[Soli]

IR receivers in TVs need to cope with all worst case scenarios for several different tv models. Interference the panel, from direct sunllight, from other devices, and havee good range and vertical/horisontal sensitivity etc etc. Your auntie expect the remote to just work, if not she, then she will complain to the store and return the TV set.

Yes but I expect the same for any IR built in device, as this could be for my auntie too, right? I complain about the IR sensitivity of Asrock's HTPC but my auntie would immediately stop using it thinking it's broken.

In an ideal world, yes. The difference is that a TV remote is a vital/primary feature, on a PC it's something that's tacked upon. When a primary feature doesnt work, that equals customers returning the item, reputation tarnished, dealers that have to sell their stocks on seriously reduced prices, then they choose another brand to sell next year, etc etc. In the end it equals losing business/money for both dealers and brands.

Case in point: Even though the IR sensor on your ASrock is less than good, you still kept it. If a primary feature like the USB ports didn't work, you would return it.

[afremont]

IR receivers are pretty much functionally all the same. There are different parts for specific operating conditions and modulation techniques and carrier frequencies. They are self contained devices that produce logic level outputs. All of the circuitry for reception gain, bandpass filtering and demodulating the received signal exist completely inside a 3-legged part. IOW, there isn't much that can vary with the sensor itself. I believe the shielding of ambient light is the key to getting good results. After that, it's the sender that is likely to be the problem. One key thing is that the sender be using the same carrier frequency that the receiver is expecting. The carrier frequency can be anywhere from 30kHz to 56kHz. Mismatching the sender and receiver carrier frequencies (don't confuse this with modulation methods such as RC5, RC6, Sony etc which are independent of carrier frequency) will result in severe performance issues.

Here is a link to the major player in the IR receiver game:
http://www.vishay.com/ir-receiver-modules/

EDIT: Some protocols do specify the carrier frequency, NEC is one. The most common carrier frequencies are 36kHz, 38kHz and 40kHz. I would venture to guess that 38kHz is the most common. It is extremely important that you not stray very far from the frequency that the receiver expects or you will lose range rapidly. They have a fairly sharp filter response.

EDIT2: One more thing. Fluorescent lights are the bane of IR communications. They can cause a lot of grief to a receiver.

[noggin]

EDIT2: One more thing. Fluorescent lights are the bane of IR communications. They can cause a lot of grief to a receiver.

As are plasma displays - as they can emit a lot of IR noise.

[bugmenot2015]

Zotac's Braswell nano model is apparently the Zbox CI323:

http://www.skytech.lt/zboxci323nanobe-zo...77473.html

[maynard76]

Hi guys,
i received the Zotac CI323 today and have a little problem with it. The exact same problem occured with the Asrock N3700 ITX board i had before.
The CI323 is connected to a Samsung TV via HDMI. Eventhough i know most of you do not like Windows, i have installed Windows 7 Pro for several reasons.
The whole system was freshly installed today and the newest drivers from the Zotac website were used.

The problem i am having is, when i move the mouse-cursor quickly to the lower left side of the desktop screen, the picture is shaking and finally turns blue or white.
I uploaded a short 20MB mp4-clip, so you can see what exactly i mean. I did not have the problem with an A6-6310 laptop connected to the tv. I do have it with the Zotac and Asrock, which both have Braswell chips.
The problem does not occur though, when moving the cursor in Kodi.

Is there any idea, what i might try to fix the problem?

http://workupload.com/file/X0qQ2mI1

[xexe]

Just an update. I purchased a Asrock N3150-ITX with two 4GB Kingston budget RAM sticks from the approved memory list.

Running OE6 stable and even thought it is pre Kernel 4.3 it works fantastically.

The speed at which this MB and CPU works on paper should not be that much better than the C2D + nvidea it replaces but in reality it is significantly faster. More noticeably and unexpectedly the picture quality from the same source material and the same cables + TV is MUCH... and I really mean MUCH better (and apparently its better again after 4.3).

