The benchmarking was done by Netbook News while at MWC, who managed to run a 3DMark03 graphics benchmark test during their hands-on time. The resulting score of 3,049 is lower than some Ion LE netbooks achieve; the Samsung N510, for example, scores a 3593.

There are possible explanations, of course, and we won’t know for sure how the Ion 2 stacks up until we’re able to test a production-ready unit. As Netbook Choice points out, it’s possible that Acer and Nvidia purposefully hamstrung the machine to keep it stable during demonstrations. Then again, it’s also possible that Nvidia has had a difficult time working with the Aspire One 532G’s Pine Trail processor. Let’s hope it’s the former: the last thing netbooks need is even more limited graphics capabilities.

[Netbook News via Netbook Choice]

Clarkdale is faster that anything dual core out there, while both mobile chips with integrated graphics are faster than their Core 2 Duo counterparts, with the extra performance having no impact on the battery life. Obviously, don’t expect to play Call of Duty 4 at full with the built-in GPU in any of these. [Anandtech on Arrandale and Anandtech on Clarkdale, Hot hardware on ArrandaleHot Hardware on Clarkdale]

Let’s get this straight right away: We didn’t test dropped voice calls, we didn’t test customer service, and we didn’t test map coverage by wandering around in the boonies. We tested the ability of the networks to deliver 3G data in and around US cities, including both concrete canyons and picket-fenced ‘burbs. And while every 3G network gave us troubles on occasion, AT&T’s wasn’t measurably more or less reliable than Verizon’s.

It was measurably faster, however, download-wise, in six of the 12 markets where we tested, and held a significantly higher national average than the other carriers. Only Verizon came close, winning four of the 12 markets. For downloads, AT&T and Verizon came in first or second in nine markets, and in whatever location we tested, both AT&T and Verizon 3G were consistently present. If you’re wondering about upload speeds, AT&T swept the contest, winning 12 for 12.

The Cities

Last year, Sprint was the big winner in an eight-city, coast-to-coast test. This year, we have results from 11 cities coast-to-coast, and even got to test (during what was otherwise a holiday) on the Hawaiian island of Maui. Also, unlike last year, we were able to test T-Mobile’s new 3G network, active in all the markets we visited (except, at the time, Maui). For being such a latecomer, T-Mo did well, and the numbers show even more promise from them.

We tried to spread the love around this year, geographically, hitting cities we didn’t get to last year (at the cost of losing a few from ‘08). Besides Maui, we hit Atlanta, Chicago, Denver, Las Vegas, Los Angeles, New York, Phoenix, Portland, Seattle, San Francisco/Bay Area and Tampa.

The Methodology

Our testing regimen was based on the same scheme as last year: We picked five locations in each city, including at least one “downtown” location that was considered a suburb. The selections were arbitrary, or fixed but logical – landmarks, residences, etc. (Note: Due to timing constraints, Chicago and Maui only had three test locations.)

Our hardware consisted of two identical stripped-down Acer Timeline laptops running Windows Vista, and four 3G wireless modems requested from the carriers. We allowed them to make the choice of hardware, simply asking for their “best performing” model. Once up and running, here are the tests we ran:

• Bandwidth & Latency: Speedtest.net – Reports upload and download bandwidth in megabits per second, as well as ping latency in milliseconds. We performed this test five times at each location on each modem.

• Pageload: Hubble images at Wikimedia – A 4.42MB web page with 200 4KB thumbnails, it was fully reloaded three times, and timed using the Firefox plug-in YSlow. The three time readings were averaged.

• Download: Wikimedia’s Abell 2667 galaxy cluster photo – This single 7.48MB JPEG is a clear test of how fast you can download stuff from the cloud, and again, we hard refreshed this file three times, and measured time using YSlow for an accurate human-error-free reading.

This was a test of 3G performance. Even though Sprint and its tech partner Clearwire have intrepidly released 4G networks in half of the tested markets – Atlanta, Chicago, Las Vegas, Maui, Portland and Seattle – we only tested Sprint’s 3G network. The reason should be obvious: While we performed the test with laptop cards on PCs, it’s supposed to serve as a test of the network’s ability to deliver service to all devices, including smartphones, dumbphones and laptops. Show us a Palm Pre WiMax edition – better yet, sell 100,000 of them – and then we’ll switch it up. And while you may argue that this 3G test still doesn’t adequately reflect your experience with your iPhone, at least it’s the same network, and may serve to rule out AT&T’s data pipe as the independent cause for all those infamous dropped calls.

(On a side note, when multiple carriers release 4G networks, we’ll definitely conduct a comparative test of them all, using new parameters, and focused around laptop use.)

