Slashdot Mirror
Intel Announces Atom E3900 Series - Goldmont for the Internet of Things (anandtech.com)
Intel has announced the Atom E3900 series. Based upon the company's latest generation Goldmont Atom CPU core, the E3900 series will be Intel's most serious and dedicated project yet for the IoT market. AnandTech adds: So what does an IoT-centric Atom look like? By and large, it's Broxton and more. At its core we're looking at 2 or 4 Goldmont CPU cores, paired with 12 or 18 EU configurations of Intel's Gen9 iGPU. However this is where the similarities stop. Once we get past the CPU and GPU, Intel has added new features specifically for IoT in some areas, and in other areas they've gone and reworked the design entirely to meet specific physical and technical needs of the IoT market. The big changes here are focused on security, determinism, and networking. Security is self-evident: Intel's customers need to be able to build devices that will go out into the field and be hardened against attackers. Bits and pieces of this are inerieted from Intel's existing Trusted Execution Technology, while other pieces, such as boot time measuring, are new. The latter is particularly interesting, as Intel is measuring the boot time of a system as a canary for if it's been compromised. If the boot time suddenly and unexpectedly changes, then there's a good chance the firmware and/or OS has been replaced.
"Inerieted"? That's an odd word. Did the writer perhaps mean to type "inebriated"?
I wonder how useful having the time it takes to boot be a measurement if a ROM is compromised or not.
For example, assuming the ROM uses Linux and has a few writable partitions, if it boots up and does a fsck, or just replays filesystem transaction logs, this will almost certainly be different each boot, especially if the system had a dirty shutdown.
However, if the timing is measured from the OS boots until it mounts the read-only RAMdrive and gets ready to load the main OS, that is a lot more predictable.
Let's see
http://saveie6.com/
That's damn hungry for IoT...
Meanwhile, ARM announces Cortex M23 potentially capable on running purely on harvested energy alone apparently.
"I bless every day that I continue to live, for every day is pure profit."
I may have missed something but isn't a target TDP of 6.5-12 watts a little too much for IOT? I often read about ARM based boards with TDP on 0,9-2 watts. Ok those Atoms may be most probably far powerful, but their TDP still looks like a little too high.
Of course. There is also ways to create secure devices in the first place. Unfortunately neither is a selling point, manufacturers of those devices are not held responsible for the damage their insecure and impossible to secure devices cause so you won't get it.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Complain about Windows 10? No way, as long as there is Windows, my job security is guaranteed.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Intel doesn't understand what businesses want: inexpensive parts.
Intel doesn't understand what hobbyists want: inexpensive parts that don't need NDAs.
Intel doesn't understand what the world doesn't need: more power hungry x86 platforms.
Intel doesn't understand that we don't need them.
Anons need not reply. Questions end with a question mark.
One slide says "Enhanced IOs", but what kind of I/Os are there ? I hope there are GPIOs, TW, SPI, etc.
A DDOS doesn't require CPU power, but maybe you were trying to be funny.
You joke, but there is a market.
Japanese washlets are quite sophisticated, and can allow uploading of audio files for the sound masking. When you "produce", it's not uncommon to have a button you can push that generates a flushing sound, or otherwise camouflages the sound by playing another sound.
In some areas where water is a premium resource, it can also be useful to monitor the number of washes and flushes. A high number of flushes compared to washes might mean installing a dry urinal could save water. Or that a better sound for "flushing" could be useful, so users use that instead of actual flushes.
Does anyone else see a "Microsoft Azure App Service" Sponsored Content entry right beneath this one on the main page?
Man, I hate the chip market. I want to have an affordable 6 to 12 core chip with 5 to 6 GHz default clock rate, not this low-powered Internet of things crap.
I don't want overkill. I want something stable, that won't need to be encased in a cubic meter of gold/lead alloy to be protected from cosmic rays because the fab die has decreased to barely usable. Something that will last for 15+ years, while delivering enough umph, but not orders of magnitude more than I need.
My main server is a PIIIs, and as it still runs the latest software, why would I need new hardware that's less reliable? It is more than enough to handle DNS, DHCP, internal web, incoming e-mail for multiple domains, and various other services, at an average load of 0.04 (and 0.03 of that is due to incessant incoming spam, mostly from IoT botnets).
Give me reliability, not bells, whistles and turbocharging I don't need.
Many of these attacks involve taking over an IP security camera, and throwing the video stream at the DDoS target. So it's the same rate and difficult to tell the difference, except maybe by looking for the lack of ACK packets.
What Intel's definition of IoT?
This looks like a human interface device (GUI) with some hardware control capability (enhanced determinism) rather then an embedded MCU.
Perhaps sth that can be use for example on a drone for higher level functions such as command, navigation, video (CV) ?
4wdloop
I feel like you completely missed the point of this platform (embedded devices do not need shittons of CPU... ) but what, exactly, are you doing that requires 12 cores at 6GHz because your raw CPU is the bottleneck in a Desktop platform or architecture??? 'Slow desktops forever' my ass... 16G RAM, SSDs, and bus-level access are commodity product specs now. If you have Desktop software that is slow because of a CPU bottleneck, you either have shitty code, or your software should never have been designed for a desktop use case. Da fuq?
