Posted on Thursday, November 23 2017 @ 10:11:03 CET by Thomas De Maesschalck
Power Compare compiled some interesting statistics about the huge electricity consumption of the Bitcoin network. With an estimated 2017 power consumption of 29.05TWh, Bitcoin accounts for roughly 0.13 percent of the total global electricity consumption.
While that doesn't sound like a lot, Power Compare points out that's more than the individual power consumption of 159 countries. If Bitcoin was a country, it would rank 61st in terms of electricity use.
Power Compare also indicates that Bitcoin power usage is growing at an alarming rate. Over the past month alone, Bitcoin miners increased their power consumption by 29.98 percent as the cryptocurrency keeps soaring to new heights.
Interestingly, if Bitcoin power consumption keeps growing at the same rate, it would consume as much electricity as the whole world by February 2020. Of course, the odds of that happening are extremely low. Clearly, something's gotta give as this sort of growth is not sustainable. The big problem here is that a single Bitcoin transaction equals about as much energy as a US household consumes in a week as the process to secure the database is very inefficient.
The huge power consumption of cryptocurrencies like Bitcoin is nothing new but it could be a potential new angle to attack the technology, from an environmental as well as resource management point of view. Cryptocurrencies like Bitcoin are already on track to become a major headache for governments and its inefficiency will probably be one of the talking points to curtail its use.
Posted on Thursday, November 23 2017 @ 09:48:56 CET by Thomas De Maesschalck
The latest Top500 supercomputer list reveals China has not only the fastest but also the second-fastest supercomputer in the world. The Titan supercomputer of the US Oak Ridge National Laboratory ranks fifth but the US has a shot of becoming great again as the lab's upcoming Summit supercomputer should propel it to the top of the list.
Summit is expected to be ready in 2018 and will have a raw computing performance of roughly 200 petaflops, double as much as the 93 petaflops delivered by China's Sunway TaihuLight. It's pretty interesting how fast things are moving along in the supercomputer world, the current second-fastest supercomputer has 33.8 petaflops and the third fastest model has just 19.59 petaflops. The Summit supercomputer is projected to be faster than the current top 5 supercomputer combined!
Made by IBM, the Summit supercomputer will feature about 4,600 watercooled nodes, an individual node packs two IBM
Power9 processors and six NVIDIA Volta GV100 GPUs. Each node also has 512GB of coherent DDR4 and HBM2, as well as 1600GB non-volatile RAM.
The system features 96 lanes of PCIe 4.0 that comes in handy for the dual-port Mellanox EDR InfiniBand adapter, which has a theoretical maximum throughput of 400Gb/s. IBM has measured throughput at 392Gb/s, which is twice the bandwidth of a PCIe 3.0 adapter.
The Volta GV100's connect via PCIe 3.0 and NVLink 2.0. The NVLink interface provides 100GB/s of throughput for CPU-to-GPU and GPU-to-GPU traffic. The GPUs are arranged in a dual-mesh design.
More specs can be found at Tom's Hardware. Summit will consume a whopping 15MW of power, a significant increase from Titan's 9MW. However, Summit delivers over eleven times the performance of Titan so there's definitely a huge increase in power efficiency. The supercomputer will be used to simulate and explore complex systems.
Posted on Wednesday, November 22 2017 @ 17:48:27 CET by Thomas De Maesschalck
Sharkoon launched its S1000, this is a new series of low-budget cases with a Micro-ATX form factor. The focus here is on providing a decent price/performance, you get a tunnel for PSU and HDDs, some cable management features, removable dust filters, and a cableless front panel design. Pricing is 39.90EUR for the most basic version, for an extra 5EUR you can get the model with the acrylic side window plus a pre-installed 120mm blue LED fan.
With the compact S1000 and S1000 Window, Sharkoon introduces two new micro-ATX cases with an integrated tunnel for power supplies and hard drives. As a result, the power supply and hard drives are installed discreetly to the case bottom and out of sight. The S1000 Window version also offers an acrylic side window and a pre-installed 120 mm blue LED fan on the rear panel.
Fresh air is supplied through the side air intakes of the front panel, behind which a detachable dust filter is also provided. Another removable dust filter is located on the bottom of the case. Both case versions include a pre-installed 120 mm fan in the front. Suitable mounting holes are also provided for additional fans. Alternatively, despite the compact design, a 240 mm radiator can be installed with a total height of 5.9 cm including fan. To facilitate the installation of 280 mm radiators, the 5.25" drive bay is modular and, if necessary, easily removed. Located on the top of the front panel are the audio connectors for the microphone and headphone as well as two USB 3.0 ports. These are incorporated directly into the body of the case, thus allowing the front panel to be cableless removed. On the top panel of the S1000 Window is an additional, pre-installed dust filter, which is secured by magnetic fasteners, plus it is also possible to install up to three 120 fans or two 140 mm fans. With the S1000, the top panel is closed and a 120 mm fan is pre-installed on the rear panel, but without LED illumination. Another optional 120 mm fan can be installed to the side panel.