The Results

Now that you know how we ran the test, here are the top finishers in each market, plus some pretty bar graphs showing you how bandwidth compares.

Though we tested for uploads and downloads, we focused our additional tests on the downstream, as it’s the more important direction, in the minds of most consumers and most carriers. The anomaly there is AT&T, which has dramatically good upload bandwidth, even when its download bandwidth doesn’t keep up. Fast uploads are a priority for AT&T, and will soon be for T-Mobile, which recently turned on faster uploading in NYC, as you can see in our test results. Meanwhile, although Verizon technically came in second in uploads as well as downloads, it doesn’t seem to treat this as a major priority.

When it came to downloads, though, the competition was markedly stiffer:

Atlanta – AT&T, followed by Verizon
Bay Area/San Francisco – AT&T, followed by Verizon
Chicago – AT&T, followed by Verizon then Sprint
Denver – AT&T, followed by Verizon
Las Vegas – Verizon, followed by AT&T
Los Angeles – AT&T, followed by Sprint
Maui – Verizon, followed by AT&T
New York – AT&T, followed by T-Mobile
Phoenix – Verizon, followed by T-Mobile
Portland – T-Mobile, followed by Verizon
Seattle – Verizon, followed by T-Mobile
Tampa – Sprint, followed by AT&T

Is That The End?

No. We’ve compiled the following gallery with all the data from each test location in the 12 markets, so you can see on a neighbourhood-by-neighbourhood level who won what. This also includes latency, pageload and download numbers, so you can track the performance in several ways. (The data above is bandwidth, though as you’ll see, that was generally representative of the overall performance. If a carrier was tops in bandwidth, it was usually tops in download time.) These tests are all just “snapshots in time”, as the carriers like to say, so feel free to bitch about where your experience doesn’t reflect our results. We stand by them, but acknowledge that network performance is changing all the time and experiences very regular hiccups.

Regarding latency, you’ll notice it didn’t appear to affect actual user experience – 3G isn’t really up for Modern Warfare 2, if that’s what you’re thinking – we will gladly show you latency averages, as well as pageload and file download averages, broken out for every market on the test.

Special thanks to all of the excellent testers we enlisted, Mark Wilson, Chris Mascari, John Herrman, Kyle VanHemert, Dan Nosowitz, Matt Buchanan and Rosa Golijan from our own team, along with Tamara Chadima and the indefatigable Dennis Tarwood. You guys were troopers, and I’m pretty sure FedEx either loves you or hates you. Thanks to John Mahoney for helping develop the initial tests that we’ve continually refined, to Chris Jacob for mapping all the locations, and to Don Nguyen for the mad number crunching – you truly are a spreadsheet pimp.

Note: Some of you may have noticed that San Diego is among the cities highlighted on the top illustration – and that Maui is not. The reason is that while we did testing in three great San Diego locations, one of the locations didn’t get any Sprint or T-Mobile service, and the already fairly thin dataset was rendered too compromised for any kind of usable report. As for Maui’s absence, Maui’s just too far out in the Pacific to make for a pretty map shot.

Of course you don’t go for an Atom for muscle, and the Pineview processors do beat the Ion in power consumption and cost (the new guys are both under $US100). But we expect performance to go up as well, and the Ion mopped the floor with the D410 and D510 in most real-world applications. The Pineview procs did win in raw processing power, however. These are early tests, so take them with a grain of salt, but we’d really have preferred to see some numbers that at least beat existing configurations like Ion. [CarTFT (German PDF) and SemiAccurate via Engadget]

Ultimately, Ubuntu Netbook Remix 9.10 did the best, and openSUSE 11.2 also did well. But Chrome OS performance was far from spectacular. That shouldn’t be a huge surprise, though. It’s not slated to ship for like a year, and its performance should pick up as builds continue. The main surprise looks to be Moblin 2.1’s comparatively slower speed, despite Intel’s efforts to optimise it for netbooks. It looks pretty, though.

All distros were tested with default configurations/packages, except for Chrome OS. They “needed to remount the root file-system in a read-write mode and add in the standard Ubuntu Karmic package repositories for which Google’s operating system is based.”

The Phoronix test suite included H.264 video playback, OpenArena, LZMA and 7-Zip file compression, IOzone, PostMark, WAV to OGG audio and H.264 video encoding. Full test results at: [Phoronix via Slashdot]

And I’m not sure that the solution is just a refresh away.

The Mac Pro was once the only viable option for an OS X lover in need of serious horsepower for tasks like editing media. Now, with the new iMac? I think it’s straight up stupid to buy a Mac Pro.

