TechieGroups

Netbooks aren't dead, but they're not quite the hip topic of discussion they were a few years ago. The focus on cost conscious computing hasn't changed since the introduction of the first netbook, but the maturity of tablets has. Intel still sees the netbook segment as a profitable one (for itself) through 2012, although I'm fully expecting the line between netbook and convertible tablet to blur after the launch of Windows 8. 

Atom has been at the heart of nearly all netbooks since the segment's inception. We've seen only one major platform shift since then: from the original 2008 Atom platform to Pine Trail. Pine Trail integrated the GPU and memory controller without significantly changing the Atom architecture. Today Intel is officially announcing its next major netbook platform shift: Cedar Trail.

While the original Atom and Pineview (Pine Trail's Atom) were built on Intel's 45nm process, Cedar Trail moves to 32nm. Cedar Trail's SoC shrinks to 56mm2, finally making it smaller than AMD's Zacate APU. The underlying CPU architecture hasn't really changed, nor have cache sizes (512KB L2 per core) or clock speeds (1.66GHz and 1.86GHz parts available), so what this is really about is a reduction in power consumption. 

There are three Atom CPUs being offered as a part of Cedar Trail: the N2600, N2800 and D2700. Just as before, the N-series are for netbooks while the D-series are for desktops. All of the Cedar Trail Atoms are dual-core parts, but they all slot into the same power envelope as the old single-core Pine Trial platforms (5 – 8W). The only exception is the D2700 which is a 10W platform. Note that this is the total TDP for the Atom SoC + the NM10 Express chipset (providing USB, LAN, PCIe, etc…). 

The spec breakdown is below:

Given the same number of cores and the same clock speeds, CPU performance shouldn't go up compared to Pine Trail. Since everything is now dual-core we should see a boost at the low end, but I wouldn't expect to see CPU performance that's better than Zacate. 

Cedar Trail now supports DDR3-800 and 1066 (up from 667MHz max data rates before). The bigger change is the GPU. The GMA 3150 used in Pine Trail was an Intel Gen graphics derivative (45nm GMA 3100), however Cedar Trail now features a PowerVR SGX 545 sourced from Imagination Technologies. At 640MHz in the N2800, we've never seen the SGX 545 run at anywhere near this clock speed before so it'll be interesting to see how well it performs. Intel is claiming a > 2x GPU performance improvement compared to the GMA 3150 in Pine Trail in 3DMark 06. The big question is Windows driver maturity, but we'll find out soon enough as systems based on Cedar Trail are in production now and are expected to ship in early 2012. Expect to see Cedar Trail netbooks from ASUS, Acer, HP, Lenovo, Samsung and Toshiba for starters.

The new graphics block also includes support for H.264 video decode acceleration (we're still digging for specifics) as well as Intel Wireless Display technology. Note that WiDi support will vary depending on the system and price point:

Intel is expecting the vast majority of Cedar Trail netbooks to be sold in the 9 – 9 price point. At 9 is where you'll likely find features like WiDi as well as potentially fanless designs. Don't expect any of those new form factors at 9 until the later part of next year, likely coinciding with Windows 8's release.

Overall the addition of HD video decode support and lower power consumption are both nice features to have, but I'm skeptical as to whether this will be enough to carry Intel based netbooks throughout the majority of 2012. Atom is in dire need of an architecture update (something we'll get in 2013) and the netbook as a platform is in need of a refresh. I do hope to see some manufacturers taking risks with slim, fanless Cedar Trail based designs next year but we'll have to wait and see if they're any good.

Earlier this week Computer World published a story about Intel adding x86 optimizations to the Ice Cream Sandwich release of Android (4.0). The story itself was fine but a quote in the article was a bit confusing:

"Ice Cream Sandwich includes OS optimization for x86, so Intel architecture-based devices can support it," said Suzy Greenberg, an Intel spokeswoman.

This statement was taken by many to assume that Intel had somehow done some optimization to Ice Cream Sandwich, however that can't be true as it flies in the face of what we know about how Google operates.

For each major Android release, Google first and foremost chooses a device OEM (e.g. Samsung for ICS, Motorola for Honeycomb). Together with the device OEM, Google selects components to be used for this Android launch vehicle – including the SoC. All of the SoC players bid. The selection process is a bit mystified however it seems that Google takes into account how well you did last time in picking you for the next round. Deliver killer performance on or ahead of schedule? Good for you. Let schedule slip or force Google to push back its launch? There may not be a next time. That's oversimplifying it a bit but you get the drift.