The S1000 has room for a CPU cooler with a maximum height of 15.5 cm; within the S1000 Window, the installed CPU cooler can have a maximum height of 15 cm. In both case versions, graphics cards can have a maximum length of 40 cm. The opaque tunnel offers enough space for decoupled mounting of a power supply with a maximum length of 24 cm. Inside the HDD cage, up to two 3.5" hard drives or two 2.5" SSDs can be decoupled mounted. The cage can also be slightly moved along the tunnel floor thanks to the oval whole cutouts and convenient thumbscrews, allowing more space for cabling. Air vents on the tunnel ceiling ensure optimal cooling of drives. Up to two 2.5" HDD/SSDs can mount to the back of the mainboard tray with help of two mounting brackets and attached thumbscrews. The cable management system with practical pass-throughs guarantees easy and discreet cabling.
The Sharkoon S1000 is now available for the suggested retail price of 39.90 euros from authorized retailers. The Sharkoon S1000 Window is also now available for 44.90 euros.
Posted on Wednesday, November 22 2017 @ 10:36:56 CET by Thomas De Maesschalck
Will chips continue to scale or will physical barriers be hit in the near future? Tsai Ming-kai, the chairman of fabless chip designer MediaTek, said this week that he foresees Moore's Law will apply to two more generations.
In a November 21 talk with Alibaba's chief technology officer Wang Jian, Ming-kai remarked process technology may hit bottlenecks after moving to the 3nm node.
Tsai made the comments in a November 21 talk with Alibaba's chief technology officer Wang Jian. He said that along with the ever-advancing computing technologies and performance and the ever-expanding data flow, new business opportunities associated with solving human problems through data applications will emerge anytime and anywhere for those pursuing innovations and tech startups.
More at DigiTimes. This does't mean we'll no longer see performance improvements, but continuing on the same pace as in the past is getting harder and harder.
Posted on Wednesday, November 22 2017 @ 10:27:48 CET by Thomas De Maesschalck
Not a lot of news today so here's a teaser announcement from Spire. The company uploaded a tiny photo of Supreme Pro, an upcoming computer case. This model has a separate PSU chamber and has "smart features". The product launch will follow soon.
Posted on Wednesday, November 22 2017 @ 10:16:53 CET by Thomas De Maesschalck
Well this is pretty embarrassing. Microsoft just confirmed that the charger that ships with its Surface Book 2 isn't powerful enough to keep the device charged under all usage conditions. The software giant said that during long gaming sessions, with the Power Mode slider set to "best performance", the charger may not be able to provide enough power to prevent battery drain.
The "best performance" setting is recommended if you want to play games or are doing video editing, to ensure you fully utilize the potential of the onboard hardware of the Surface Book 2. The Verge points out some data about the power consumption and the maximum power delivery capacity.
It appears that the Surface Book 2 has been designed to supply 95 watts of power from the charger to the device, which isn’t enough to run the processor, graphics card, and all other hardware components at max. The processor alone draws 25 watts in high-power mode, and will even burst to 35 watts. Microsoft’s Nvidia GTX 1060 variant draws between 70 and 80 watts, bringing the total to 105 watts at peak. Microsoft works around this by aggressively throttling the Nvidia chip during games at “better performance” and “best battery” settings.
Overall, it does't appear this is problematic. The maximum peak power consumption of the device is about 105W, and the power delivery system supports reloading of up to 95W. The Verge crunched the numbers and said the battery drainage occurs at a rate of at least 10 percent per hour, depending on the game and load. This means you can game many hours at full performance before you fully drain the battery. So not a huge issue, but definitely funny.
Posted on Wednesday, November 22 2017 @ 10:01:24 CET by Thomas De Maesschalck
Most of the tech giants of today are in the business of data so it's probably not a surprise that they capture as much as they can. Google is in the news this week as Quartz discovered that the search giant is keeping close tabs on users of its Android operating system.
The site found out that Google receives data about your location and your movement, even if you've turned off location services, haven't used any apps, and haven't even inserted a SIM card. This is because since the start of this year, Android started collecting data of nearby cellular towers, and this information is send back to Google.