The $4499 Mac Pro, desperately in need of a refresh, gives you a 2.66GHz Quad-Core (i7), 3GB of RAM (triple channel, but seriously?), 640GB hard drive (again, seriously?) and a nominal graphics card. Spend $1400 more and you’ll get a another processor and 3GB more RAM.

The $2599, 27-inch iMac obviously includes a screen, plus you get a 2.66GHz Quad-Core (i7), 1TB drive, 4GB of RAM and a nominal graphics card.

But beyond those clock speeds, the Mac Pro’s i7 processor is the more premium Bloomfield edition, while the iMac uses the Lynnfield. (More on those differences here.)

Still, the bottom line is that the iMac’s Lynnfield processor is newer, and it shows in performance.

Macworld benched the new iMacs against the latest Mac Pros. And, you know what? The i7 iMac more than held its own. It basically defeated the quad-core Mac Pro across the board.

And other than a few specific tasks in which the most expensive Mac Pro’s eight cores proved beneficial (Handbrake, Cinebench, etc), the iMac outperformed the competition or kept things close enough not to be relevant, plus it straight-up won in the eyes of Speedmark 6.

Performance-wise, the base Mac Pro makes no sense at all. The eight-core Mac Pro offers a touch more power, sometimes, and other times (in many day to day tasks) even it is outgunned.

Of course, any Mac Pro still allows multiple internal hard drives, three PCI slots, more FireWire ports (four vs one) and more room for RAM expansion (32GB vs 16GB). But once again, even in the worlds of professional media creation, that’s a pretty questionable upsell, especially with external storage solutions and the fact that most high, high end media pros (like special effects artists) turn to dedicated render farms to do their heavy number crunching anyway.

With the new iMac, Apple has shrunk the Mac-Pro-needing niche even smaller. And I can’t tell anyone with a straight face that a handful of expandability is worth $US300-$1100 with no monitor, no matter how deep their pockets are.

Apple needs to re-examine their pricing model. Even with an inevitable processor refresh (i9, anyone?), it’s time for a price drop and/or some free with purchase displays. Just because you’re a pro doesn’t mean you’re a sucker.

The i9 doesn’t extract significant advantages from its pumped core count (which brings processing thread count up to 12) in a lot of day-to-day tasks, so don’t expect to see an increase in game performance, Windows startup speed or other single-core optimised tasks. It’s when you start rendering video or doing 3D modelling – tasks that are suited to parallelisation – that the i9 flexes its muscles.

That’s roughly a 50 per cent increase in video-encoding performance over a similarly clocked i7 – already no slouch by any existing standards.

The i9 processors won’t ship until sometime in early to mid 2010, and when they do, expect them to be a bit on the expensive side. But man, 50 per cent. I think I can stand to save up a few more bucks, honestly. [PCLab via Electronista]

The Basic Differences in Chips

First off, I should note that the Core i7 chip has what Intel calls a “turbo mode”. That is, when it’s not utilising all of its cores, it can dynamically overclock itself up to 3.4GHz on whatever single cores are in use. It can, as shown in this video, work in steps. So you get turbo benefits from partial use of the four cores in this iMac’s chip, but also, you get turbo benefits when each core is only being partially used. For example, if four cores are running but only at a fraction of their total capacity (less then 100 per cent), the cores can still run a little faster. This should theoretically make up for the difference between the two-core 3.06GHz chip and the hyper-threaded quad core chip at a base of 2.8GHz.

The other thing to realise about these newer Core i7 chips are that they have no northbridge — or bus — between the memory and CPU. The memory controller is built right into the processor, and there’s a new tech called Quickpoint interconnect which connects the cores in a point-to-point architecture. Core i7 supports triple channel memory (which would use three banks at once), but this iMac only came loaded with two banks of RAM filled. Like our other iMac, that’s a 2GB + 2GB arrangement.

Matt explains more about i7 here. (And yes, there are differences between i7 and i5, besides clock speed.)

*Note that this machine also had a faster ATI Radeon 4850 video card with 512MB of RAM (versus the 4670 card in the other iMac) which may have impacted performance in several apps. I have no idea which of these apps uses the GPU to accelerate its tasks under Snow Leopard. (For example, Preview may use it to help render JPGs faster, or it may not. Apple could not tell me. In Adobe After Effects, the Radeon series of cards apparently is not supported for OpenCL acceleration. )

Performance with Multithreaded Apps

In short, any task we tried that expressly was written to either a) take advantage of multiple cores, or b) take advantage of multiple cores through Snow Leopard’s multicore middleware, Grand Central Dispatch, were 2-3 times faster. (More on that here.) These results include:

• 64-bit versions of Geekbench, which focus on CPU and memory tests.
• Adobe After Effects benchmarks
• Opening 20 images of Tokyo Tower that are 2000×2000 pixels and 35MB each.

