HP Z1 All-in-One Workstation - The Stylish & Upgradeable Workhorse

Introducing the World's First AIO Workstation

In February, HP announced the Z1, the world’s first all-in-one (AIO) workstation that features a 27-inch display and is also easily serviceable, allowing users to swap and upgrades parts.

At first glance, comparisons with the Apple iMac and HP’s Omni and TouchSmart AIOs are unavoidable. However, while the iMac, Omni and TouchSmart AIOs might have the same design philosophy of cramming all the hardware into the display, the above mentioned products are targeted at consumers. The HP Z1, on the other hand, is a proper workstation designed for prosumers and professionals engaging in graphics, 3D rendering and imaging work. As a result, the HP Z1 is specced with workstation-class components such as Xeon processors, error-correcting memory (ECC) and the latest NVIDIA Quadro graphics.

As mentioned in our Z1 Teardown article, HP’s latest AIO is based on the Intel C206 chipset which offers support for the latest Sandy Bridge Xeon processors. The Z1 can be specced with three different processor options which are as follows: 

 Additionally, the Z1 can also be equipped with one of four different NVIDIA Quadro solutions:

• NVIDIA Quadro 500M
• NVIDIA Quadro 1000M
• NVIDIA Quadro 3000M
• NVIDIA Quadro 4000M

Aesthetically, the Z1 is certainly a stylish-looking piece of kit and would definitely up the hip-quotient of any work desk. Moving on, the overall look is clean and minimalistic and with its 27-inch, 2560 x 1440 pixels resolution IPS display, it is big and imposing too. It weighs every bit as much as it looks. The display itself weighs close to 20kg, which is actually not that surprising considering it houses all the components - power supply, processor, memory, graphics, heatsinks and blowers.

The heft of the unit is supported by a cleverly engineered stand that can be tilted forwards and back by around 25 degrees. It can also be raised or lowered by 10cm. The stand is also what makes it so easy to upgrade and swap parts in the Z1 as it can be folded to hold the Z1 in a completely flat position. In this position, the innards of the Z1 can be accessed simply by reaching for two latches located at the bottom of the screen. A damper then helps lift the heavy screen up. This is as idiot-proof as it can get; if you can open a door, you can open the Z1. 

The HP Z1

Part of the reason why HP decided to go with a large 27-inch display for the Z1 is because of the real estate it offers in terms of where to place and position the components. This is crucial because in top-spec, the Z1 uses pretty high-end components that can generate quite a bit of heat.

One of the most amazing things about the Z1 is its flexible stand. The stand allows the Z1 to be adjusted for height and tilt and is impressive because the display itself weighs close to 20kg. And as we will see later, the stand is also part of what makes the Z1 so easy to upgrade.

Peeling open the unit, we can see that the components of the Z1 are separated into four main zones so that heat can be effectively managed. In accordance to the photo below, from left to right, the four zones are the 400W power supply, Quadro graphics card, memory/storage, followed by the CPU which are all hidden and shielded behind fans and panels to keep them cool.

Test Setup

Our Z1 test unit came with workstation-class specifications and it is as follows:

• Intel Xeon E3-1245 (3.3GHz)
• Intel C206 chipset
• 16GB DDR3-1600MHz ECC memory
• 1TB 7200rpm HDD
• NVIDIA Quadro 4000M

With the specifications in mind, we are going to evaluate the HP Z1 from a couple of standpoints. Firstly, we are going to see how the HP Z1 performs against against a comparable DIY system with completely standard, easy to find off-the-shelf products. Next, we are going to ramp things up a little by pitting the Z1 against that same DIY system but with an NVIDIA Quadro card instead.

However, bear in mind that this is by no means a direct comparison, because a) the Z1 comes in an entirely different and unique form factor and b) support and warranty (crucial for workstations) for the two are completely different as well. Rather, we are doing this to ascertain the Z1's performance and also to provide a bigger picture of what you can expect if you are buying a workstation right off the shelf as opposed to building one yourself.

Here’s the specifications of our DIY system.

• Intel Core i7-2600K (3.4GHz)
• Intel DZ77GA-70K (Intel Z77 chipset) motherboard
• 4 x 2GB DDR3-1600MHz
• 1TB 7200rpm HDD
• NVIDIA GeForce GTX 560 1GB GDDR5 / NVIDIA Quadro 4000

The Pricing Equation

Configured to the specifications of our Z1 AIO test unit, the Z1 will cost around US$4932. Our DIY system, including a similar 27-inch IPS display, keyboard, mice and a decent casing comes up to around US$2300, which is less than half the price of our Z1 test unit. Here's a breakdown of the cost of our DIY system.

