Sign in with
Sign up | Sign in
HP Z1 Workstation: High-End Hardware In An All-In-One
By ,
1. The Looks Of An All-In-One...

What comes to mind when you think of an all-in-one desktop or workstation? A low-power CPU? Maybe even a mobile-oriented processor or ARM-based SoC? Integrated graphics, almost certainly. Or maybe a mobile GPU with severely limited upgrade options, at best. System memory is either soldered-down or in a SO-DIMM form factor, right? And storage is limited to 2.5" drives, starting with inexpensive notebook disks and topping out with SSDs. But as power users, regardless of all other options, we have to be almost certain that all-in-ones aren't serviceable.

In most cases, you'd be right on all accounts. But HP's Z1 AiO defies those stereotypes. Yes, you read that correctly: this is an all-in-one workstation. The Z1 is a different kind of machine, aimed at a very different market than most other all-in-ones (specifically, the ones from a certain fruit-named competitor).

Note that this is a first-generation Z1, and not the second-gen model launched in February. It's still available and still relevant though (Update: An HP representative let us know that the first-gen Z1 is no longer available; it has been replaced completely by the Z1 G2). We'll talk about what makes each version unique as we go through the speeds and feeds.

The back of the Z1 features a brushed-aluminum panel, matching the industrial motif of most other Z-series workstations. Unfortunately, the edges protrude slightly, so if you aren't careful, the aluminum will scratch other objects. Although the metal facade does make the enclosure more resilient than other all-in-ones we've handled, it also adds to the system's overall weight. Our test sample weighs in at 47 pounds as a result. Sure, that seems excessive. But it's still less than most pedestal-based workstations and separate 27-inch displays together.

The base attaches to a VESA mount, so you could conceivably also secure it to a mounting arm with enough strength to hold the chassis. To simplify your search, HP offers a list of arms able to support the Z1. There's a good chance you won't even need to mount the workstation like that, though, since the Z1's base is fairly flexible. It offers 100 mm of height adjustment, five degrees of forward tilt, and 30 degrees of tilt the other direction.

In addition, the Z1 can be reclined fully onto its base. Two latches on the underside of the machine allow the display panel to rise, held in place by a hydraulic support arm and allowing the machine's internals to be accessed. The second-gen Z1's base is slightly different; it can be pivoted into a serviceable position more easily.

In addition to an optical drive, the right side of the unit has the power button (top), media card reader, FireWire port, two USB 3.0 ports, and the microphone and headphone jacks. The second-gen Z1 lacks the FireWire port and has an option to replace the optical drive with a Thunderbolt 2 module.

Underneath the back of the machine is where most of the remaining ports hide. Four USB 2.0 ports, gigabit LAN, line in and out, optical S/PDIF, and DisplayPort 1.1 connectivity that operates bi-directionally (yes, once this product outlives its usefulness as a system, it will still make a perfectly good monitor).

While a plethora of ports is always appreciated, hope that you don't have to connect anything to them while the machine is in its normal operating position; locating and identifying them can be quite difficult.

Among the biggest features of the Z1, one that differentiates it from competing systems in the same segment and even its successor, is its display. The Z1 sports a DreamColor 2560x1440 IPS screen. It's a true 8-bit panel, and HP employs FRC (Frame Rate Control) in the same way that other manufacturers do in their 6-bit displays to get a closer approximation of 8-bit. In this case, HP uses FRC to get better color bandwidth than 8-bit. The screen is addressed by Windows as a 30-bit display, requiring non-standard drivers that allow it to toggle through various color spaces. HP's second-gen Z1 has a regular IPS display with an optional matte finish or a glossy, 10-finger, touch-enabled version. HP states that the DreamColor panel will "eventually" become available in this workstation's follow-up as well.

2. ...The Performance Of A Workstation

HP's Z1 takes standard DDR3-1600 desktop memory, and four available slots allow for up to 32 GB if you use 8 GB ECC-capable DIMMs.

To the right of the RAM, a slot-loading optical drive is positioned vertically on the right-hand side of the case.

An LGA 1155 processor interface sits under a custom shroud and heat sink. Standard CPU options include Ivy Bridge-based Core i3, i5, and Xeon 1200 v2 models. The second-gen Z1, in comparison, uses Haswell-based CPUs.

