Category Archives: hardware

Network Troubleshooting Part 1

Command Line

Even as enterprise infrastructure is moving towards the cloud, physical terminals, clients, devices ect… need to connect to that cloud. Even with wireless networking those wireless access point still need a wired connection. So this is how i go about checking for network connectivity.

Step 1: Check the Physical Layer. If a machine is not connecting to the network. I always pull it out and look at the link state lights. If i see green or amber flashing or steady. I then know at least the machine is connecting to something on the other end. If i don’t see any lights or red lights then i know the problem is somewhere in the physical layer.

Ethernet Link Lights

Often when i pull out the client box i’ll look for any bad kinks in the Ethernet. It not a common occurrence but is its possible that a really bad kink has cause internal rubbing and cross talk noise or even complete failure of the twisted pairs inside the Ethernet.

Step 2: Check the network device settings. You can read more about network device settings here. There is various ways of doing this if you’re on a windows based device with command line “ipconfig /all” you bring up all the relevant adapters and their current configurations. If you’re on a linux based system “lfconfig” will do the same.

ipconfig output

It is important to check the site’s documentation to see if the gateway is correct or if there is a possible ip address conflict. Also it very important to see if the machine is on the right VLAN. I’ll talk more about VLANs and sub netting in another blog article.

Thank your for reading please keep an eye out for part 2.

CCTV Systems, Costs

CCTV

Choosing a Closed Circuit Television System is important and can be very complex, and costly if you do it wrong. Nowadays there’s lots of different choices for cameras and types of cameras. I like to break down these product in to three simple categories. Some of these camera have features that overlap each other for example there are cloud based cameras that have local storage, but for simplicity’s sake i’ll group them with more common solutions for easier understanding.

 

Cloud Based Cameras– these are camera that often sold single or in a small pack with few dozen units, many of them are wifi based.

 

BNC Systems– BNC is a type of analog connection used to wire the cameras back into a main set to box somewhere that usually runs a linux based custom interface (please change the default password asap). They typically support anywhere from 4-16 cameras. Some may even support 20 but not much more.

 

IP/POE Based Systems– these cameras are connected using power over ethernet. They are usually hooked up to a server that runs software to manages all the video. These systems can range up to hundred of cameras.

 

Costs

Cloud based camera are the most affordable but a majority of them require monthly fees that can add up quickly especially if start getting more then one. Most cloud based cameras are designed to be installed the end user, and often are wireless and just need a power cable.

 

BNC set-top boxes don’t require a monthly fee and the cameras themselves are fairly cheap. The one fee that might pop up is if you don’t have a static IP address and need to get one from your ISP or sign up for a dynamic DNS service. The camera all require power and Analog video cable also certain models of camera a audio capable, but all of these connection usually are bundled together in a single cable. End-users can install it themselves but will probably opt for a professional installer.

 

IP/POE based systems are probably the most expensive solution they usually don’t require monthly fees except for certain software licenses and fees, and static IP fees. You need to have a proper network setup and PC/server to record and access the footage. Most likely end-users would have to contact a professional to design and setup this solution.

 

If you are interested in setting up a CCTV please let me know.

ThunderBolt 3, and eGPUs A New Form Factor in Computing

hardware

Thunderbolt 3 is a relatively new standard by Intel in late 2015. It uses the USB-C standard for the physical connections. Which makes it a little confusing, but it makes the physical cables needed cheap and gives the devices ability to be compatible with the USB 3.1 standards. Basically Thunderbolt 3 inherits all the ability and standards of USB 3.1 including the 10 Gbps that USB 3.1 can transfer. The place where Thunderbolt 3 stands out is that is is capable of 40 Gbps speeds and on Intel platforms links directly with the PCI Express bus. This allows its one standout feature the ability to link laptops to external GPUs using a small USB-C cable. This opens up a whole new market. Now Consumers can have a ultraportable laptop that they can hookup to and eGPu to game while they are at home.

