Leaked Slide details Intel's Lynx Point Chipset
With Haswell set to release in 2013, we continue to learn more about Intel's new platform with leaked information on the Lynx Point chipset.
Ivy Bridge isn't set for release until April, but we have already begun to see leaked information on its successor, Haswell, which is set for release in 2013. In a recently leaked slide from Chinese website Zol.com.cn, we get an early overview of the Lynx Point chipset, or the Intel 8-series, as it'll be known at final release.
As we have learned, Haswell will be on the new LGA 1150 socket and is set to be released around the March to June 2013 time-frame. Now, let's take a little look at the chipset associated with Haswell. First, we see some of the display port features have been transferred over to the CPU under the new architecture for Haswell. As discussed here, the Lynx Point chipset will only handle analog displays, while the CPU will be handling the digital displays. In theory, this makes the Lynx Point not as advanced as Panther Point in that regard but it does still have its advancements.
Lynx Point marks Intel's switch to full SATA 6.0 Gb/s support for all native SATA ports (no more native SATA 3.0 Gb/s SATA ports). Intel has added two additional USB 3.0 ports for a total six ports, which is two more than Panther Point. Intel is introducing I/O port flexibility feature with Lynx Point. As the feature suggests, it should give users the ability to allocate which ports are USB 3.0 and which are USB 2.0. Additional features for Lynx Point include: SSD performance/power optimization, new Rapid Storage Technology driver & Intel vPro Technology enhancements, lower TDP and power usage, a smaller package (22x23mm FCBGA), and support for quad-read SPI devices. Based on the slide, it looks like Intel has still not moved to a faster DMI interface, as the chipset's PCIe ports are still running at PCIe 2.0.
It does look like the new LGA 1150 socket will follow suit like the LGA 1155 socket and last two CPU generations. It is expected that both Haswell and Broadwell, the 14 nm shrink of Haswell, will be on the LGA 1150 socket. Please keep in mind, of course, that these specifications are from a leaked source. We won't know for sure until Intel shares official information. Stay tuned!
Update: After reviewing some of the comments and feedback received on PCIe 2.0 vs. PCIe 3.0 for Lynx Point, I felt it was necessary to clarify my statement on the chipset's PCIe ports still running at PCIe 2.0.
First and foremost, Haswell / Lynx Point will support PCIe 3.0 just like Ivy Bridge will in coming months. If you look at the leaked slide in the Intel Haswell iGPU to support DirectX 11.1, OpenGL 3.2 news story, you'll see that the Haswell processor supports PCIe 3.0 on the CPU, not through the chipset. This was a move made by Intel to remove the bottleneck seen with the DMI interface by transferring PCIe support to the CPU. With Lynx Point, Intel still has not been able to solve the bottleneck issue with the DMI interface, so the PCIe ports running off the chipset (not the CPU) will still be running at PCIe 2.0. The ports running of the chipset are typically PCI ports, PCIe x1 ports, or USB 3.0/2.0 ports for example.
- Reports of Counterfeit Parts Increase 4X Since 2009
- Specialty DRAM in Tight Supply, Prices Likely to Climb
- Todd Howard Explains Morrowind and Cyrodiil in Skyrim
- Deals Feb. 15: 24'' Planar 1080p LCD Monitor $139.99 FS
- NASA Shuts Down Last Mainframe, Signals End of an Era
- ViewSonic Releasing Windows 8 Tablet in 2Q12? No.
- Firefox to Get a New Interface in Second Half of 2012
- SanDisk Intros Extreme and X100 Lines of SSDs
- Microsoft Rumored to Change Windows Logo in Windows 8
- Deals Feb. 16: Dell Inspiron 15 Core i3 2.66GHz Laptop $399
- Fair Labor Association Says Apple Factories Are ''First Class''
- Razer Producing Mass Effect 3-Themed PC, X360 Gear
- Nvidia Reports 17 Percent in Profits in Latest Financial Report
- Micron/Elpida Could Challenge Samsung's DRAM Dominance
- Apple Gives Mac OS X a Dose of iOS with Mountain Lion
- SSDs Have Bleak Future, Says Researchers
- Sony's PS Vita Gets the Teardown Treatment
- HP CEO: Android Will Become Closed Source
Is it *really* necessary to change sockets every other year? LGA 1156, LGA 1155, now LGA 1150?
:|
Wait, PCI2? After Ivy Bridge is supposed to have PCI3 this year?
Remember how long intel had socket 478, early-late P4 and Celerons, then Socket 775 for Pentium 4/Celeron/Pentium-D/Pentium Dual-Core/Core 2 Duo/Core 2 Quad. I'm surprised they have been changing so much. You used to be able to actually upgrade your CPU, now you have to buy a new board.