Only Google, the OEM selected and the SoC parter get access to the next Android code at this point. The rest of Google's partners are free to work on the Android Open Source Project (AOSP), but they don't get access to the new version of Android until it's added to the project.

For Ice Cream Sandwich, only Google and Samsung have access to the source. Sometime very soon, ICS will be added to the AOSP and then all of the SoC guys get to have at it. But where does this leave Intel and the statement above about ICS including OS optimization for x86?

It's not as exciting as it sounds.

Intel has been contributing x86 patches to the AOSP and Google's internal developer branch for the past two years. Today if you're to download Android, you can build a version that should run on x86 just fine. All of Intel's x86 support should be included as of Android 2.3.7. 

When Google goes about developing a new version of Android, it takes the current AOSP and branches off from there. Any patches/optimizations that have been committed at the point of the branch make their way into the next version of Android. See where I'm going?

Any patches/optimizations, including those by Intel, that were present at the time of the branch made their way into ICS. Therefore, ICS does technically include OS optimization for x86.

Intel doesn't have access to ICS, nor is it any closer than all of the other SoC vendors to putting out an optimized ICS build. When ICS is open sourced, Intel along with NVIDIA, Samsung, Qualcomm, et al. will begin work on optimizing the OS for their platforms.

Five years ago Intel announced its ambitious tick-tock release cadence. We were doubtful that Intel could pull off such an aggressive schedule but with the exception of missing a few months here or there tick-tock has been a success. On years marked by a tick Intel introduces a new manufacturing process, while tock years keep manufacturing process the same and introduce a new microprocessor architecture. To date we've had three tocks (Conroe, Nehalem, Sandy Bridge) and two ticks (Penryn, Westmere). Sampling by the end of this year and shipping in the first half of next year will be Intel's third tick: Ivy Bridge.

Read on for our analysis of Intel's 2012 Core microarchitecture.

A month ago, Intel introduced its newest quad Xeon E7 series, the successor of the Xeon 7500. We gave you an overview of the Xeon's E7 lineup, but we didn't have benchmark results yet. According to Intel, the new Xeon established 16 new world records, beating similarily configured IBM POWER servers in some workloads and SPARC servers that are twice as expensive by a wide margin in almost any workload.

Today, we'll look at the available SAP S&D measurements, along with our results from weeks of virtualization benchmarking on ESX 4.1 update 1. By looking at both response time as throughput in our own virtualization benchmarking and power measurements with vApusMark, we get a very realistic idea what the the Xeon E7 is exactly capable off. We'll compare the best Xeon with AMD's Opteron 6174 (Dell R815) and its predecessor, the Xeon 7560.

Brooke Crothers broke a very important story today – he published the name Silvermont. Atom's first incarnation came to us in 2008 as a Pentium-like dual-issue in-order microprocessor. The CPU core was named Bonnell, after the tallest point in Austin at around 750 feet. Small mountain, small core. Get it?

Bonnell and the original Atom were developed on a 5-year cadence, similar to how Intel ran things prior to the Core 2 revolution (the P6 to Netburst/Pentium 4 move took 5 years). With the original chip out in 2008, five more years would put the next major architecture shift at 2013, which happens to be exactly when the Cnet report mentions Silvermont will be introduced.

When I first met with the Atom design team they mentioned that given the power budget and manufacturing process, the Bonnell design would be in-order. You get a huge performance boost from going to an out-of-order architecture, but with it comes a pretty significant die area and power penalty. I argued that eventually Intel would have to consider taking Atom out of order, but the architects responded that Atom was married to its in-order design for 5 years.

Intel's Moorestown – same Atom core, just more integrated

Since 2008, Atom hasn't had any core architecture changes. Sure Intel integrated the GPU and memory controller, however the CPU still communicates with both of them over an aging FSB. The CPU itself remains mostly unchanged from what we first saw in 2008. Even Intel's 32nm Atom due out by the end of this year doesn't change its architecture, this is the same dual-issue in-order core that we've been covering since day 1. The 32nm version just runs a bit quicker and is paired with a beefier GPU.

Silvermont however changes everything. It is the first new redesign of the Atom architecture and it marks the beginning of Atom being on a tick-tock cadence. Say goodbye to 5 year updates, say hello to a new architecture every 2 years. Read on for more!

Although not quite the Intel SSD announcement we were expecting in Q4, today Intel unveiled its first mSATA SSD: the Intel SSD 310.

