Archive

Posts Tagged ‘3dmark’
Orange Pi Development Boards

Vorke V1 Plus Celeron J3455 Mini PC Review with Windows and Ubuntu

Most Intel based mini PCs use processors classified as ‘Mobile’ as these have lower thermal design power (TDP) ratings which is the maximum amount of heat generated by the processor:

However, the new Vorke V1 Plus has incorporated a ‘Desktop’ processor namely the Intel Celeron J3455. On paper this processor looks like it should perform similar to the Intel Pentium N4200 processor but with a tradeoff between being a cheaper processor to purchase but more expensive to run due to the increased power requirements.

Geekbuying provided a Vorke V1 Plus for review so let’s start by taking a look at the physical characteristics.

Click to Enlarge

The device comes in a plain box and was supplied with the ‘right AC Adapter’ for my country.

The first observation is that it is quite a large device. At just over 6” square (153mm) and nearly 1.5” tall (38mm) it is the biggest mini PC I’ve seen with an Apollo Lake processor.

It has a large (white) power button on top which is very ‘soft touch’ making it easy to accidentally switch off the device simply by a glancing contact for example when picking up or moving the device.

There are four USB ports with the front ones being 2.0 and back ones 3.0. Design-wise mixing these to include one of each front and back might have been better as connecting a wired keyboard either means using a ‘valuable’ rear 3.0 port or having untidy cabling from the front 2.0 port.

The front also has an IR receiver and the IR Remote Control is an optional extra.

Click to Enlarge

Otherwise the specification is interesting for not having an eMMC card but a replaceable mSATA SSD of 64 GB together with the ability to add a full sized 2.5” SSD as well. The HDMI is 2.0a and so it supports [email protected]

Click to Enlarge

Booting the device and Windows asks the familiar basic set-up questions before displaying the desktop. A quick look at the hardware information shows it is aligned to the specification.

Click to Enlarge

Unfortunately the installed version of Windows is old (version 1703) and is missing the ‘Fall Creators Update’.

Click to Enlarge

Whilst it is ‘activated’ it also includes some setting changes (e.g. the computer name) and additional icons are present on the desktop.

There is also a device without a driver showing up in the ‘Device Manager’. As a result I decided to install the latest Windows ISO (version 1709) from Microsoft making sure it was fully updated with the latest patches:

Click to Enlarge

And whilst the resultant Windows was still correctly activated:

Click to Enlarge

several devices were missing drivers. Fortunately, a full set of drivers is available from the Vorke support page, and it is simply a case of downloading and unzipping the file and updating each of those devices:

Click to Enlarge

which results in one device still missing a driver similar to how to mini PC first came:

Click to Enlarge

Interestingly the missing drivers relate to the ‘Intel Dynamic Platform and Thermal Framework’ including the ‘Fan Participant’ driver and this may explain an issue with Ubuntu covered later below.

Once everything was updated a healthy amount of disk space remains available:

As usual I ran my standard set of benchmarking tools to look at performance under Windows:

which confirms the performance to be similar or better than the N4200 SoC although this in part may be attributable to the improved disk performance because of using an mSATA SSD:

Click to Enlarge

Next I shrunk the Windows partition and created new a 10 GB partition so I could install and dual boot Ubuntu. I used a standard Ubuntu desktop ISO however I needed to change the OS ‘selection’ in the BIOS:

Click to Enlarge

I first ran some basic commands to look at the hardware in more detail:

which shows the memory as dual-channel.

Running my usual suite of Phoronix tests generated mixed performance results compared with N4200 devices again likely being affected by the faster mSATA disk:

Ubuntu’s Octane result was slightly better than in Windows:

Click to Enlarge

Looking at the device’s performance against other Intel Apollo Lake devices:

Click to Enlarge

shows that overall the device performs well.

Playing videos under Windows using a browser (either Edge or Chrome) worked without issue:

 

Click to Enlarge

Click to Enlarge

I also tried playing a [email protected] video which played fine in Edge:

Click to Enlarge

but resulted in dropped frames in Chrome:

Click to Enlarge

although the number of dropped frames was lower than when the same video was played on the N4200 Intel Compute Card which has HDMI 1.4b:

Click to Enlarge

Under Ubuntu the previously seen issue of playing 4K videos in Chrome was again encountered and playing the video at 1080p resolved stuttering and frame loss:

Click to Enlarge

Click to Enlarge

And it was a similar situation with [email protected] videos in Chrome although playing at 1080p now results in dropped frames:

Click to Enlarge

Click to Enlarge

Running Kodi on Windows with a VP9 codec encoded video uses software for decoding resulting in high CPU usage and a slightly jerky playback:

Click to Enlarge

compared with a H.264 codec encoded video which uses hardware to decode and plays smoothly:

Click to Enlarge

as do videos encoded with H.265 or HEVC:

Click to Enlarge

Kodi on Ubuntu uses hardware to decode all three codecs:

 

Click to Enlarge

Click to Enlarge

with no issues with the playback of the videos. However some H.265 videos resulted in a blank (black) screen just with audio whereas others played without issue:

Click to Enlarge

Click to Enlarge

The ‘elephant in the room’ with this device is the noise from the internal fan. Maybe as a result of running a desktop processor rather than a mobile one means a larger more powerful fan is required or maybe it is just the type of fan used. However it can be loud. Under Windows the fan’s running speed (and therefore loudness) is dependent on internal temperatures i.e. workload. Under Ubuntu the fan runs continuously. The fact that Windows required specific drivers for the ‘Intel ® Dynamic Platform and Thermal Framework’ including a ‘Fan Participant’ driver might indicate a fan driver issue with Ubuntu. Even trying the latest Ubuntu by running the daily ‘Bionic Beaver’ ISO updated with the latest v4.15.1 kernel did not fix this issue.

I’ve tried to make a video to demonstrate the fan’s noise by including a battery-powered clock next to the device to act as a reference in comparing how audible the fan actually is. In the video initially the device is in the BIOS boot menu and the fan is running at low speed and is just audible. As the device boots into Ubuntu initially the fan stops and then after loading the kernel the fan comes back on at high speed and is noticeably audible in a normal operating environment:

Albeit noisy the fan was able to prevent any thermal throttling:

Click to Enlarge

and kept the external temperature below 30°C.

which is not surprising given the fan is quite a large component in the device:

Click to Enlarge

Another two typical ‘pain’ points with Ubuntu on mini PCs are the micro SD card reader and headphone audio. However with this device, both worked without issue:

Click to Enlarge

 

Just for reference the headphones work under Windows:

Network connectivity throughput was measured using ‘iperf’:

with the wifi performance being similar to comparable mini PC devices.

Power consumption was measured as:

  • Powered off – 0.4 Watts
  • Standby* – 0.9 Watts
  • Boot menu – 5.7 Watts (no fan running) 6.4 Watts (fan running quietly)
  • Idle – 4.7 Watts (Windows) and 4.9 Watts (Ubuntu)
  • CPU stressed** – 14.3 Watts (Ubuntu)
  • Video playback*** – 8.1 Watts (4K in Windows) and 9.2 Watts (HD in Ubuntu)

* Standby is after Windows has been halted.
** Initially there is a high power demand before reducing to a constant rate.
*** The power figures fluctuate so the value is the average of the median high and median low power readings.

