CHIP DESIGN HOUSE AMD has released a benchmark selector tool that's designed to allow customers to sort and filter benchmark and performance data for servers using its four-core and six-core Opteron processors.
Andy Parma, an AMD product marketing manager for servers and workstations said the tool will be enhanced soon to include data for servers using the 8-core and 12-core AMD Opteron 6100 Series processors.
The tool can filter for integer throughput performance of servers using low-power six-core AMD Opteron processors, sort data for mainstream processor models, higher performance SE processor models, and energy efficient HE and EE processor models.
It can also benchmark data by workload application, including virtualisation and compute-intensive high performance computing (HPC) applications.
An updated AMD website also includes data on the energy efficiency of servers using AMD Opteron processors and feature set comparisons and performance summaries.
Parma could not resist reminding people that some compilers, which he would not name but said began with the letter I, had a nasty habit of identifying some AMD chips as faster than others when they clearly were not.
He also said there is more involved in selecting a new server than just looking at the benchmark numbers. With that in mind, he encouraged customers to also consider the information available on other parts of the company's website, including its latest virtualization webinar.
AMD's benchmarking tool can be found here. µ
About that compiler thing, most software like games and utilities are of course compiled on the computer of the developers, and when compiling they select CPU options that are present on the majority of CPU's for compatibility reasons, and in fact in the past there was some lamenting because they found out that often they were VERY conservative not even using SSE often, and I'm talking SSE 1 the original, while we are at version 4.x now.
When that discussion hit the scene things improved a bit but it's still conservative, and only a subset of companies (and many open source projects) actually use some clever detecting to detect and select optimizations on the run, but in those cases the compiling foul play doesn't matter one would think.
Also there are more compilers out there than intel's offering.
HPC Is Going to Bigger Grounds with 1 Gb/s Ethernet to isp interconnect.Fibre is Old, yet best to home., its really echo star & faster huge statelit machines of today that make HPC possible on theREADERS' Desktop.Much Much FASTER.
Once When there Where Performance Artist to Distract Viewer & Artist Has Zero Value or Less, BAD Arts therefore Gone Back To SLOP.. Now With Digital Repetitive Clowns Can Draw SAD FACE, for unneeded, HPC Demands more faster & Clowns Step up From Barberry Coast with Scorne, Only to be Lashed to MaSTHEAD..
Fst Work of Fast Machne. AMD Score 2.34 Clowns pr Second. Score is once deal, it shouldn't change within hat system, its reference number, test it & know you in normal score range. top,upper,middle or lower. don't try to make lower system into more. score shows what does in system buildt. only newer or bettter system can out do old scores. just be happy its another thng user can know.
signed:Goldeneye. Hows You Creative Score, Hummmm.NEXT.
Most of my family use PCs based on AMD processors as well as those based on Intel ones.
After the fuss about the compiler CPU dispatcher problem at Christmas 2009, I thought something might happen, and that AMD users would get a performace improvement on many games.
However, things seem to have gone very quiet, even at www.ixbt.com and http://www.agner.org/optimize/blog/read.php?i=49
All the new articles talk about the open source or AMD compilers as if the Intel one will not be changed.
Is this story still alive or has it been killed off?
@Zio
Nick must be really tired, or his sense of humor got to an entirely new level.
Considering that he wrote a completely garbled article to Fudzilla today, I think he needs a well deserved vacation.
AFAIK intel compilers crippled all non-intel cpu, making them slower rather than faster...