While it is true that the DIY system is considerably more affordable, we must not forget that the Z1 comes packaged as an upgradeable AIO system and that in itself is why the Z1 commands a considerable premium. Whether or not the premium is justified, we'll find out soon in the coming pages where we report the results.

Benchmarks

The benchmarks we are running can be divided into two broad categories - CPU and GPU. The CPU benchmarks will pit the workstation-class Xeon E3-1245 processor against a suitable consumer desktop equivalent - in this case, a Core i7-2600K. Both are quad-core processors with HyperThreading and both run at nearly comparable clock speeds, so it should be interesting to see the results.

The GPU benchmarks will see the Z1’s Quadro 4000M GPU going against the closest desktop equivalent we could find - in this case, a GeForce GTX 560. The GeForce GTX 560 has the same number of CUDA processors as the Quadro 4000M but has significantly higher clock speeds. We expect the GeForce GTX 560 to get the upper hand in games, but for workstation-related tasks, the Quadro 4000M should have the advantage. To make thing more interesting, we will also swap out the GeForce GTX 560 for a desktop-class NVIDIA Quadro 4000 graphics card to see how it fares.

The list of benchmarks used are as follows:

CPU

• SPEC CPU2000 v1.3
• Black Scholes v3.5
• SunGard Adaptive Analytics v3.5
• Lightwave 3D 9.0 (64-bit version)
• Cinebench 11.5

GPU

• Cinebench 11.5
• 3DMark 11
• Cadalyst Systems Benchmark 2011

SPEC CPU2000 v1.3 Results

SPEC CPU2000 version 1.3 consists of two benchmark suites for measuring highly compute-intensive integer and floating point performance. Not surprisingly, in single-threaded performance tests, both the HP Z1 and our DIY system recorded pretty competitive scores, with our DIY system posting a moderate lead. Overall, on both floating point and integer workloads the Intel Core i7-2600K recorded scores that were about 6% greater, which is not surprisingly once you consider its higher clock speeds.

SPEC CPU's rate tests run multiple copies of the speed test to load all available processing cores. We configured the benchmark for four-user and eight-user workloads to fully evaluate the two processors. Expectedly, the Core i7-2600K processor in our DIY system was once again the better performer.

Lightwave 3D 9.0 Results

On Lightwave 3D, the Core i7-2600K in our DIY system was once again the better performer, completing the rendering tasks about 10% quicker than the HP Z1. Again, we can attribute this performance difference primarily to the Core i7-2600K processor’s higher clock speeds.

Cinebench 11.5 Results

Cinebench is a benchmark that favors multi-threaded CPUs and with both the HP Z1 and DIY system both having comparable CPUs, the result was expectedly close. However, like the preceding benchmarks, our DIY system and its Core i7-2600K CPU achieved moderately higher scores.

Black Scholes v3.5

Black Scholes is a model used for financial engineering and risk management for pricing current and future stock options. It is a multi-threaded benchmark and thus scales well with the number of processing cores available. We configured the benchmark for both four and eight threads to check out the performance scaling of both systems. It was a close fight for both systems, but our DIY system with the faster clocked Core i7-2600K once again managed to gain the upper hand. Although technically similarly, the Xeon E3-1245 processor in the Z1 simply cannot get on par with the DIY system due to its marginally lower clock speeds.

SunGard Adaptive Analytics Results

This is another multi-threaded financial analysis tool which performs risk analysis by using Monte Carlo simulation of financial futures. It was more of the same here as our DIY system was about 10% quicker overall than the HP Z1.

Cadalyst Systems Benchmarking 2011

Cadalyst Systems Benchmarking 2011 tests performance on AutoCAD 2011. We expected the two workstation-class Quadro systems to take the lead here, but that certainly wasn’t the case. Looking at the results from our two DIY systems, the results were pretty comparable despite switching from the consumer GeForce GTX 560 to the workstation Quadro 4000. Clearly, whether or not the workstation Quadro cards can perform up to its potential greatly depends on driver optimizations, and in this case, even with the latest drivers, the Quadro cards were clearly not properly optimized for AutoCAD 2011. There are a few other Quadro driver variants on NVIDIA's website such as these, and unfortunately we didn't see any gains either. Had the Quadro drivers been well optimized, we would have seen a notable gain in results that would have been equivalent to what we've tested in the past.

3DMark 11 Results

3DMark 11 is a popular synthetic benchmark from FutureMark that measures the graphics capabilities of a graphics card. In this latest iteration of FutureMark’s benchmarking utility, all of DirectX 11’s features are extensively tested. As we had expected, our DIY system which was fitted with the GeForce GTX 560 recorded the highest scores. Although the GeForce GTX 560 is technically similarly to Quadro 4000 and Quadro 4000M GPU in that it has a comparable core and shader configuration, it blitzes ahead because of its significantly higher clock speeds on the consumer graphics card.

