Intel to segregate Mainstream and Enthusiast Platforms even more with Sandy Bridge

[scooterlibby]

I don't really want a 6-core. I was just hoping for a pumped 32nm quad (from either company) at the beginning of next year that would offer a good gain over C2Q. Nehalem just didn't seem worth it to me (in any incarnation) as a gamer.

 

[VirtualLarry]

I still don't see lightpeak as a panacea. You are still going to want all of those connections simultaniously, PS/2 keyboard, USB mouse, VGA/DVI/HDMI, LAN, audio, and whatever additional accesories.

So does that mean that the mobo will have to have 5 lightpeak connectors on it, as a minimum, or does that mean that the mobo will have one lightpeak connection on it, and then we will have an ugly breakout box, with all of the "legacy" connectors on it, ala laptop docking stations/breakout boxes.

 

[Modelworks]

I still don't see lightpeak as a panacea. You are still going to want all of those connections simultaniously, PS/2 keyboard, USB mouse, VGA/DVI/HDMI, LAN, audio, and whatever additional accesories.

So does that mean that the mobo will have to have 5 lightpeak connectors on it, as a minimum, or does that mean that the mobo will have one lightpeak connection on it, and then we will have an ugly breakout box, with all of the "legacy" connectors on it, ala laptop docking stations/breakout boxes.

Probably breakout boxes in the interim. Lightpeak like everything will take time, probably 10 years before it is a common interface . Running many devices over one fiber isn't a problem. Intel demoed it streaming a blu-ray at full resolution to the monitor, doing a file copy, interfacing with keyboard and peripherals, and it was using only 1 fiber for all of them sharing the fiber with zero performance hit. It starts at 10Gbps both ways simultaneously and can scale to 100Gbps so plenty of room to grow.

I think the thing people are forgetting is that while USB3 is faster then USB2, there isn't much use for it in most devices that use USB2. A keyboard /mouse/ phone has no need for anything faster because it doesn't use the capacity of USB2. USB3 really is a competitor for ESATA , everything else is already handled fine by USB2 . USB2 now is so cheap to implement that most manufacturers are not going to up the production cost just so they can claim USB3 support if it adds nothing to the device functionality.

 

[VirtualLarry]

USB2 now is so cheap to implement that most manufacturers are not going to up the production cost just so they can claim USB3 support if it adds nothing to the device functionality.

But if there is native USB3 support in the system chipset, it doesn't really up the production cost at all. Certainly far less than including a seperate NEC controller chip on the board.

 

[VirtualLarry]

Intel demoed it streaming a blu-ray at full resolution to the monitor, doing a file copy, interfacing with keyboard and peripherals, and it was using only 1 fiber for all of them sharing the fiber with zero performance hit. It starts at 10Gbps both ways simultaneously and can scale to 100Gbps so plenty of room to grow.

How did it physically connect, if there was only one fiber? Daisy-chained like firewire?

 

[ZimZum]

Board mounted lasers are now as cheap as copper connections, fiber optic cables are now as cheap or cheaper than copper in most cases.

Toslink has been around for 30 years, and yet a small Toslink optical transmitter is still going to cost you more than an SATA port. And Toslink is a fairly simple device that only transmits audio. Lightpeak is a great deal more complicated and I just dont see how it will be able to compete with the options that cost a fraction as much to implement. The cost differences may not seem like a lot but they matter to manufacturers especially when your talking putting a dozen or so lightpeak ICs on a MObo.

 

[ExcaliburMM]

I think I, and many other 920 owners, will be holding onto our setups for a couple of years. We'll see what Intel and AMD can offer me Q2 2012.

 

[Modelworks]

How did it physically connect, if there was only one fiber? Daisy-chained like firewire?

It has end connectors that are very similar to usb connectors only smaller. So all the devices can plug into a device like a usb hub then from the hub 1 fiber carries that data back to the controller. Inside the USB like cable is a fiber and the outer wall of it is a copper foil covered with rubber to protect the fiber. The connections inside the hub device are passive optical , sort of like those toslink splitters that use mirrors, so it is pure light all the way back to the controller. So someone could have many devices connected as long as you could keep the light levels correct.

 

[OCGuy]

......gpu's and a new ssd seems to be a much better upgrades these days..

Exactly. If you are using s775 and arent on a SSD, you should think about that WAY before upgrading your CPU.

