Unisys may have written Itanium's epitaph on Wednesday--at least for some of the largest server vendors.
Colin Lacey, vice president of Systems and Storage at Unisys, discussed in a phone interview why Unisys--one of the top 10 U.S. server vendors--doesn't see a future for Itanium, including the long-delayed quad-core Itanium "Tukwila" processor.
Lacey said Itanium's appeal has almost vanished for many vendors in server industry. "It's appeal has certainly narrowed down. It's almost exclusively down to a single vendor," he said, referring to Hewlett-Packard. "The current shipping platform is overdue for a technology refresh (and) it's been delayed a couple of times already," he said. In short, Itanium's chronic delays and underwhelming performance mean it's not a viable option anymore for Unisys--which offered Itanium-based servers in the past.
Lacey went on to say that Xeon can be "harnessed" in a high-end server environment to deliver performance that surpasses Itanium with the same reliability. Unisys announced on Wednesday that its newest enterprise server--the ES7000 Model 7600R--has set a record for price/performance in the Transaction Processing Performance Council's (TPC) TPC-H benchmark test. "The performance of the Unisys server, which uses the latest Intel Xeon 6-core processors, shows the increasing superiority of Xeon-based systems for mission-critical applications such as business intelligence over those based on Intel Itanium processors." the Blue Bell, Penn.-based company said in a statement.
Unisys' comments uncannily echo a statement made more than 10 years ago (in 1998) by an analyst who was discussing the long-delayed "Merced" chip--the processor jointly developed by HP and Intel that eventually became Itanium. At that time, an analyst said, "if the performance is similar to Merced," server vendors will opt to "squeeze more profit out of Xeon" instead of adopting Itanium.
Lacey spelled out how Unisys did its testing. "It's not like we're loading the dice. I know that Itanium has only two processors, but the configurations we're comparing have exactly the same number of processor cores. We're comparing 64-core to 64-core. There is no compute engine deficit between one and the other. So, it's looking at the architecture and what works."
With latest generation of Xeon technology "we can deliver pretty compelling raw performance as well as a very significant cost reduction by migrating over to a Xeon architecture. We're talking about a Windows SQL database environment and there's no real pain, if you will, from doing that on Itanium to doing that on Xeon," he said.
At Unisys, there is no significant difference in reliability either, one of Itanium's purported marquee features. "We track the unplanned downtime of our customer base and we track pretty much identical results between Xeon and Itanium architectures with respect to downtime. We don't see any material difference whatsoever and we keep very detailed tracking on that," he said.
Intel, on the other hand, believes both processors offer distinct advantages. "Both platforms offer unique advantages for different needs and applications," Intel spokesman Patrick Ward said Wednesday. "It's great that Unisys is being so aggressive with Xeon's price/performance strengths. Itanium offers terrific scalability and reliability strengths that are a better fit for some of the most mission critical workloads," he added.
Jointly, HP and Intel, not surprisingly, have a different take on Itanium's value for the large enterprise customer. In this video, Intel CEO Paul Otellini and HP CEO Mark Hurd discuss the technology. "Itanium is our architecture for enterprise-class machines. And that's a big segment of the market. About $28 billion. This is a very, very critical architecture for us," Otellini said in the video. "There are things that we're putting into the architecture, the RAS (Reliability, Availability and Serviceability) features, the reliability characteristics, the power-performance characteristics. These are things that we're tuning for the enterprise at the highest end," he said.
"The alliance around Itanium is also unique," Otellini continued. "Multiple operating system environments; 13,000 applications. It's outgrown every other mainframe architecture on the planet over the last five years," he said. "Itanium is about 10 years old now. It's really hitting its stride. It's at the point where we outsell other architectures in Asia. (Sun Microsystems) SPARC and (IBM) Power-based machines," Otellini said. "There are over a thousand silicon engineers at Intel working on this product line," he added.
More details on the Unisys test results can be found in the Unisys February 18 news release.
Updated at 3:45 p.m. PST with correction of Motorola, IBM executives' names.
Even the biggest chip companies churn out their share of flops. But the hype that surrounds these chips is more fascinating than the failures.
It's been almost a year since I posted A brief history of chip hype--and flops (part 1). Consider this Part 2.
Itanium
First, I have to revisit Intel's Itanium. Simply because it's still around and still missing production target dates.
Intel's Itanium has been relegated to obscurity if not practical oblivion
(Credit: Intel)The hype: "This design philosophy will one day replace RISC and CISC. It is a gateway into the 64-bit future." This copy was, at one time, posted on Hewlett-Packard's Web site. And analysts were drinking the Kool-Aid too. "I expect Itanium to replace Xeon, but not until 2003," one analyst said back in 2001. (Xeon is Intel's successful, lucrative line of server processors that doesn't include Itanium.)
