A quick look at Arrandale performance against desktop CPUs
Before diving into the first Arrandale-based notebook to arrive in the Benchmarking Sweatshop, it's probably a good idea to see how the performance of Intel's latest mobile CPU compares with what's available on the desktop. The Asus K42F notebook we used for testing is equipped with a Core i5-540M clocked at 2.53GHz, whose performance we've presented in the context of the results from our Clarkdale desktop coverage. The i5-540M is most comparable to the desktop Core i3-540, which lacks the mobile chip's Turbo Boost capability, but whose 3.06GHz base clock speed exactly matches the peak Turbo Boost speed of the i5-540M. Of course, the i5-540M is stuck with slower 1066MHz memory and only 3MB of cache.
I should also note that our notebook is using the Intel IGP, which carves out a slice of system memory for graphics. Having slightly less than 4GB of usable system memory shouldn't impact the workloads we're testing, but the Graphics Media Accelerator will consume some of the memory bandwidth otherwise available to the CPU. The desktop CPUs are all paired with a discrete graphics processor that just wouldn't fit into our notebook's 14" chassis. Still, I'm curious to see how the i5-540M fares against some admittedly stacked desktop competition.
The i5-540M shadows the i3-540 in Sandra's cache and memory bandwidth benchmark, but remains a step behind, likely due to its lower base clock speed. I didn't see the i5-540M hitting its Turbo Boost peak all that often with single-threaded workloads.
With its SO-DIMMs running at only 1066MHz, the i5-540M has about 30% less memory bandwidth than its closest desktop equivalent. Still, you get loads more bandwidth than with old-school Core 2 designs that don't feature integrated memory controllers.
The i5-540M lags behind the i3-540 in 7-Zip. This Arrandale implementation is still faster than a Core 2 Duo E860, though, and by quite a margin in the decompression test.
TrueCrypt doesn't appear to take advantage of the new AES-NI instructions built into Westmere CPU silicon. However, the i5-540M still fares reasonably well, just edging out the E8600.
The i5-540M's relative position in the field doesn't change when we switch to Cinebench. Interestingly, the 540M is slower than the E8600 in the single-threaded rendering test. Hyper-Threading gives Arrandale the advantage when we move to the multi-threaded rendering test, though.
Although it completes The Panorama Factory's stitch operation a couple of seconds faster than a Phenom II X2 550, the i5-540M is a ways behind the rest of the competition here.
The i5-540M and E8600 trade blows in the x264 HD video encoding test. With both passes, the mobile chip trails the i3-540 by a decent margin. These results were obtained using the latest 3.03 revision of the x264 benchmark. We have some additional x264 results a little later in the review that were obtained using an older 2.0 revision, so the scores aren't directly comparable.
|Early Unreal Tournament concept art reminds us how far we've come||6|
|Report: Intel targeting larger, pricier Android tablets||17|
|AMD's Mullins APU appears in $250 HP netbook||42|
|Core i7-4790K 'Devil's Canyon' overclocking revisited||34|
|Steam controller gets an analog stick||52|
|Delays strike Battlefield: Hardline, Dragon Age: Inquisition||18|
|It's official: Microsoft will consolidate Windows development||68|
|Microsoft's 2014 revenue up 11.5%, but income stagnates||28|
|Tuesday Night Shortbread||49|
|The new new name for the UI is called Retro.||+37|