Intel researchers put WiFi inside—the processor, that is(arstechnica.com)
arstechnica.com
Intel researchers put WiFi inside—the processor, that is
http://arstechnica.com/information-technology/2012/09/intel-researchers-put-wifi-inside-the-processor-that-is/
6 comments
Will CPUs access HDDs via WIFI in the near future? Will it matter if they're in the same case anymore?
I presume that you mean block level access over wifi? Surely iSCSI already works over WiFi, if a little slowly. Why would you want to do this in preference to accessing files using NFS or HTTP?
Although if you had a stupidly fast internet connection it would be possible to hold devices images in RAM centrally and may turn out to be quicker than booting from local storage.
Although if you had a stupidly fast internet connection it would be possible to hold devices images in RAM centrally and may turn out to be quicker than booting from local storage.
One reason why it's a clear cut no: latency. Maybe for file servers but not for operating system and application files. Expect always to have a few GBs accessible by wire (even if one day it is included in the SOC).
Wi-Fi is about 1/10th the speed of SATA, so no.
I think Wi-Fi will speed up faster than both SATA and disk drives, and SATA already is faster than portable hard disks.
For example, http://en.wikipedia.org/wiki/IEEE_802.11ad aims at 7Gb/s. http://www.tomshardware.com/charts/2012-mobile-hdd-charts/-0... gives max read speed for mobile disks at less than 1.5Gb/s.
Form that viewpoint, I can see a laptop disk drive that only needs a power connection. Battery-operated, you could even keep it in your bag while using it,
Big questions, however, are a) whether people will still want external drives for thei laptops. SSDs and the cloud may supplant them, b) whether the bandwidth is reliable enough (what if I sit next to someone who also has a Wi-Fi external drive?) and c) whether it is a wise idea to make it so easy to detach your disk. Software will have to be prepared to handle the case "user picks up laptop and walks out of range of external disk"
Especially for reasons a and b, I do not see this happen.
For example, http://en.wikipedia.org/wiki/IEEE_802.11ad aims at 7Gb/s. http://www.tomshardware.com/charts/2012-mobile-hdd-charts/-0... gives max read speed for mobile disks at less than 1.5Gb/s.
Form that viewpoint, I can see a laptop disk drive that only needs a power connection. Battery-operated, you could even keep it in your bag while using it,
Big questions, however, are a) whether people will still want external drives for thei laptops. SSDs and the cloud may supplant them, b) whether the bandwidth is reliable enough (what if I sit next to someone who also has a Wi-Fi external drive?) and c) whether it is a wise idea to make it so easy to detach your disk. Software will have to be prepared to handle the case "user picks up laptop and walks out of range of external disk"
Especially for reasons a and b, I do not see this happen.
I don't know the latency of SATA off the top of my head, but I would assume it would be less then that of Wi-Fi. With the way the OS currently controls hard drives, this proposal probably just won't work. Since this pretty much rules out an OS drive over Wi-Fi (lots of random reads), we could instead think of a NAS setup for mass storage. So we have a small NAS with Wi-Fi, with samba or NFS protocols, and use it as a storage drive.
Personally though, I don't mind having an internal SSD in my laptop (although it is a bit small, really should get a bigger one). I also don't mind cables for external storage drives. As for Wi-Fi, I usually have a hard time hitting the maximum bandwidth of my AP (where maximum bandwidth is slightly less then half the connection speed). Wi-Fi is also prone to interference. One interesting thing about the 802.11 ac spec is that it now supports channel bonding on upto eight channels, which severely limits the 5GHz band, which was supposed to take away all the congestion we have on 2.4GHz.
Personally though, I don't mind having an internal SSD in my laptop (although it is a bit small, really should get a bigger one). I also don't mind cables for external storage drives. As for Wi-Fi, I usually have a hard time hitting the maximum bandwidth of my AP (where maximum bandwidth is slightly less then half the connection speed). Wi-Fi is also prone to interference. One interesting thing about the 802.11 ac spec is that it now supports channel bonding on upto eight channels, which severely limits the 5GHz band, which was supposed to take away all the congestion we have on 2.4GHz.