The reality: Yes, Itanium is still warm, still breathing in the rarefied very-high-end server market--where it does have a limited role. But its architecture will never live in a desktop or laptop or even 99 percent of the servers as once thought. And it certainly hasn't remade the computer industry. And it is still chronically late. This time it's Tukwila that's tardy. The quad-core version of Itanium is late because Intel had to make "some engineering enhancements to the Tukwila platform," according to an Intel statement earlier this month. I can only guess that one day Intel will finally let this failed research project go cold and die quietly.
PowerPC
IBM's original PowerPC platform never lived up to the hype. Even when Motorola and IBM processors populated Apple computers.
The hype: "The PowerPC G5 changes all the rules. This 64-bit race car is the heart of our new Power Mac G5, now the world's fastest desktop computer," said Apple CEO Steve Jobs back in 2003. Jobs, a master of hype (also referred to as a Svengali-like reality-distortion field), continued with this precious quote. "IBM offers the most advanced processor design...and this is just the beginning of a long and productive relationship." (Emphasis added.)
The reality: Apple dumped IBM, Motorola, and the PowerPC in 2005 and it was revealed later that the Mac operating system had been leading "a secret double life" for about five years. But the PowerPC platform had really failed long before 2005. Look no further than these comments from an IBM marketing manager in this 1997 Electronic News article: "Many business school case histories will be written about this failure," Jesse Parker, marketing manager at IBM Micro, said at that time. "No one of the three companies involved in PowerPC executed on their plans. IBM didn't. Motorola didn't. And Apple didn't," he said.
The original PowerPC project was conceived by John Sculley, president and CEO of Apple, and Jack Kuehler, vice chairman of IBM. Phil Hester, an IBM manager at the time, and David Mothersole, a Motorola executive, where also instrumental in starting the project, known initially as "Somerset." But as the PowerPC came to market, Mr. Sculley was pushed out of Apple (and) Mr. Kuehler retired. Their replacements did not have the same enthusiasm for the PowerPC alliance, dooming the project.
In short, the PowerPC failed to challenge Intel in the PC market in a big way. (Though it has been reincarnated as IBM's Cell processor that powers Sony's PlayStation and the architecture still powers IBM servers.)
And I have my own vignette to relate that illustrates one reason why Apple eventually dropped the PowerPC. When Apple first began to crow about the dual-processor Power Mac (circa 2003), a neighbor of mine at the time bought into the hype and purchased an Apple Power Mac tower with two IBM G4 processors (this preceded the dual-processor G5 tower that followed soon after). This thing was a furnace. It quite literally raised the temperature in the room it was in, had about five fans too many, and was deafening, to boot. That was the first time I fully understood the magnitude of Apple's fabrications about IBM's "superior" PowerPC designs. (IBM's less-than-impressive--at that time--chip manufacturing process that was used for PowerPC processors also contributed to the problem.)
AMD Puma
Lastly, turning to Advanced Micro Devices, I'll try to look beyond the botched Barcelona launch in September of 2007 (as I've already covered this in Part 1) and focus instead on AMD's mobile "Puma" platform. Though I can't leave Barcelona entirely out of the discussion because there are some disturbing parallels. (Note: AMD's upcoming Yukon and Congo platforms offer some hope for mobile redemption.)
The hype: Like Barcelona, AMD had too much to say about Puma too long before it was real. AMD started pumping Puma back in April 2007 when the company did one of its many (infamous) soft launches (a PR strategy that it has thankfully ditched). This prompted some editorializing from me as well as other publications. "It is questionable whether...Puma will meet the hype AMD is currently trying to generate though these early announcements," according to a rare editorial from DigiTimes in 2007.
Things got even more dicey at the financial analyst day in December 2007 when AMD said Puma would be delayed until the second quarter of 2008.
It's not so much that Puma (aka Turion X2 Ultra coupled with ATI graphics) is a failure of epic proportions like Itanium, it's that the CPU component (separate from the ATI GPU component) fell so far short of the long, ballyhooed build-up it got.
And it is beaten consistently by Intel in the mobile marketplace. Here's an October 2008 CNET review of laptops with AMD's Turion X2 Ultra. "Turning to AMD's 2.0GHz Turion X2 Dual-Core RM-70 CPU might be an option if you're looking to keep costs down and have only basic computing needs," the review said. "The Acer Aspire 5735-4624 costs only $499 and uses a 2.0GHz Intel Pentium Dual Core T3200; it completed our multitasking benchmark test in one-third the time the HP G60 did. The HP G60 wasn't the last-place performer in our mainstream midprice holiday retail laptop roundup. That dubious distinction goes to the Toshiba Satellite L355D-S7825, also an AMD-powered system."