Physics is nothing like religion. If it was, we'd have an easier time trying to raise money!
I currently have a bidet seat on my toilet. A bidet is a device that washes my privates after I do my thing.
This bidet seat does have a microprocessor in it that controls both the heater for the water jet as well as the heater for the air jet used to blow dry my privates.
Right at the moment, I see no reason for this bidet seat to be on line. Perhaps there might be some reason in the future. Oh, it could be something to do with sensing that I have some sort of contagios stomach disease and can page/im DCD or NIH. Be as it may, there is enough real estate inside that bidet seat unit the accomodate either a 4G modem or a WIFI device.
Most Respectfully Yours Mark Allyn Bellingham, Washington
That's true that generating packets doesn't require a lot of cpu power.
> consider that your average 20 buck switch has a routing capacity of several Gigabits per second (on paper, at least)
Switches actually don't route, they switch. Routing is level 3, IP packets. Switching is level 2, ethernet frames. For $5,000 you can get a "level 3 switch", which is actually a router combined with a switch.
> and that *is not* running any top of the line CPU.
It is, however, running a purpose-built switching chip, which runs about $6 in small quanities. The CPU (mcu) isn't involved in switching frames at all. The cpu/mcu only gets involved when there are CHANGES to the switching rules, at which point it sends new tables to the chip that does the switching. Switching frames is completely different from generating new IP packets from scratch.
Yes, but it won't be useful. Part of the nature of the DDOS attack is that no particular host sends packets at a high rate (too easy to detect and shut down). It's the sheer number of slow devices that floods the target.
Well, how else is the webcam going to publish a stream?
Log in or piss off.
Flight simulation, audio processing with huge number of tracks and soft synths, compiling large software projects, stuff like that. It's really not hard to get to the limits of off-the-shelf PCs and I don't have the money for a 5k workstation either.
This isn't really "IoT". A home router is not IOT either. Yes it's a thing, and it's on the internet, but so is your computer and your phone. This new CPU sounds like a power hog focused on being fast, whereas I'm working on something that has to run on a non-rechargeable battery for twenty years. Intel is focusing on the consumer IoT fad I suspect.
What an IoT chip *should* have: low power and security. Meaning average only a handful of microamps, the fewer the better as it allows customers to add their own exrtras. And for security you want secure key storage (private keys never in ram) and support for crypto that allows PKIP (you need more than just AES). Too many devices seem oriented to wifi which is just an awful way to go; high power and bad security.
They should be held responsible. That means the customers must be the ones demanding those features. Sometimes the manufacturers have to inform the customers about why security is important (ie if your SCADA or utility customers think security is an afterthought then bring up the worst case scenarios and scare them into getting a better product).
Too often time to market trumps security - especially in the consumer market where fads can vanish suddenly and your market for IoT refrigerators dries up overnight.
These attacks rely on security shortcomings of said devices. Whether they can make one or a million requests per second doesn't change the game, the problem is that there are many such compromised devices, not that a single one of them is causing a lot of traffic.
The problem is actually less the amount of traffic. That amount did increase, yes, but until the IoT became a part of the attack, most of the high volume attacks were reflected DNS attacks or similar that could easily be filtered at scrubbers. You simply run your traffic through a scrubber with a fat pipe, it filters all the DNS replies and presto, instant solution. Doesn't work anymore now that these devices are not reflecting, they actually are numerous enough to run attacks that look like genuine http(s) requests. No chance to filter that. And it doesn't matter whether this single device can be limited to X requests. The problem is the number of compromised devices.
So unless you're willing to cripple the devices to the point where they become essentially useless, this is the wrong approach.
The right approach is to disable the more damaging properties of the device until they are properly set up. The very least this must consist of is a change of the default password. The current batch of attacks is mostly relying on IoT devices connected to the internet with the default password still valid. Most of them because the users never bothered to change it, but sadly there are even devices where "changing" the default password only adds another valid password to the list and the default credentials remain valid.
This is the core of the current problem.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
And now explain that to the mom that just bought a cam to monitor her baby. I'll bring the popcorn.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
12 or 18 EU configurations of Intel’s Gen9 iGPU
What is an EU in this context? Execution Unit?
The customers will not demand these features. For them, these aren't features. To them, they're at best useless, at worst a nuisance. Where's the benefit for the user if his device doesn't harm him, only harms others, and he's not responsible for it?
Yes, he should feel responsible. But people are not that way.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Well, if your customer is a utility, and they've got an angry mob outside your office wondering why all their data has been stolen... A city manager may care that the traffic lights aren't hacked. And so on. Those people ARE responsible. The home hipster though probably isn't concerned that his wifi coffee maker is being spied on, he just wants a cool gadget to prove to his friends that he's not a luddite. But if home hipster causes damage to others through his own negligence he will be held liable in a courtroom.
I'm less concerned with the home hipster, he will quickly be deconverted as soon as one of his computer savvy friends points out that the device is insecure and that he's now part of the bot herd. Because if there's one thing a hipster cannot stand it's being part of the herd.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.