IDF 2011: Knights Ferry Shown 8-Deep Running Ray Tracing

Subject: Graphics Cards, Processors, Shows and Expos | September 15, 2011 - 06:17 PM |
Tagged: ray tracing, knights ferry, idf 2011, idf

Very few things impress like a collection of 256 processor cores in a box.  But that is exactly what we saw on our last visit to the floor at the Intel Developer Forum this year when I stopped by to visit friend-of-the-site Daniel Pohl to discuss updates to the ray tracing research he has been doing for many years now.  This is what he showed us:

rt01.jpg

What you see there is a dual-Xeon server running a set of 8 (!!) Knights Ferry many-core processor discrete cards.  Each card holds a chip with 32 Intel Architecture cores running at 1.2 GHz on it and each core can handle 4 threads for a total of 1024 threads in flight at any given time!  Keep in mind these are all modified x86 cores with support for 16-bit wide vector processing so they are pumping through a LOT of FLOPS.  Pohl did note that only 31-32 of the cores are actually doing ray tracing at any given time though as they reserve a couple for scheduling tasks, operating system interaction, etc.

rt02.jpg

Each of the the eight cards in the system is using a pair of 6-pin PCIe power connectors and they are jammed in there pretty tight.  Pohl noted this was the only case they could find that would fit 8 dual-slot add-in cards into it so I'll take a note of that for when I build my own system around them.  Of course there are no display outputs on the Knights Ferry cards as they were never really turned into GPUs in the traditional sense.  They are essentially development and research for exascale computing and HPC workloads for servers though the plan is to bring the power to consumers eventually.

rt03.jpg

To run the demo the Knights Ferry ray tracing server was communicating over a Gigabit Ethernet connection to this workstation that was running game processing, interaction processing and more and passed off data about the movements of the camera and objects in the ray traced world to the server.  The eight Knights Ferry cards then render the frame, the Xeon CPUs compress the image (8:1 using a standard Direct 3D format) and send the data across the network.  All of this happens in real time with basically no latency issues when compared to direct PC gaming. 

rt04.jpg

While the ray tracing game engine projects might seem a little less exciting since the demise of Larrabee, Pohl and his team have been spending a lot of time on learning how to take advantage of the x86 cores available.  The Wolfenstein demo we have seen in past events has been improved to add things like HDR lighting, anti-aliasing and more.

rt05.jpg

Though these features have obviously been around in rasterization based solutions for quite a long time, the demo was meant to showcase the fact that ray tracing doesn't inherently have difficulty performing those kinds of tasks as long as the processing power is there and alotted to it. 

rt06.jpg

I am glad to see the ray tracing research continuing at Intel as I think that in the long-term future, that is the route that gaming and other graphics-based applications will be rendering.  And I am not alone - id Software founder and Doom/Quake creator John Carmack agreed in a recent interview we held with him

Source: PCPer

IDF 2011: Intel Many Integrated Core (MIC) Knights Corner

Subject: Processors, Shows and Expos | September 15, 2011 - 01:54 PM |
Tagged: idf, idf 2011, knights ferry, knights corner, mic, terascale

During Justin Rattner's closing keynote at the Intel Developer Forum he discussed the pending changes to the Many Integrated Core Architectures (MIC) that we previously knew as the Terascale projects.  While we have heard about the Knights Ferry component for some time, and it was basically used a software development platform for Intel's many-core initiative. 

02.jpg

Impressive to see at this stage, the upcoming Knights Corner product will actually be built on the new 22nm tri-gate transistors and with more than 50 cores.  They haven't posted more details on what exactly ">50" refers to but it does mean that Intel continues to progress down this path and is going to be pushing the terascale computing projects into the future. 

Rattner also indicated that not all of the cores on the many-core projects have to be identical and we will soon see designs that combine more than the x86 processors to make for truly heterogeneous computing platforms. 

03.jpg

Research into the program continues including things like stacked and shared memory, new communications protocols like optical interconnects, etc.  We are just as eager to see the fruits of this research as we were for its application to gaming and graphics that eventually failed.

Source: PCPer

IDF 2011: Live Blog of Keynotes, Technical Discussions

Subject: Editorial, General Tech, Processors, Chipsets, Systems, Storage, Mobile, Shows and Expos | September 15, 2011 - 12:15 PM |
Tagged: live blog, Intel, idf 2011, idf

PC Perspective is all over the 2011 Intel Developer Forum and we'll be covering it LIVE here all week.  Expect to hear news about Ivy Bridge, Sandy Bridge-E, SSDs, X79 chipsets, 22nm tri-gate transistors and more!  We will have specific news posts about the major topics but if you want to keep up with our information to the minute, then you'll want to migrate to this page throughout Tuesday, Wednesay and Thursday morning.  