When confronted with the findings, Google claims this data was never used or stored. A Google spokesperson promised this feature will be turned off by the end of this month:
The cell tower addresses have been included in information sent to the system Google uses to manage push notifications and messages on Android phones for the past 11 months, according to a Google spokesperson. They were never used or stored, the spokesperson said, and the company is now taking steps to end the practice after being contacted by Quartz. By the end of November, the company said, Android phones will no longer send cell-tower location data to Google, at least as part of this particular service, which consumers cannot disable.
“In January of this year, we began looking into using Cell ID codes as an additional signal to further improve the speed and performance of message delivery,” the Google spokesperson said in an email. “However, we never incorporated Cell ID into our network sync system, so that data was immediately discarded, and we updated it to no longer request Cell ID.”
Quartz says it's not clear how this data could be used to improve message delivery, but notes the privacy threat is pretty clear as addresses of cell towers can be used to pinpoint a user's location. In rural areas this triangulation works out to an approximation of a quarter-mile radius, but more accurate pinpoints are possible in urban areas.
Posted on Wednesday, November 22 2017 @ 09:49:03 CET by Thomas De Maesschalck
With Thanksgiving approaching the news stream will probably dry up a bit this week as a lot of people have left the office. Today we have news about a new EVGA GTX 1080 Ti K|NGP|N Hydro Copper video card, this is a new single-slot flagship model that costs a whopping $1,249.99.
As the naming suggests, this version is equipped with a Hydro Copper waterblock. The card has 10+3 phase power, has some RGB LEDs, comes with the EVGA iCX technology, features nine thermal sensors and has a pair of side-mounted 8-pin PCIe power connectors.
By default, the card runs at 1582MHz base clock and 1695MHz Boost. However, EVGA guarantees this version can hit a frequency of at least 2025MHz.
Posted on Tuesday, November 21 2017 @ 16:15:58 CET by Thomas De Maesschalck
There are a lot of ways to cool computer chips. The more commonly used techniques include passive cooling with just a heatsink, active cooling with a heatsink+fan or watercooling. I always found immersion cooling to be one of the fancier ways to cool a computer, the concept has been around for a long time but it's not really suitable for the mainstream market as there are a lot of downsides. But it looks really cool, when executed properly it gives stunning visuals.
PC Perspective reports Allied Control had an immersion cooled mining rig on display at the SuperComputing 2017 conference.
You can view it in the tweet below, it's a two-phase immersion cooling system with a water cooled condensor coil and the 3M Novec fluid. Novec is a non-conductive fluid with a boiling point of just 41°C. This fluid costs over $100 per liter and needs to be hermetically sealed so it's easy to see why this isn't going to go mainstream. But still, it looks extremely cool!
Nick Knupffer (@Nick_Knupffer) posted a video (embedded below) of the cooling system in action cooling a high end processor and five graphics cards. The components are submerged in a non-flamable, non-conductive fluid that has a very low boiling point of 41°C. Interestingly, the heatsinks and fans are removed allowing for direct contact between the fluid and the chips (in this case there is a copper baseplate on the CPU but bare ASICs can also be cooled). When the hardware is in use, heat is transfered to the liquid which begins to boil off from a liquid to a vapor / gaseous state. The vapor rises to the surface and hits a condensor coil (which can be water cooled) that cools the gas until it turns back into a liquid and falls back into the tank. The company has previously shown off an overclocked 20 GPU (250W) plus dual Xeon system that was able to run flat out (The GPUs at 120% TDP) running deep learning as well as mining Z-Cash when not working on HPC projects while keeping all the hardware well under thermal limits and not throttling. Cnet also spotted a 10 GPU system being shown off at Computex (warning autoplay video ad!).
Posted on Tuesday, November 21 2017 @ 15:41:50 CET by Thomas De Maesschalck
Some details about Toshiba's HDD roadmap have been made public. The company recently started shipping 9th generation perpendicular magnetic recording (PMR) based HDDs, this is a 1TB model with a 2.5" form factor and a thickness of 7mm. This new generation enables 2.5" platters with a capacity of 1TB and 3.5" platters with a capacity of 1.8TB. It will allow Toshiba to launch 14TB HDDs in 2018.
According to SDK, 9th generation PMR media for 3.5” hard drives will have capacity between 1.5 and 1.8 TB. Typically, platters of the same generation produced by SDK, Seagate, and Western Digital have similar areal density with some minor differences. Therefore, we can expect the 9th gen PMR technology from the aforementioned HDD makers to feature similar capacities. Showa Denko plans to start mass production of its 9th gen PMR 3.5” media in early 2018, but, neither Seagate nor Western Digital (who produce their leading-edge platters in-house) have announced their timeline for similar discs. Typically, production schedules for advanced media also tend to be very similar for various makers, but we do not have any official data in hand right now.