

Personal blogging site for technology, business and everyday musings.
Warren East, CEO of ARM Holdings says there will be over 10 ARM-based Netbooks on the market by year end. That puts a lot of pressure on Microsoft to port Windows to the ARM processor, or does it? Here's the dilemma. If Microsoft ports to the ARM, Intel could optimize x86 CPUs like Atom for Linux (and Android). Even though Microsoft has embedded versions of Windows for kiosks & cash registers, it's not clear to me the ARM processor could handle a scaled down version of Windows. Not to mention how it would perform against a lightweight Linux ARM implementation. Intel wins either way. Apple's not playing this game; they've segmented the processor platform of the Intel MacBooks from the iPhone. They will most likely develop their own MPU/GPU combo chip as I've discussed here before. Microsoft should stay the course and stick with Intel.
With Cisco's acquisition of Tidal Software last week, the company continues to beef up its software portfolio beyond its core networking business. Tidal makes performance monitoring software for enterprise applications like Oracle and SAP that run in large data centers. While $105 million is not a big bet for Cisco, I see this as further convergence of computing and communications in both the enterprise data center and ultimately in the cloud. There is innovation at the core OS level to optimize performance for Intel's new Xeon 5500 processors. Microsoft Hyper-V, Sun Solaris and Red Hat Linux have been extended to enable performance and energy efficiency improvements available using Intel's extended page table memory access. This allows the hypervisor to bypasses kernel software codepaths altogether and map directly to virtual guest instances. Advanced memory management using Intel's QuickPath enables the ability to virtualize previously uncommon I/O workloads such as database & file serving. This is how we will see network layer virtualization evolve, reducing power consumption while creating efficient memory use along the way. All these features, combined with Cisco's California architecture, are coming together to disrupt the computing core. Moving up the data center management stack with tools such as Tidal makes a lot of sense for Cisco, because at the end of the day, the key will be application performance.
Dell introduced the new Dell Precision workstations based on the Xeon product line from Intel today. As the master of PC industry road-map, Intel tells IBM, Dell, HP and other PC manufacturers what they are building next and when. These processors also power the new Mac Pro and Apple has a nice summary of the architecture on their site. Nehalem (Xeon 3500/5500) utilizes the 45nm process and features the latest micro-architecture. The next stop will be the shrink of the die called "Westmere" in a 32nm process. Intel has a well thought out roadmap - shrink, then innovate, then shrink again (Tick/Tock). This gives OEMs a good runway for product planning and price/performance curve forecasting. Now with Cisco's entry into the datacenter space, they will also be a key stakeholder in the future generations. Much has been said about virtualization enhancements with this architecture but the real key is on-chip memory control. Improved memory management is the key to supporting more VM instances on the bare metal. We'll see the rest of the industry show their hands on Monday.
Cisco is taking virtualization from the network layer up as an alternative approach from a data center architecture perspective. These are not ordinary blade servers being announced today. The new Unified Computing System from Cisco will have a simplified infrastructure, meaning fewer management modules, I/O bays and a more open backplane. Nehalem-EP processors can support half-blade or full-width blade configurations providing a great deal of flexibility. The release states "patented extended memory technology" to support applications with large data sets. They've taken an approach to economically address large logical memory requirements with better physical DIMM utilization (think lower cost). Supporting iSCSCI, Fiber channel and Fiber Channel over Ethernet gives customers the option of NAS or SAN support and a built-in upgrade path. Much of the news most likely will focus on HP, Dell and IBM and what this means for Cisco's large strategic partners. The blades could cut both ways.
MetaRAM, a new startup recently announced new technology to produce less expensive DRAM by packing four times the number of 1 Gbit DRAM chips per module. To give some perspective, a high-end PC server with 256 Gbytes of RAM in a four-socket server can cost up to $500,000. Fred Weber, the CEO was the former CTO from Advanced Micro Devices. The cost savings by some manufacturers of 8 Gbyte DIMMs, such as Smart Modular, are significant. They expect to ship new modules using MetaRAMs chips for $1500 vs. $5000. This can cut the price of an end system by a factor of 10. An eight-way super PC server with 256 Gybtes of RAM can be had for about $50,000 vs. half a million. Why is this interesting? As you can affordably pack and dense real memory, you can run an entire enterprise database in RAM versus from a storage device. You reduce the latency and performance issues of data access between the microprocessor and the I/O device. What the processor is looking for would be held in memory. This also makes virtualization a better value proposition because you can increase the amount of virtual machines running various applications side by side within the same physical platform. This reduces power consumption, heat and maintenance from a data center perspective. These chips are made at TSMC in their 180nm technology (huge by nm standards). Sun recently announced they picked TSMC to build the SPARC chips they design. Sun joins TI in the move away from internal fabrication of chips. Historically, this was seen as a major disadvantage as merchant semiconductor manufacturers were always one or two generations behind U.S. semiconductor companies, if we were sub-micron, they were 1 micron or more.
My illustrious CEO at one time, Jerry Sanders from AMD once said, "real mean have fabs." The wafer fab has long since gone offshore now along with contract manufacturing and other services. Intel's new ultramobile microprocessor are 45 nm, TI stopped advancing at 65nm. Many analysts believe the foundries can achieve large production volumes quickly by serving multiple chip designers. TMSC is down the 45nm learning curve today. The only problem is that learning insights from process technology accrue to the manufacturer, not the designer. Chip design used to be a closed-loop, fully end-to-end design process where the nano-world belonged to the designers. The transistor logic, lithography, steppers, reticles, wafers etc. all came together in a beautiful process that began with taping out the first silicon and testing the output of a once sophisticated wafer fab process. Has the wafer fab gone the way of refineries? No one wants to build one here anymore.