If you are on the fence jump over it because for the money this kit is amazing.

[kjhambrick]

All --

Sorry if this is off-topic for the forum but it's about the site that mentions the NUC and the Zotac ZBoxes.

I've been testing the new generation boxes for a headless Data Conversion Appliance we supply to our customers.

The App is IO Bound ( insane amount of Disk IO ) and most CPU Load is String Compares with more than a few Integer and Floating Point to String Conversions.

These are the Boxes:

1. Zotac ZBox ID18 ( Celeron 1007U @ 1.5 GHz )
2. NUC NUC5PPYB ( Pentium N3700 @ 1.6 GHz )
3. Zotac ZBox BI323 ( Celeron N3150 @ 1.6 GHz )

All three Boxes have the same HDD ( HGST HTS721010A9E630 ) and each has 8 GB of Crucial Ballistic RAM

The OS is CentOS 6.7 64 bit

The two ZBoxes have 2 - 4GB Sticks of 9-9-9-24 / 1.35 Volt RAM, the NUC has a single 8 GB Stick 10-10-10-30.

The newer Braswell Chips run over 2-times SLOWER than the older ID18 !

1. ID18 - 85 sec
2. NUC - 179 sec
3. BI323 - 187 sec

Any insight out there as to why this might be the case ?

Thanks.

-- kjh

[noggin]

All --

Sorry if this is off-topic for the forum but it's about the site that mentions the NUC and the Zotac ZBoxes.

I've been testing the new generation boxes for a headless Data Conversion Appliance we supply to our customers.

The App is IO Bound ( insane amount of Disk IO ) and most CPU Load is String Compares with more than a few Integer and Floating Point to String Conversions.

These are the Boxes:

1. Zotac ZBox ID18 ( Celeron 1007U @ 1.5 GHz )
2. NUC NUC5PPYB ( Pentium N3700 @ 1.6 GHz )
3. Zotac ZBox BI323 ( Celeron N3150 @ 1.6 GHz )

All three Boxes have the same HDD ( HGST HTS721010A9E630 ) and each has 8 GB of Crucial Ballistic RAM

The OS is CentOS 6.7 64 bit

The two ZBoxes have 2 - 4GB Sticks of 9-9-9-24 / 1.35 Volt RAM, the NUC has a single 8 GB Stick 10-10-10-30.

The newer Braswell Chips run over 2-times SLOWER than the older ID18 !

1. ID18 - 85 sec
2. NUC - 179 sec
3. BI323 - 187 sec

Any insight out there as to why this might be the case ?

Thanks.

-- kjh

Passmarks :

1007U in the ID 18 = 1419 (Single thread 768)
N3700 in the NUC = 1824 (Single thread 529)
N3150 in the BI323 = 1431 (Single thread 386)

The 1007U is dual core (Ivy Bridge?), the N3700 and N3150 are quad core (Braswell?). If your code is not massively optimised for multithreading, then the older 1007U is likely to perform better. As you can see it's Passmark score in a single thread is twice that of the N3150. Or in other words, the 1007U cores benchmark at twice the speed as the N3150 cores. (4 N3150 cores only just about keep up with 2 1007U cores, and a single N3150 core has approx half the performance of a 1007U core)

Unless you software is hugely optimised for multithreading then the single core/thread figure is worth keeping an eye on.

Obviously benchmarks don't tell the whole story - but the Celeron 1007U is based on an Ivy Bridge (?) Core-series core (i.e. the same as used in the Core i3/5/7 of the same era). The N3700/N3150 are effectively based on (Braswell?) Atom-type cores, even though they are now marketed as Celerons and Pentiums. The inherent CPU architecture in the Braswell isn't as powerful as the Core-series CPU architecture used on the 1007U (and the following 2955/2957U and current 3205U etc.) However they are also a lot more power efficient.

For comparison here are the Passmarks for the newer Core-series based Celerons :
3205U Broadwell Celeron (based on Core not Atom CPU architecture) = 1712 (Single thread 863)
2957U Haswell Celeron = 1488 (Single thread 804)

I'd expect both of the above CPUs to outperform - albeit slightly - the 1007U?