The results show a slightly higher power consumption than comparable mini PC devices which is in line with expectations from using a ‘Desktop’ processor.

The BIOS seems to be unrestricted:

Finally I installed an SSD using the supplied mounting kit:

Click to Enlarge

The SSD SATA port is accessed by removing the single screw on the base plate underneath the device and after fixing the bracket to the SSD it is then secured in place with a screw at the top of the SSD as the base plate will also secure the SSD by using the hole on the right:

I then successfully installed and booted Intel’s Clear Linux OS by selecting the SSD from the ‘F7’ boot menu:

Click to Enlarge

Overall the device performs well with the exception of the noisy fan and for some including Ubuntu users this may not be acceptable. It is a rather large mini PC and this needs to be considered before purchasing. Performance is comparable with an Intel Pentium N4200 mini PC although it will cost more to run due to increased power consumption. If you’re interested in Vorke V1 Plus , you can purchase it on GeekBuying for $159.99 including shipping [Update: using GKBPC1 coupon should bring the price down to $149.99].

Intel Compute Cards Review – Windows 10 and Ubuntu 17.04 on CD1C64GK, CD1P64GK and CD1M3128MK

The Intel Compute Stick revolutionized the mini PC market through the introduction of x86 based processors making Windows available as an OS option. However, for Intel the biggest target market turned out to be business rather than consumer with digital signage being a key user. As a result Intel have responded with the introduction of the Intel Compute Card. So far they have released four versions of card:

Click to Enlarge

and they they differ from compute sticks by no longer being standalone mini PCs but dependent on a dock or host device.

The card itself is relatively small with a footprint slightly larger than a standard credit card:

Click to Enlarge

and is distinguished by the back being printed with details about the card including the model:

Click to Enlarge

The lack of emphasis on the consumer market is also evident in the rather unobtrusive plain packaging:

On the end that inserts into the dock or host device is a connector which is separated into two sections: a Type C-compliant portion and an extended portion. The Type C portion supports Type C-compliant connections including video with audio and USB. The extended portion supports video with audio, USB, and PCIe. Power is supplied to the card from the device the Compute Card is plugged into using the Type C portion of the connector.

The card uses bidirectional authentication to authenticate a compatible device and card. The authentication uses digital keys which are provisioned by default during manufacturing ensuring only correctly provisioned card and devices work together.

As the card can get hot during heavy workloads it totally relies on the dock or host device for cooling. It is designed so that direct conductive contact with the card surfaces provide heat dissipation. This means the card is capable of operating within all critical component temperature specifications and will produce surface skin temperatures that may violate typical safety guidelines or requirements. To stop the user being burnt when handling the card immediately after use requires the dock or host device to delay the card being ejected if additional cooling is needed to reduce the skin temperature to below 55 °C.

Although the cards now targets OEMs, manufacturers, distributors and channel partners Intel have also released an Intel Compute Card Dock allowing consumers to use a card as a mini PC.

The key specifications of the dock include:

Click to Enlarge

and it comes with a small power brick with international plug adapters together with a two meter/six foot long power cable and the dock can be used with any of the cards.

Intel kindly provided a dock and three cards (CD1C64GK, CD1P64GK and CD1M3128MK) for review.

After connecting the power cable, a monitor using the HDMI port, a wireless keyboard and mouse that connects through a USB dongle and an ethernet cable, the basic operation requires sliding the card into the dock followed by firmly pushing it in to ensure connectivity.  The card can be removed by pressing the eject button which only works while power is connected. Then depending on the BIOS setting the card will either boot immediately or after the power button is pressed.

As the cards do not come with an OS I first installed Microsoft’s Windows 10 Enterprise product evaluation ISO in order to run my standard set of benchmarking tools to look at performance under Windows:

  • CD1C64GK Compute Card
  • CD1P64GK Compute Card
  • CD1M3128MK Compute Card

The results show the improvement the newer SoCs have given the cards over the sticks:

Click to Enlarge

and shows comparable performance with devices using similar SoCs:

Click to Enlarge

The results for the Core m3 card are significantly better due to the internal storage being an NVME device rather than eMMC however the fan was noticeably audible when running some benchmarks. Interestingly the eMMC performance of the Celeron card was better than the Pentium card and this is attributed to a tolerance in manufacturing of the eMMC rather than a device characteristic and this difference is reflected in some of the benchmark scores.

Next for each device I shrunk the Windows partition and created new a 10 GB partition so I could install and dual boot Ubuntu. I used a standard Ubuntu desktop ISO however whilst the installation completed successfully the Ubuntu NVRAM entry failed to be created correctly on the Core m3 card and needed to be fixed by manually using the ‘efibootmgr’ command.

For each card I ran some basic commands to look at the hardware in more detail:

  • CD1C64GK compute card

  • CD1P64GK compute card

  • CD1M3128MK compute card

Running my usual suite of Phoronix tests shows a similar performance improvement of the cards over the sticks in Ubuntu:

Click to Enlarge

with Ubuntu’s Octane result being slightly better than in Windows.

Looking at the individual performance of the Intel Apollo Lake cards against similar devices:

shows the cards performed the best:

Click to Enlarge

Playing videos under Windows using either a browser (Edge or Chrome) or KODI worked without issue on each device:

  • CD1C64GK Compute Card
  • CD1P64GK Compute Card
  • CD1M3128MK Compute Card

Under Ubuntu the previously seen issue of playing 4K videos in Chrome was encountered even on the Core m3 card and playing the videos at 1080p resolved stuttering and frame loss:

    • CD1C64GK Compute Card
    • CD1P64GK Compute Card
  • CD1M3128MK Compute Card

And again some HECV videos played properly under Ubuntu while some videos resulted in a blank (black) screen just with audio. I also noticed for the first time that one of the working HECV video was actually very slightly jerky in parts on the Apollo Lake cards but played perfectly on the Core m3 card. The drawback however was that the fan is also noticeably audible when playing the video on the Core m3 card.

The internal temperature when playing videos using KODI on the Apollo Lake cards is very similar whereas it is much higher on the Core m3 card although the dock’s fan was able to prevent any thermal throttling:

and the external temperature did not exceed 33/35°C.

Interestingly the ‘temperature cost’ of KODI is very significant on the Core m3 and was obvious after exiting the application:

Network connectivity throughput measured using ‘iperf’ was similar across the cards:

with the wifi performance measuring much better than comparable mini PC devices.

