Single page Print

A hands-on session with Woodcrest
We marched into Intel's reviewer's workshop with a shedload of workstation benchmarks, ready to test out Woodcrest as advertised, and found out we'd be testing the Bensley server platform. Bensley is very similar to the Glidewell workstation platform for Woodcrest, but it has no real capacity for PCI Express graphics—you only get one PCI Express lane with Bensley. We're not generally comfortable with testing in a remote location, outside the confines of our own labs, anyhow, but without any pre-prepared server benchmarks in hand, we were limited to a relatively small set of non-server-oriented tests. I apologize in advance for the test selection. Hopefully, the numbers you do see will offer some sense of how the Woodcrest-based Xeons will compare to the Opteron.

Our dual Woodcrest test system was running at 3GHz on a 1333MHz front-side bus, with 4GB of FB-DIMM memory installed. Here's what CPU-Z had to say about it:


These Woodcrest processors were too early to have all of the requisite P-states enabled that allow Intel's DBS clock throttling tech to do its thing. They did, however, have the basic C1E halt state that many newer Intel processors do. When the system was completely idle, the Woodcrest processors would clock themselves down to 2GHz in order to conserve power, even without DBS.

Pull off the cooling shroud over the motherboard, and this is what is exposed:


The CPU coolers and the FB-DIMMs, all of which are helped by active cooling in this server box.


Our system's FB-DIMMs packed 1GB of DDR2 533 ECC memory with 4-4-4 timings

These early Woodcrest systems were reasonably stable in our experience, although some others around us experienced BSODs when attempting to insert USB flash drives and the like.

After the workshop, we returned to our labs and attempted to configure an Opteron-based test system as closely to the Woodcrest system as possible for comparison. We were able to get pretty close, but we wound up using different hard drives and a few things like that, as you'll see on the next page. I don't believe these differences are likely to have affected the results of the few benchmarks we ran, fortunately.

The other bit of good news is that Intel shipped us the Woodcrest box after we returned from the workshop, and it's now humming away happily in Damage Labs, so we can conduct a follow-up round of tests under more controlled conditions soon. In fact, if you have requests for a specific benchmark (that doesn't require the use of a 3D graphics card), let us know, and we may give it a shot.