Opteron performance
This article “No More Apple Mysteries, Part Two” at Anandtech is a second focus on the PowerPC. Of greater interest to me was the numbers between the Intel and AMD processors.
At the moment I’ve been reviewing the purchase of a new 1U machine and trying to decide between a Sun Fire v20z or a Intel SR1400 Server Chasis. Basically the using an Opteron 248 2.2 GHz or a Xeon Irwindale 3.2 GHz 2 MB L2 800 MHz FSB.
One of my issues is this current colo machine is a single processor 2.8GHz IBM x335. Unfortunately as this machine is now EOL adding a second processor will cost me $4400. Almost the cost of a new machine. So I’m planning to push the old machine into fileserver duty at the backend, with the main operation of the new machine as a NX thin client desktop server. Something I’ve have been planning to do for a long while. Using Xen I’ll also be running other applications.
Considering the two CPUs and even at the lower clock-rate the Opteron is definitely the faster processor. The FPU performances almost beats the G5 PowerPC. In main tests the Opterons smashes the Xeons.
The Opteron 248 setup managed to outperform Intel’s fastest, largest cache Xeon MP by a whopping 45%. Boasting 141 ms request times, the Opteron 248 system was 12% faster than the Opteron 244 setup, indicating very good scaling with clock speed — a 50% increase in performance for every 100% increase in clock speed.
Of more interest to me is the FSB Impact on Performance:
In the case of a 4-way Xeon MP system with a 400MHz FSB, each processor can be offered a maximum of 800MB/s of bandwidth to the North Bridge. If you try running a single processor Pentium 4 3.0GHz with a 400MHz FSB you’ll note a significant performance decrease and that’s while still giving the processor a full 3.2GB/s of FSB bandwidth; now if you cut that down to 800MB/s the performance of the processor would suffer tremendously.
Each Opteron CPU features three point-to-point Hyper Transport links, delivering 3.2GB/s of bandwidth in each direction (6.4GB/s full duplex). The advantage is clear: as you scale the number of CPUs in an Opteron server there are no FSB bottlenecks to worry about. Scalability on the Opteron is king, which is the result of designing the platform first and foremost for enterprise level server applications.
Overall because of this the Opteron has lower latency, which can be see in low-level system benchmarks, like process creation, IPC and signal handling.
Other features are also compelling.
Although the IMPI is an Intel product, Sun is better documented . Of great interest is console access via SSH with SOL . Being able to ssh spipaddr -l spuser platform console will likely save much heartache and airplane time if I happen to be overseas when a kernel upgrade goes sour.
On the cost side, factoring in the self-support costs, the SR1400 works out at about 20% cheaper than the v20z. In further comparison the nearest comparable IBM xServer, a x336 3.2GHz, system is about 5% cheaper than the v20z.
Even with the large price difference, overall I think the v20z is the best machine.