 

[Modelworks]

Lightpeak is a great deal more complicated and I just dont see how it will be able to compete with the options that cost a fraction as much to implement. The cost differences may not seem like a lot but they matter to manufacturers especially when your talking putting a dozen or so lightpeak ICs on a MObo.

http://electronicdesign.com/article...ransceiver_targets_light_peak_technology.aspx

Optical-fiber interfaces are the fastest but still expensive, especially for computers. Mountable on a printed-circuit board (PCB), Luxtera’s OptoPHY optical transceiver paves the way to breaking the one dollar per Gigabit/s cost barrier for faster interfaces in the near future. The OptoPHY transceivers use Luxtera’s CMOS process, which integrates the optics and electronics on a single chip.
The LUX6001 is single-channel transceiver that can operate at data rates from 1 to 10.3125 Gbits/s full duplex. It uses a 1490-nm laser and mates up with standard single-mode fiber. Its maximum reach is 4000 m, and its link bit error rate is 10–15. Its signal latency is only 1 ns for both transmit and receive. And, its short pigtail fiber cable lets users place the devices anywhere on the PCB, eliminating the need to edge-mount the device (see the figure).
Key applications include 10-Gbit/s Ethernet, optical backplanes, system interconnects, rack interconnects, and board interconnects. The transceiver is a good choice for 2- to 10-Gbit/s Fibre Channel storage-area network (SAN) equipment. Also available, the LUX6004 includes four 10-Gbit/s transceivers and provides up to 40-Gbit/s throughput with the same specifications as the single-channel unit.
The LUX6001 and LUX6004 are scheduled for production in mid-2010. A 12-channel unit is also expected next year.

Cost is around $15 per module . Down from the intel $39 price just 6 months ago.

 

[ZimZum]

Cost is around $15 per module . Down from the intel $39 price just 6 months ago.

Lets go one step further and cut that in half. So $7.50 per module. If you intend these to replace SATA and USB ports lets figure 10 for an average mobo (6 internal 4 external). Now you're up to $75.00 just on ICs alone, compare that to the same amount of SATA ICs that cost around $1 a piece. I just dont see this solution being able to compete on price. I could see maybe 1 or 2 ports. But not as a ubiquitous replacement for everything. And as such it becomes just another competing standard in an already crowded field.

 

[Modelworks]

Lets go one step further and cut that in half. So $7.50 per module. If you intend these to replace SATA and USB ports lets figure 10 for an average mobo (6 internal 4 external). Now you're up to $75.00 just on ICs alone, compare that to the same amount of SATA ICs that cost around $1 a piece. I just dont see this solution being able to compete on price. I could see maybe 1 or 2 ports. But not as a ubiquitous replacement for everything. And as such it becomes just another competing standard in an already crowded field.

You don't need 10 IC . One IC can handle multiple protocols at the same time. For example a mobile phone can have one port. That port can do ethernet, external monitor, keyboard over the one fiber where it now requires wireless/ethernet cable , video cable, usb connector to keyboard or mouse or the use of some sort of dock.

For motherboards you could use one IC for video ,sound, data and a second one for external connections. Think of it like a hub where the interface is 10Gbps to the motherboard for every IC installed and as many devices as the capacity can handle can use that hub without degredation. You don't need an IC for every device, just one for every 10Gbps of capacity.

 

[pjkenned]

Make me a $15/1gbps Ethernet adapter with $50 in misc IC's/ Parts, and I'll buy a bunch. I'd love to have that bandwidth for my clients and servers. Not that sticking 2x Intel Pro/1000 PT Quads onto a motherboard with 2 onboard Intel and 1 management NIC doesn't give you decent bandwidth, but not dealing with link aggregation would be great.

 

[Rezist]

I guess I'll get a 1356 board when these come out.

 

[Makaveli]

Since I just built my 920 rig in oct 09, I will probably be skipping sandy and waiting for the next refresh or whatever amd is offering 2012-2013. I should easily get 3-4 years out of my current rig with only GPU upgrades.

As for it not having USB 3.0 don't care I use ESATA!

 

[rahvin]

Make me a $15/1gbps Ethernet adapter with $50 in misc IC's/ Parts, and I'll buy a bunch. I'd love to have that bandwidth for my clients and servers. Not that sticking 2x Intel Pro/1000 PT Quads onto a motherboard with 2 onboard Intel and 1 management NIC doesn't give you decent bandwidth, but not dealing with link aggregation would be great.