There are other issues that also may play a part - and skew things in favour of the Ivy Bridge/Haswell/Broadwell - is I/O. It could be that the Core-series have a better I/O implementation and are using dual channel memory more effectively? Could also be that IO software/driver support is faster on the more powerful cores?

[kjhambrick]

All --

Sorry if this is off-topic for the forum but it's about the site that mentions the NUC and the Zotac ZBoxes.

I've been testing the new generation boxes for a headless Data Conversion Appliance we supply to our customers.

The App is IO Bound ( insane amount of Disk IO ) and most CPU Load is String Compares with more than a few Integer and Floating Point to String Conversions.

These are the Boxes:

1. Zotac ZBox ID18 ( Celeron 1007U @ 1.5 GHz )
2. NUC NUC5PPYB ( Pentium N3700 @ 1.6 GHz )
3. Zotac ZBox BI323 ( Celeron N3150 @ 1.6 GHz )

All three Boxes have the same HDD ( HGST HTS721010A9E630 ) and each has 8 GB of Crucial Ballistic RAM

The OS is CentOS 6.7 64 bit

The two ZBoxes have 2 - 4GB Sticks of 9-9-9-24 / 1.35 Volt RAM, the NUC has a single 8 GB Stick 10-10-10-30.

The newer Braswell Chips run over 2-times SLOWER than the older ID18 !

1. ID18 - 85 sec
2. NUC - 179 sec
3. BI323 - 187 sec

Any insight out there as to why this might be the case ?

Thanks.

-- kjh

Passmarks :

1007U in the ID 18 = 1419 (Single thread 768)
N3700 in the NUC = 1824 (Single thread 529)
N3150 in the BI323 = 1431 (Single thread 386)

<snip>

Unless you software is hugely optimised for multithreading then the single core/thread figure is worth keeping an eye on.

Thank you VERY much noggin !

The Data Conversion App is pretty much a 'top-down, sequential' process so while there may be some multi-threading going on at the OS-Level, the App itself is not.

That explains it all, I think so I'll stop looking for BIOS / OS-Level optimizations to 'fix' the performance on systems with the newer CPUs.

Thanks again.

-- kjh

[noggin]

All --

Sorry if this is off-topic for the forum but it's about the site that mentions the NUC and the Zotac ZBoxes.

I've been testing the new generation boxes for a headless Data Conversion Appliance we supply to our customers.

The App is IO Bound ( insane amount of Disk IO ) and most CPU Load is String Compares with more than a few Integer and Floating Point to String Conversions.

These are the Boxes:

1. Zotac ZBox ID18 ( Celeron 1007U @ 1.5 GHz )
2. NUC NUC5PPYB ( Pentium N3700 @ 1.6 GHz )
3. Zotac ZBox BI323 ( Celeron N3150 @ 1.6 GHz )

All three Boxes have the same HDD ( HGST HTS721010A9E630 ) and each has 8 GB of Crucial Ballistic RAM

The OS is CentOS 6.7 64 bit

The two ZBoxes have 2 - 4GB Sticks of 9-9-9-24 / 1.35 Volt RAM, the NUC has a single 8 GB Stick 10-10-10-30.

The newer Braswell Chips run over 2-times SLOWER than the older ID18 !

1. ID18 - 85 sec
2. NUC - 179 sec
3. BI323 - 187 sec

Any insight out there as to why this might be the case ?

Thanks.

-- kjh

Passmarks :

1007U in the ID 18 = 1419 (Single thread 768)
N3700 in the NUC = 1824 (Single thread 529)
N3150 in the BI323 = 1431 (Single thread 386)

<snip>

Unless you software is hugely optimised for multithreading then the single core/thread figure is worth keeping an eye on.

Thank you VERY much noggin !

The Data Conversion App is pretty much a 'top-down, sequential' process so while there may be some multi-threading going on at the OS-Level, the App itself is not.

