Stephen O’Grady, a founder at the technology analyst company RedMonk, said the technology industry often has swung back and forth between more standard computing systems and specialized gear.
A little tip of the hat to Andrew Feldman, CEO of SeaMicro the startup company that announced it’s first product last week. The giant 512 cpu computer is being covered in this NYTimes article to spotlight the ‘exotic’ technologies both hardware and software some companies use to deploy huge web apps. It’s part NoSQL part low power massive parallelism.
Comment With Intel sending its “Larrabee” graphics co-processor out to pasture late last year – before it even reached the market – it is natural to assume that the chip maker is looking for something to boost the performance of high performance compute clusters and the supercomputer workloads they run. Nvidia has its Tesla co-processors and its CUDA environment. Advanced Micro Devices has its FireStream co-processors and the OpenCL environment it has helped create. And Intel has been relegated to a secondary role.
Intel’s long term graphics accelerator project code-named “Larabee. It’s an unfortunate side effect of losing all that money by time delays on the project that forces Intel now to reuse the processor as a component in a High Performance Computer (so-called Super Computer). The competition have been providing hooks or links into their CPUs and motherboard for auxiliary processors or co-processors for a number of years. AMD notably created a CPU socket with open specs that FPGA’s could slide into. Field Programmable Gate Arrays are big huge general purpose CPUs with all kinds of ways to reconfigure the circuits inside of them. So huge optimizations can be made in hardware that were previously done in Machine Code/Assembler by the compilers for that particular CPU. Moving from a high level programming language to an optimized hardware implementation of an algorithm can speed a calculation up by several orders of magnitude (1,000 times in some examples). AMD has had a number of wins in some small niches of the High Performance Computing market. But not all algorithms are created equal, and not all of them lend themselves to implementation in hardware (FPGA or it’s cousin the ASIC). So co-processors are a very limited market for any manufacturer trying to sell into the HPC market. Intel isn’t going to garner a lot of extra sales by throwing development versions of Larabee out to the HPC developers. Another strike is the dependence on a PCI express bus for communications to the Larabee chipset. While PCI Express is more than fast enough for graphics processing, an HPC setup would prefer a CPU socket adjacent to the general purpose CPUs. The way AMD has designed their motherboards all sockets are on the same motherboard and can communicate directly to one another instead of using the PCI Express bus. Thus, Intel loses again trying to market Larabee in the HPC market. One can only hope that other secret code-name projects like the CPU with 80 cores will see the light of day soon when it makes a difference rather than suffer the opportunity costs of a very delayed launch of Larabee.
Intel’s executives were quite brash when talking about Larrabee even though most of its public appearances were made on PowerPoint slides. They said that Larrabee would roar onto the scene and outperform competing products.
And so now finally the NY Times nails the coffin shut on Intel’s Larrabee saga. To refresh your memory this is the second attempt by Intel to create a graphics processor. The first failed attempt was some years ago in the late 1990s when 3dfx (bought by nVidia) was tearing up the charts with their Voodoo 1 and Voodoo 2 PCI-based 3D accelerator cards. The age of Quake, Quake 2 were upon us and everyone wanted smoother frame rates. Intel wanted to show its prowess in the design of a low cost graphics card running on the brand new AGP slot which Intel had just invented (remember AGP?). What turned out was a similar set of delays and poor performance as engineering samples came out of the development labs. Given the torrid pace of products released by nVidia and eventually ATI, Intel couldn’t keep up. Their benchmark was surpassed by the time their graphics card saw the light of day, and they couldn’t give them away. (see Wikipedia: Intel i740)
1998 saw the failure of the Intel i740 AGP graphics card
The Intel740, or i740, is a graphics processing unit using an AGP interface released by Intel in 1998. Intel was hoping to use the i740 to popularize the AGP port, while most graphics vendors were still using PCI. Released with enormous fanfare, the i740 proved to have disappointing real-world performance, and sank from view after only a few months on the market
Enter Larrabee, a whole new ball game at Intel, right?! The trend toward larger numbers of parallel processors on GPUs from nVidia and ATI/AMD led Intel to believe they might leverage some of their production lines to make a graphics card again. But this time it was different, nVidia had moved from single purpose GPUs to General Purpose GPUs in order to create a secondary market using their cards as compute intensive co-processor cards. They called it CUDA and provided a few development tools at the early stages. Intel latched onto this idea of the General Purpose GPU and decided they could do better. What’s more general purpose than an Intel x86 processor right? And what if you could provided the libraries and Hardware Abstraction Layer that could turn a larger number of processor cores into something that looked and smelled like a GPU?