Being that Thunderbolt 3 uses USB-C i think it is important to let consumers, end-users know that not all USB-C ports will support thunderbolt 3 speeds and capabilities. And not all devices will have thunderbolt 3 compatibility if the have USB-C. This is due to the fact that Thunderbolt 3 is an stander developed by intel in collaboration with Apple. While USB is a standard developed by a joint non-profit group made up of leading tech companies. The USB-C is fairly cheap to license sometimes it can even be free if you order a certain amount from manufacturer. Where as thunderbolt 3 is license from intel. Intel has claimed it will make Thunderbolt 3 free to license, but as of july of 2018 has not done so.

External GPUs or eGPUs have been on the market for a while. Before thunderbolt 3 one of the solutions was a dock for the GPU that used a cable to link to and internal or external PCI-E mini like those used for wireless LAN cards these were not ideal and lack driver support from many laptop manufacturers. But since thunderbolt 3 many major manufacturers have been making external enclosures to support the newest GPUs and supply the immense power that they require.

 

One of the major issues that eGPU users have to contend is the overhead that Thunderbolt 3 needs to compress or process the signal from the PCI-E 16x lans into the Thunderbolt 3 signal and then over to the GPU and to the monitor. This clocks in at around ~20% performance drop if you are using a monitor connected to the eGPU. If you are using the eGPU and the laptop’s screen the Performance hit can be around 20%-30%.

Another major issue is that the GPUs need extra hardware to be utilized as a eGPU they often need in excess of 300 watts and are called upon to charge the laptop while docked. They eGPUs also need thunderbolt 3 chipset with a link to an PCI-E 16x connector form the GPU itself. Also many offer extra usb connections for external devices that the laptop can connect to. This raises the costs significantly and causes segmentation in the offering for these enclosures. Thunderbolt 3 is a game changer when it comes to Form factors but price and performance are considerations when deciding to go this route.

Portable Gaming and Power Users

hardware

Gaming and Power Users now have a lot of options in regard to portable laptops. The biggest difference in these regards to these 2 type of users is the inclusion of a high end discrete graphics chips. When referring to PC a discrete graphics card means the computer has a chip or card that is specifically designed to process graphics as opposed to a graphics solution embedded in the cpu like AMD APUs, or Intel’s HD graphics. There are some processor that integrate “discrete” GPU but they usually aren’t suitable to high end gaming or programs that heavily use gpu processing. That being said there are solutions that lie in between both use cases with lower TDP discrete graphics options. TDP is referring to Thermal Design Power/Point which is essentially the amount of cooling you need to run the chips optimally.

 

High end graphic solutions of often require large heat pipes/sink and larger fans to perform well. Often adding 1-3 lbs to a laptops total weight. We often see these high end gaming laptops weight in around 5-8 lbs. For example and Gigabyte Aorus X9 GT sports a i9-8950HK and a GTX 1080 and a 17.3” screen and it weighs in around 8 lbs. Similarly a Acer predator with a i7-7700HQ and a GTX 1080 weigh in at 5.29 lbs, and a 15.6” screen. The main difference between these two besides the cpus are the screen sizes. Larger screens need more battery capacity to run around the same time.

All in all i find 7lbs to be even a little on the heavy side for carrying around, often they come with very large power bricks that tip the scales into too heavy. But on the extreme side of gaming laptops you can find SLI GTX 1080s in laptops, but these laptops push the weigh and noise envelope of laptops. SLI is a configuration of laptops that utilizes 2 identical GPUs that render in unison theoretically doubling the graphic rendering power of a system. For example the MSI GT83 Titan with a core i7 8th gen 8850H and 2x Nvidia 1080 is SLI configuration weighs in around 12.13 Lbs sports a 18.4” inch screen and costs $4,699.00.  Unfortunately i don’t have 5000$ lying around to buy this obscene machine. But it requires 2x 330 watt power adapters weighing in at 3 Lbs each. So if you’re going to go this route expect to carry around 19 lbs just for this machine.

