Posts Tagged ‘Intel’

Massively Parallel Computing – Code as Architecture

Posted: August 6, 2011 in Intel, microprocessors, nVidia, parallel computing
Tags: , , , , , , , , , ,
0

A friend sent me a link to an article on changes coming in microprocessors. The article is The Lifer: Why Your Core i7 Processor May Be Obsolete Sooner Than You Think. It got me thinking about writing this post not because the article has any great insight but because of the opposite. The article is too shallow.

One of the topics mentioned is specialized computing. This is nothing new. While it wasn’t the beginning, many people may remember the Intel 8087 floating point coprocessor that offloaded the 8086. Earlier there was the less well know 8031A. I have linked to a copy of the datasheet if you want to see how things used to be. The 8031A paired with the 8080 microprocessor. Interestingly, considering the two companies today, the 8031 and 8031A were licensed versions of AMD’s   AM9511 and AM9511A introduced in 1977. Today, we take it for granted that this floating point capability is built into the processors we use.

Throughout computing history the research agencies have driven the need for large, somewhat specialized computers. From the CDC 6600 (1964), to the Cray 1 (1976), to Nebulae (2010) floating point performance has driven a class of supercomputers designed for scientific and military research.  Originally these designs employed vector processors. Today, machines like the Nebulae use off the shelf graphics processors as general purpose computing engines (GPGPU). In particular, nVidia has started marketing to this area. The problem is that modern GPU’s are basically SIMD machines and bring along many of the limitations found with a SIMD architecture. Working with the limitations of SIMD and mitigating those limitations is a big topic with a large body of work so I won’t address it in depth here. For restricted problems such as graphics rendering it is a very effective approach. At the top end, the AMD 6990 graphics card contains two processor chips which together yield 3072 Stream Processors, 192 Texture Units, 128 Z/Stencil ROP Units, and 64 Color ROP Units. For graphics rendering this gives amazing performance. What it is not good at is general computing. In summary, specialized computing is nothing new and has been with us for a long time. Massively parallel specialized computing is here today.

Myslewski talks about large numbers of general purpose computing cores. We have made great progress utilizing four core and even eight core system. There are restricted problems such as design rule verification of large chip designs which are amenable to massively parallel systems. However, general purpose computing has trouble utilizing even four cores effectively. More interesting than the straight forward approach Myslewski mentions are approaches which reconsider the very nature of what a processor is. I have been thinking about this lately after watching a talk by Steve Teig of Tabula.

http://www.c-eda.org/IEEE-CEDA-DAC-061510/IEEE-CEDA-DAC-061510.html

Steve mentions Haskell as a language of choice. This is a transition that is needed and is fundamental. We currently force fit a one CPU ecosystem onto multiCPU processors. We patch language structures and manually work to make task division successful. In graphics this is somewhat straight forward. You tell the different cores “Care 1 you work on this area of the scene, core 2 you work over here, core 3 …” Except for specialized areas such as graphics, this model does not fit what we do today when we get beyond four cores. Right now we can, at a very simplistic level, say, “Core 1 you handle operating system commands, core 2 you run the program, core 3 you take care of the anti virus background tasks, core 4…” What is wrong here is the process and mindset itself. That’s why Steve mentions Haskell. The mental process I just outlined is forcing the code onto the processor. What is needed is a new paradigm of code as architecture. I am not talking about the Tensilica approach but something closer to the work discussed here. If you read through the various papers you will see a common theme related to the problem of limited FPGA size. The idea of time as a third dimension opens the door to a possible solution. What needs to be worked out is an interface that gets around the von Neumann memory bottleneck and allows continuous reconfiguration of the FPGA. Once that is achieved, arbitrarily large code can be executed with a three dimensional FPGA (X, Y, time) as the direct instantiation of the code. For an example of this type of FPGA check out Tabula. Be careful to not get lost in the hardware although that is a key component. The main advantage of the hardware is the ability to latch its state and rapidly reconfigure. More important that that functionality is compiling down into the FPGA in a way which allows a mapping of code to circuitry that continuously reconfigures as code is executed rather than execute, save state, load code, reconfigure, execute. Let me know what you think of the concept of code as architecture.

Advertisement

Will FinFETs Save Intel?

Posted: May 31, 2011 in ARM, Intel, Tablet
Tags: , , , , ,
0

I have said before that Intel’s big ace in the hole is its process technology. At the beginning of this month, Intel made an announcement regarding their Tri-Gate technology. They make it sound completely unique to Intel but in reality it exists at several companies although the industry usually refers to it as FinFET technology. Never mind that others also have this technology. The key is that Intel can out process those guys.  Intel consistently has superior real world (volume production ready) process technology compared to anyone. When the world moved to 90nm and below, leakage current became a big deal. Many would say it became the dominant design factor if you are at all concerned with power. This meant that going smaller didn’t necessarily reduce power. It meant complex design choices just to save power. You could design with a low leakage transistor or a low power one. Today designers often mix transistor types to get the best balance of speed and power. FinFETs offer a move back to the old world where smaller meant both faster and lower power.