The GPU is also covered by a custom shroud and heat sink assembly, under which you'll find an MXM module. Whether the graphics processor operates at second- or third-gen transfer rates depends on the capabilities of the CPU installed.

GPU in its MXM slotGPU in its MXM slot

Aside from its orientation, the cooler is similar to what you'd find in a desktop system, employing a long radiator and heat pipes to draw heat from powerful GPUs.

Implementing such a sizable cooler on a mobile GPU means that the Z1's fans can spin more slowly, enabling quiet running.

Underside of the GPUUnderside of the GPU

The entire system is designed to move air from the bottom to the top, which conveniently keeps the LED-backlit display cool, since there's a bit of a gap between the platform and display panel.

As for storage, there is a drive cage in the middle of the machine able to hold either one 3.5-inch desktop hard drive or, in the case of our test machine, two low-profile 2.5-inch drives (like SSDs). Address them as separate repositories or set them up in RAID 0/1.

Below the drive cage is the remaining USB port, inside the machine. It houses the wireless keyboard/mouse dongle. However, the chassis can be locked shut and an application dongle can be left in its place. If you think that's unlikely, then you haven't had to go check out a Pro Tools or Avid hardware key at school in order to participate in class. Make no mistake: the education market (for both science and art) is a large segment of the Z1's target market.

3. Test System Configuration And Benchmark Settings

Test Machine

The baseline machine remains the same, with the same limitations affecting its performance.


iBuyPower P500X
HP Z1
CPU

Intel Xeon E3 1270 v2, 3.5 GHz, LGA 1155, 8 MB Shared L3, Hyper-Threading enabled, Power-savings enabled

Intel Xeon E3 1280 v2, 3.6 GHz, LGA 1155, 8 MB Shared L3, Hyper-Threading enabled, Power-savings enabled

Cooler

Asetek 550LC

HP air-cooler

Motherboard

Asus P8B WS, BIOS 2009, Intel C206 series chipset

HP 3561h, BIOS A08J52 v02.10, Intel C602 series chipset

RAM

2 x Kingston KVR1333D3E9S/4G, ECC DDR3-1333 CAS 9

4 x Micron 18JSF51272AZ-1G6M1, ECC DDR3-1600 CAS 11

Graphics

PNY Quadro 2000 1 GB, 625 MHz GPU, 1300 MHz memory, 128-bit GDDR5, 192 CUDA cores

Quadro K4000M 4 GB, 600 MHz GPU, 700 MHz memory, 256-bit GDDR5, 960 CUDA cores, MXM module recognized as PCIe 3.0 x16

RAID Controller

N/A

Intel C206 integrated SATA/RAID controller

SSD

Kingston Hyper-X SH100S3/120G, 120 GB MLC NAND SSD

2 x Intel SSD 320 SSDSA2BW300G3H, 300 GB MLC NAND SSD

Hard Drives

HGST HDS732020BLA642 2 TB, 7K3000 7200 RPM

N/A

Optical Drive

Lite-On iHAS124-04(C) 24x Dual-layer DVD±RW Writer

HP BD DRV BD-5841H5 6x BD-ROM/8x DVD±RW Writer/24x CD Writer

Sound

Echo Digital Audio AudioFire 2 (not included in price)

Echo Digital Audio AudioFire 2 (not included in price)

Networking

Integrated Intel 82574L

Integrated Intel 82579LM

FireWire

Integrated VIA 6308S

Integrated JMicron JMB 38X

Power Supply

Corsair TX650 V2, 80 PLUS Bronze, 650 W

Delta Electronics DPS-400AB-15A, 90% efficient, 400 W

Chassis

Cooler Master Silencio 550

ZP Z1 All-in-One

Monitor

LG E2250T-PN, 22”, 1920x1080 (Not included in price or testing)

HP Z1 integrated, 27”, 2560x1440

LG E2250T-PN, 22”, 1920x1080 (used as solo monitor for the 1920x1080 benchmarks, not in price)

Operating System

Windows 7 Professional x64

Windows 7 Professional x64

Graphics Driver

Quadro Driver 320.49

Quadro Driver 307.74 (latest available when testing was done)

Audio Driver

5.8

5.8

ASIO Driver

Included in audio driver

Included in audio driver

Warranty

Three-year labor; one-year parts

Limited one-, three-, four-, and five-year options. Mon-Fri 8-5 next business day, parts, labor and 24x7 phone support, terms and conditions may vary.