I hadn't heard of it yet when I wrote my earlier reply, but intel demoed a WiFi solution a few days earlier: http://www.computerworld.com/s/article/9231264/Intel_demos_7...
http://techreport.com/news/23303/marvell-wilocity-partner-up... writes about it:
"We're told latency will be in the order of "microseconds, not milliseconds"
So, latency will not be much of a problem with this solution.
I do have the impression that this is called Wi-Fi only on commercial grounds, not on technical ones.
http://techreport.com/news/23303/marvell-wilocity-partner-up... writes about it:
"We're told latency will be in the order of "microseconds, not milliseconds"
So, latency will not be much of a problem with this solution.
I do have the impression that this is called Wi-Fi only on commercial grounds, not on technical ones.
Do you always need maximum speed?
Always no, often yes. Just think about copying about anything from/to HDD. Now let it take ten times as long. Game installations, torrents, everything. Currently HDD is often a bottleneck.
So we'll copy directly HDD to HDD?
Yeah, DMA makes it "not that bad" at first glance, but it's just a workaround (hdd to hdd) and you'll see that you need a workaround on workaround next to a workaround when you go that alley, instead of just getting just bandwidth.
In most regards, yes. Even at the best speeds possible, going over a wifi connection would be leagues slower.
The title is somewhat misleading. Intel developed a "processor" (digital chip) to handle WiFi. The title implies that the chip + antennas are integrated within the CPU.
Sure, in the future the digital chip may end up being integrated, but this isn't major news, in my opinion.
Are you sure? Because if you were right, I don't think Intel would use money and resources to make a big announcement that they have caught up with technology that has been around for 13 years. From all other news and PR releases I've read, the wifi transceiver is integrated on the intel cpu dye. I believe you still need an antenna, but at these frequencies those are quite small (3mmx3mm inverted f).
That's exactly what they did. That's what SOC means. Basically instead of having to have a separate chip on the board for the radio, it's within the processor. Of course the board still needs an antenna, but this is just another step towards x86 cell phones.
This is basically Intel announcing that they are ready to compete in the mobile phone/tablet SoC market. The other vendors already had integrated wifi, 3G/LTE, but used ARM processors, locking out Intel.
I think that other vendors are using two chips by putting them inside one package. I don't think anybody else made the cpu and rf sections on the chip yet. also nice is that the rf is made using digital technology which benefits from moore's law.
RF and logic on the same chip is not new. See http://www.fwdconcepts.com/dsp032112.htm .
tldr: more efficient WIFI = longer battery life
I think maybe some of you missed the part where they say:
While the back-end elements of wireless have been digitized long ago, the front end—phase modulation, frequency synthesis, and RF power amplification, for example—have largely been dependent on analog components.
The problem posed by those analog components is that while digital components can be scaled down with improvements in silicon die manufacturing, the analog parts can't—as they get smaller, they get worse, Ratter said. Yorgos Palaskas, the research leader in Intel's radio integration lab, said that because analog components generally performed much better when manufactured on a larger scale, the analog components for WiFi transceivers and cell phones "are typically made on a separate fab."
...
Some of the components already existed in a digital form, but needed to be significantly improved. Intel had digital phase modulators already that were developed for satellite and mobile communications, but they only handled enough frequency channels for 3G communications. "We needed much wider channels for WiFi, up to 40 MHz of bandwidth," Palaskas said. "It required some very creative mathematical manipulation."
I think maybe some of you missed the part where they say:
While the back-end elements of wireless have been digitized long ago, the front end—phase modulation, frequency synthesis, and RF power amplification, for example—have largely been dependent on analog components.
The problem posed by those analog components is that while digital components can be scaled down with improvements in silicon die manufacturing, the analog parts can't—as they get smaller, they get worse, Ratter said. Yorgos Palaskas, the research leader in Intel's radio integration lab, said that because analog components generally performed much better when manufactured on a larger scale, the analog components for WiFi transceivers and cell phones "are typically made on a separate fab."
...
Some of the components already existed in a digital form, but needed to be significantly improved. Intel had digital phase modulators already that were developed for satellite and mobile communications, but they only handled enough frequency channels for 3G communications. "We needed much wider channels for WiFi, up to 40 MHz of bandwidth," Palaskas said. "It required some very creative mathematical manipulation."
Hopefully it'll be linux-compatible...