machine offers the highest performance per dollar both at stock clocks and when overclocked. While the sub-$2,000 system does poorly at stock clocks and offers decent value when over clocked, the sub-$4,000 system never achieves more than 50% of the performance/dollar of the sub-$1,000 system.
In the real-world however, the sub-$4,000 system makes good sense, as you might be in a position where the time you’ll save is worth convenience or money. Not to mention it can simply do some things that the cheaper sub-$1,000 system can not accomplish, such as playing a demanding game at a playable frame rate.
To illustrate this, we’ve added a new chart, High Resolution Gaming per Dollar. The idea is that if a Marathon PC cannot deliver playable performance at the 1920x1200 resolution, it is not viable at all and the performance per dollar should reflect this. To this end, we’ve produced the following chart using ONLY 1920x1200 frame rates, and if a PC cannot produce a ‘playable’ frame rate it gets zero in that particular benchmark. Since ‘playable’ is a relative term, we’ve erred on the side of generosity and for the purpose of this benchmark we’ve defined playable as 40 frames per second average in a twitch game like a first person shooter, and 30 frames per second average in a real-time strategy game. Here are the results:
As you can see, the sub-$1,000 system was hurt by some lower frame rates in Crysis and as a result the stock sub-$2,000 system is a better high-res gaming value than the stock sub-$1,000 system. When overclocked however, the sub-$1,000 system games well at high resolution and regains it’s position as the value leader.
Very surprisingly, the sub-$4,000 system did no better than it’s results in the regular performance per dollar score. We had though concentrating on high-resolution performance would give it an edge, but its price is just too high to allow it to take advantage of this favorable scenario.