Price As Configured
$1999 (2012)
$6601

The SSD configuration in our sample is no longer available, so the pricing we're using corresponds to a pair of Micron C400 256 GB drives.

Test Suite

The test suite is also unrevised since last fall, with the minor addition of some tests for the Z1's display.

Compression/Archiving
7-Zip

Version 9.28: THG-Workload (1.3 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=5"

WinRAR

Version 4.2: THG-Workload (1.3 GB) to RAR, command line switches "winrar a -r -m3"

WinZip

Version 17.0 Pro: THG-Workload (1.3 GB) to ZIP, command line switches "-a -ez -p -r"

Content Creation
NewTek LightWave 3D 11.5

Custom workload: High-polygon-count Tom’s Hardware logo, Modeler test: Scripted cloning of surface details across a segment of the logo, Render test: 1920x1080 render of logo with photoreal motion blur, ray-traced shadows, global illumination, OpenGL Test: Generate OpenGL preview of animation for real-time playback on screen

Blender

Version: 2.68a, Syntax blender -b thg.blend -f 1, Resolution: 1920x1080, Anti-Aliasing: 8x, Render: THG.blend frame 1, Cycles renderer and internal tile renderer (9x9)

e-on Software VUE 11 Infinite PLE

Custom workload: Landscape (generated in Vue 8 full version and imported into PLE)

Autodesk 3ds Max 2014

Space Flyby Mentalray, Frame 248, 1440x1080, Tom’s Hardware Logo render in V-Ray 2.4.04, 1920x1080, global illumination, photorealistic motion blur, ray-traced shadows, Create Nitrous preview of logo scene, to Y: RAM drive, Autodesk chair scene in iray, 1920x1080, 250 passes, GPU (CUDA)-only rendering, Car render in V-RayRT, 1920x1080, 256 passes, CUDA-only

Autodesk Maya 2014

Tom’s Hardware Logo render in mental ray, 1920x1080, global illumination, photorealistic motion blur, ray-traced shadows, OpenGL Test: Generate Playblast (OpenGL preview) animation to Y: RAM drive

Maxon Cinebench r11.5

3D Rendering and OpenGL Benchmarks, built-in benchmarks with default settings

Adobe Premiere Pro CC

Custom Workload: Edit of 59.94 FPS 720p DVCProHD footage, with transitions and some color correction, Render To H.264 720p

Adobe Photoshop CC

Filter 15.7 MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates filters

Adobe After Effects CC

Custom Workload: SD motion graphics sequence with 3 picture-in-picture frames sourced from 720p HD Quicktimes, Same scene rendered using a frame sequence instead of from Quicktime sources, HD redo of the project using frame sequences, to 1080p

Reaper v4.402

DAWBench Universal 2012: Test number of simultaneous copies of ReaXComp that the system can effectively run, Custom Workload: Render and mix down to .wav custom score project, multiple tracks of audio, VST synthesizers and effects

Visual Studio 2010

Compile Chrome project (1/31/2012) with devenv.com /build Release

Encoding
HandBrake CLI

Version: 0.9.9
Video: Big Buck Bunny (720x480, 23.972 frames) 5 Minutes, Audio: Dolby Digital, 48 000 Hz, Six-Channel, English, to Video: AVC Audio: AC3 Audio2: AAC (High Profile)

TotalCode Studio 2.5

Version: 2.5.0.10677: MPEG-2 to H.264, MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG-2), Audio: MPEG-2 (44.1 kHz, 2 Channel, 16-Bit, 224 Kb/s), Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV

LAME MP3

Version 3.98.3: Audio CD "Terminator II SE", 53 min, convert WAV to MP3 audio format, Command: -b 160 --nores (160 Kb/s)

Synthetics
SPECviewperf 11

Default GUI options; Workloads: CATIA, EnSight, LightWave, Maya, Pro/E, SolidWorks, Teamcenter Visualization Mockup, Siemens NX

