Processor News

The executives at server and operating system maker Sun Microsystems have been uncharacteristically quiet since the $5.6bn Oracle deal was announced back in April. And they've been silent since Sun's shareholders approved the deal last Thursday. This - from one of the most aggressive, PR-driven firms on the planet - is a bit disturbing. But Oracle is calling the shots, which is why the IT trade press had to figure out for itself that Sun has actually done a good thing and boosted the clock speeds on its 'Niagara' family of Sparc T2 and T2+ processors. 

The preliminary program for the annual Hot Chips conference says a lot about the state of the microprocessor industry both for what's on it, and for what's not.

All the top server CPU vendors--Advanced Micro Devices, IBM, Intel, Fujitsu and Sun Microsystems--will present on their latest or next-generation chips. Competition is expected to be intense between Intel's Nehalem EX, AMD's Magny Cours, IBM's Power7 and Sun's Rainbow Falls.

However, Sun did not submit a paper on Rock, its high-end server CPU first described in February 2008, leading to speculation the company may have canceled the chip. Sun lost Rock's chief architect, Marc Tremblay, to Microsoft earlier this year and the merger with Oracle cast uncertainty over Sun's future Sparc efforts.

Rock provided hardware support for two advances in parallel processing, transactional memory that would eliminate today's inefficient data locking techniques and scout threads that could pre-fetch data to cache in anticipation of program branches. "The signs are not looking good, so we may be writing an epitaph for Rock soon," said Nathan Brookwood, principal of market watcher Insight64 (Saratoga, Calif.).

Fujitsu may help Sun fill a hole if it cancel's Rock. The company will describe an eight-core Sparc processor at Hot Chips, and Fujitsu and Sun have a long established relationship of selling each other's systems.

Also missing from the Hot Chips program is Intel's Tukwilla, the first Itanium CPU to use Intel's Quick Path Interconnect and an on-board memory controller. Intel deferred the Tukwilla launch from this fall to early next year.

"My reading on that was they probably found some places where it was not scaling as good as they thought it should," said Brookwood.

Nvidia founder and chief executive Jen-Hsun Huang will give a visionary keynote at Hot Chips. However there are no graphics processor papers on the program.

The next-generation graphics chips are expected to launch following the release this fall of Microsoft's Windows 7 and its DirectX 11.0 application programming interface. It supports a new DirectX Compute ability to manage multiple cores on graphics and host processors, handling similar jobs as the OpenCL software supported by Apple and Nvidia's CUDA parallel programming environment.

Classic rivals, software in spotlight

In a departure from chip-oriented papers, organizers also have planned a session where leading research labs will report on their efforts to define parallel programming models for tomorrow's multicore chips. The lack of parallel programming tools for future many-core architectures is seen as the most pressing problem in computer science today.

The conference will also play host to classic rivalries in chip architecture. Intel will present details of its Nehalem EX, a dual-threaded eight-core version of its Xeon 5500 with enhanced memory bandwidth suitable for four-socket systems. Archrival AMD will counter with a paper on Magny Cours, a single-threaded 12-core part supporting four memory controllers.

I think Nehalem EX is likely to be a higher performing chip than the Magny Cours," said Brookwood, but the competition between the two "may be close," he added.

For its part, IBM will present two papers on Power7, a follow on to its 65nm dual-core Power6 chip that has been shipping since 2007. IBM has confirmed Power7 will sport 8 cores. It is expected to be made in a 45nm process and ship in systems next year.

Among other processors at Hot Chips, Sun will describe Rainbow Falls, a third generation of its aggressively multi-threaded Niagara architecture aimed at Web servers.

Intel and Texas Instruments will face off in mobile processors. Intel will describe Moorestown, its next-generation mobile platform including the Atom CPU and a core logic chip. TI will describe the OMAP 4430, its next-generation applications processor.

FPGA rivals Xilinx and Altera will also detail their next-generation parts. Xilinx will discuss the Virtex6 and Altera will present on its Stratix IV GT.

Read the original article: http://www.eetimes.com/news/semi/showArticle.jhtml?articleID=218000214

Sun Microsystems may have dropped a bit of weight by the time Oracle officially acquires the company. According to two people briefed on Sun’s plans, the company has canceled its Rock chip project, putting an end to one of its biggest revitalization bets.

Sun has been working on the Rock project for more than five years, hoping to create a chip with many cores that would trounce competing server chips from I.B.M. and Intel. The company has talked about Rock in the loftiest of terms and built it up as a game-changing product. In April 2007, Jonathan Schwartz, the chief executive of Sun, bragged about receiving the first test versions of Rock.

But the two people familiar with Sun’s plans say Rock has met with an unceremonious end. The people requested anonymity, as they are not authorized to speak with the press about Sun’s plans.

