http://www.engadget.com/2010/05/04/i...bps-speeds-ar/

http://www.pcworld.com/article/19422...cceed_usb.html

http://gizmodo.com/5377178/light-pea...apple-products

Pretty cool stuff on the horizon.
Im seriously putting off building my next computer because Id like to get in on this technology, as it is significantly faster than SATA.
Hopefully my hopes wont be smashed by prices or other unforeseen issues.

Im curious what some of you hardcore computer guys think of this, Im just an enthusiast, but it seems transferring data between drives is probably a serious bottle neck on disk access.

My thoughts are:
Light Peak + Solid state HD = epic win

 

it means less than you think. signalling is easy. cheap, low latency, and massive throughput is the problem. consumer SSD's are just barely reaching the upper bounds of sata 2. hdd's are limited by their rotational speed and arm speed, but for price per density they're awesome. Gotta remember how USB/Firewire got their start - as simplified, cross-platform, dynamically installed devices. USB 1.1 was what, 11Mbps iirc? And most USB devices today (cameras, audio, mice, keyboards, wireless) barely push the bounds of USB throughput.

If you want to see an awesome implementation of SSD, look up "sandforce". They have an awesome hack as well, in the form of a pcie card, that acts as a pci-x bridge, then uses a *cheap* pcix sata raid controller to go to 2 ssd banks, acting as an all in one, very cheap, pcie-interface raided SSD.

http://www.anandtech.com/show/3788/o...dable-pcie-ssd

 

What the **** does that mean?

 

Nah, I'll stick with SATA 3.0, since I think it will be half a decade or even a full decade or more before HDD technology is able to even touch the upper limits of 3.0, being 6 Gb/s. My HDD currently on SATA 2.0 is not even close to taxing.

 

Agree with above. I have over 5TB of data on my network and even if my HDD could read/write at 10Gbps (~1.25GBs), to utilize that transfer speed, as a home user I don't really need to transfer anything over 10-12GB at a time. So I'm not too excited about moving large files faster.

 

Eh... 10Gb/s is not going to come cheap. Problem is that you would probably still use Windows on your system with some crappy *** drivers that will artificially limit the throughput of the system to like 1Gb/s (which is still fast as **** but definitely not 10Gb/s)

I run several file servers in our data center that are based on Atom D510 and 6 500GB WD RE3 drives in RAID 10. I see 200Mb/s and enough IOPS to sustain ~10 VM servers per each file server (5-14 VMs per server, all use NFS/gluster as main storage platform.) Total cost with drives is about $800. How much is that 10Gb/s thing going to cost you?

Oh, and the fastest SSD I've played with personally only went up to ~140Mb/s on reads and still cost me ******* $100 for a 30GB version (OCZ Vortex).

Also, what's wrong with using Infiniband for fast data transfer? Technology has existed for a while and is widely implemented.

 

In for pricing.

 

*cough*50gb/s*cough*

http://www.techspot.com/news/39790-i...transfers.html

 

Hmmm
I assumed that the whole point in them doing this was that it was.. you know.. NEEDED
But I guess not yet
Although, the way Intel is backing it its sounding like USB 3.0 is never going to happen, so this is probably going to be a good platform for future growth in hardware.

 

usb 3.0 has already happened, intel is just holding back making chipsets for some reason, probably because of lightpeak