LuxMark 2.0

OpenCL-based rendering benchmark, default settings, “Ball” and “Room” tests

CASE Euler3D

CFD simulation over NACA 445.6 aeroelastic test wing at Mach 0.5

Thesycon DPCLat

DPC Latency Checker, run with default settings

SiSoftware Sandra

CPU Test=CPU Arithmetic/Multimedia, Memory Test=Bandwidth Benchmark, Cryptography

Iometer 1.1.0

Workers = 1, 16 GB repeating data, 4KB random, 128 KB sequential

AS SSD 1.7.4739

Seq, 4K random, and 4K-64 THRD tests

Display
Argyll CMS 1.6.1, DisplCalGUI 1.5.3.1

Running with a Spyder Elite 4 sensor.

4. Results: Adobe And Autodesk

Adobe After Effects CC

This is the same After Effects test you've seen us use in prior workstation reviews, slightly revised with its footage as PNG frame sequences rather than QuickTime files.

The Z1's additional memory and faster storage allow it to nearly double the speed of our baseline system in the HD version of this test. In the standard-def version, the Z1 is still 38% faster. The additional memory allows the Z1 to use all of its CPU cores for rendering, and its faster storage speed means that the system can keep the memory subsystem fed.

Adobe Premiere Pro CC

Despite the Z1's significantly faster GPU, our Premiere Pro results end up just slightly ahead of the baseline machine, since the CPU-based H.264 encode accounts for a majority of this test's overall processing time.

Adobe Photoshop CC

A slightly faster CPU, faster memory, and a faster GPU give the Z1 a slight edge over the baseline system. HP's workstation is 13 percent faster in the CPU-based test and 14 percent faster in the OpenCL-accelerated test.

Essentially, these benchmarks demonstrate that the well-integrated all-in-one is still competitive with a more conventional desktop system in Adobe's application suite.

Autodesk 3ds Max 2014

The Z1's faster CPU and greater memory bandwidth allow it to beat the baseline machine by 10 seconds.

Likewise, V-Ray opens up and makes better use of the Z1's processor cores, giving the all-in-one a 32 to 41 percent lead over the baseline workstation.

The Z1's better GPU renders V-Ray RT's Modena test almost twice as fast as our baseline.

In the iray chair scene, HP's Z1 is slightly slower. Perhaps this is the result of a slightly older driver. Right after our testing was completed, HP made newer software available for the Z1. Unfortunately, that was after the system had gone back to the company.

All of our 3ds preview tests are run at the monitor's native resolution and on a separate 1080p display.

At 1920x1080, the 3ds Max preview is 20 percent faster on HP's Z1. Stepping up to the workstation's native QHD resolution slows performance substantially.

Autodesk Maya 2014

In mental ray, at least within Maya, and given a very polygon-heavy scene, the two machines are very close, with the P500X actually finishing six seconds (1.2 percent) faster than the Z1. In reality, though, both workstations are within a margin of error when we execute the run multiple times.

Playblasting in Maya generates a preview of an animation to storage, which we render to the RAM drive, and run at 1920x1080 and 2560x1440. Unlike we saw in 3ds Max, HP's Z1 is 25 percent faster in the 1080p test. Even at 1440p, the Z1 still enjoys an 8.5 percent advantage over the baseline system at 1080p.

5. Results: 3D Modeling And Digital Audio

NewTek LightWave 3D 11.5

Even though LightWave 3D 11.6 is now available, we're running 11.5 to keep our comparison data consistent for a bit longer.

The LightWave rendering results are very close; HP's Z1 manages a lead between 0.5 and 1.5 percent. Considering that the 100 MHz speed difference between the two CPUs only accounts for about 2% of their total frequency, the outcome is in-line with what we'd expect.

Difference between both GPUs affects preview speeds in LightWave as well. The Z1's higher-clocked CPU gives it an edge in this largely single-threaded test, allowing us to clearly see the difference between GPUs without a bottleneck in the way.