You can also hit up http://www.pcper.com/category/tags/idf to see all of the posts relating to and coming from IDF this week!

Feel free to leave comments for me on what exactly you want to know and I will do my best to address your questions as the day progresses. 

Source: PCPer

Bloggers and techies descend on the IDF

Subject: General Tech | September 14, 2011 - 01:36 PM |
Tagged: Intel, idf, idf 2011

Ryan wasn't the only one madly recording the Intel Developers Forum keynote address by Mooly Eden, The Tech Report was also there.  Drop by their record of the live blog that they created here, complete with pictures from a different angle than Ryan's and with different content in some cases. There is even a hacker ninja!

TR_haswell.jpg

"Our own Scott Wasson and Geoff Gasior live blogged Mooly Eden's keynote (complete with pictures) at the Intel Developer Forum this morning. The keynote centered on Intel's mobile endeavors, including Windows 8 tablets and Ivy Bridge-powered ultrabooks. Eden also gave a sneak preview of Intel's next-gen Haswell processors, which will succeed Ivy Bridge in 2013."

Here is some more Tech News from around the web:

Tech Talk

 

IDF 2011: New Ivy Bridge Details from Mooly Eden Keynote

Subject: Editorial, General Tech, Processors, Shows and Expos | September 14, 2011 - 01:25 PM |
Tagged: mooly eden, Ivy Bridge, idf 2011, idf

Today is day 2 at the Intel Developer Forum and with the first keynote out of the way, we can share a few short details about Ivy Bridge that we didn't know before.  First, the transistor count is 1.48 billion - a hefty jump over Sandy Bridge that had less than 1 billion.

m05.jpg

There was also mention of a new power management feature that will allow interrupts from other hardware devices to go to other cores than Core0, which it had ALWAYS done in the past. This means that it can route it to a core that is already awake and doing some work and not wake up a sleeping core unless necessary.

We also saw the Ivy Bridge processor running the HAWX 2 benchmark, now with support for DX11.

m12.jpg

If you look at the die image at the top of this post, you will also notice that it appears more of the die has been assigned to graphics performance than was allocated to it on Sandy Bridge.  Remember that on AMD's Llano about 50% of the die dedicated to stream processors; it would appear that by adding support for DX11, nearly doubling performance and including required support for things like DirectCompute, Intel was forced to follow suit to some degree. 

Mooly laughed at press taking pictures of the die as he had purposely modified the image to hide some of the details or distort them to prevent precise measurements.  Still, it looks like about 33% of the new Ivy Bridge processor is dedicated to graphics and media.  This is good news for consumers, but potentially very bad news for the discrete GPU market in notebooks and low end PCs.

Finally, Mooly Eden ended with a brief look at future Ultrabooks that will be based on the Ivy Bridge processor.

m13.jpg

If you thought the current generation of Ultrabooks was sexy (as I do) then you will really like what is coming up next.

Source: PCPer

IDF 2011: ASUS UX21 Ultrabook Still Sexy, I Still Want It

Subject: General Tech, Processors, Mobile | September 14, 2011 - 11:48 AM |
Tagged: idf, idf 2011, asus, ultrabook, ux21

Yes, I realize the ASUS UX21 was first shown at Computex in June, but this was my first chance to get my hands on it and I have to say after using it for just a few minutes and comparing it to the aging Lenovo X201 that I am typing this on, I am in love with the form factor.

ux21-01.jpg

I don't have anything else to report yet - no performance metrics, no real-world testing, but I couldn't pass posting these few pictures of it.  Enjoy!

ux21-02.jpg

ux21-03.jpg

ux21-04.jpg

ux21-05.jpg

Source: PCPer

IDF 2011: MSI Shows Off Upcoming X79 Motherboards

Subject: Motherboards, Processors | September 14, 2011 - 10:59 AM |
Tagged: x79, sandy bridge-e, msi, idf 2011, idf

Many of Intel's partners are on hand at IDF to showcase upcoming products and I was able to stop by the MSI booth yesterday to get a peek into the future of the X79 chipset.  This chipset will launch with the upcoming Sandy Bridge-E processors (for enthusiast) sometime later this year and introduce a new processor socket (Socket 2011) as well as some new features like dozens of PCI Express connections and quad-channel memory.