In regard to naming schemes in laptop processors 7th gen intel product have suffix at the end of them to denote special features of each. The HQ stands for high performance graphics, quad core, HK stand for high performance graphics, and an unlock overclockable chip. There are also U designation for Ultra-Low power, and Y for extremely low power. Most gamers and power users are going to go for the H variants of the chips as the U,Y are clocked at lower speeds to save power.

But there does seem to be some interesting products that will allow you to do both lightweight computing and gaming with some caveats i’ll go into more of that next time.

Cross-Over PC Products and Ultra Portability

hardware

In my last article i discussed the emerging ARM base RISC style laptops equipped with the snapdragon 850 processor. In this article i’d like to highlight some of the more interesting aspects of emerging light and ultra light solution companies are PC manufacturers are coming up with to address the different segments of PC users.

 

In my option the PC market has segmented into several main user groups with some overlap, but mostly spending distinct amount of money on PCs. On the mainstream part of the spectrum we have light powered users i believe these make up the bulk of PC users. People who use their computers mainly for consuming video, surfing web pages, or doing office workloads on their PC. Light users options range greatly they usually anywhere form $300 – $1000 on a PC, and may even opt for a tablet. Then there are gamers these user often spend a good amount of money on a computer with most of it going towards a good graphics card. Gamers main use of their computers are of course gaming. You can read more about gaming builds here. Gamers can spend anywhere from $700 – $2000. Then there are heavy computing productivity users, these users usually work on video rendering and 3D effects and/or modeling.

 

In mid 2018 the light PC user segment has a wide range of form factors to choose from. The tablet style PC is very good option for weight vs power ratio. The microsoft surface line of tablet style PCs which released in 2013 had reach a good spot in portability vs power. They can playback all types of 1080p media, surf the web with no problem, come with all the latest communication connectivity hardware 802.11 AC, Bluetooth, and incertian models cellular LTE advanced. But one of the issue for some users is the keyboard they use case style flat keyboard, that may not work for some heavy typist. This can be mitigated with a bluetooth or USB keyboard but this would compromise the portability of it. You can Go to any microsoft store near you and see if you like their case cover/keyboard. Microsoft also offer laptop style tablet that dock with a extra battery and keyboard/trackpad. Also Microsoft has released a pen that users can use to draw/write on the touchscreen. The battery on the surface pro they claim can last upto 13 hours but most reviewer pegged it closer to 6-7 hours.

Light PC users also have the Ultra-portable laptops available to them. While Intel has trademark Ultrabook and specified that they use intel low power processors. I usually see these offering as any laptop usually weighing in as less than 3 lbs. Most of the intel ultrabooks do come with thunderbolt 3 with PCIe lanes that can be used with External Graphics Cards. Because of the weight limit imposed this ultrabooks are usually limited to 13”-14” screens. There are some i7 ultrabooks out there but mostly i5 intel processors. Also like i noted before in my computex 2018 highlights there are some new ryzen based laptops coming. Include some ultrabooks while they most likely won’t have thunderbolt 3 Intel owns the trademark and specification.Intel has said they would open source the specification but so far there hasn’t been any action. I doubt it will happen before any ryzen laptop come out.

In the next article i’ll go into portable options for gamers and maybe power users.