Power consumption for the dock (DK132EPJ) alone was measured as:

  • Powered off – 0.3 Watts

Power consumption for the Celeron card (CD1C64GK) in the dock was measured as:

  • Powered off – 0.8 Watts
  • *Standby – 1.0 Watts
  • BIOS menu – 5.4 Watts
  • Boot menu – 4.8 Watts
  • Idle – 3.9 Watts (Ubuntu) and 5.2 Watts (Windows)
  • **CPU stressed – 8.3 Watts (Ubuntu)
  • ***Video – 7.4 Watts (HD in Ubuntu) and 7.7 Watts (4K in Windows)

Power consumption for the Pentium card (CD1P64GK) in the dock was measured as:

  • Powered off – 0.8 Watts
  • *Standby – 1.0 Watts
  • BIOS menu – 5.1 Watts
  • Boot menu – 4.5 Watts
  • Idle – 3.8 Watts (Ubuntu) and 5.0 Watts (Windows)
  • **CPU stressed – 8.2 Watts (Ubuntu)
  • ***Video – 7.8 Watts (HD in Ubuntu) and 7.5 Watts (4K in Windows)

Power consumption for the Core m3 card (CD1M128MK) in the dock was measured as:

  • Powered off – 0.8 Watts
  • *Standby – 1.0 Watts
  • BIOS menu – 9.7 Watts
  • Boot menu – 7.8 Watts
  • Idle – 4.8 Watts (Ubuntu) and 5.0 Watts (Windows)
  • **CPU stressed – 13.0 Watts (Ubuntu)
  • ***Video – 7.7 Watts (HD in Ubuntu) and 7.9 Watts (4K in Windows)

*Standby is after the OS has been halted and card is available for removal.

**The dock’s fan initially creates a high power demand and before reducing to a constant rate.

***The dock’s fan speed changes due to the temperature and consequently the power figures fluctuate. The value is the average of the average high and low power readings.

Finally the BIOS for each card only has a few key settings available:

One issue I encountered when removing a Sandisk Ultra Fit USB from the front port on the dock is that it is very easy to accidentally press ‘eject’ or catch the ‘power’ button resulting in the card shutting down.

The lack of a USB Type-C port on the dock is also a noticeable omission given a DisplayPort is provided. Neither is there an SD or micro SD card slot.

Overall the card and dock combination works well and the performance is as good or better than equivalent mini PCs. The design is well executed and an the card is a great innovation for computing.

The cards come with a three (3) year warranty and the dock comes with a one (1) year warranty no doubt limited because of the internal fan. The support that Intel offers is very good with regular BIOS updates and drivers available from their support website and RMA for defective devices under warranty in the country of purchase.

However for consumers who are less risk-averse they are expensive especially when compared to other mini PCs using the same Apollo Lake SOCs and when the cost of support is not factored into the purchase price.

The price also reflects the premium of the form-factor. Whilst the card and dock fulfill the functions of a mini PC the cost of ‘portability’ is hard for consumers to justify given the alternatives to the dock such as a card based laptops or a card based all-in-ones have so far failed to materialize. Equally the Core m cards and dock are competing both on price and better configurability with Intel’s own NUC range. From a consumer perspective the Intel Core m3 Compute Stick with pre-installed and fully licensed Windows 10 is actually a better option purely because it is cheaper than the overall cost of the cheapest card (Celeron), dock plus the cost of the Windows 10 software and would then offer a far superior performance than the compute card package.

With Gemini SOC mini PCs already announced it seems unlikely the card and dock will be popular with consumers unless manufacturers can offer products which use the cards at price competitive points. Which is a shame as they are very good products with very good support.

Review of GOLE 10 Mini PC with 10.1″ Touchscreen Display – Part 2: Windows 10 Pro

January 17th, 2018 6 comments

Depending on your point of view, GOLE 10 is a mini PC with a touchscreen display, or a really thick tablet with an inclined display.  I’ve already received a sample, and had a look at the hardware in the first part of the review, so in the second I tested the performance and stability, and thought about and test some use cases for this type of products.

GOLE 10 (aka F6) Setup and System Info

There are various way to use the mini PC, either as a standalone screen without any peripheral connected using the touchscreen, or as a mini PC with USB keyboard and mouse and potential other accessories, or in a dual display setup with the device connected to an HDMI TV or monitor.

I decided to connect it to my “test” TV, add a USB 3.0 drive, USB keyboard and mouse, Ethernet cable, and of course the power supply. Note that contrary to other similar model, GOLE 10 does not come with a battery. I pressed the power button on the right setup, and the first time it started in portrait mode, but after moving the PC (to take the photo below), it automatically switched to landscape mode.

Click to Enlarge

This made me wonder if you was practical / feasible to use it in portrait mode, and while it’s not really designed for this, the left side does not come with any connector, so you could rotate it 90-degres anti-clockwise to switch to portrait mode.

As you can see from the first picture, the system is configured to use Clone displays by default when HDMI is connected, but you can obviously switch to Extended Desktop mode. You may want to see the HDMI TV/monitor as the primary display if you plan to use the touchscreen display as a control console, as some apps like Kodi will apparently start on the primary display by default.


I could not find a way to force the display orientation, and a few times GOLE 10 started in portrait. You can either move it around, or go to the settings and set orientation to 270 degrees.

We can find some basic info about the computer and operating system in Control Panel -> System and Security -> System. The hardware info is as expected with an Intel Atom x5-Z8350 processor and 2.00 GB RAM (GOLE 10 is also sold with 4GB as option), but the operating systems is sort of a disappointment as we have an unactivated Windows 10 Pro 32-bit OS installed.

I tried to see if I could activate it by click on Activate Windows, but unsurprisingly it did not work. So you’d have to get your own key. [Update: If you click on Troubleshoot in the screenshot above, it will activate Windows successfully. See comments.]

Windows 10 recognized the NTFS and EXFAT partitions in my USB 3.0 expansion drive, and we got 15.3 GB free of 28.4 GB in the eMMC flash partition (C:).

At this point, I got a pop-up window saying a Windows update was available, so I went ahead thinking it would not take that long, but it ended being a rather large Windows 10 update that took close to 4 hours to complete…

Click to Enlarge

I wanted to check the installed update to check for Meltdown / Spectre, but none would show, so I clicked on Uninstall an update to get a list. KB4056892 does not show up, and all updates are older, and somehow Windows 10 did not ask me to update any further, maybe because it’s not activated? I don’t know.

Click to Enlarge

Device Manager show further information about the hardware, and surprisingly a TPM 2.0 module appears to be installed. HWiNFO32 shows about the same information as other Atom x5-Z8350 mini PCs.

Click to Enlarge

I’ve been using a mouse and keyboard so far, but I wanted to see how it would feel to remove them, and instead use the touchscreen and software keyboard instead. It did not started so well. as the software keyboard would not show up automatically by default.

Click to Enlarge

So I had to go to Ease of Use-> Use the computer without a mouse or keyboard, and tick “Use On-screen keyboard“. The keyboard works but the keys are really tiny on a 10.1” 1920×1200 display, so you’d have to resize it to make it suitable, or use a capacitive touch stylus. I could not make the keyboard automatically when I select a text input, e.g. a search box in a web browser, so I pinned the on-screen keyboard app to the taskbar to call it whenever I needed it, and minimize it when not. This is not ideal, but that’s the only way I found.

GOLE 10 Benchmarks

Let’s go through some benchmarks – without HDMI display just in case – to check the system runs normally, starting with PCMARK 10, and oops “To run this test you need to run PCMark 10 in 64-bit mode on a 64-bit operating system”, so it looks like PCMARK 10 does not work on 32-bit systems. Good to know.

Click to Enlarge

I’ll try PCMARK 8 later, but in the meantime, I went with Passmark PerformanceTest 8.0, with a score of 601.0 points compared to 698.8 points for MINIX NEO Z83-4 Pro, but keeping in mind that the 3D graphics test completely failed.

Click to Enlarge

So a direct comparison of the total score is not possible, but if we go to the detailed score, and compare it to the results for MINIX NEO Z83-4 Pro, we can see something is really wrong.