Meanwhile, HP's Z1 blows the doors off of the baseline machine in LightWave's Modeler test. We've seen this pattern before. A slight increase in memory bandwidth and a better GPU make all the difference. In essence, we're illustrating the impact of memory bandwidth and graphics performance on user-interface operations within LightWave 3D's Modeler.

E-on Vue 11 PLE

As we saw in LightWave 3D, the landscape render results in Vue show the Z1 two percent faster than our baseline configuration. The workload is entirely processor-bound; the way we'd push higher performance would be through a six- or eight-core processor operating at higher clock rates.

Blender

Meanwhile, Blender indicates slightly less than a 0.5 percent difference.

Thesycon DPCLat

Thesycon's DPCLat measures total system latency. These are very good results, suggesting the Z1's potential suitability as a Digital Audio Workstation (DAW).

Tom's Score

The outcome of this test is split. At 44.1 kHz, the Z1 comes out faster, while at 192 kHz, it's slightly slower. We're reminded just how similar the CPUs in both machines really are.

DAWBench 2012 RXC

DAWBench tells a different story. The baseline machine outperforms the Z1 at all three latency settings. At higher latencies, the two systems get closer together, again reflecting the similarities between their CPUs.

The Z1's poor DAWBench results can likely be attributed to HP's use of a JMicron FireWire chipset instead of the VIA silicon inside iBuyPower's baseline machine (or other brands known for solid performance, such as Texas Instruments). This is made more ironic by HP's reliance on TI for enabling USB 3.0 support.

When we take these results and consider that an audio professional won't need the Z1 display's expanded capability, the second-gen version of this all-in-one starts looking like a better option. Plus, the newer model sports a Thunderbolt option, which can be connected to either a Thunderbolt audio interface or a Thunderbolt-to-FireWire adapter.

6. Results: Media And Encoding

LuxMark 2.0

This OpenCL-based rendering test shows the Z1 with a roughly 10 percent lead in both tests. The margin is tighter than what we saw from the V-Ray RT test, but HP still has an edge.

Cinebench

Cinebench shows the Z1 in front by 8.4 percent running on one core; it leads by two percent in the threaded version of the test. Why the difference between quad-core processors? More than likely, greater memory bandwidth helps propel performance a bit compared to the Xeon.

In the Cinebench OpenGL test, the Z1's faster GPU pulls away from the baseline machine, speeding off to a score that's 1.7 times faster.

SPECviewperf 11

The very long chart below includes all of the viewperf results. Similar to the OpenGL- and DirectX-based real-world preview tests, we ran viewperf on the Z1 at 1920x1080 and its native 2560x1440.

Even at its native resolution, the Z1 outperforms our baseline system. And as we've seen before, certain tests perform slightly better with AA on.

Euler3D

Euler3D's wing simulation, being entirely CPU-bound and with a rather small memory footprint, gives us predictable results due to the Z1's slightly faster processor and greater memory bandwidth.

TotalCode Studio

The Z1's slight improvement in our TotalCode benchmark (less than three percent) is in-line with other benchmarks where CPU speed is the deciding factor.

HandBrake

The HandBrake results are inconsistent with our other tests. Despite multiple runs to check the results, the Z1 loses to the baseline system by a hair.

LAME

The same thing happens in LAME, which is single-threaded and should favor the Z1's 100 MHz-higher peak Turbo Boost frequency. Our baseline system still ends up on top though, and the only thing we can think of is lower memory latencies that might help compensate for the clock rate difference.

Visual Studio

In our Chrome compile test, the Z1 enjoys a nice lead. Our best guess to explain the difference is the Z1's two-drive RAID configuration preventing a storage bottleneck. The HP workstation's additional memory capacity and bandwidth might also come into play.

7. Results: Synthetics, Compression, And Storage

Sandra 2013

The Z1 posts slightly better results in the Dhrystone metric (as it should, given a slightly higher CPU clock rate). However, the Whetstone numbers are exactly the same.

In Sandra's Multimedia module, the two machines are neck and neck. The behavior of their processors is so similar that calling Intel's Xeon E3-1280 v2 faster would be mis-characterizing its advantage.

The finish isn't as close in Sandra's Cryptography test. We know the AES benchmark to be memory bandwidth-limited on platforms with AES-NI support, since instructions are fed to the processor as quickly as possible from system RAM. Because the Z1 benefits from higher memory transfer rates, it serves up the best encryption/decryption performance.