Is it *really* necessary to change sockets every other year? LGA 1156, LGA 1155, now LGA 1150? :|
Yes if you want awesome new features that are not compatible with older chip generations. I find it funny that so many people complain about this. How often do you upgrade the CPU and not the MB? I know that I tend to always upgrade the 2 at the same time.....
Yes if you want awesome new features that are not compatible with older chip generations. I find it funny that so many people complain about this. How often do you upgrade the CPU and not the MB? I know that I tend to always upgrade the 2 at the same time.....
The point he was making was that there's no choice in that matter. How many times did you have the choice to keep your older motherboard when you switched chips?
These days? Never, because it is not a real option.
"I tend to always upgrade the 2 at the same time.....". AMD saved my backside twice on this front in 10 years, and Intel once. So in 10 years, I saved on 3 mobo purchases whereas you would of purchased new ones for features. I'll drink to the $100+ x 3 mobo's I saved over you, sir. IMHDO(drunken) Good day!
Yes if you want awesome new features that are not compatible with older chip generations. I find it funny that so many people complain about this. How often do you upgrade the CPU and not the MB? I know that I tend to always upgrade the 2 at the same time.....
You clearly know nothing about chip design. There was little reason for 156-55, and no reason for 55-50, besides Intel padding their profits. Intel is still using the same basic bus it introduced with the first Core i7/i5/i3.
Yes if you want awesome new features that are not compatible with older chip generations. I find it funny that so many people complain about this. How often do you upgrade the CPU and not the MB? I know that I tend to always upgrade the 2 at the same time.....
Thats is BS you can put the same new features on an existing socke for factor, you don't need to a scoket change every other year to do that, all you need is a chipset update.
Now its AMD's turn to leak something
Is it *really* necessary to change sockets every other year? LGA 1156, LGA 1155, now LGA 1150? :|
It's business, and praticality. If you get a new chip then you need the board that supports all of it's features in order for the end user to have the fullest experience, changing the pinout forces that. Plus you want different pinouts for different product levels so that people will buy the more expensive products... even if you can OC the cheaper products to perform at a similar level. And lastly, it is good business relations for mobo manufacturers. When the manufacturers know that people will be buying more products, because they will be upgrading more often, then they are going to be more willing to put out better/more interesting products for your chips, and push your chips more than 'the other guy'.
The point he was making was that there's no choice in that matter. How many times did you have the choice to keep your older motherboard when you switched chips?
On my first 775 board, I went from a E6400, to an E6600, to a Q6600.
Remember how long intel had socket 478, early-late P4 and Celerons, then Socket 775 for Pentium 4/Celeron/Pentium-D/Pentium Dual-Core/Core 2 Duo/Core 2 Quad. I'm surprised they have been changing so much. You used to be able to actually upgrade your CPU, now you have to buy a new board.
Thats BS 478 might have the same ammount of pins but you cant run say a prescott chip in an early board, or a wilamette in the last gen boards, same as 775 -- a core 2 Q6600 prime example all the core 2 supported boards (early 965 chipsetset based boards) couldnt use it, P4 775 boards wouldnt touch core 2 chips etc.
second, why use a rubbish prehistoric board with new high end modern cpu? they revise it and change the socket (most of the time) so they can improve things etc
well
Wait, PCI2? After Ivy Bridge is supposed to have PCI3 this year?
There is the PCIe that goes to your graphics card which in the latest chipsets is directly connected to your CPU, and then there is the chipset PCIe that can be used as additional slots (PCI, PCIe x1, x4, etc...) as well as expanding additional IO onto the board with. For example because there is no current USB3 support, chipset PCIe bandwidth is used to add USB3 ports to your board. Also in current boards although you have PCIe v2 for graphics, you have PCIe v1 for your chipset connectivity. In Ivy Bridge (or maybe it already happened in Sandy?) that chipset bandwidth was upgraded to PCIe v2. So basically it won't upgrade in Haswell either.
My notation is I don't see a path to a overall processor speed increase. Mores Law is dead though some still use it without knowing what it really means that speed doubles every 18 months, that's quit happening. A new law has come about and I don't remember the name but every 18 months the transistors on the die doubles basically.
Personally I like to be able to just upgrade the chip. To me its easier, and I don't feel the need to always have the latest and the greatest. An improvement in speed and performance is all I usually wan.t
My notation is I don't see a path to a overall processor speed increase. Mores Law is dead though some still use it without knowing what it really means that speed doubles every 18 months, that's quit happening. A new law has come about and I don't remember the name but every 18 months the transistors on the die doubles basically.
