In an attempt to collect as much info on AMD’s Trinity results as possible, German blog site Citavia has compiled a list of results obtained from working with Berkeley Open Infrastructure for Network Computing (BOINC) software.
According to Wikipedia, BOINC is “an open source middleware system for volunteer and grid computing. It was originally developed to support the SETI@home project before it became useful as a platform for other distributed applications in areas as diverse as mathematics, medicine, molecular biology, climatology, and astrophysics.
The intent of BOINC is to make it possible for researchers to tap into the enormous processing power of personal computers around the world.”
The compilation focuses on comparing the per-GHz performance of “Bulldozer”-based CPUs to the per-GHz performance of the next generation “Piledriver”-based processors.
This is a major issue for potential AMD buyers because everybody in the market for an AMD CPU was very disappointed with the results of the "Bulldozer" launch last year.
The problem was not that AMD’s top performing CPU wasn't offering better results than Intel’s flagship, but rather that it was showing lower performance per clock than AMD’s “Stars” architecture.
What made things even worse was that, despite the increased clock speed, the new “Bulldozer”-based CPUs demonstrated lower performance than “Thuban”-based Phenom II X6 CPUs, when Floating Point performance was measured.
Therefore, everybody is hoping that “Piledriver” will have better FPU performance than “Bulldozer” and, hopefully, better than “Stars.”
The BOINC benchmarks used in the compilation were single threaded so that the single core performance of the architecture could be analyzed. They show an increase in Floating Point performance of 6% to 8% when the CPUs are clocked at 2.2 GHz but, if the frequency is increased towards 4.1 GHz , Trinity’s advantage is quickly diminishing towards zero.
What everybody should keep in mind is that a CPU's performance doesn’t scale upwards perfectly when the frequency is increased. The superior results that Trinity shows at lower clocks demonstrate that Trinity is an overall more efficient architecture with lower latencies than the ones that plagued "Bulldozer."
We’re more than one month away from the official tests with Trinity, so we can only hope that the reality will confirm the rumors circling on the Internet lately, and PC enthusiasts will have something to work with.
0 comments:
Post a Comment