For Intel it seemed like a win/win/win everybody wins. The manufacturing lines using older design rules at the 45nm size could be utilized for production, making the graphics card pure profit. They could put 32 processors on a card and program them to do multi duties for the OS (graphics for games, co-processor for transcoding videos to MP4). But each time they did a demo a product white paper and demo at a trade show it became obvious the timeline and schedule was slipping. They had benchmarks to show, great claims to make, future projections of performance to declare. Roadmaps were the order of the day. But just last week rumors started to set in.
Similar to the graphics card foray of the past Intel couldn’t beat it’s time to market demons. The Larrabee project was going to be so late and still was using 45nm manufacturing design rules. Given Intel’s top of the line production lines moved to 32nm this year, and nVidia and AMD are doing design process shrinks on their current products, Intel was at a disadvantage. Rather than scrap the thing and lose face again, they decided to recover somewhat and put Larrabee out there as a free software/hardware development kit and see if that was enough to get people to bite. I don’t know what if any benefit any development on this platform would bring. It would rank right up there with the Itanium and i740 as hugely promoted dead-end products with zero to negative market share. Big Fail – Do Not Want.
And for you armchair Monday morning technology quarter backs here are some links to enjoy leading up to the NYTimes article today:
It’s interesting to see how the whole iTunes U structure works. I’ve been reading documentation about the ‘web services’ enabled within iTunes U. It completely replicates the GUI functions but through a semi-automated interface. Reminds me a little of how you can change the underlying LDAP directory structure using LDIF commands or LDIF files with all the changes embedded within it. In iTunes U, you do an HTTP PUT securely with a signed token, and the iTunes U Web service sucks that up and executes all the commands embedded within your XML file that you put. Very powerful, but very scary too as these changes are made to your production environment. So there’s no real easy way to test the results of your commands without just taking a big risk, leaping in and seeing what happens. This is like SQL commands where you DROP TABLE, not a fun thing to do. DROP TABLE is a big black whole that makes your data disappear in an unrecoverable way. iTunes U has similar functions where you delete the structure AND the data at the same time. You may restore the structure (by backing up your data tree in XML format), but the data embedded within the tree, well that’s gone. So restoring stuff is going to be impossible if you get the syntax wrong in your XML file. The only real benefit to me now is the ability to get a listing of the whole site structure using the Tree command and then forcing an update to any groups that are of the type RSS Feed. The update will be necessary if anyone adds files to a podcast being hosted on servers within our institution.
I discovered or re-discovered a tool called Woolamaloo which was introduced to me during the Apple iTunes training. University of Illinois Champaign-Urbana created it to allow you to use a GUI to control the Web services from a desktop OS. This is good as I was at a loss to adapt the sample code into anything like a reliable generator of tokens to send to iTunes U web services. I couldn’t figure out what parts of the java example to comment out and recompile. Starting this Monday I’m going to put Woolmaloo through it’s paces. If I can force the RSS feeds to update on demand when somebody has a problem updating their Podcast feeds, I can at least speed things up. But I’m still very leery of deleting or merging any section. I will copy so I can make a course appear in more than one place without using the iTunes multi-click interface. But I will not delete or merge.
And just today I also discovered there are Apple Automator scripts readily available that add a graphical layer on top of all the web services goodness. So now I can integrate a bunch of steps from uploading bunches of files, forcing RSS feeds to update to merging/rename whole sections all from Automator. I’m going to test it and se how good it really works.
Intel is finally going to ramp up it’s newest production lines to include Flash memory chips, thereby shrinking the design rules down to 34nm. Density of the new Flash memory chips is going to allow even larger Solid State Drives (SSD) and in some cases the prices may be less for the newer drives than the equivalent preceeding generation of SSDs. Price points quoted in the article are projected to be around $276 possibly as low as $261 for the 80GB/34nm based SSD from Intel. The closer to $200 the better, that’s the point at which you can buy some of the higher capacity traditional HDD’s from Seagate, and Western Digital. The day of the $200 Flash Drive is coming soon.
A Canadian RedFlagDeals technology website expects an announcement within a week and says there will be 80GB, 160GB and 320GB models.
Imagination is the name of the of company making the wicked cool low power graphics chip called PowerVR SGX. In the handheld manufacturing market Imagination scored two huge design wins. First was the Apple iPhone and iTouch. Second is the Palm Pre. It is very encouraging that the designers at Palm chose the PowerVR in order to create the iPhone killer. No doubt Palm benefited directly from inside knowledge of the iPhone when they hired former Apple VP Jon Rubinstein to head this new iPhone killer project at Palm.