PassMark 8.0 – GOLE 10 vs MINIX NEO Z83-4 – Click to Enlarge

Every single score is significantly lower on GOLE 10, except Disk Mark. I include HWiNFO32 sensor data in the Passmark scrrenshot, and while throttling is not officially detected we can see temperatures went really high, and the system most likely throttled, something I’ve not seen in a long time by just running this benchmark.

Then I tried 3DMark, and somehow I had no benchmarks installed so I tried to click on Install Fire Strike, but it said “Your license does not allow installing this update”.

Click to Enlarge

Then I tried to uninstall it, and decompress the downloaded file again, and this time I noticed the USB 3.0 hard drive – where I had downloaded the benchmarks – would from time to time disconnect during decompression, and then I remembered that during download I had some beep, following by a message asking me to resume download. I realized that connection my USB 3.0 drive to the USB 3.0 port ofthe mini PC was not such a good idea after all, and the system’s power supply or circuitry could not handle it. So instead I connected it to the adjacent  USB 2.0 port, and I could extract, install, and run 3D benchmarks.

Click to Enlarge

I only ran two:

Again this does not look good, as MINIX NEO Z83-4 Pro with the exact same processor, but proper cooling and a Window 10 Pro 64-bit OS, achieved respectively 233 and 20,284 points in those same benchmarks

 

Click to Enlarge

I went back to PCMARK 8, and the system got 1,006 points, against 1,445 points, so poor performance is confirmed. I should probably also point out that it is winter here, so room temperature is now 20 to 22°C instead of the usual 28 to 30°C.

The eMMC flash is however faster than in the one in MINIX NEO Z83-4 Pro, as shown in CrystalDiskMark 6.0.0 above, as well as Passmark above.

Stress Testing / Power Consumption

Based on the results above, I was not expecting too much from AID64 Extreme system stability test, and after 12 minutes I decided to stop since I had enough data.

Click to Enlarge

The good news is that neither AID64 nor HWiNFO32 detect thermal throttling nor power limit exceeded, but the bad news is that this info is obviously bogus as temperatures run really high, and we can see frequency drops to as low as around 480 MHz during the test, averaging around 1,170 MHz for a CPU with a 1,440 MHz base frequency.For reference, MINIX NEO Z83-4 Pro averages 1680 MHz for the same test for a 2-hour period.

Since there’s no battery, I could take power consumption numbers (without HDD connected unless otherwise stated) from a kill-a-watt clone connected to the mini PC only:

  • Power off – 0.0 Watt
  • Sleep – 1.3 Watts
  • Idle (100% brightness) – 9.1 to 10.1 Watts
  • Idle (50% brightness) – 6.4 to 7.2 Watts
  • Idle (0% brightness) – 5.0 to 5.3 Watts
  • AIDA64 Stability Test (100% brightness) – 13.2 to 14.2 Watts
  • Kodi 1080p H.264 video playback to HDMI TV from USB 3.0 HDD (100% brightness) – 16.0 to 16.4 Watts

Potential Use Cases for GOLE 10 mini PC

I’ve been thinking about potential application for GOLE 10, and similar type of mini PCs with inclined display, and based on the results above it’s clear that anything involving multitasking or heavy loads is clearly not well-suited to the GOLE 10 due to the poor thermal design. Here’s a non-exhaustive list:

  • Single Display Setup
    • Home automation control panel
    • 3D printer control panel
    • USB microscope display
    • WiFi or USB oscilloscope or logic analyzer display
  • Dual Display Setup
    • Point-of-sales with touchscreen for cashier, and secondary screen for customer, with barcode reader connected to DB9 serial port
    • VLC for music or video with media shown on HDMI TV, and playlist on touchscreen display.
    • Kodi for music, photos or vidoes with media shown on HDMI TV, and web based interface on touchscreen display

I’ve tested four of the use cases above using my USB microscope to check out a Raspberry Pi Zero board, IkaScope WS200 wireless oscilloscope, as well as VLC and Kodi is dual display configuration, all of which can be seen in the video below.

As a side note, at first I tried to find Windows 10 Kodi remote control in the Windows Store, and I discovered Windows Store had a limit of 10 devices, so I was taken to the page below to remove some of my devices.

Click to Enlarge

After I could find some remote for Kodi, and tried mrRemote for Kodi, and XBMCee, but was unable to connect to Kodi, while it worked just fine from Firefox as shown in the video.

Conclusion

I find GOLE 10 to be much more useful than GOLE1 mini PC with a 5″ display I reviewed in 2016, as the display is now much bigger and usable in Windows 10. You’ll have to do without battery however, and the system has some serious limitations as thermal design is poor so it will almost immediately throttle, and the 5V/3A power supply provided with the device is underpowered if you plan to use a USB 3.0 hard drive connected to the USB 3.0 port. I managed to work around the latter by connecting the drive to the USB 2.0 port, or you could lower brightness which also greatly impacts the power draw. Using a better power supply might help too, provided the board does not have other limitation with regards to power draw.

The device can still be used for applications that are not too demanding, and the best fit would probably be point-of-sales, followed by control panel for automation / 3D printer, and display for USB or wireless tools such as microscopes or oscilloscopes. VLC and Kodi works, but I’m not sure it’s very useful, as surely controlling the players from a smartphone should be more convenient.

If you are interested, HIGOLE GOLE 10 can be purchased for $168.99 on GearBest (2GB RAM).

MINIX NEO N42C-4 Mini PC Review – Part 2: Windows 10 Pro

January 7th, 2018 9 comments

MINIX NEO N42C-4 is the first Apollo Lake mini PC from the company, which also happens to be their first one with a fan, using internal antennas for WiFi and Bluetooth, and offering user-upgradeable storage and memory thanks to M.2 and SO-DIMM slots. The device also features three video output via HDMI 2.0, mini DiplayPort, and USB Type C  ports supporting up to three independent display.

I’ve received a sample and already checked the hardware, and showed how to install an M.2 SSD and SO-DIMM RAM to the device in the first part of the review entitled  MINIX NEO N42C-4 Triple Display Capable Mini PC Review – Part 1: Unboxing and Teardown, so I’ll report my experience with Windows 10 Pro in the second part of the review, and there should also be a third part specifically dealing with Linux support.

MINIX NEO N42C-4 Setup, System Info, BIOS

The device is basically an update to MINIX NEO Z83-4 Pro mini PC, also running Windows 10 Pro but on a Cherry Trail processor instead, and maybe of the part will be similar so I’ll refer to that review from time to time.

I first connected the mini PC with the usual peripherals and cables including USB keyboard & mouse, USB 3.0 hard drive, HDMI cable to my 4K TV, Ethernet cable, and since the computer also comes with a USB type C port supporting DisplayPort Alt mode, I also connected Dodocool DC30S USB-C hub in order to get a second HDMI display.

Click to Enlarge

Time to connect the power supply, press the power button for two or three seconds to get started, with the blue LED on the front panel turning on, and shortly after getting to the Windows 10 Pro Welcome screen, where you could use Cortana voice assistant (or not) in order to go through the setup wizard to select the country and keyboard, accept the EULA, sign-in or create a user, set privacy settings and so on.

