Dual Opteron SFF PC Tested
Steve writes "Back in May, IWILL announced the ZMAXdp (slashdot article), a dual Opteron SFF PC. Sept 23rd saw a further press release with more details of this intriguing system. At HEXUS.net, we've had the exclusive chance to get one of these systems in our test lab, obtaining pictures and specs along with our own analysys and benchmarks. The system runs suprisingly quiet and cool considering what's under the hood. This could become the system of choice for high-end workstation users who don't want a huge machine taking up their desk-space, or perhaps the toy of choice for those of us who hunger for so much power in such a small system."
Raw CPU speed helps the Opteron 250 system, some 600MHz faster than the Opteron 244s in the ZMAXdp systems, take the easy win. The ZMAXdp ends up just faster than the Xeons. The performance difference is about right for a combination of faster memory but lower CPU speed.
3DMark 2001SE Finally, 3DMark 2001SE to round things off in these basic tests. NVIDIA's GeForce 6800 GT does all the hard work, GPU wise, in each system. The Opteron systems use the AGP variant, the Nocona gets a PCI Express version.
3DMark 2001SE
Again, the ZMAXdp is some 30% slower in scanline rendering benchmark, indicating we're almost completely CPU bound with that test too. The scanline rendering process is an inherently parallel process, so executing as many render threads as possible is the basic key to good performance, so the Xeon's are comfortable winners here.
Auto Gordian Knot using XviD Natalie Portman naked and petrified. This test uses Auto Gordian Knot to create high quality XviD output of a 10 minute chop of the opening from Star Wars Episode 1 DVD. It's multi-threaded, supporting the multiple processors in each test system.
Auto Gordian Knot using XviD
Just over 32% separates the 244-equipped ZMAXdp from the Opteron 250 system, leaving it last in this particular test, just behind the 3400MHz Xeon system. Again, we're scaling nicely with CPU speed in this media encoding test. Fast FPUs are the key here, so we can see why we scale as we do.
Performance Summary While I wasn't able to run a massive array of tests, the tests I did manage to run show that performance from the ZMAXdp, using nForce3 250 core logic, is pretty much exactly where you'd expect it. The CPUs hold most of the performance cards with Opteron, since the memory controller is resident on the CPU, so the core logic doesn't have to do that much in terms of performance. Strong performance, given the CPUs, an indication that IWill are ready as far as performance goes.After we broke those spy shots back in May, a lot of people asked me whether I thought the ZMAXdp would ever get released. It's one thing to work on an engineering concept, but dump hot grits down my shorts, well-engineered retail product based on that concept. I think we can lay any fears that the ZMAXdp won't see the light of day to rest. IWill showed it off to the public in Japan and they're obviously sampling nearly-ready units to their launch partners.
Highly impressive. My only wish now is to test it with full-voltage Opteron 250s, to see if it works well there. With the right CPUs, it's an awesome product.
As a parting thought, imagine it with cool-running, dual-core Opterons next year, with a high-end graphics accelerator, plenty of memory and a pair of your favourite big hard disks. Start saving.
Thanks Armari for the unit and assistance in setting it up correctly.
So I guess we must deduce from reading the article that "SFF" means "Small Form Factor". It wasn't worth the effort it took to find out.