You might want to take another look at Moores Law. It states "number of transistors that can be placed inexpensively on an integrated circuit doubles approximately every two years." Not that frequency, and/or speed, and/or performance increases every 2 years or 18 months. So, this is not the "new law" but the performance/speed, etc doubling every 18 months was stated after Moores Law, by David House, an Intel executive. Read all about it at Wikipedia by searching on: Moores Law
love being voted down for a valid point
THERE IS NO SUCH THING AS AN UPGRADE PATH
I agree with changing sockets - to prevent all the people buying a new cpu and trying to put it on a board it will not run on - I can not tell you how many people with an old HP or Dell who saw it had a 775, (originally a Prescot or Pentium D, or celeron) and bought a quad core 9xxx on ebay or some place, and then got pissed at me for trying to tell them it would not run, and they just smoked something trying..
make the socket for only those CPU that work on it, and be done.
love being voted down for a valid pointTHERE IS NO SUCH THING AS AN UPGRADE PATH
You didn't read this in Moores Law you only quoted what you wanted.
"(being a combination of the effect of more transistors and them being faster)."
The transistors are not being faster. Hence no speed doubling 18 months or 2 years of the processor.
"However, the 2010 update to the International Technology Roadmap for Semiconductors has growth slowing at the end of 2013, after which time transistor counts and densities are to double only every 3 years."
"In multi-core CPUs, the higher transistor density does not greatly increase speed on many consumer applications that are not parallelized. There are cases where a roughly 45% increase in processor transistors have translated to roughly 10–20% increase in processing power."
You didn't read this in Moores Law you only quoted what you wanted. "(being a combination of the effect of more transistors and them being faster)."The transistors are not being faster. Hence no speed doubling 18 months or 2 years of the processor."However, the 2010 update to the International Technology Roadmap for Semiconductors has growth slowing at the end of 2013, after which time transistor counts and densities are to double only every 3 years.""In multi-core CPUs, the higher transistor density does not greatly increase speed on many consumer applications that are not parallelized. There are cases where a roughly 45% increase in processor transistors have translated to roughly 10–20% increase in processing power."
you didnt read what i responded to
Remember how long intel had socket 478, early-late P4 and Celerons, then Socket 775 for Pentium 4/Celeron/Pentium-D/Pentium Dual-Core/Core 2 Duo/Core 2 Quad. I'm surprised they have been changing so much. You used to be able to actually upgrade your CPU, now you have to buy a new board.
Thats BS 478 might have the same ammount of pins but you cant run say a prescott chip in an early board, or a wilamette in the last gen boards, same as 775 -- a core 2 Q6600 prime example all the core 2 supported boards (early 965 chipsetset based boards) couldnt use it, P4 775 boards wouldnt touch core 2 chips etc.second, why use a rubbish prehistoric board with new high end modern cpu? they revise it and change the socket (most of the time) so they can improve things etc
Thats is BS you can put the same new features on an existing socke for factor, you don't need to a scoket change every other year to do that, all you need is a chipset update.
this is what AMD does -- look how far they get in performance and efficiency.
When it comes to the upgrades, it's all a matter of business.
AMD allows a customer to upgrade there cpu on the last generation mob and then allows the customer to upgrade the mob when they choose for better features and performance. This is because AMD target customers that are DIY, budget minded people who are generally only going to be able to afford one at a time or are tight fisted and don't want to spend that much up front.
Intel on the other hand is not a budget cpu (although their prices are starting to come down); with this in mind Intel wants the best features and performance out of every new chip and to do that you need to upgrade mob and cpu togather. Since Intels target customer is the professional and home user (who will general replace a whole computer at time not a cpu at a time) they probly will not be changing this model anytime soon.
I agree with apache-lives on
this is what AMD does -- look how far they get in performance and efficiency.
but still it is annoying to have to buy a new board each cpu upgrade even though the newer socket is part of the performance improvement some how with the last 2 or 3 they got the less pins to give out a higher performance.
I agree with apache-lives on but still it is annoying to have to buy a new board each cpu upgrade even though the newer socket is part of the performance improvement some how with the last 2 or 3 they got the less pins to give out a higher performance.
Why are you upgrading your cpu so often then? I don't get it.
A year ago i bought my i7 2600k rig, a year later here now its still top end cause i spent a bit extra and in ~2 years ill probably end up buying a whole new rig once its obsolete -- i don't want to use an old crappy motherboard that's 3 years old to bring down my new rig (if it would even fit) why would i do that?