I won’t into details since the procedure is exactly the same as their previous moduel, and you can check the Windows 10 Pro setup wizard section of NEO Z83-4 Pro review to get more photos about the initial setup. What was different this time is that a large Windows update (2 to 3GB) was available, and I waited for it to complete before accessing Windows desktop, but as you can see from the photo below there’s also an option to “go to my desktop while my PC updates”.

Now we can get to the desktop, and check info about the Windows 10 license, and basic hardware info in Control Panel-> System and Security -> System. The mini PC runs an activated version of Windows 10 Pro 64-bit, and is equipped with an Intel Pentium N4200 CPU, 4GB RAM as advertised.

Click to Enlarge

I took a screenshot of the “Devices and drives” section in “This PC” right after initial setup, and the 32GB eMMC flash has a 28.1 GB Windows partition (C: drive) with only 7.72GB free, but later on a pop-up will show up asking whether we want to delete the old Window 10 update files, and free space will increase a lot.

The D: drive is the 240GB M.2 SSD I installed myself, but since I partitioned it for another review with EXT-4 and NTFS, only the 59.6GB NTFS shows up. E: and F: drives are the NTFS and EXFAT partition on the USB 3.0 drives, so all my storage devices and (Windows compatible) partitions have been detected and mounted properly.

Click to Enlarge

I tool a Device Manager screenshot for people wanting more technical details, and since we can see Trusted Platform Module 2.0 shown in security devices, I also launched tpm.msc “Trusted Platform Module Management” program to confirm the TPM was indeed ready for use.

Click to Enlarge

HWiNFO64 shows the same information as for other Intel Pentium N4200 systems, except for the CPU microcode (μCU) which has been updated to version 24, and hardware specific items like the motherboard name, and BIOS date and version.

Click to Enlarge

Since I don’t own any DisplayPort capable display, so I could not test triple display support, but I could still work with a dual setup display using the HDMI 2.0 port and USB type C port via my USB-C hub as shown in the photo below. You may want to read the video output ports limitation listed in the first part of the review to make sure the system meets your requirements if you plan to use three displays.

Click to Enlarge

Pressing “Esc” at boot time will allow you to access Aptio Setup Utility, often referred to as “BIOS”.

Click to Enlarge

In previous models, MINIX had added several extra features in Advanced->MINIX Feature Configuration, but when I went there I could only find EarPhone Standard selection, no more restore AC power loss, wake-on-lan, etc…But then I found the other extra MINIX options had moved to Advanced->Power Management Configuration, and we still have WoL, resotre AC power loss, RTC wakeup, etc.. functions. So all is well…

MINIX NEO N42C-4 Benchmarks

The performance of Intel Apollo Lake processor is now well know, but let’s still go through the usual benchmark to make sure everything is working as expected.

Click to Enlarge

A PCMARK 10 score of 1,568 points is actually quite better than the score I obtained with MeLE PCG03 Apo (Celeron N3450 – 1,334 points),  and MeLE PCG35 Apo (Pentium J3455 – 1,391 points), both quad core fanless Apollo Lake mini PCs, so it looks like the fan may be helping, as well as the faster storage as we’ll see below. For reference, NEO Z83-4 Pro’s PCMARK 10 score was 896 points, so there’s a clear performance benefit here.

Click to Enlarge

NEO N42C-4 passmark 9.0 score: 768.3 points. In this case, the mini PC is slower than MeLE PCG35 Apo with 790.7, which should be expected since Pentium J3455 (1.5/2.3 GHz, 10W TDP) is supposed to be a bit faster than Pentium N4200 (1.1/2.5 GHz, 6W TDP). However, if we compare to Voyo (V1) VMac Mini‘s score (1087.0 points) also based on Pentium N4200 processor, then it’s disappointing. But there’s an explanation, as PassMark attributes a significant share of the score to storage performance, and Windows 10 is install in the faster SSD in the Voyo mini PC, breaching Microsoft’s low cost license agreement in the process… However, there’s also another element of the score that is weak in N42C-4: 3D graphics mark (132.2 vs 325.8), and both systems were configured to use 1080p60.

I’ve run the 3G graphics mark manually again to make sure the issue was reproducible (it is), and get some data to compare to similar system with better score in the future.

Click to Enlarge

However, switching to 3DMark’s 3D graphics benchmarks, MINIX NEO N42C-4 performs better than Voyo V1 with respectively 366, 1,567, and 2,658 points for respectively Fire Strike 1.1, Sky diver 1.0, and Cloud gate 1.1, against 267, 1,384, and 2,347 points for the Voyo mini PC. Ice Storm benchmark failed to complete on NEO N42C-4 after three tries, even after a reboot, so there may be a problem with the 3D graphics drivers.

Links to results:

MINIX used a pretty good 32GB eMMC flash with sequential read up to 307.5 MB/s and writes around 81 MB/s, nearly twice as fast as their MINIX NEO Z83-3 Pro for reads, but random I/Os are roughly the same.

I also benchmarked KingDian N480 M.2 SSD, and results were even better than in MeLE PCG03 Apo mini PC with significantly better sequential and random speeds in most tests.

USB 3.0 NTFS write speed was rather poor (35 to 45MB/s) in MINIX NEO Z83-4 Pro mini PC, but NEO N42C-4 has no such problem getting over 100 MB/s for both read and write.

Full duplex Gigabit Ethernet performance is excellent:

802.11ac WiFi performance is also very good, and much better than MINIX NEO Z83-4 Pro:

  • Download

  • Upload

So in my case, having switched to internal antennas do not negatively affect performance at all.

I’ve compared MINI NEO N42C-4 mini PC to other Apollo Lake mini PCs, as well as Cherry Trail based MINIX NEO Z83-4, and an Intel Core M3-6Y30 Compute Stick, whenever scores are available. First, there’s a clear advantage of upgrading from the Cherry Trail model to the Apollo Lake one, N42C-4 has the best eMMC storage performance (although systems run Windows 10 on an SSD will be faster), and usually performs better than other Apollo Lake mini PCs, except for Passmark 9.0 due to poor 3D graphics issues in that benchmarks.

Click to Enlarge

Note that the values above have been adjusted with different multipliers for each benchmarks (e.g. 3DMark Fire Strike multiplied by 5) in order to display all benchmarks in a single chart.

MINIX NEO Z83-4 Stress Testing, Power Consumption, and Fan Noise

 

I also use the device as a desktop computer, doing my usual tests such as multi-tasking with Thunderbird, LibreOffice, Firefox, etc, as well as multitab web browser, YouTube, playing Aaphast 8: Airborne game, etc… It works well with a user experience similar to most Apollo Lake mini PC, and the usual caveat like YouTube 4K working better in Microsoft Edge, but usual in Chrome/Firefox as long as you disable VP9. Whether Kodi 17.6 works suitably well with depending on your requirements. Automatic frame rate switching, HDMI audio pass-through for Dolby Digital 5.1, and 4K H.265 / H.264 are usually all working, but VP9 is using software decode and is quite slow, pass-through for TrueHD and DTS HD is not working, and from time to time some H.265 videos just show a black screen.

I stress-tested the mini PC using Aida64’s stability test for two hours, and CPU temperature never exceeded 60°C, so no thermal throttling problem at all, and it should make a good mini PC in relatively hot environments. CPU frequency averaged 1.8 GHz.