The beauty of the optical interface is you don't have to worry about electrical interference. Most copper connections these days are limited to 100m at the very longest due to electrical interference from radio signals, tv's, microwaves, cell phones, cordless phones, and the earths magnetic field. Light peak tops out at 4000m. This makes it a network engineers wet dream, no repeaters are required in long distance installations.

USB was developed in the late 90's and the initial meetings took place in the mid 90's. It was designed at the time to replace a very limited number of ports on MB's. Since it was developed though Firewire, SATA, HDMI, DVI, and recently displayport and dozens of other high bandwidth ports were invented to compensate for the limited bandwidth of USB.

Light peak will remove all those limitations. Even today a reliable and inexpensive 10G network interface is non-existent nearly a decade after it was proposed and even today there are still at least 5 competing standards using copper and fiber. Even if lightpeak is unsuccessful at replacing all the ports initially if anything they will succeed very quickly in network integration. It has has already been mentioned that if they can get chips and interfaces down to under $10 and the cables are cheap enough, a fiber network interface will finally be here in the retail consumer market for under $1000. With the protocol carrying capacity it then becomes trivial to start using the fiber for everything else. Much like USB this is going to take a decade or more to become ubiquitous, but is the future and I think its pretty damn good with 10Gbs to start and easily upped to 100Gbs and farther down the road without problem with multiphase lasers.

 

[Dadofamunky]

intel paid them a billion dollars for their part in amd's inability to leverage P4's craptasticness.

AMD did the best they could, considering the market collusion.

That LGA1155 is what looks craptastic to me. The only interesting thing is the GPU directly on the CPU die. Interesting to see how that behaves. But others are right, this seems to leave a big opening for AMD. And the way AMD has been going lately, I think they have a chance to make up some ground.

 

[Dadofamunky]

Exactly. If you are using s775 and arent on a SSD, you should think about that WAY before upgrading your CPU.

Just saw that, that's also good advice. Unfortunately I need a fairly beefy one and that's outta my price range still... I admit, I'm still pretty happy with my cheap RAID0 rig.

 

[Golgatha]

Since I just built my 920 rig in oct 09, I will probably be skipping sandy and waiting for the next refresh or whatever amd is offering 2012-2013. I should easily get 3-4 years out of my current rig with only GPU upgrades.

As for it not having USB 3.0 don't care I use ESATA!

I agree about not upgrading my platform for awhile, but I disagree on the eSATA comment. I said the same thing, but there is a lot of convenience in being able to plug in my laptop's external HDD to a single port (instead of data + power) and not having to reboot to attach an eSATA HDD. Not to mention that in a pinch, the USB 3.0 port is backwards compatible with all the USB 2.0 ports floating around. Most people don't have eSATA ports, but nearly everyone has USB 2.0 ports on their machine. I just installed an ASUS U3S6 card last night and I absolutely love USB 3.0 now. Works exactly like USB 2.0 external drives but connects with all the speed the HDD is capable of. Getting 118 MB/sec from my 2TB WD Green drive.

 

[Anubis]

Since I just built my 920 rig in oct 09, I will probably be skipping sandy and waiting for the next refresh or whatever amd is offering 2012-2013. I should easily get 3-4 years out of my current rig with only GPU upgrades.

As for it not having USB 3.0 don't care I use ESATA!

heh i feel the same way an i have a USB 3 mobo, SSDs and vid card upgrades woo woo

now i may pick up a 6 core CPU if intel makes a cheaper one then the 980. but i have my doubts

 

[RussianSensation]

now i may pick up a 6 core CPU if intel makes a cheaper one then the 980. but i have my doubts

http://techreport.com/discussions.x/18800

But I doubt it'll be < $800

 

[AdamK47]

It looks like the X58 may live longer than the 440BX did. I had a few of those boards with several CPUs starting from the Celeron 300A to the Pentium 3 Coppermines.

 

[Munky]

LOL @ all the peeps waiting for a "real" upgrade from core2.

 

[Anubis]

LOL @ all the peeps waiting for a "real" upgrade from core2.

going from a core2 to an i7 sure felt like a real upgrade to me

 

[yh125d]

LOL @ all the peeps waiting for a "real" upgrade from core2.

Yeh, I finally decided to upgrade from penryn to lynnfield, instead of waiting for SB. With my i5 750 in the mail I shouldn't need to upgrade til the SB/BD refreshes instead of soon after release (think upgrade to penryn instead of conroe)