Cinebench 11.5 (OpenGL) Results

OpenGL is a standard specification multi-platform API for writing applications and physics simulation. It is widely used in computer-aided design, virtual reality and scientific visualizations. Here, our two workstation-class Quadro GPUs took a marginal lead. However, truth be told, the difference in performance - barely a single frame - is going to be unnoticeable to the human eye. 

SPECviewperf 11.0

Those wondering why one should spend so much more for a workstation-class graphics card, SPECviewperf 11.0 reveals some insights. On most professional rendering tasks, the optimized Quadro cards managed significantly higher scores. Looking at our results here, the HP Z1 with its Quadro 4000M and our DIY system equipped with a Quadro 4000 managed scores that were many times higher than our DIY system when it was equipped with a GeForce GTX 560. This is the kind of performance difference you can expect with performance optimization done on the professional series of graphics cards.

Lastly, note that the difference in specifications between the Z1's Quadro 4000M (a mobile GPU part) and the desktop's regular Quadro 4000 GPU was the reason performance differences exist between them as captured in our testing:-

Power Consumption

Although not a clear apples to apples comparison, we can see that the HP Z1 is drawing more power than our DIY system. However, bear in mind that the Z1 is full-fledged system consisting of speakers as well as a large 27-inch IPS display. The large display is one of the main reasons why the HP Z1 is drawing so much more power. With this in mind, the power draw figures of the HP Z1 is acutally not too excessive.

Matching Style with Substance

The HP Z1 forges new ground by being the first AIO workstation in the market. Overall, we think it’s a neat concept and it carries it off with style and suitably decent performance.

As we’ve mentioned from the start, our DIY system was meant to serve as a reference point for the HP Z1 and not a direct apples to apples comparison. And if we look back at the results, the HP Z1 managed to hold its own on most accounts, despite the slightly lower-clocked processor. So on the performance front, it’s safe to say that the HP Z1 has got it mostly covered.

Bear in mind however that looking at the performance and design of the HP Z1, it's a system that is best used for light rendering tasks and graphics work. For users looking for more taxing workloads such as render farms and high performance computing should stick to traditional desktop workstations. Furthermore, these higher-end desktop workstations have support for multiple processors and have massive expansion possibilities with add-on processing cards, which the HP Z1 can't compete against.

Nevertheless, the HP Z1 crams a lot of performance into its relatively compact form factor and the ability to upgrade is a real novelty where AIOs are concerned. Granted, HP recommends that all upgrades should be performed by a qualified HP technician and some components, such as the graphics cards for instance (which uses the MXM form factor), are not readily available off the shelf, but the Z1 is still technically upgradeable. Previously with AIOs, you're pretty much stuck with whatever specifications it came with unless you were adventurous, determined and maybe even creative enough to conduct your own DIY operation.

Moving on to price, there's no denying that the HP Z1 is pretty exorbitant. If you spec a Z1 with our specifications, the cost you’re looking at will be close to US$5000. A similarly specced desktop system with mix-and-match components, a 27-inch IPS panel and an uber-expensive NVIDIA Quadro 4000 desktop graphics card will only run up to around US$2400 - that’s half the price! Even if you spec up a traditional desktop HP Z420-class workstation to comparable levels, that system will still be considerably cheaper than the Z1 - around US$3700. No matter what, be prepared to pay a premium for the Z1.

However, like we said, this is not an apples to apples comparison and although the DIY system does represent significant savings and bang-for-buck, it doesn’t come in a sleek and sexy AIO form factor that the HP Z1 does. In the same way how companies like Google and Apple take great pains and costs to beautify the workplace in a bid to boost creativity and productivity; the same could be same for the HP Z1, where an aesthetically-pleasing system could provide similar intangible benefits at the workplace. With a compact footprint and a large screen, space savings and improved productivity are some of its key selling points - especially for a workstation class product. Don't forget that the price you pay also includes warranty and support, important for a business organization, but may not be a necessity for a DIY-wiz in a SOHO environment.

HP also offers the Z1 is various other trims. For those who are in love with the Z1’s form factor but have no need for a workstation-class processor and graphics card, the the most basic Z1 can be comes with an Intel Core i3-2120 processor with integrated Intel HD Graphics 2000 for a more palatable US$1899 (but it's still expensive considering the configuration).

In closing, the HP Z1 is by no means what you’d call affordable, but you do get your penny’s worth. In some ways.

Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.