Michelle Parkinson, a Sun spokeswoman, said the company had no comment.

Rock was at one point scheduled to ship last year, but Sun has delayed it a number of times, and over the past few months company insiders have complained about various glitches hurting the product.

This marks the second high-end chip in a row that Sun has canceled before its release. These types of products cost billions of dollars to produce, and Sun now has about a 10-year track record of investing in game-changing chips that failed to materialize.

Sun has been relying on chips from Fujitsu for its larger servers while it waited for the Rock development to be finished. Now it is likely to just continue using Fujitsu chips, which should lower research and development costs. That’s probably good news for Oracle, which is in the process of acquiring Sun.

Sun does still design a line of chips for its smaller servers, and those products have proved popular in recent years.

Read the original article: http://bits.blogs.nytimes.com/2009/06/15/sun-is-said-to-cancel-big-chip-project/

The word on the street today is that server maker Sun Microsystems - which is in the midst of being eaten by software giant Oracle for $5.6bn - has formally killed off the 16-core "Rock" UltraSparc-RK processor that has been in development for more than five years.

The official response from Michelle Parkinson, the spokesperson for Sun's Systems group, was that Sun had "no comment re: Rock," which is exactly what you would expect from a company that is in the middle of being acquired and which is cleaning house before Larry Ellison and a whole bunch of Oraclers move in.

Sun has two more weeks before its fiscal year ends and another four weeks before Sun shareholders vote on the Oracle takeover, and it is hard to believe it will close out the year on a financial high given its woes, the economy, the Oracle acquisition, and all of the uncertainties that has introduced in the customer base.

A faithful source - a former Sun employee who requested anonymity and who still has deep ties with the server maker - said that he was told by a high-ranking Sun exec this morning that the Rock chip was canceled. The New York Times is also running a story that cites two people familiar with Sun's plans (and who also requested anonymity) as saying that Sun has canceled the Rock chip project.

If the Rock processors and their affiliated "Supernova" servers are indeed dead, they are dead because Sun has to kill it off. If Sun kills it off before Oracle completes the deal, Oracle can still say that it believes in and will invest in the Sparc hardware business. That's what Larry Ellison, Oracle's chief executive officer, told Reuters in a recent interview. By the way, as Sun revealed in filings with the Securities and Exchange Commission, Oracle didn't originally want Sun's hardware business.

If Rock is dead, it's a safe guess that Oracle will continue to partner with Fujitsu for midrange and high-end Sparc64 servers and that it will probably try to position the future "Niagara-III" 16-core Sparc T series chips - if they survive the knife - as the future for Sparc computing. Expect Oracle to emphasize Solaris on x64 chips, particularly with 64-core, 128-thread Intel "Nehalem EX" servers due early next year using Intel's own "Boxboro" chipset. Oracle will be able to show off some pretty scalable Solaris iron - even if it does force some customers into doing a recompile.

The death of the Rock processor will come after plenty of trials and tribulations for Sun in the chip racket. Sun's UltraSparc-II processors, the real dot in the dot.com Internet revolution, were excellent, competitively priced, and came to market without too much drama. The UltraSparc-III processors, which were developed in the late 1990s, came to market late and underpowered in the early 2000s, giving IBM's dual-core Power4 processors a chance to start eating market share as Sun dropped support for x86 chips with its Solaris platform and kept its Sparc-based server prices too high to compete.

A few years later, the UltraSparc-V chip was killed, maybe because it was too ambitious for Sun's declining development budget. Sun threw together a stop-gap dual-core UltraSparc-IV chip and started investing heavily in the multicore and multithreaded Niagara and Rock processors. To bridge the gap between UltraSparc-IV chips and the Rocks, Sun partnered with Fujitsu to resell its Sparc64 SMP boxes. Wouldn't you know it: Those were late coming to market too, as Fujitsu drove off its Sparc64 roadmap. The Rock chips and their servers were supposed to come to market in the second half of 2008, but were delayed at the end of 2007 and pushed out to the second half of 2009.

Considering how close this product was supposed to be to launching - within the next four to five months - and how much money Sun had invested already, something must be truly wrong with the Rock chip or the servers that use it for Sun to cancel it.

Now, Sun doesn't really have a second half of 2009 - unless by some warping into bizarroworld, Sun's shareholders don't approve of the Oracle deal. And it seems likely that Rock chips and their Supernova boxes don't have a second half either.

Read the original article: http://www.channelregister.co.uk/2009/06/15/sun_kills_rock_sparc/

Companies demonstrate multi-core emulator based on Virtex(R)-5 at grand opening of Advanced Research & Teaching Facility for Embedded System Design at Beijing University of Technology