msi02.jpg

The flagship board on display was the X79A-GD65 (8D) that is one of the few X79 boards I have seen sporting 8 DIMM slots (hence the name) and capacities as high as 64GB!  Most of the MSI features we have come to love on current motherboards are going to be on this line as well including Military Class components, OC Genie II and the much updated and improved ClickBIOS II.

msi04.jpg

The board is completely 3-Way SLI ready (and CrossFire as well) and sports three total PCI Express 3.0 slots at x16 bandwidth but also adds in three more PCIe 2.0 slots for good measure. 

msi03.jpg

There are 4x USB 3.0 ports, 12x USB 2.0 ports, 4x SATA 6G ports and 4x SATA 3G ports.  Needless to say the X79 platforms are going to be an enthusiast's dream.

msi05.jpg

MSI is also likely to include a new breakout box with X79 motherboards that will reside in a 5.25-in bay slot on your case (and honestly I need more things to use up there these days) and give you a couple of USB ports, an OC Genie button that will double as a BIOS reset button when held down, and even integrated WiFi and Bluetooth.  The above photo is just a mock up but the plans are in place to deliver them with these new boards.

msi01.jpg

Finally, MSI did have the much improved and updated ClickBIOS II UEFI on display and without a doubt it kicks the first version's ass.  We are working on a review that includes this new implementation so expect more on that soon.

Source: PCPer

IDF 2011: Lucid HyperFormance Technology Improves Game Responsiveness

Subject: General Tech, Graphics Cards, Motherboards | September 14, 2011 - 02:12 AM |
Tagged: virtu, mvp, lucid, idf 2011, idf, hyperformance, hydra

Lucid has a history of introducing new software and hardware technologies that have the potential to dramatically affect the PC gaming environment.  The first product was Hydra shown in 2008 and promised the ability to use multiple GPUs from different generations and even different vendors on the same rendering task.  Next up was Lucid Virtu, a software solution that allowed Sandy Bridge processor customers to take advantage of the integrated graphics features while also using a discrete graphics card.  Lucid added support for AMD platforms later on and also showcased Virtual Vsync earlier this year in an attempt to improve user gaming experiences. 

mvp04.jpg

That is a nice history lesson, but what is Lucid discussing this time around?  The technology is called "HyperFormance" (yes, like "High-Performance") and is included in a new version of the Virtu software called Virtu MVP.  I'll let the Lucid press release describe the goals of the technology:

HyperFormance, found in the new model Virtu Universal MVP, boosts gaming responsiveness performance by intelligently reducing redundant rendering tasks in the flow between the CPU, GPU and the display. 3D games put the greatest demands on both the CPU and GPU. And as the race for higher performance on the PC and now in notebooks never ends, both CPUs and GPUs keep gaining performance.

First, a warning.  This software might seem simple but the task it tries to accomplish is very complex and I have not had enough time to really dive into it too deeply.  Expect an updated and more invasive evaluation soon.  There are a couple of key phrases to pay attention to though including the idea of boosting "gaming responsiveness performance" by removing "redundant rendering tasks".  The idea of boosting responsiveness pertains to how the game FEELS to the gamer and should be evident with things like mouse movement responsiveness and the stability of the on-screen image (lack of tearing).  Lucid's new software technology attempts to improve the speed at which a game responds to your actions not by increasing the frame rate but rather by decreasing the amount of time between your mouse movement (or keyboard input, etc) and what appears on the screen as a result of that action. 

How they do that is actually very complex and revolves around the Lucid software's ability to detect rendering tasks by intercepting calls between the game engine and DirectX, not around dropping or removing whole frames.  Because Lucid Virtu can detect individual tasks it can attempt to prioritize and learn which are being repeated or mostly repeated from the previous frames and tell GPU to not render that data.  This gives the GPU a "near zero" render time on that current frame and pushes the next frame through the system, to the frame buffer and out to the screen sooner. 

To think of it another way, imagine a monitor running at 60 Hz but playing a game at 120 FPS or so.  With Vsync turned off, at any given time you might have two to four or more frames being rendered and shown on the screen.  The amount of each frame displayed will differ based on the frame rate and the result is usually an image some amount of visual tearing; you might have to top 35% of the screen as Frame1, the middle 10% of the screen as Frame2 and the bottom 55% as Frame3.  The HyperFormance software then decides if the frame that is going to take up 10% of the screen, Frame2, has redundant tasks and if it can be mostly removed from the rendering pipeline.  To replace it, the Lucid engine just uses 65% of Frame3. 

mvp05.png

The result is an output that is more "up to date" with your movements and what is going on in the game engine and in "game time".  Like I said, it is a very complex task but one that I personally find very interesting and am looking forward to spending more time visualizing and explaining to readers.