That explains it all, I think so I'll stop looking for BIOS / OS-Level optimizations to 'fix' the performance on systems with the newer CPUs.

Thanks again.

-- kjh

No worries. The Braswell Celerons (which are Atom-based) are gaining popularity in these parts because they finally have a decent GPU with enough EUs to be useful, and have decent hardware codec support for non-CPU video decoding (I think Braswell does HEVC - though not sure if 8 or 10 bit)

For pure CPU stuff the Core Celerons (i.e. Haswell and Broadwell-based) are probably still a better bet, and definitely are for single-threaded stuff, but they draw more power, and are less widespread as commercial motherboard+SOC deployments. So whilst you can buy barebones and systems based on them, you can't buy many (any?) mini-ITX or micro-ATX motherboards with them on. Which is very annoying. You CAN buy plenty of Braswell motherboards though... It's one reason the Chromebox has been so popular - as it has a great sweet-spot of CPU and GPU performance and was available at a reasonable price. If a motherboard were available with the same SoC/CPU it would have been popular too at the right price!

[Cob2]

Ok. After a lot of lecture from fritsh and other now I know what is all about with hevc 10 bits and all future stuff. I would take advice from fritsch amd buy something under 200$. Beebox and nuc5pphy are on the favor. I have read at one review at nuc can not play 4k x254 content propertly and beebox can because of dualchannel memory. Other hand nucblog says nic can almost everything with x254 but I do no know can I trust him. What Is true? I am so close to make a choice. Only that one particurarly thing left. Can anyone help me?

[fritsch]

I don't know about x254, sorry :-)

Get the beebox 3150 with 2 * 2 GB memory and use OpenELEC and throw it in 12 months and get the next generation.

[falc410]

I want to replace my Intel 1037U HTPC with something that can play back 4k@30. I already bought my father a Beebox with N3150 and so far he is quite pleased - plays content from his 4k Camcorder without a hitch. Although I should mention that it's only encoded with h264 - I would require it to handle HEVC h265 8-Bit 4k@30. I'm not 100% sure if this already works today. Apparently when Braswell came out it didn't, but the improvements to vaapi during the summer lead me to the conclusion that Braswell is now in fact capable of h265 4k@30 - is this correct?

I just noticed that some mainboards are also available with N3700 - would I have any advantages when running Openelec on this one compared to a N3150?
I'm also still a bit undecided if I should instead move from my beloved HTPC case (with IR and Logitech Harmony) to a Android Box. I'm kinda scared since Audio Passthrough to my external Receiver might be a problem (although fritsch said more than once to use LPCM - have to test this) and also automatic refresh rate change might not work with all devices apparently.

So would you go for a Intel Braswell solution or an Android Box? I've tried to read through all the recent threads but I haven't really found a comparison (except that Android Boxes with HDMI 2.0 can play back 4K@60 which I don't have content for anyway).

Getting a new mainboard for my HTPC would be a lot cheaper than any other solution (cheapest N3150 board is just 63 Euros atm).

[pr0xZen]

If OpenElec is your flavor, personally I'd go with the braswell rather than current android boxes in the lower price segment. Or, wait it out for next gen NUCs.

As of yesterday, skylake units (i3/i5) are shipping - celerons expected Q1 2016. And according to Intel reps, skylake celerons should finally have the HDMI HBR audio passthrough issues (under windows) worked out. I still maintain my (fairly well covered) stand that its a matter of "won't" from Intel, rather than "can't" (in regards to former and current celeron nucs) - but I digress.

[mwake]

I just got my hands on NUC5CPYH kit and want to install openelec on it. I am an absolute noob with regard to Linux and openelec. However, I have skimmed through some of the forum articles and have been seeing that there were/are some installation/usage problems with the latest official build of openelec and braswell based nuc. I heard that Fritsch build is more stable, but would like to know what Fritsch build can/can't do compared to official build and where I can find the Fritsch build as I can't seem to find it on the internet. Many thanks for any help on this.