RISC Vs. CISC

hardware

Snapdragon’s new 850 processor is looking to slowly take over the windows productivity users. As i was talking about in the computex 2018 post Qualcomm is looking to create super energy efficient laptops. This brings up a very interesting debate and possible peek into the future of PCs. So a RISC or reduced instruction set computer works kind of like what it sounds like where the computer uses less complex processes of computing cycles to solve and problem. Where as CISC per cycle has much more complex cycles. Back when intel first release the x86 architecture this was a revolution in computing because the chip integrated complex processes into 8 bit architecture. This allow cpus to run on less ram but perform just as well. An oversimplified example of this is a programer would say to a CISC type cpu “open door A”, whereas in a RISC type CPU he would have to say something like “Grab DoorHand A; Twist DoorHandle A; Open Door A;”. These extra instructions would take up more memory and cost more, whereas a CISC CPU knows in it instruction set to grab and twist the door handle therefore only need a short instruction. But for many years we relied on the x86 architecture for our PC needs. And many of the productivity software and programs are based on the x86 architecture. This was fine because we weren’t worried about all the power it took for the x86 to process grabbing/twisting door handles. But as we enter and age where portability and energy efficiency are becoming more important we are looking to RISC to reduce the size and heat in our devices. Which brings me to Snapdragon 850 and the rumored Snapdragon 1000.

The Snapdragon series of processor are Qualcomm’s all in one package from cell phones and tablets. They use ARM’s RISC style architecture and also put many of the modern hardware modules onto one chip. Like in the Snapdragon 845 the have packaged in wifi, cellular data, GPU, system memory, and audio processor, security cryptographic processor, a processor for camera enhancements, and a processor for enhance instructions. They hold about 40% market share in the android market. Also they are the de facto chips for flagship phones. It makes sense that they are trying to push into the productivity market many folks already use their phone and tablets for work often paring a table with a bluetooth keyboard to work outside of the office.

So why don’t we use RISC processors in our PCs now that we have memory to spare. Well there are some roadblocks, but RISC architecture is slowly overcoming them. Of the issue was that RISC runs slower for many modern computational tasks because it doesn’t have the twist instructions built in. Many times to do complex float point complex integer math it could have a co-processor, and this would cause delays as packets of computing were moved back and forth between these to processors. Also many of the programs we use on windows would have to be rewritten to perform efficiently on RISC processors. But these are just roadblocks next i’ll look deeper into new and emerging devices that will probably replace laptops in the next few years.

Computex 2018 Highlights

hardware

Computex is an industry trade show for PC hardware manufacturers held in Taiwan. This year’s computex brought some exciting and interesting news and omissions of news. Trade show are interesting in that they basically forecast the next 6 months of products that companies will release.

 

This year the CPU vendors are fighting for the high end workstation market segment. AMD showed Thread Ripper 2 cpu with a whooping 32 cores on a single package. The details are scarce, but its is rumored to be around $1,500 price point. If this price is accurate this would really push Intel offering in this price range off the table. Intel showed a 28-core 5 ghz cpu, but this was essentially a marketing stunt to run a cpu at 5ghz requires phase change cooling and wouldn’t be reasonable for a consumer product. They said they’re going to release a 28-core chip but didn’t really have any details or a product to show.They didn’t show anything that could compete with threadripper 2 and at the 1.5K$ threadripper 2 rumored price point their current closest offering would be a core i9 extreme at 1.9k$ with 18 cores. AMD also announced they would start releasing 7nm node gpu, and server cpu chips. This is very exciting because a die shrink mean more performance for less heat and energy drain. You can read more about die shrinks here. Hopefully it eventually trickle down in to the consumer markets. While on the Intel side there was no news about 10nm, so we might have to wait a while for that. Also another thing that caught my eye is laptop makers are beginning once again to put AMD cpus into their line ups. Many of them come with Vega graphic compute units integrated into the ryzen cpus. The vega graphics bundle cpus are significantly better than Intel’s HD graphics offerings. This is significant for AMD because they had been largely left out of the laptop market for the past 10 years. Intel also release a limited edition i7 cpu to celebrate 8086 intel cpu. I don’t understand why they decided it would be a good idea. I just hope they don’t start trying to randomly put gold plated cpus and call them CPU loot boxes. Also Qualcomm announced their snapdragon 850 chip that is design for use with windows for ARM processors. This is interesting and i’ll talk more about ARM laptops later. That is the gist of CPU news from computex 2018.

In the GPU arena Nvidia has stayed mum, but AMD has announced their road map for their 7nm vega chips which should be shipping in the 2nd half of 2018.