Interestingly, this first implementation of HyperFormance does require the use of a multi-GPU system: the integrated GPU on Sandy Bridge or Llano along with the discrete card.  Lucid is working on a version that can do the same thing on a single GPU but that application is further out.

mvp01.png

Frame rate without HyperFormance 

There is a side effect though that I feel could hurt Lucid: the effective frame rate of the games with HyperFormance enabled are much higher than without the software running.  Of course, the GPU isn't actually rendering more data and graphics than it did before; instead, because HyperFormance is looking for frames to report at near zero frame times, benchmarking applications and the games themselves *think* the game is running much faster than it is.  This is a drawback to the current way games are tested.  Many gamers might at first be fooled into thinking their game is running at higher frame rates - it isn't - and some might see the result as Lucid attempting to cheat - it isn't that either.  It is just a result of the process that Lucid is trying to get to work for gamers' benefits.

mvp03.png

Frame rate with HyperFormance

Instead, Lucid is attempting to showcase the frame rate "increase" as a responsiveness increase or some kind of metric that indicates how much faster and reactive to the user the game actually feels.  It might be a start, but claiming to have 200% responsiveness likely isn't true and instead I think they need to spend some time with serious gamers and have them find a way to quantify the added benefits that the HyperFormance application offers, if any. 

There is a LOT more to say about this application and what it means to PC gaming but for now, that is where we'll leave it.  Expect more in the coming weeks!

Source: PCPer

IDF 2011: Other Foundries Falling Further Behind Intel Technology

Subject: General Tech, Processors | September 13, 2011 - 06:07 PM |
Tagged: TSMC, idf 2011, idf, GLOBALFOUNDRIES

While learning about the intricacies of the Intel tri-gate 22nm process technology at the Intel Developer Forum, Senior Intel Fellow Mark Bohr surprised me a bit by discussing the competition in the foundry market.  Bohr mentioned the performance advantages and competitive edge that the new 22nm technology offers but also decided to mention that other companies like TSMC, GlobalFoundries, Samsung and IBM are behind, and falling further behind as we speak.

22nm18.jpg

When Intel introduced strained silicon in 2003, it took competition until 2007 to implement it.  For High-K Metal Gate technology that Intel brought into market in 2007 it wasn't until 2011 that AMD introduced in its product line.  Finally, with tri-gate coming in 2011, GlobalFoundries is talking about getting it implemented in the 2015 time frame.

Obviously those are some long delays but more important to note is that the gap between Intel and the field's implementations has been getting longer.  Three years for strained silicon, three and a half for high K and up to four years for tri-gate.  Of course, we could all be surprised to see tri-gate come from a competitor earlier, but if this schedule stays true, it could mean an increasing advantage for Intel's products over AMD's and eventually into ARM's. 

This also discounts the occasional advantage that AMD had over Intel in the past like being the first to integrate copper interconnects (on the first Athlon) and the first to develop a Silicon-on-Insulator product (starting with the 130nm process); though Intel never actually adopted SOI. 

Source: PCPer

Intel & McAfee submerging their DeepSAFE deep into the Core

Subject: General Tech | September 13, 2011 - 05:05 PM |
Tagged: mcafee, Intel, idf 2011, idf

As the Intel Developer Forum commences we finally learn a little bit about what Intel is attempting to do with the acquisition of McAfee among other tidbits. Malware is one of the banes of computing existence. Information is valuable, security is hard, and most people do not know either. Antimalware software remains a line of defense between you and infections in the event that your first three lines of defense (patching known security vulnerabilities in software; limiting inbound connections and permissions; and common sense) fail to help. While no antimalware software is anywhere near perfect Intel believes that getting protection a little deeper in the hardware will do a little more to prevent previously unknown exploits.

IDF-McAfee.jpg

Great Norton’s Ghost!

According to McAfee’s website, DeepSAFE is a platform for security software to see more of what is going on in the hardware around the Operating System itself. They are being very cagey about what technology is being utilized both on their site as well as their FAQ (pdf) which causes two problems: firstly, we do not know exactly what processors support or will support DeepSAFE; secondly, we do not know exactly what is being done. While this is more details than we knew previously there are still more than enough holes to fill before we know what this technology truly is capable of.

Source: McAfee