Now Apple realizes it needs to protect it’s competitive advantage. They are sinking some several million dollars in Imagination stock to prevent any hostile takeover of their strategic partner. Even more interesting than this move on Apple’s part is Intel has already staked a huge claim on Imagination without having a single design win to announce. There’s some word out that future netbooks will use an integrated PowerVR chip. But the next revision of the Atom CPU and chipset will definitely have PowerVR integrated in, scoring some bigger more strategic design wins on the low power front. Intel hopes to best Apple at the low end, low power, long battery duration category.
The investment is considered important for Apple, which uses only PowerVR graphics in its iPhone and iPod touch devices. Its most recent launch, the iPhone 3GS, uses a PowerVR SGX video core now believed to be the SGX535.
There’s a new video trend in personal home video. Companies are lining up to provide aftermarket tools to process and provide corrections to camera phone video. Pure Digital’s Flip! camera line has some tools available to do some minor cutting to video clips and publish it to sharing websites. All of which presents an entrepreurial opporunity to provide pay for tools to help improve poorly shot video.
Some tools are provided within video editing suites like Apples iMovie (it corrects camera shake). Now on the PC there are two new products, one of which is designed to take advantage of the nVidia GPU acceleration of parallel programming. The product is called vReveal
While vReveal works with Windows XP or Vista (and not with Macs), it will make its enhancements much faster if the machine contains a recent graphics processing card from Nvidia, Dr. Varah said. Nvidia is an investor and a marketing partner with vReveal; a specific list of cards is at vReveal’s Web site.
I’ve seen some claims that newer SSDs coming out are implementing the SATA TRIM command. This development is hailing a new era in SDD performance, something we have all wished for since the introduction of SSDs back in 2005. In the last 4 years, performance gains have usually been obtained by using RAID controllers within the SSDs. Worse yet, some SATA disk controllers on the SSDs were known to be total dogs when it comes to performance. Enter the hero of our story: Indilinx
Indilinx decided after multiple requests to enter the market and show that SSDs are worthy of some real product development. Patriot is the one of the first manufacturers to adopt the Indilinx disk controller. Given announcements from Microsoft recently over the addition of full OS support for the SATA TRIM command and now the Indilinx controller,…
One can only hope that Windows 7 will allow SSDs to finally equal or surpass their HDD counterparts. Finger crossed, hoping the Indilinx takes the market by storm and Microsoft will fully embrace and improve its support for the TRIM command
Despite the huge performance gains, two major things plague SSDs:
About a year ago I wrote an article about nVidia’s attempt to use it’s video graphics cards to accelerate transcoding. H.264 was fast becoming the gold standard for desktop video, video sharing through social networking websites, and for viewing on handheld devices. In the time since then, Badaboom entered the market and has gone through a revision of it’s original GPU accelerated transcoding software. Apple is now touting OpenCL as the API through which any software can access the potential of using all those graphics pipelines to accelerate parallel operations off of the CPU. nVidia is supporting OpenCL whole hog and I think there is some hope Microsoft won’t try to undermine it too much though it’s standing strong with DirectX as the preferred API for anything that talks to a graphics card for any reason.
So where does AMD with it’s ATI card fit into the universe of GPU accelerated software? According to Anandtech, it doesn’t fit in at all. The first attempts at providing transcoding have proved a Big Fail. While Badaboom outlcasses it at every turn in the transcoded video it produces. Hopefully OpenCL can be abstracted enough to cover AMD and nVidia’s product offerings with a single unified interface to allow acceleration to occur much more easily as citizen of the OS. Talking directly to the metal is only going to provide headaches down the road as OSes are updated and drivers change. But even with that level of support, it looks like AMD’s not quite got the hang of this yet. Hopefully they can spare a few engineers and a few clock cycles and take Avivo out of alpha prototype stage and show off what they can do. The biggest disappointment of all is that even the commercial transcoder from Cyberlink using the ATI card didn’t match up to Badaboom on nVidia.
A few months ago, we tested AMD’s AVIVO Video Converter. AMD had just enabled video transcode acceleration on the GPU, and they wanted to position their free utility as competition to CUDA enabled (and thus NVIDIA only) Badaboom. Certainly, for a free utility, we would not expect the same level of compatibility and quality as we would from a commercial application like Badaboom. But what we saw really didn’t even deliver what we would expect even from a free application.
Carpet Bomberz Inc. 