Treeduck

Intel put itself far ahead of AMD technically last year with its Core i7 desktop CPUs, but the high-end prices for the Core i7 900-series made Intel's most advanced chip architecture more of a luxury than an industry standard. Monday's announcement of Intel's new, more affordable Core i7 800-series chips, as well as an even cheaper Core i5 CPU, will likely lead to Intel's most advanced chip penetrating the mainstream retail market.
Intel has three new chips to announce, as well as the new Intel P55 Express motherboard chipset to support them. The new Core i7's include the $562 2.93GHz Core i7 870, the $284 Core i7 860 at 2.8GHz, as well as the $196 2.6GHz Core i5 750 chip. Each is essentially a stripped-down version of its counterpart from the Core i7 900-series, the most affordable of which, the 2.66GHz Core i7 920, starts at about $280.
The technical sacrifices in the new chips are relatively minor. The new Core i7's have a double-channel memory interface, as opposed to triple-channel RAM in the Core i7 900's. That means new Core i7-based PC owners won't have quite as much RAM throughput, but they also save money by only having to buy two sticks of DDR3 at a time, as opposed to three with Core i7 900.
(Credit: Intel)
The sole Core i5 chip has the same two-channel memory limit, and Intel has also stripped out the Hyperthreading capability. Hyperthreading is an Intel technique that effectively doubles the number of processing threads (adding four virtual threads to the four physical CPU cores) depending on the workload. Heavy multitaskers and those who use multithreaded software will feel the loss here, although Intel's current mainstream Core 2 Quad family, which the Core i5 may replace, has no Hyperthreading either.
To build a desktop PC around either new chip, you'll also need a new motherboard using Intel's P55 Express chipset. We've already mentioned the change to the memory interface. The next most significant change has to do with the graphics bandwidth.
Like the old Core i7-compatible Intel X58-chipset motherboards, the P55 Express boards support multiple graphics cards via Nvidia's SLI and AMD's CrossFireX technologies. The difference with P55 is that you only get half the graphics data bandwidth as with X58. You still get full 16x PCI Express throughput if you use a single card on P55 Express, but install two cards and the graphics slots become two 8x slots, as opposed to full dual-16x on X58.
PC gamers on the cusp of purchasing a mainstream or a higher-end gaming PC may face a conundrum due to the graphics bandwidth limit. But for most price-sensitive gamers, one decent midrange 3D card will provide a more than adequate gaming experience, so the graphics bandwidth limit isn't a major loss for the new chipset's intended market.
So how fast are these new CPUs? We tested a PC from Falcon Northwest Talon with a Core i7 860 overclocked from 2.8GHz to 3.39GHz and found it competed well against Core i7 920-based PCs that cost about $500 more. You can read the full review here. We also have a $1,300 Core i7 860-based system from Velocity Micro on deck for later this week.
For standalone chip reviews, PC Games Hardware found the Core i7 860 as fast or faster than either the Core i7 920, or AMD's Phenom II X4 965 on most of its benchmarks. The Core i5 even squeezed out a few wins of its own. The other enthusiasts sites haven't posted their coverage yet, and we'll add the appropriate links as other reviews come online, but the early results indicate that the new Core i7 and Core i5 will achieve Intel's goal of bringing its latest architecture available to mainstream consumers in a competitive package.

The R2 update for Windows Server 2003 was mostly about bug-fixes plus the odd feature tweak, but that's not the case with Windows Server 2008 R2. Expected around October, WS 2008 R2 is a far more important release with a lot more in the way of new features and functionality than you might normally expect from this kind of mid-term update.

With WS 2008 R2, there's no 32-bit version, which clearly means that a lot of customers will require new hardware before they can upgrade. However, as Microsoft points out, it's more than two years since 32-bit servers disappeared from vendors' portfolios so this shouldn't be a major issue. Moreover, the 64-bit platform can deliver a lot more scalability, with support for up to 256 logical processor cores in the R2 server itself and 64 logical cores per virtual machine (VM) in the updated Hyper-V hypervisor — about which more shortly.

Another change is the use of the same core software for both Server 2008 R2 and the upcoming Windows 7 desktop OS. Indeed, the server product was originally referred to as Windows Server 7. Unfortunately the name didn't stick, but the resemblance is clearly visible in the GUI, with a lot of shared code under the surface too.

New from the start
We've been evaluating Windows Server 2008 R2 release candidate for a while, and our impressions have all been good. Installation is straightforward, with — as in the 2008 release — a choice of a full graphical install, a headless Server Core setup or a web server setup. We opted for the full install, and the process took just under 30 minutes on our test servers, with minimal operator intervention required. In fact, it was very much like installing Windows 7 and — if it's at all possible with a server OS — really quite fun.

As with previous releases of Windows Server, you need to configure optional server roles before you can do any real work.