For decades we have lived in a world where processing speed and cost were the big drivers. What hasn’t been a huge issue has been power. Yeah, we would give it lip service and laptops would have to take it into account but it still took a back seat to speed and cost. That has changed. That is how ARM went from being a niche architecture to threatening the behemoth of the CPU world. Now power is a major focus of Intel. If all things were equal it would be a case of realizing the future too late to stop where things are headed. But… all things aren’t equal and it isn’t a level playing field. Intel has huge resources and access to proprietary process technology. The world won’t move from ARM to Atom just because Intel wants them too. Companies like the fact that they can bake their own chips by licensing an ARM core. The ARM architecture is now standard in both the smartphone and tablet worlds. However, all phone and tablet companies are pushing up against power consumption that runs into the brick wall of battery technology. Batteries are improving but not as fast as the demand for mobile processing power. If Intel can make Atom based chips clearly superior to ARM based approaches when it comes to power then they just might be able to win companies over. They are trying very hard. Stay tuned.

The Real Importance of Google Wallet

Posted: May 28, 2011 in Convergence, Google, Intel, Microsoft, NFC, Transparency, Trends
Tags: , , , , , , , , , , ,
0

It may seem like I have been mostly regurgitating news. Look deeper. I am trying to point out the trends of convergence and transparency and how they are reaching everywhere. On the surface Google Wallet is a nice tweak to how you pay for what you buy. In terms of those affected it is easy to see the retailers, banks and credit card companies. If you look on the surface at semiconductor companies you might just think about those chips which enable NFC. This is part of something much bigger that affects many more companies. NFC services like Google Wallet will make transactions more transparent i.e. easier and more convenient. They also converge services into the phone and continue pushing the phone towards becoming your dominant computing platform. This is what I started this blog off with. It doesn’t matter if Google Wallet in it’s present form becomes big or not. It’s a symptom of a larger movement. No matter what business you are in you need to evaluate your strategy with convergence and transparency in mind. How will your business play out when the phone is the dominant computing platform? Intel and AMD are reacting to this today. For once the interests of AMD and Intel are aligned. They need to bring the X86 architecture to tablets and then mobile phones. Microsoft is also reacting as they worry about Windows being marginalized. Think how differently this would have been had the iPhone and iPad been based on the Atom processor. For the other chip companies there is the increasing importance of LTE and the cloud. Flash memory will continue to be pushed to grow in density and decrease in price. The world is moving towards one gigabyte of storage in the phone. Remember reading about how over built the global network is? Think again. OLED screens will finally become a mainstream technology driven by the phone. Eventually they will grow to be the dominant technology in both laptops and TV’s.  This shift affects media. The RIAA and  MPAA continue their vain attempts at protecting intellectual property rather than embracing the technology trends and profiting from them. That’s an entire blog (or two or three) in and of itself. Is your company preparing for the upcoming changes? More importantly, have you looked deep to see how convergence and transparency will change your business landscape?

Random Bits and Some Good Reading

Posted: May 18, 2011 in Apple, Convergence, Google, Intel, Microsoft, Transparency, Trends
Tags: , , , , , ,
1

Did Intel read my blog? Interesting posts about Intel here and here and here.

Of the Android tablets, the Xoom was the first to look reasonable but the iPad 2 made it look dated. The first real iPad 2 challenger appears to be the Samsung Galaxy Tab 10.1. Don’t confuse this with the thicker 10.1v. Engadget has a good review here. They also have a review of the HTC Flyer posted here. While not as positive as the Galaxy Tab review, I think the integration of audio and notes on the Flyer has a lot of promise.

Does anyone else think Microsoft buying Nokia’s smartphone business is a bad idea? That would make Microsoft like Apple but still trying to be like Google. Pick an approach! If Microsoft tries to straddle the fence then they will fail. Why should HTC or anyone else make a Windows Mobile phone if their direct competition on the Windows Mobile space will be Microsoft. Microsoft could wind up being the sole manufacturer of Windows Mobile phones. At that point I give a big advantage to Apple.

I haven’t talked much about Chrome and the Chromebook. A good article on them is here.

If you are following what I am saying about transparency and convergence I suggest you read  Sarah Rotman Epps’ Blog.

Winners and Losers – Imagination Technologies is on a winning streak

Posted: May 11, 2011 in Intel
Tags: , , ,
0

As reported by VR-Zone, it looks like Intel’s upcoming Cedarview Atom platform will sport PowerVR graphics from Imagination Technologies. This is a big deal. PowerVR is also used in Apple’s chips that power the iPhone and iPad. PowerVR is set to be the standard for graphics across a broad range of hardware. This will make it difficult for other technologies to enter the space as PowerVR becomes the industry standard. The loser here is Intel. For Cedarview, they are passing over their own graphics technology in favor of PowerVR. That means paying royalties to Imagination Technologies. One of those cracks in Intel’s armor just got a little bigger.