 

Click to Enlarge

While there was not thermal throttling, the power limit was exceeded during spikes to burst frequencies, but I’d assume this may be normal behavior.

Click to Enlarge

While the mini PC comes with a fan it is incredibly quiet, and if my main computer – which I admit is rather noisy – completely overwhelm whatever noise comes from NEO N42C-4. When I turn off all other equipment, I cannot hear anything while idle, unless I place my hear close to the device, in which case I hear some low level noise, either the fan turning slowing, or another source of noise. Under load, it’s possible to hear the fan, but again noise is very low.

I used GM1352 sound level meter, placing the device about 2cm above the enclosure (since I don’t happen to own an anechoic chamber), and as you can see from the table below measured sound levels are really low compared to a device like Voyo VMac Mini.

Noise Level (dBA)
Ambient voise (aka Silence) 38.5 to 38.9
MINIX NEO N42C-4 Idle 39.1 to 39.5
MINIX NEO N42C-4 Stress test 39.7 to 40.4
Voyo Vmac Mini – Idle 52.3
Voyo Vmac Mini – Stress test 52.5 to 57.5

Finally some power consumption numbers without USB-C hub, nor USB 3.0 expansion drive unless otherwise noted:

  • Power off –1.1 to 1.2 Watts
  • Sleep – 1.2 Watts
  • Idle – 6.4 Watts
  • Aida64 stress test – 13.4 Watts
  • Kodi 4K H.264 from HDD – 14.3 to 16 Watts
  • Kodi 4K H.265 from HDD – 15 to 17.1 Watts

Conclusion

If you’ve been using MINIX NEO Z83-4 Pro mini PCs, MINIX NEO N42C-4 will offer a nice upgrade with significantly better performance, and all some problems I found in the Cherry Trail  device are gone: USB 3.0 NTFS write speed is normal (100 MB/s), and 802.11ac WiFI performance is excellent, the best I’ve tested so far (with iperf). Compared to other Apollo Lake mini PCs, the performance is also a bit higher, running temperature is very low (< 60 °C) thanks to the quiet fan, and you’ve got a TPM 2.0 chip, VESA mount, support for triple display setup, an activated Windows 10 Pro OS, all features normally not found in other cheaper models. The low running temperature should make it ideal in hot climates where room temperature may be 35 to 40ºC.

The mini PC has some of the same limitations as other Apollo Lake mini PCs, with Kodi 17.6 handling VP9 codec with software decode, and no TrueHD, nor DTS-HD pass-through), and watching online videos for example with YouTube works better in Microsoft Edge. The only small issues I found are low 3D graphics performance in Passmark 9.0 – but no such performance issues in other benchmarks – and 3DMark Ice Storm benchmark would not complete successfully.

MINIX NEO N42C-4 Pro mini PC sells for $299.90 and up on various sites including Amazon US, Amazon UKGearBest, GeekBuying, etc…

Intel Apollo Lake Windows 10 Benchmarks Before and After Meltdown & Spectre Security Update

January 6th, 2018 36 comments

So this week, there’s been a fair amount of news about Meltdown & Spectre exploits, which affects all major processor vendors one way or another, but especially Intel, and whose mitigations require operating systems and in some case microcode updates that decrease performance for some specific tasks.

Microsoft has now pushed an update for Windows 10, and since I’m reviewing MINIX NEO N42C-4 mini PC powered by an Intel Pentium N4200 “Apollo Lake” processor, and just happened to run benchmarks before the update, so I decided to run some of the benchmarks again to see if there was any significant difference before and after the security update.

First I had to verify I had indeed received the update in the “installed update history”, and Windows 10 Pro was updated on January 5th with KB4056892, which is what we want, so let’s go ahead.

Benchmarks before Update

PCMark 10 is one of my favorite benchmark since it relies on typical program that many people would use on their desktop computer.

Click to Enlarge

Link to full results.

Let’s through 3DMark Sky Diver to get some 3D graphics performance.

Click to Enlarge

Link to 3DMark result.

Finally, I’ve run CrystalDiskMark to test I/O performance of the internal eMMC flash.

Benchmarks after Update

Let’s see if there are any significant differences, bearing in mind there’s always some variation for each benchmark run.

Click to Enlarge

Link to full results.

Right the score is lower, but it’s really insignificant, and represents at 0.63% decrease in performance, which should likely have nothing to do with the update. So no difference before and after update here.

Click to Enlarge

Same story for 3DMark Sky Diver 1.0, basically the same score as before the update. Link to 3DMark result.

Before (left) and After (right) – Click to Enlarge

There’s normally a lot more variation for I/O benchmarks like CrystalDiskMark, so results are a bit  more difficult to analyze, and have both screenshot side-by-side. We can safely say there’s no difference for sequential read/write (Seq Q32T1 & Seq), and I even got slightly better numbers after the updates. Random I/O look fairly good after the update, except for “4K Read” test. I repeated it several times, and always got 14 to 17 MB/s after the update (23 to 37% slower), while the “4K write” was always higher. This should not matter to most use cases.

At this stage, I was expecting to draw a table showing a 5% difference after the update, but I won’t, because there’s no clear performance hit after the update, despite Apollo Lake architecture being impacted by Meltdown and Spectre. Maybe some other database specific tests would have shown a difference, or the security fixes may mostly impact the performance of higher-end processors.

Zotac ZBOX PI225 Review – SSD-Like Mini PC Tested with Windows 10 & Ubuntu

What makes the Zotac ZBOX PI225 so interesting is that this is the first true ‘card’ form-factor mini PC. It is a mini PC that looks like a SSD. Whilst Intel replaced the ‘stick’ form-factor with a similar ‘card’ form-factor for their next generation mini PCs they also required a ‘dock’ in order to use them. The difference with the PI225 however is that it actually is a standalone mini PC and includes all the necessary input/output ports.
Intrigued by this new form-factor I decided to purchase one and the following is my review of its performance and capabilities.
The Zotac ZBOX PI225 is a fanless device which features an Apollo Lake N3350 SoC with 32GB of storage pre-installed with Windows 10 Home, 4GB RAM, 802.11ac WiFi, Bluetooth 4.2, two USB Type-C ports, a micro SD card reader and a power connector.
Importantly it comes with all the accessories you need to get up and running:

Click to Enlarge

including a Windows OS recovery disk although perhaps this could have been better provided on an SD card for ease of access. The twin USB/HDMI adapter means the device’s built-in Type-C USBs make the PI225 future-proof whilst removing the need to purchase new cables from the outset. Adding a VESA mount is a nice touch and emphasizes the size or lack thereof given the device is marginally smaller than a regular SSD.
The device once booted starts Windows which becomes fully activated after connecting to the Internet:

Click to Enlarge

The basic hardware matches the specification:

Click to Enlarge

with just under half the storage used after Windows updates:
Running my standard set of benchmarking tools to look at performance under Windows:

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

reveals the performance is much lower than expected for a N3350 SoC device. Checking the BIOS reveals that ‘Turbo Mode’ is disabled resulting in the clock speed being restricted to its based frequency of 1100 MHz and preventing it bursting to its top frequency of 2400 MHz.