Just like the previous release, you end up with a minimal OS installation and must configure optional roles and features before the server can do any real work. Again, we found this easier and quicker to do in the R2 release. In particular, the enhanced Server Manager application is both slicker and more capable. Also, it can now be installed on a PC to configure and manage multiple servers remotely — a long-overdue option.

The Server Manager tool is enhanced in the R2 release and can now also be run remotely.

Better domains
For our evaluation we configured a number of different roles, including Active Directory Domain Services, making our server a Domain controller. This used to be quite an involved process, but proved a lot easier and quicker on the WS 2008 R2 server. We also liked the new Active Directory Administrative Centre, which delivers a kind of one-stop shop for all AD management and eliminates the need to trawl through multiple MMC snap-ins. It takes a little getting used to, but should prove invaluable.

The Active Directory Administrative Centre provides a one-stop shop for all your AD management needs.

There's also a new Active Directory recycle bin, to undo accidental changes to the directory. However, if you're expecting a graphical tool like that on the Windows desktop you'll be disappointed — it's there, but you have to interact with it via the command line.

Virtually better
Of course, the headline feature in Windows Server 2008 R2 has to be the new and improved Hyper-V hypervisor with its Live Migration support, whereby virtual machines can be moved from one server to another without having to be shut down.

Windows Server 2008 R2 includes an updated Hyper-V role with the long-awaited Live Migration feature to enable VMs to be moved around without being shut down.

The current release features Quick Migration — which is, admittedly, pretty quick already, requiring just a few seconds of downtime. Live Migration, however, is even faster, taking just milliseconds. It's a long-awaited feature, not least because the market leader VMware has had this capability for years. Unlike VMware's offering, Live Migration doesn't cost extra and isn't particularly difficult to configure. However, it does require servers with processors from the same vendor, and these have to be configured as a failover cluster with a common storage location created using WS 2008 R2's new Cluster Shared Volumes (CSV) feature. Unlike Microsoft's previous attempt at clustering, this is surprisingly easy to configure and therefore much more likely to be implemented.

We've already mentioned the ability to support up to 64 logical processors per VM with the new Hyper-V. Performance is further enhanced by support for a processor technology called Second Level Address Translation (SLAT). Each VM can also address up to 64GB of memory, storage can be added without a reboot, and physical as well as virtual machines booted from a common virtual (.vhd) disk image.

Other enhancements
A lot of customers are likely to adopt Windows Server 2008 R2 simply to get the new virtualisation features, but there's a lot more on offer. IIS 7.5, for example, is the next generation of Microsoft's web server; there's also support for both IIS and .NET on Server Core, and a new Active Directory module for Windows PowerShell.

Terminal Server/Services also comes in for a revamp in WS 2008 R2, with yet another new name — Remote Desktop Services — to reflect the emphasis on Virtual Desktop Integration (VDI). To this end, remote desktops get better graphics support and audio facilities plus the ability to publish shared applications in such a way that users will find it hard to tell programs executed remotely from those installed on the local PC.

Terminal Server/Services get a makeover in WS 2008 R2 and a new name — Remote Desktop Services.

Another welcome addition is a new kind of remote access technology called DirectAccess, which should prove more seamless and easier to manage than a conventional VPN server. The only drawback is that it's only available for use with Windows 7 clients and requires the use of IPv6.

Basic document/file management is available in WS 2008 R2 using the new File Server Resource Manager tool.

Finally, at the very last minute, Microsoft has managed to include its own nascent document management technology into the WS 2008 R2 release candidate. A selectable component within the File Services role, File Classification Infrastructure (FCI) includes a new File Server Resource Manager tool to classify documents and files and use that classification to move files around and automate management. You're a little limited in what you can do in this first implementation, and there's little opportunity to automate file classifications, for example. However, it's a promising start that's likely to worry a lot of the specialist document management vendors.

There really is a lot in Windows Server 2008 R2 — much more than we can fit in here — that will be of interest to both small and large businesses. Some of the newer technologies introduced for the first time need further development, such as FCI and the Active Directory Domain Administrator, but that shouldn't put you off. Indeed, our only major concern would be customers assuming the R2 release to be a simple service pack-style update. It's not, and those wanting to upgrade will need to plan for deployment in much the same way as for a completely new OS.

IBM's Roadrunner and Cray's Jaguar have retained their No. 1 and No. 2 rankings on the Top500 supercomputer list, which is drawn up twice a year.

Roadrunner, the first machine to break the petaflop barrier a year ago, registered 1.105 petaflops to top the June 2009 list, published on Tuesday. (One petaflop is 1,000 trillion floating-point operations per second.) Built in 2008, the system is housed at the US Department of Energy's (DOE) Los Alamos National Laboratory.