And there are more unfavorable comparisons. This review at Hexus.net of a Toshiba Satellite 300D with AMD Turion X2 Ultra ZM-80 said the Turion X2 Ultra CPU was "found wanting when compared to Centrino 2." The one bright spot was the ATI Mobility Radeon HD 3650 graphics card--but this speaks more about solid ATI technology than AMD's shaky Turion processor.
Intel has delayed a high-end server chip, billed as the world's first 2 billion transistor microprocessor, originally expected as long ago as 2007.
Tukwila is a quad-core update to the Itanium processor, which has had a less-than-promising run since the original version was announced back in 2001. The chip's architecture--based on explicit instruction-level parallelism--is a radical departure from the x86 design used in PCs today. It was believed at one time that Itanium would replace x86 chips in many Intel-based computers.
So, what's delaying it this time? Intel has "made some engineering enhancements to the Tukwila platform," according to a statement Thursday from Intel. As one enhancement, Tukwila and its follow-on processors on the Itanium roadmap, Poulson and Kittson, will be socket compatible. Intel is also introducing a technology called "scalable buffered memory" to Tukwila platforms. Scalable buffered memory enables higher memory capacities and uses newer DDR3 (Double Data Rate, third generation) memory.
The number of CPU pins that can be dedicated to a memory interface is limited. Scalable buffered memory acts as a memory expander by connecting to multiple DDR3 RDIMMs (Registered Dual In-line Memory Modules) for each CPU memory interface, thus increasing the total memory capacity compared to natively attached memory configurations, according to an explanation provided by Intel.
"Validation testing of the Tukwila processor with the new memory will move the Tukwila platform launch from early '09 to mid '09," and Intel spokesperson said.
Current and past users of Itanium processors include Hewlett-Packard, SGI, NEC, Fujitsu, Unisys, and Hitachi.
The processors are targeted at high-end enterprise servers and high-performance computing systems.
UPDATE An alliance to promote the use of Intel's Itanium processor said a quad-core version of the chip will come out in early 2009. The alliance also disclosed more than a 30 percent jump in volume growth year to year and an update to its partnership with Microsoft, among other announcements.
Itanium is an Intel 64-bit architecture for high-end servers. It differs from the x86 architecture, widely used in PCs. One principal difference is the way the compiler is used. Because the compiler makes decisions about the parallel execution of instructions, the processor can execute up to six instructions per clock cycle.
(Credit:
Itanium Solutions Alliance)
The quad-core version of the Itanium, code-named Tukwila, will come out in early 2009 and be one of the first "monolithic" quad-core designs from Intel, said Rob Shiveley, worldwide marketing manager of the Mission Critical Server Platform Group at Intel. A monolithic design puts all four cores on one piece of silicon (called a die). To date, Intel has built its quad-core processors by combining two dual-core processor dies. AMD's "Barcelona" Opteron quad-core design puts all four processors on one die.
This will give system vendors the ability to deliver eight-socket systems with up to 32 cores (4 cores per socket), said Mike Mitsch, General Manager, Enterprise Servers, IT Platform Group, NEC Corporation of America.
Tukwila will also have an integrated memory controller and QuickPath Interconnect technology, Shiveley said. This will increase the data transfer rate within the processor. Intel's "Nehalem" processor--due in the fourth quarter of this year--will also be a monolithic design with an integrated memory controller and QuickPath Interconnect technology.
On-chip cache memory for Tukwila will also be increased from 24MB to 30MB. "It is focused on database performance"--that's what the large 30MB cache is for, Shiveley said. Windows Itanium platforms are used, for example, to consolidate a larger number of SQL database servers.
Using two billion transistors (the bulk of the transistors are allocated to the large cache memory), Tukwila will be based on Intel's 65nm process technology and will initially have a clock speed of up to 2GHz at both 170 watts and 130 watts.
NEC's Mitsch said Itanium also supports extremely robust error detecting and error correcting memory. "We're in the process of qualifying up to two terabytes of logical (memory)," he said.
On another front, the Itanium Solutions Alliance announced that worldwide annual Itanium-based factory system revenue and system volume continued to grow in 2007, with a year-over-year increase of 30.8 and 36.3 percent, respectively. The Asia-Pacific region led the way, with year-over-year growth in factory system revenue and system volume of 61 percent and 45 percent, respectively. In Japan, Itanium-based revenue exceeds all other non x86 server platforms, the alliance said.
The alliance also said it is collaborating with Microsoft to deliver new programs and tools to help businesses migrate from "costly legacy RISC systems and mainframes" to Itanium-based platforms. Currently, a fourth of the more than 13,000 Itanium-based applications are Windows Server-based, the alliance said.