Click to Enlarge

This was obviously set to limit the heat produced by the CPU and assist in the thermal design which makes use of the device’s outer metal case to dissipate heat in its role of passive cooling.
After enabling ‘Turbo Mode’ and ‘Active Processor Cores’

Click to Enlarge

the benchmarks were repeated:

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Interestingly the results for CrystalDiskMark noticeably improved after enabling ‘Turbo Mode’ and ‘Active Processor Cores’ as well:

Turbo Disabled

Turbo Enabled

which is reflected in all the benchmarks including on Linux (see later) and highlights the need to ‘interpret’ the results as indicative comparisons rather than definitive and accurate measurements.
So with this in mind the full results can be compared with other devices such as Beelink AP34 Ultimate or BBEN MN10.

Click to Enlarge

Next I installed Ubuntu to the eMMC as dual-boot. The BIOS includes the ‘Intel Linux’ as an ‘OS Selection’ under Chipset/South Bridge/OS Selection:

Click to Enlarge

However I found it wasn’t necessary to change it when using a standard Ubuntu ISO and it also wasn’t necessary to respin an ISO using my ‘isorespin.sh’ script.
Similar to Windows there is a significant performance gain when enabling ‘Turbo Mode’ and ‘Active Processor Cores’:

Click to Enlarge

Octane without ‘Turbo Mode’:
recorded a result of nearly half that of Octane with ‘Turbo Mode’:
With ‘Turbo Mode’ enabled the performance is as expected when compared to other devices with the N3350 SoC:

Click to Enlarge

and can be compared with other Intel Apollo Lake devices:

Click to Enlarge

Revisiting the hardware using Linux commands additionally shows that the memory is faster at 1866 MHz and configured as quad-channel and that the micro SD card is running the faster HS400 interface:
The device doesn’t have a headphone jack so audio is only available over HDMI:

Before looking at real-world usage examples it is worth discussing the thermal limitations of the device. From running the benchmarks alone it would seem obvious that keeping ‘Turbo Mode’ enabled would ensure maximum performance from the device. But as previously mentioned this setting is originally disabled and in part the reason for this can be demonstrated using the Octane benchmark. Without ‘Turbo Mode’ the benchmark runs without issue:

Before Octane 2.0 – Click to Enlarge

After Octane 2.0 – Click to Enlarge

However with ‘Turbo Mode’ enabled (note the CPU speed below the graph on the right):

Before Octane 2.0 – Click to Enlarge

After Octane 2.0 – Click to Enlarge

the power limit (see ‘Maximum’ column on left) is exceeded.
When the device with ‘Turbo Mode’ enabled was put under continuous load, for example playing a 4K video, this causes the temperature to continually rise and then thermal protection cuts in and the device effectively crashes. The following screenshot was taken shortly before this occurred during testing and shows that the CPU speed had already been throttled although the core CPU temperatures are still rising:

Click to Enlarge

So the findings are that with both Windows and Ubuntu it is impossible to watch a 4K video of any length without the device crashing when ‘Turbo Mode’ was enabled.
The good news is that 4K videos play as good as any similar device without ‘Turbo Mode’. Starting with Windows the first test was watching a 4K video using Microsoft Edge which worked perfectly:

Click to Enlarge

The same video when watched using Google Chrome resulted in the very occasional dropped frame:

Click to Enlarge

and changing the video quality to high definition (1080p resolution) results in fewer dropped frames:

Click to Enlarge

Running on Ubuntu the same video at 4K in Google Chrome was unwatchable with excessive dropped frames and a stalled network connection after a short while:

Click to Enlarge

At 1080p the video is watchable but does suffer from dropped frames:

Click to Enlarge

Running Kodi on Windows with a VP9 codec encoded video used software for decoding resulting in high CPU usage:

Click to Enlarge

compared with a H.264 codec encoded video which uses hardware to decode:

Click to Enlarge

and similar for videos encoded with H.265 or HEVC:

Click to Enlarge

with no issues playing the videos.
On Ubuntu hardware is used to decode all three codecs:

Click to Enlarge

Click to Enlarge

Click to Enlarge

however some H.265 videos resulted in a blank (black) screen just with audio whereas others played without issue:

Click to Enlarge

During testing without ‘Turbo Mode’ the device heats up playing videos but reaches a point where the passive cooling prevents the device from overheating:

Click to Enlarge

Click to Enlarge

But when the inside warms up so does the outside. Included within the packaging is a single slip of paper warning how the outside temperature can reach up to 57°C during continuous video playback:
Even allowing for a margin of error this temperature was reached during testing:
and with ‘Turbo Mode’ enabled the surface temperature can get very hot:
so that is a very good reason why this settings should not be enabled by default. For comparison a single walled paper cup of freshly poured coffee will be a similar temperature and for most people this is too hot to hold.
For WiFi connectivity, the 2.4 GHz throughput measured using ‘iperf’ shows 42.2 Mbits/sec for download but only 22.3 Mbits/sec for upload. However 5.0 GHz throughput is consistent with download measuring 152 Mbits/sec and upload of 142 Mbits/sec.

Click to Enlarge

I also purchased a small hub that connects through a Type-C connection and provides an HDMI port for video, a USB port for keyboard/mouse and a Gigabit Ethernet port for networking (‘iperf’ confirms 940 Mbits/sec for both upload and download). Using this hub means I still have the second Type-C port on the device for using a USB etc.
Power consumption for the device was measured as:
  • Power off – 1.0 Watts
  • Standby – 0.8 Watts
  • Boot menu – 5.0 Watts
  • Idle – 3.8 Watts (Ubuntu) and 4.3 Watts (Windows)
  • CPU stressed – 4.1 Watts (Ubuntu)
  • 4K video – 6.6 Watts (Ubuntu) and 6.4 Watts (Windows)

Finally the BIOS is very flexible with all the key settings available:

It may seem that this device is overly restricted by its thermal design. However, I’ve not found that to be the case once the limitations are known. The 4GB of memory is sufficient to run Windows or Linux OS and having a BIOS that supports Linux means that you are not restricted in what OS you can install. Storage can be expanded by using an SD card and the Type-C ports provide flexibility in how the device is connected. The ability to select ‘Turbo Mode’ means you can use this device as a mini PC although it should be disabled if using as an HTPC.  Zotac could have removed the setting from the BIOS, but kudos to them in leaving it and letting the user use the device and be responsible for how it is used. As shown the setting is not required for watching 4K videos, and this makes the device perfect for digital signage. Including the dual USB/HDMI adapter, VESA mount and the Windows recovery disk with detailed documentation is particularly noteworthy. Overall it is a very commendable effort given the new form-factor and challenges it presents.


Zotac ZBOX PI225 mini PC can be purchased for a little over $200 on websites such as Amazon or eBay.

Amlogic S905X vs Rockchip RK3328 vs Allwinner H6 Processors – Benchmarks & Features Comparison

November 27th, 2017 46 comments

Rockchip, Amlogic and Allwinner are all battling for the lower and mid range segment of the TV box market, so it may be interesting to compare their solutions. We won’t look into the ultra low-end market with 32-bit ARM Cortex A7 processor, but instead compare some of the recent quad core 64-bit ARM Cortex A53 processor for 4K HDR TV box from the company with respectively Amlogic S905X, Rockchip RK3328, and Allwinner H6 SoCs.