Cracks in the Wintel Armor

Posted: May 9, 2011 in Broadcom, Intel, Qualcomm, Trends
Tags: , , , , , ,
1

During its peak no one ever thought Rome would fall. Similarly, in the world of technology, no one has thought the pairing of Windows and Intel could be challenged. For decades Microsoft has dominated the computer software space while Intel, with its X86 processor line, has dominated the chip side of high tech. This Wintel hegemony has been unassailable. It hasn’t been for lack of trying. IBM tried with OS2. Apple had some limited success once they brought out OSX. Motorola tried with the 6800 and 68000 chips. Sun tried with the SPARC processor and there was MIPS. These are just a few. HP, DEC and Fairchild had processors. IDT had the Winchip. Some survive today but only AMD, with its X86 clones, ever had much success as measured by volume shipped. Today things have changed. After WWI, the French built the Maginot Line. Everyone agreed it was unassailable. What France neglected was the rise of mobile warfare. Since the Germans couldn’t defeat the Maginot Line they went around it. The same is happening today in both the operating system area and in microprocessors. The attack on the Wintel hegemony is both real and potentially fatal. The outcome hasn’t been determined but the battle has started. Just like Germany flanked the Maginot Line, the Wintel alliance is being flanked. For Germany the enabler was mobile warfare and the development of the blitzkrieg strategy. Today the enabler is the smartphone together with iOS and Android.

In Q4 2010, according to IDC, there were more smartphones sold than PC’s. This is an important statistic. It means that the smartphone is set to become the dominant computing platform. Few smartphones run a Microsoft OS. Currently Microsoft’s share is at 7.5% which is down from 8.4% just a few months ago (see here). As far as the CPU, most smartphones use the ARM processor (see here). The tablet is more closely aligned to the smartphone than to the desktop. iOS and Android are the dominant operating systems. Microsoft won’t have a contender on the market till 2012. How about another shocker. Apple has surpassed Microsoft in both revenue and profit (Businessweek). No longer is Apple the little weakling hoping Microsoft won’t notice them. During the 90’s Microsoft was threatened by Netscape. Once Microsoft awoke to the threat it wasn’t a fair fight. The sheer size of Microsoft allowed them to give away Internet Explorer and destroy Netscape. Today the battle is different. Both Apple and Google have massive resources.

I’m a chip guy. I’ve been talking software. What about chips? Specifically, what about Intel? On the plus side Intel has vast wealth and controls the best semiconductor fabrication technology in the world. Intel continues to push R&D so, at least when it comes to semiconductor process technology, they can’t be accused of sitting still. They make the Atom processor line so they have an entrant in the low power, low cost processor market. Their problem is convergence at the chip level. Recall that I said convergence is much bigger than people think. Convergence is the dominant trend in the semiconductor business. More and more functionality is being integrated onto a single chip. Chip count is going down. Look inside a DVD player and you will see that there isn’t much there. Intel has made its money selling discrete microprocessors and selling them at high margins base on a near monopoly in the desktop CPU space. Back in 2009, Intel made the Atom core available through TSMC. However, the main core used today is ARM. The Apple A5 chip contains a dual-core ARM Cortex-A9 MPCore CPU and a dual-core PowerVR SGX543MP2 GPU. Samsung is the foundry for the chip although Apple may be moving to TSMC. LG is getting into the act and will be designing its own chip. The future is clear. Companies are buying IP and integrating it into one chip. This is a huge threat to Intel. With ARM becoming the dominant processor, Intel is losing its grip on the processor market. Even if Atom cores sell well the profits will be nothing like what Intel has seen in the past. None of this spells near term doom for Intel. They continue to mint money in the desktop and laptop space. Netbooks are dominated by Atom processors. Long term the outlook is different. Netbooks are being hurt by ARM based tablets. The smartphone is driving convergence. Soon it will be the smartphone, tablet and TV vs. the notebook and desktop. One rumor already has Apple moving from X86 to ARM for their laptops. This move seems premature to me but something that will eventually happen. In and of itself this wouldn’t threaten Intel. Apple isn’t their major client. However, imagine if the Android market were to progress in that fashion i.e. ARM based Android phones drive an ARM based Android tablet world which eventually drives a Chrome/Android notebook market. The force behind this? Convergence! Intel sees this and is pushing its own smartphone solution. For this reason the outcome isn’t certain. What is certain is that there will be a battle royal with Intel no longer having a monopoly.

If you are in the chip business there is something broader to be taken from this. Integration continues to increase. The system on a chip, SoC, is the dominant trend. That may seem obvious but I continue to see companies ignoring this trend. Analog companies fail to see that they will be integrated into an SoC. RF companies don’t see how they are being isolated as surrounding functionality gets integrated. Consider the GPS chip. The market for stand alone GPS chips continues to diminish as better and better solutions get integrated into chips like the Quallcomm MSM8660. Either you are a Quallcomm or Broadcom and doing the integration or you are being marginalized in the GPS marketplace. Expand this to the chip business in general. Sound chips (are you listening Wolfson), power sequencers, and others will be integrated. Consolidation is happening in semiconductors just like it happened to the automobile industry during the 1900’s.