Jaguar, installed at DOE's Oak Ridge National Laboratory, reached 1.059 petaflops. The result was a repeat of the rankings in November, when the machines led with the same results.

Eight of the top 10 systems in the June list are housed in the United States, while two machines are based in Germany.

In third spot was another Big Blue machine — a new BlueGene/P system called Jugene, installed at the Forschungszentrum Juelich (FZJ) in Germany. It achieved 825.5 teraflops and has a theoretical peak performance of just above 1 petaflop, according to the Top500 Web site.

FZJ is also home to Juropa, the new No. 10 system, which registered 274.8 teraflops. It is built from Bull Novascale and Sun Microsystem SunBlade x6048 servers.

Two other high-performance computing (HPC) systems in the top 10 are new: the Kraken, a Cray XT5 machine installed at the National Institute for Computational Sciences at the University of Tennessee, at No. 6; and anIBM BlueGene/P system called Dawn, installed at the DOE's Lawrence Livermore National Laboratory, at No. 9.

The highest-ranked UK supercomputers are two machines at the European Centre for Medium-Range Weather Forecasts in Reading, which took 25th and 26th place.

Overall, there were significant gains in performance, said the Top500 team. The last system on the June list would have been listed at No. 274 in the November rankings, the team said.

The United States' share of supercomputers stood unchanged from the previous list at 291, while the European share dropped to 145 from November's 151. Asia was home to 49 of the HPC systems, up from 47.

While IBM is ahead by overall installed performance, its share of the total number of systems on the latest Top500 list was 24 short of rival Hewlett-Packard's 212 systems.

A total of 399 systems, up from 379 previously, are powered by Intel processors. According to an Intel statement on Tuesday, 33 of these machines tap into the Xeon 5500, launched in March.

IBM Power processors were used in 55 systems, down from 60 last November, while AMD's share of the systems decreased by 16, to the current 43.

IBM and the Swiss Federal Institute of Technology at Zurich plan to build a water-cooled supercomputer whose surplus heat will be re-used to heat the university's buildings.

The Aquasar supercomputer will be located at the ETH Zurich facility, and it will start operations next year, the partners said in an announcement on Tuesday.

The supercomputer will combine two rack-mounted IBM BladeCenter servers, each containing multiple blades with a mixed population of IBM PowerXCell 8i and Intel Nehalem processors. It is expected to deliver a peak performance of about 10 teraflops.

The installation will re-use heat directly for in-building heating. IBM estimates that the watercooling scheme will reduce the system's carbon footprint by up to 85 percent and save up to 30 tons of CO₂ annually, compared with standard cooling approaches. The comparison calculations are based on average yearly operation of the system and on in-building heating energy being produced by fossil fuels, the company said.

The energy-consuming refrigeration units used by almost every datacentre consume about half of the a datacentre's energy. Aquasar will need no such equipment. As a result, it should reduce overall energy consumption by 40 percent, according to IBM.

"Energy is arguably the number-one challenge humanity will be facing in the 21st century. We cannot afford anymore to design computer systems based on the criterion of computational speed and performance alone," Professor Poulikakos of ETH Zurich, the leader of the Aquasar project, said in a statement. "The new target must be high-performance and low-net power consumption supercomputers and datacentres. This means liquid cooling."

The system is the product of an extended joint research project between ETH and IBM scientists, focused on chip-level water-cooling. It also encompasses a concept for "water-cooled datacentres with direct energy re-use" proposed by scientists at IBM's Zurich Lab.

Aquasar's use of warm water rather than cold water for cooling is unique and IBM-patented, a spokesman for the company said. Water, which is about 4,000 times more efficient as a coolant than air, will enter the system at 60 degrees C. This will keep the chips in the system at operating temperatures below their maximum of 85 degrees C, according to IBM.

The high input temperature of the coolant results in an even higher-grade heat as an output, which in this case will be about 65 degrees C, the company said.

The system uses jet impingement cooling, which means that water makes direct contact with the back of the chip via micro-channels in the heatsink, according to research papers by the IBM and ETH scientists involved in the Aquasar project. "This method incurs neither the thermal resistance overhead of a base plate, nor the overhead and reliability problem of thermal interface materials, and thus is promising for removing highest-power densities," according to one paper.

Pipelines from the individual blades link to the server rack's waterpipe network, which in turn is connected to the main water transportation network. Aquasar will need about 10 litres of water for cooling, pumped at some 30 litres per minute, IBM said. The cooling system is a closed circuit: the water is heated by the chips and cooled to the required temperature as it passes through a passive heat exchanger, delivering the removed heat directly to the heating system of the university.

Aquasar will be used by the computer science department at ETH Zurich for multiscale flow simulations related to nanotechnology and fluid dynamics. Researchers plan to show that solving scientific problems efficiently can be performed in an energy-efficient manner.