 

Asus show some very interesting and unique products. They announced their ROG or Republic of Gamers branded phone which is a binned snapdragon 845 phone. The phone has a extensive array of accessories including a fan that will attach to the phone in landscape mode and cool the device, a device that docks with the phone that has a battery and second screen that can be used as a touch controller. Also a laptop that has a second screen instead of a keyboard, and a laptop that has a second small touch screen instead of a trackpad.

This is the main news from computex 2018 there are some interesting this from other companies that i will probably discuss as they are release.

PC Power Supplies (Part 6 Noise & Troubleshooting)

hardware

Most modern PSU have fans that cool the components inside the case. There are PSU that use very large fans to move more air at lower speed to negate some of the noise. There are  true “passive” cooled meaning the don’t use “active” fans. Most of the passive units still need some airflow from the case so being truly silent is difficult. And most Passive cooled PSUs are in the 450-600 watt range so if you have a dual graphics card setup it is probably not feasible. But the Quiet PSU with large 120mm+ fans will provide the power but if your PC rig need 600+ watts your graphics card fans or radiator fans will probably be making more noise anyway. I’ve seen truly silent pc cases that are very large aluminum heat sinks that would be able to be cooled passively. You can read more about them here. They’re very expensive specialty products but if you need a true silent or fanless set-up they’re probably the only ones. But even then the components themselves might emit a slight hum. It’s possible that the best option is to have the PC isolated in a different room if you need truly silent cooling.

This is very important don’t open a computer power supply unless you know what your doing the charge held inside can kill a person.

 

Troubleshooting a PSU is either very easy or very hard. The easiest way to tell if your PSU isn’t working is to try to turn on the PC. If it doesn’t power on at all then the first thing you need to check is the PSU. To turn-on the PSU you need to momentarily short the 14the pin. That will give the PSU the signal to power on. If it doesn’t respond  chance are you PSU is dead. If your not confident in shorting pins they sell these very inexpensive 20+4 pin PSU testers online for less than 10$. I recommend you get on that can test 4 pin cpu, molex, Sata, and 8 pin GPU plugs as well. They usually test the 12V, 5V, and 3.3V rails in your PSU to see if everything is operational. Also the ester is the best was to test those other plugs. In some cases a PSU might have a seperate rail for the pherperals. If you had a tester you could just power on the PSU and plug in each plug to see if they are working. The symptoms of a rail that just won’t turn on the peripherals don’t work.

There are other more difficult to diagnose issues a PSU can have unfortunately. One of the worst is when a rail momentarily stops working due to overheating or faulty components. I have seen PSU that boot a computer fine but when a load is applied the computer would just hang. I believe this was caused by a bad capacitor that burst. Because when it was booting it didn’t need a heavy load but when you would launch a game that needed the extra energy it wouldn’t be able to filter that load and the rectifier would just stop functioning. This can cause the computer to hang or just shut off randomly. The issue can be intermediate and seem like issue with other components. And it can be difficult to catch. There are probably tools that can measure and simulate loads but i think he easiest and cheapest was is to buy a $50 spare PSU and plug that in run the computer and if everything functions well you have fixed the problem.

 

Those are the simpler issues that you may have to deal with when working with PSUs.

PC Power Supplies (Part 5 Aesthetics)

hardware

As i have said before PC gaming enthusiast have driven a lot of innovation in the pc hardware market, but also have created a kind of branch evolution of hardware aesthetical upgrades. It’s kind of funny some of the options that have emerged. A fairly recent feature that hardware manufacturers have started adding have been “RBG” or red green blue LEDs that add lighting effects to pc hardware. The reason i’m filing this in power supply category is that RBG power supplies have existed for a while, but recently Lian Li has been showing their Strimer branded RBG power cables. While they haven’t modified the cable themselves that have created a kind of shroud of RGB wires that go over a set of white cable that give the effect of having RBG cables. I think they look pretty good, but i don’t really go for the RGB aesthetic. Some higher end motherboards have started including RBG headers which will allow the user to change the color schemes of their RGB parts. The Lian Li Strimer prototypes come with pci slot controls but i would imagine they would eventually interface with these RGB headers, but we won’t know for sure until they release them.