Hashing, on the other hand, is CPU-bound, so both processors demonstrate identical scores.

The dedicated bandwidth test quantifies what it means to sport DDR3-1600 memory instead of DDR3-1333.

WinZip 17 Pro

Once again, the Z1's slightly faster processor, higher memory bandwidth, and faster storage combine to give it the lead over our baseline machine in the WinZip tests. The OpenCL-accelerated metric reflects a similar lead as the CPU-based test: about 10 percent.

WinRAR

WinRAR goes the other direction, and the baseline system secures a win, though we'd have a difficult time attributing this to any specification in particular.

7-Zip

In 7-Zip, the Z1 regains its small lead.

AS SSD

HP Z1HP Z1

iBuyPower P500XiBuyPower P500X

In AS SSD, the Z1's read times are mostly better than the baseline workstation's, though the write times are a mixed bag. Sequential writes are definitely better on the Z1, but the HP posts worse scores than the baseline system in the other two tests.

Iometer

Iometer shows us more of what is going on in the Z1's storage subsystem. In sequential reads, the Z1's configuration is virtually flat at around 500 MB/s. This tells us that the two-drive array is sufficient to saturate the Z1's SATA 6Gb/s interface. Meanwhile, its sequential writes flatten out at a much lower 300 MB/s. One of the main reasons for using a RAID array in a workstation is for sequential reads and writes (such as capturing and editing video).

In random performance, the RAID starts out lower, climbs slowly, and continues climbing even when the single drive of the baseline machine flattens out. The array looks like it's just about to catch the single drive in random reads at a queue depth of 32.

8. Results: Display Measurments

Since the Z1 has a built-in monitor, we want to evaluate that as well, especially since this workstation's display is one of its biggest selling points. We use a workflow similar to what you see from our mobile device testing, which employs a Spyder4 sensor. Since it uses different hardware and software, don't compare these numbers to Christian's monitor reviews.

Brightness

Factory Brightness

288.2

Minimum Brightness

75.48

Maximum Brightness

331.15

Calibrated Brightness

200.09

Out of the box, the Z1 is set too bright for our target of 200 nits. Yet, it was only halfway to the system's maximum brightness. With the level set to just less than 50 percent, we get a calibrated value of 200.09.

Black Level

0.36

50% Level

43.41

White Level

200.09

Approximate Gamma

2.21

Contrast

559:01:00

White Correllated

6388 K

With brightness set to 200 nits, the Z1 provides decent, though not outstanding, values for gamma and contrast.

Our measurement for the Z1's gamut is 116.1 percent of sRGB and 80 percent of Adobe RGB.

9. Pricey, But Sleek, Unique, And Powerful

The HP Z1 is a unique all-in-one system that, despite its otherwise-limiting form factor, is fast enough to replace a pedestal-based workstation. You could even say that the sleek-looking platform is misleading. Rather than packing mobile-oriented parts to hit a low thermal ceiling, HP engineers the hardware intelligently to enable desktop-class processors, four DIMMs, potent graphics modules, and even multiple SSDs. Performance-wise, the only way to surpass it is with an LGA 2011-based box with much more expensive hardware.

Of course, part of the allure is HP's high-quality display. The Z1 is certainly capable of yielding professional image quality and better color than many monitors, making it suitable for business-class applications. But it isn't quite as good as some of the higher-end wide-gamut screens we've seen, some of which can do 100 percent of Adobe RGB.

On the other hand, the Z1 sports a far better display than most other all-in-ones, including competitors in the same market segment. When you think about what the Z1 costs (at least the first-gen model we're reviewing), remember that you'd need to spend at least $700 to equip another workstation with a comparable 2560x1440 monitor.

There are certainly form factor-imposed concessions. You're limited to mobile graphics, for instance. And there isn't a ton of room for storage. But HP's Z1 is still a surprisingly capable system made all the more impressive because it does fit into an all-in-one. The company further complements the hardware with service and warranty options, along with certifications in professional applications that other all-in-ones lack, all while facilitating the compactness you won't find in more conventional workstation builds.