Impressive stuff, but honestly, those tests were kind of uninteresting to me. I mean, those tests don’t really have any correlation to my daily computing use. So on a whim, after benchmarking, I tested handbrake, the DVD ripping software I love. It, too, was freaking fast.

I know the app is multi-threaded, but I did not know what level of optimisation it was written for. I was blown away by a 3x speed multiplier with the i7. On the Core i7 iMac, it took 43 minutes to rip a DVD, Storm Riders, a surfing film from the ’70s featuring Gerry Lopez (my favourite) and others. On the Core2Duo machine, it took 147 minutes! I know this is basically a DVD read test coupled with decoding and video conversion, but the results have me excited because this is a real task that takes my computer a long time to do, performed by a program that hasn’t been revised in a year.

Performance With Single-Core Optimised Apps (Otherwise Known as Reality)

Unfortunately, there are still very few applications that take advantage of multiple cores directly or via Snow Leopard’s GCD, not even video-based, let alone general-purpose computing.

Photoshop CS4 on the Mac, which is not set up to handle multicore processors, showed almost less than a 3 per cent improvement using the Driver Heaven benchmark. Basic tasks, like booting and shutdown, saw virtually none. Playing the 1080p Quicktime trailer of Avatar consistently showed that the i7 was using 3 per cent less of its total CPU than the Core2Duo, but I wonder if that’s a result of the faster graphics card kicking in using CoreCL. Xbench, the old program that does a more comprehensive job of benchmarking a system from disks to processors, showed almost no difference.

I think Xbench, which hasn’t been updated in years, is a solid benchmark for that old program that you depend on but has been long abandoned or at least ignored by its developer.

These scores, again, are in relation to the top line 3.06GHz Core2Duo iMac we tested. Some benchmarks have come in from the web comparing the i7 to the i5. Here’s one that claims a 30 per cent jump using Geekbench. Now we know Geekbench likes and does well with more cores and is a synthetic CPU test. But if the i5 is 30 per cent slower, and the i7 pulls even with the 3.06 GHz Core2Duo chip in single threaded activity — most day-to-day activity — does that mean the i5 is slower than the cheaper Core2Duo? Maybe. Probably not 30 per cent, since Geekbench is strictly CPU/memory and likes more cores, and this stuff does not translate so literally in the real world. But we can assume the i5 will have 30 per cent less jump from the top tier Core2Duos, translating into a mere 1.3-2 times speed increase from last gen chips on programs that like cores.

Value

For the most part, in our review, I said that you should stick to the preconfigured options, upgrading to Apple’s next recommended config before considering upgrades to the lower tier models. How does that advice change now that we’ve seen the i7? I don’t know! I guess it depends if you’re a betting man. If you think programs for Snow Leopard using GCD are coming, paying $US200-$US500 bucks more from the top line Core2Duo chip for an i5 or i7 might make sense. The probability of you getting programs that can use those extra cores goes up if you are a graphics or video professional who expects to see support from Adobe, Apple, etc. (Apple already claims big jumps in Aperture that we weren’t able to test.) Or if you rip a lot of DVDs! The rest of you? The Core2Duo stuff could be fine for today and fine for tomorrow. But the Core i7 is not worse for today and will definitely be faster tomorrow. It just costs more.

Me personally? I’d opt for the Core i7. I just might wait ’til the new iMacs refresh a bump and the i7 is cheaper and part of a standard build. But I’m patient like that.

[iMac Review]

Specifically, using his charts and mine, it wasn’t hard to recognise the jump in the multi-threaded, 64-bit results from geek bench in the categories of integer, floating point and memory-streaming tests, as well as the threaded tests. (Memory tests were slightly faster; the others were drastically so.) Interesting, as the Core i5 chip is clocked at 2.66GHz and the Core2Duo iMac I tested runs at 3.06GHz.

(The turbo boost function, which overclocks the Core i5 chip to up to 3.2GHz when running non-multi-threaded apps, should be kicking in performance here, too.)

Interesting, but two things to remember: Core i7 chips are coming out for the iMac shortly and will run at 2.8GHz and have hyperthreading so the four cores emulate eight. And there are still not many (if any at all) major OS X apps that can take advantage of Snow Leopard’s multicore support. [Electronista's tests, Gizmodo's iMac Review]