Athlon FX-60 Review - 7
ViewPerf 8.1 Continued
The first thing to note is that there is no performance benefit of SLI over non-SLI. The second thing is that there is a very minute increase in performance in dual core CPUs, however, monitoring the Windows task manager / performance tabs showed that only in some rare cases, the CPU usage went higher than 50% and never above 61% which was reached only in extremely short transients
The overwhelming impression is that AMD processors rule here. The FX57 is at its default settings slightly faster than the FX60, which is conceivable because of the higher clock speed, however, a small overclock suffices to correct this picture. Keep in mind that we were using the overclocking for the sole purpose of underscoring the point that our benchmarks really reflect the CPU performance. A notable exception is UGS-04, which does not make any sense since regardless of the memory configuration, CPU or GPU power, all results came out within the margins of error.
Update
It turns out that in order to enable SLI for professional OpenGL applications, a separate "multi-GPU Rendering" tab needs to be selected in the nVidia Control panel. This setting did not affect the 3dsmax, catia, light and maya results but had a dramatic impact on the ensight, proe and especially ugs results. We'll follow up shortly with a detailed story.
Final Thoughts
There isn't really much to say, the FX60 has stolen the crown in almost every single benchmark we ran, and did so at power consumption levels that were partially below that of its own, lower clocked brethren. Officially, as we mentioned earlier, there is no change in silicon between the earlier revisions and the FX60. During some of the benchmarks, we thought we had seen some more dramatic increase in performance, however, retesting and retesting over and again showed that even a different driver revision for one or the other system component can easily fool us into wishful thinking, while at the end of the day, the carefully construed card house collapses under the burden of repetitive benchmark results. Which showed us that there are still no miracles in the AMD camp.
On the other hand, the FX60 does not need any miracles, it is fast enough as it is and essentially without competition in raw performance. If a performance per Watt metric is applied, naturally, single core CPUs are somewhat better since they don't carry the overhead of a second core under reduced load. On the other hand, even in that case, The FX60 has beaten all our expectations, running cool, quiet and at low power while delivering stellar performance. What else could we want? Well, there is this little problem with the price of US$ 1031,- at launch....
Dr. Michael Schuette - Lost Circuits |