Based on the 34nm Intel X25-M G2 controller, the 310 will be available in both 40GB and 80GB capacities. The 80GB version should perform a bit slower than an 80GB X25-M G2 while the 40GB version will perform like a 40GB X25-V.  

Intel SSD 310 Comparison

 

Intel SSD 310

Intel X25-M G2 (34nm)

Intel X25-V (34nm)

Codename

Soda Creek

Postville

Postville

Capacities

40/80GB

80/160GB

40GB

NAND

IMFT 34nm MLC

IMFT 34nm MLC

IMFT 34nm MLC

Sequential Performance Read/Write

Up to 200/70MB/s (80GB)

Up to 170/35MB/s (40GB)

Up to 250/100 MB/s

Up to 170/35MB/s

Random 4KB Performance Read/Write

Up to 35K/6.6K IOPS (80GB)

Up to 25K/2.5K (40GB)

Up to 35K/8.6K IOPS

Up to 25K/2.5K (40GB)

Typical Power Consumption Active/Idle

150mW/75mW

150mW/75mW

150mW/75mW

Size

50.8mm x 29.85 mm x 4.85 mm

100.5mm x 69.9 mm x 7mm or 9.5mm

100.5mm x 69.9 mm x 7mm or 9.5mm

The 310 isn’t about performance, rather form factor. The SSD in Apple’s new MacBook Air is just the beginning – OEMs are beginning to shed the limits of traditional hard drive form factors as SSDs don’t need to house a circular platter.

The mSATA interface is physically a mini PCIe connector (similar to what you’d see with a WiFi card in a notebook) but electrically SATA. The result is something very compact.

The full sized mSATA 310 measures 50.8mm x 29.85mm and is less than 4.85mm thick. Total weight? Less than 10 grams. 

The Intel SSD 310 is OEM only at this point. Lenovo has already announced it will offer the 310 in ThinkPads in the future, while DRS Technologies will show off a tablet PC next month with the 310 inside. The 40GB drive is priced at while the 80GB version will run you 9 in 1000 unit quantities. 

As for the rest of Intel’s SSD updated lineup? While internal roadmaps showed a Q4 release for the 3rd generation X25-M based on 25nm NAND, that product is obviously delayed. We’re also hearing that new SandForce drives are still months away so those of you eagerly waiting for new drives at the high end will have to wait a bit longer.

What's this? The long awaited specs for Intel's third generation SSD? Indeed.

Internally it’s called the Postville Refresh (the X25-M G2 carried the Postville codename), but externally it carries the same X25-M brand we’ve seen since 2008. The new drive uses 25nm IMFT Flash, which means we should get roughly twice the capacity at the same price. While Intel is sampling 25nm MLC NAND today it's unclear whether or not we'll see drives available this year. I've heard that there's still a lot of tuning that needs to be done on the 25nm process before we get to production quality NAND. The third generation drives will be available somewhere in the Q4 2010 – Q1 2011 timeframe in capacities ranging from 40GB (X25-V) all the way up to 600GB.

Read on for the full specs!

Late last week we pulled back the covers on Intel's next-generation Core architecture update: Sandy Bridge. Due out in Q1 2011, we learned a lot about Sandy Bridge's performance in our preview. Sandy Bridge will be the first high performance monolithic CPU/GPU from Intel. Its performance was generally noticeably better than the present generation of processors, both on the CPU and GPU side. If you haven't read the preview by now, I'd encourage you to do so.

One of the questions we got in response to the article was: what about Sandy Bridge for notebooks? While Sandy Bridge is pretty significant for mainstream quad-core desktops, it's even more tailored to the notebook space. I've put together some spec and roadmap information for those of you who might be looking for a new notebook early next year.

 

NVIDIA’s ION brought a tremendous appeal to mini-ITX last year, but over the past six months Clarkdale has established itself as the natural and more capable choice for small form factor builds. ZOTAC are today attempting to reinvigorate appeal for ION by teaming up Intel’s CULV processors with NVIDIA’s aging GF9400 chipset.  We take a look at the IONITX-P-E, and aim to find out how it fits into the HTPC landscape.

Take the fastest desktop microprocessor in the world, lock its multiplier, shave off 133MHz and drop its price by 0. That's basically what you get with the Core i7 970. Despite the name this is the second 6-core 32nm Gulftown processor from Intel. 

If you're tempted by the 980X but wanted something slightly more affordable, read on to get introduced to the new Core i7 970.