RGB aside power supply cables themselves have been another area of innovation in the PSU markets. In the early days power supply cables have been simply wrapped in a standard color scheme. The scheme is pretty simple where black was ground, yellow was 12 volts, red was 5 volts, and orange was 3.3 volts. They used fairly standard material this is of course to insulate the cables from shorting out. The components are fairly protected from interference from these plastic insulators. While these material are vulnerable to wear and tear they rarely move at all so that is not really an issue. I’ve seen an instance where cables were pniched so bad that it created a short the PSU refused to power on at all until that issue was fixed. To prevent some wear and tear/fraying and shorts cables have been sleeved in other materials nylon, stainless steel for real durability. But there are now companies that will do this for aesthetic reasons. I don’t think they really provide and practical benefit though normal PVC or PE wrapped cable will serve you just fine. You can even do this yourself by buying kits and crimping tools you can also change the “molex”-style connectors to different colors or UV reactive materials.

    

PSU themselves offer RGB fans inside their cases. I believe the reason there hasn’t been transparent PSU cases is because of possible interference from the PSU electricity leakage. The metal casing should provide good shieling for the processing components. I’m sure that manufactures will eventually find a material that will allow them to RGBize their PSU fully soon enough. It reminds me of a funny story where camera flashes where interfering with older raspberry pi’s because their processors weren’t shielded from the photoelectric effect you can read more about it here.

I find it slightly amusing but also exciting that innovation is pushing these type of products.

PC Power Supplies (Part 4 efficiency)

hardware

Power supplies smooth, filter, and rectify electricity that comes from wall outlets into a form that PCs can use. You can read more about that here. In doing so some of that electricity is lost in the form of heat. Older power supplies used to do so at much poorer rates like 60%-70%. That means out of the 100 watts your putting in 40%-30% was being wasted. This is an inescapable effect of physics for now. But some of that waste can be mitigated by using higher quality components more conductive paths for the electricity to flow and overall design changes. That why the 80 Plus certification was developed.

The 80 Plus Certification was developed to promote more efficient power supplies. I’ve included a chart above that gives you the efficiency at certain load points. For example the 80 Plus Bronze level at 20% load 82% of the electricity coming from you wall should be used to power the components of your PC and only 18% is lost. Load essentially mean if your power supply is rated at 100 watts your computer is demanding 20 Watts it is at 20% load. But as you can see in the chart there is no efficiency rating for less than 20% which mean at lower loads you power supply may be less efficient that is why it is important to buy a power supply with proper wattage. You can read more about that here. The price for 80 Plus rating don’t necessary follow the rating system for example you may find a 600 watt 80 Plus Gold for less then a 600 watt 80 Plus bronze. Many factor go into the pricing of a PSU like branding, components, accessories, modularity, and cosmetics. The important thing is to verify the PSU certification you can check a PSU 80 Plus rating here.

How important is efficiency? That is a tricky question for most consumers who may only use their PCs 3 hours or less a day it probably won’t make much difference on their power bill. But business and servers who run computer for 8-10 hours a day or even 24 hours would definitely benefit from more efficiency. The difference is significant from 80% efficiency to 90% means 10% less draw from the wall. If you are running a 500 watt server that mean you would draw 50 watts less on average. That would equate to nearly 438 kWh or Kilowatt hours a year at an average electicical rate of .12 per kWh that is about 52.56 dollars a year. That would make a significant difference if you had multiple machines. Also that would mean you wouldn’t have to pull that 50 watts out of the air when cooling that room. Efficiency is a tricky thing but if you have hundreds of users and dozens of servers it is an important consideration. And like i said a platinum power supply may not be much more expensive than a bronze so shop around and buy the right power supply.