Other recent developments include: an NEC enterprise server with dynamic hardware partitioning functionality for Microsoft Windows Server 2008--the NEC Express5800/1320Xf. Sun Microsystems previewed Itanium-optimized Java SE 6 running on an Itanium-based server in several different SOA (service-oriented architecture) scenarios.
The biggest flops flop big because of hype. Supposedly sure bets get massively pumped up, then poop out fast. Ishtar couldn't lose with Dustin Hoffman and Warren Beatty, but it lost Columbia Pictures $40 million. Heaven's Gate was an "epic film" that lost $44 million, forcing Transamerica to sell off United Artists.
Boondoggled technology can be even more of a financial disaster, marketed year after year, at great expense, until one day the company either pulls the plug or relegates it to practical oblivion. With this in mind, I have come up with a few chips that have been hyped far beyond what was delivered. Some are still in the process of flopping (but may ultimately redeem themselves, somehow, as 20th Century Fox's Cleopatra did).
Cyrix 5x86 "M1" processor
(Credit: Photographer: Dirk Oppelt )First, a short preface. A post was brought to my attention a few weeks ago. The point was essentially this: A decade or so ago, Cyrix's M1 and M2 processors were hyped as the fastest, most powerful chips when they were really no such thing. Reality caught up with the hype, and, to use the author's own words, "Cyrix imploded and National Semiconductor blew I-don't-know-how-many-billion dollars cleaning up the mess."
The author also implies that the press too often buys the hype. Let's see. When I'm faced, for example, with a new chip that looks like a dog, I don't write something the next day saying so. Even if the company's marketing department is knowingly putting lipstick on the dog (or the pig, take your choice) I won't necessarily know this right away. So, I give them the benefit of the doubt. Until I know otherwise of course.
This is best illustrated by, not coincidentally, the Cyrix M1 processor. When I covered Cyrix, I'm sure I wrote stories overstating the threat the M1 posed to Intel. But it didn't take long to figure out that the M1 was not what Cyrix said it was. The moment of truth happened at a San Francisco chip conference. The die for the M1 had recently been released. And it was big. As in oversize. I remember discussing this with a chip analyst and while the exact phraseology escapes me, I believe concepts such "big" and "hot" and "waffle iron" were bandied about. My reaction was this: OK, so the marketing is fraudulent. But the company is small and they seem to be struggling. Plus, they do have one large customer planning to use their chips (why, I don't know) but I won't write a piece tomorrow slamming the waffle-iron-size die, I just won't take them seriously anymore.
So, how many M1s have come and gone over the years? And how many are out there now? That's what I will try to address here, the first of a series of posts focusing on flops. With one very important disclaimer: This is opinion, and opinion only.
Cyrix M1/M2: In addition to what is stated above, let me say this: Even back in the mid-'90s when I didn't know how bad the M2 (the successor to the M1) would be, I would have never purchased a computer with a Cyrix processor no matter how inexpensive. Why? Simple, the chips were slow.
The Intel Itanium : The hype: "This design...will one day replace RISC and CISC. It is a gateway into the 64-bit future" or "I expect Itanium to replace Xeon, but not until 2003." The reality: Development took place over 11 years, from 1989 to 2001. Despite this, when it was released it was not competitive. Not many were sold (some claim only a few thousand original Itaniums) because of relatively poor performance, in addition to poor yields and high cost. Then the other shoe dropped with Itanium 2. In September 2005, Dell said it would phase out its remaining models based on the Itanium--"another sign of the waning interest in a chip that cost an estimated several billion dollars to develop" (The Wall Street Journal, Sept. 15, 2005). In short, AMD's Opteron was a lot better.
The original version of the "Barcelona" quad-core Opteron: This is a work in progress so I don't want to be too harsh. Barcelona may ultimately succeed as the B3 stepping is adopted over the coming months. So here I will simply focus on the initial hype. The hype: "Barcelona doesn't get us back in the game (with Intel), it puts us in a leadership position" (CEO Hector Ruiz, Sept. 10, 2007). The reality: the chip was announced in September but systems (as of March 6, 2008) have yet to appear from Hewlett-Packard, IBM, Dell, or Sun.
Let's be perfectly frank: the hype surrounding Barcelona before it came out was nothing short of scandalous. Better, faster, more innovative. Let me paraphrase AMD's marketing: the Barcelona is a native quad-core design as opposed to Intel's kludgy chip that cobbles together two dual-core die. Barcelona has an on-die memory controller, Intel does not... I'll stop there. (If you want more, go to AMD's Web site.)
Advice for AMD: Hold the superlatives. First deliver in quantity the actual, viable physical chip that's supposed to do all these things better than the shipping Intel chip (shipping since October 2006). The adage "talk is cheap" has special meaning to journalists. And, I would imagine, special meaning to AMD's waiting customers.
- prev
- 1
- next