We’ll compare some of the benchmarks obtained with Android TV boxes, as well as other features like video support, USB and Ethernet interfaces.

Benchmarks

Let’s start with results for popular Android benchmarks: Antutu 6.x, Vellamo 3.x, and 3DMark Ice Storm Extreme v1.2 with results obtained from 3 TV boxes: Mini M8S II (Amlogic S905X), A95X R2 (Rockchip RK3328), and Zidoo H6 Pro (Allwinner H6). A score is highlighted in green is there’s a clear winner, and in red for a clear loser.

Amlogic S905X Rockchip RK3328 Allwinner H6
CPU (1) Quad core Cortex A53
@ 1.51 GHz
Quad core Cortex A53
@ 1.51 GHz
Quad core Cortex A53
@ 1.8 GHz
GPU (2) ARM Mali-450MP3 ARM Mali-450MP2 ARM Mali-720MP2
Antutu 6.x
Overall 33,553 33,117 40,467 / 36,957 (2)
3D (1920×1080) 3,099 1,475 6,292 / 2,782 (2)
UX 12,365 16,426 13,360
CPU 12,438 10,486 16,395
RAM 5,651 4,730 4,420
Vellamo 3.x
Metal 910 937 930
Multicore 1,491 1,464 836 (3)
Browser 1,855 (Browser) 1,943 (Chrome) 2,546 (Browser)
3DMark – Ice Storm Extreme v1.2
Total score 4,183 2,252 3,951
Graphics score 3,709 1,871 3,643
Physics score 7,561 7,814 5,608

(1) Those are the frequencies reported by CPU-Z, and the actual maximum frequency may be different. For example, it appears Allwinner H6 can only run at 1488 MHz in a sustained manner, and possibly only reach 1.8 GHz during short bursts (TBC).
(2) Allwinner H6 is the only SoC to include a GPU supporting OpenGL ES 3.1, which means it is the only one to complete Marooned 3D graphics test (Antutu 3D test has two 3D benchmarks), and the other boxes just got zero since it did not run. So I’ve included two scores for overall and 3D Antutu results: actual score / score minus Marooned benchmark.
(3) Vellamo multicore had a warning on Zidoo H6 Pro, so it may not represent the actual performance of the device.

Allwinner H6 has a slight advantage, but during use it will be really hard to notice the difference between TV boxes with one of those processors, and other factor like RAM capacity and storage performance will have more influence.One exception is 3D performance, as Rockchip RK3388 is clearly slower here, and I could notice it while playing games.

Features

But SoC performance is only one side of the equation, so let’s have a look at some of the features from the SoCs, which may or not be implemented in some TV boxes. For reference I also included USB 2.0 or 3.0 storage (HDD NTFS partition), and Ethernet performance numbers. Those numbers may vary a lot with further software optimization, configuration tweaks, so they should only be used for reference. I used the same TV boxes as for the benchmark section, except for Gigabit Ethernet relying instead on iperf results from ROCK64 development board (RK3328) and K1 Plus (Note S905, no X, for reference only, but in my experience all Fast Ethernet interfaces have about the same performance), and NEXBOX A95X for the USB storage performance.

Amlogic S905X Rockchip RK3328 Allwinner H6
Video
– 4K 10-bit HEVC Up to 60 fps
– 4K VP9 Up to 60 fps
– 4K H.264 Up to 30 fps (8-bit only) Up to 30 fps (8-bit and 10-bit) Up to 30 fps (8-bit only)
USB 2.0 / 3.0 USB 2.0 USB 3.0 USB 3.0
– A1SD Bench (R/W) 37/37 MB/s 94.52/90.73 MB/s 59.07/42.12MB/s
Ethernet 10/100M only Gigabit Ethernet MAC Gigabit Ethernet MAC
– iperf (full duplex) 91.6/91.8 Mbits/s 815/344 Mbits/s 758/350 Mbits/s
RAM Capacity (Max) 2GB 4GB 2GB
Misc  TS, Smartcard interface TS, Smartcard interface, PCIe

I did not include audio, as all those SoC are supposed to support Dolby TrueHD and DTS HD audio codec pass-through. but implementation varies a lot between devices.

Amlogic S905X is the weakest of the lot based on the two tables above, but it’s also the cheapest SoC among the three, and in my experience, one with the best support in Kodi, for example. Rockchip RK3328 is not much more expensive, and have many benefits, except when it comes to 3D graphics performance, but it usually only matter if you plan to play games on the platform, the GPU is usually good enough for user interfaces. Allwinner H6 has more interfaces, a Mali GPU with OpenGL ES 3.1 and OpenCL support, and lightly more interfaces. The few devices that are based on the Allwinner processor are currently quite more expensive with all other features being equal.

 

BBen MN10 TV Stick Review – Windows 10, Ubuntu 17.04, Benchmarks, and Kodi

The BBEN MN10 is the second Apollo Lake device to be released in the stick form-factor and on paper looks to have a lot to offer:

It features an Apollo Lake N3350 SoC, an unusual 3GB of RAM, 64GB of storage and is cooled by a ‘mute’ fan. The devices comes in a plain box with a power adapter, and a leaflet style manual.

Click to Enlarge

It also included a three-pin UK power adapter, as this was advertised as the ‘BBen MN10 Mini PC  –  UK PLUG  BLACK’.

Looking at the detail specifications:

Click to Enlarge

We can immediately see discrepancies as the device does not have a ‘RJ45 Port Speed: 1000M LAN’ port, and was not supplied with ‘1 x HDMI Cable’ nor ‘1 x Remote Control’.

Powering on the device and the ‘mute’ fan is also a miss-representation as it starts immediately and is noticeably noisy. It also runs at full speed regardless of workload so the noise is a constant reminder that the device is switched-on:

Starting Windows and the disappointment continues with a message informing that ‘We can’t activate Windows on this device because you don’t have a valid digital license or product key’:

Click to Enlarge

also ‘Intel Remote Keyboard Host App’ is pre-installed (see icon top left) and the computer name is already been set as ‘BBEN’.

As a result I tried installing Microsoft’s Windows 10 Home ISO but because of the confirmed lack of license, I then installed Microsoft’s Windows 10 Enterprise product evaluation ISO in order to review the device.

The basic hardware matched the specification:

with plenty of free-space available post installation:

I then ran some standard benchmarking tools to look at performance under Windows. These are a new set of benchmarks as I’ve updated the tools and releases specifically for devices running Windows version 1709 and later:

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Click to Enlarge

Octane 2.0 with Chrome – Click to Enlarge

As can be seen the performance is as expected for the N3350 SoC and is comparable with other devices such as ECDREAM A9 or Beelink AP34 Ultimate:

Click to Enlarge

Next I installed Ubuntu as dual-boot using my ‘isorespin.sh’ script, which includes installing the rEFInd bootloader to enable booting on Apollo Lake devices when the BIOS doesn’t support Linux:

Performance is again as expected:

Click to Enlarge

Click to Enlarge

And can be compared with other Intel Apollo Lake and earlier Intel Atom devices:

Click to Enlarge

Click to Enlarge

Revisiting the hardware using Linux commands additionally shows the micro SD card is running the slower HS200 interface: