In addition to the workstation-specific benchmarks on previous pages, we also ran the memory benchmark from SiSoftware’s Sandra 2010. We’re definitely going to expand the synthetic tests to better reflect the bank of synthetic tests used on non-workstation machines.
So, the memory performance on this system is nothing special. We've certainly seen higher numbers from more aggressive-tuned subsystems on the desktop. But we can be relatively sure that this result isn't going to bottleneck the Bloomfield-class processor in HP's Z400-series workstation.
Workstations aren’t gaming PCs.
There, we said it. They aren’t. And they need to be held to a different set of standards than gaming PCs. They need to be tested differently. And to that end, HP supplied us with a moderately-priced all-around workstation.
For pure 3D work, the onboard RAID isn't necessary, especially if the user has a file server on his network for centralized storage. That being the case, video files and animation work can consume large amounts of drive space, and someone working alone may be storing files on the workstation and editing them as well. A professional working purely with video editing will likely want an even larger and faster array, which would necessitate one of the larger workstations, if only for the additional drive bays.
This system is well put-together. However, we did have repeated problems with the desktop manager for AMD's FirePro V5700, which didn't correctly restore window sizes and positions. Both HP and AMD worked with us to resolve this problem, but it was really only fixed when a much later driver revision was used under Windows 7.
Compatibility-wise, all of the hardware HP chose for the Z400 worked well together. As stated earlier, professional components inside a workstation are incredibly important for the folks who make money with their PCs. This is a bit of a hold-over from the days when non-linear editing required expensive cards (like the Velocity and Altitude from Harris/Leitch/DPS, and the Matrox DigiSuite cards) as these were known to be finicky about motherboard selection. Editing cards still exist (the Black Magic Design Decklink series and the Avid Nitris series come to mind), and the Avid DS system is based around a preconfigured HP Z800, but most non-linear editing these days is purely software-based. These are real substantial reasons why you won’t see workstations that are slightly overclocked just to give the machine a performance edge over competitors. Doing this could substantially harm the ability of the machine to perform as intended for certain market segments.
In looking at the tests, you should be able to tell that some of the benchmarks aren't well-threaded (or in some cases, threaded at all). Of course, Intel's Hyper-Threading technology only demonstrates a benefit when the application is threaded. In a piece of software that can't take advantage of multiple cores, as is the case with LightWave 9.6 Modeler, having HT enabled actually hurts performance. LightWave Layout is a different story, as the main application doesn't take advantage of threading, while the renderer does (and quite well, we can add), in addition to the physics and cloth simulation modules. Of course, if you are multitasking between various content creation programs having Hyper-Threading on will likely benefit.
With all of that said, the Z400 is a well-designed system for its market segment, and it was nice to work on a professional-class machine that didn't like it was on a runway before takeoff. The case design is reasonably attractive, and the smart layout means that swapping internal storage and expansion cards is a breeze. Aesthetically, it's nice to have a box in something other than ‘basic black,’ which has replaced ‘basic beige.’
The system worked well for developing our new workstation benchmarks, and is representative of an affordable mid-range single-socket workstation, which we'll use moving forward as a baseline to judge other workstations.
- Introduction: Evaluating Workstations
- HP's Z400 Workstation
- The Z400 And Benchmarking Notes
- Benchmark Results: 3D Animation Tests
- Benchmark Results: Maxon Cinebench 11.5
- Benchmark Results: Landscape Rendering Test
- Benchmark Results: High-Definition Editing Test
- Benchmark Results: Digital Audio Workstation
- Benchmark Results: Scientific Tests