Hasta La Gadget!

It’s always a good idea to take benchmarks with a grain of salt — their synthetic tests don’t always match up with real-world performance. But, we wouldn’t blame you if you wrote them off completely after spying these results from Android Community forum member Simms22. Simms took his Nexus S, blessed it with a little Cyanogen Mod 7 pixie dust, overclocked it to 1.544GHz, and made a few other tweaks for good measure. The results? An absolutely unbelievable score of 10,082 in Smartbench2011. To put that in perspective, the 1.2GHz dual-core Exynos powering the Galaxy S II hit only 3,053 — and remember, the Nexus S is working with a one core handicap. The creator of Smartbench has acknowledged there are bugs to be worked out (did besting the Xoom by 400 percent give it away?) but we’re not quite ready to dismiss the numbers game completely — then what excuse would we have for publishing copious amounts of bar charts?

Update: The creator of Smartbench2011 confirms he’s working on a new version, 1.2.1, which should fix the bugs.

Nexus S hacked and tweaked to slaughter benchmarks, reality be damned originally appeared on Engadget on Tue, 21 Jun 2011 16:16:00 EDT. Please see our terms for use of feeds.

Permalink   |  Android Community, Smartbench  | Email this | Comments

We know that Intel is gunning for ARM with its Oak Trail platform, and indeed there were a few early tablets at Computex following that very path to Honeycomb. The question on everyone’s minds, of course, is how well this dual-core 1.5GHz platform can compare to the Tegra 2 competition. If you believe the results from a suite of tests that tweakers.net ran on a prototype Compal unit, then the answer is “not very well.” On benchmarks like CaffeineMark, Linpak, and Quadrant the platform was largely left in the dust by ARM competition with bigger biceps, but the Oak Trail machine did clean the floor with everyone else on the SunSpider browser benchmark. What does it all mean? Not a heck of a lot at this point, we’re afraid. It’s far too early to be drawing performance conclusions about a platform based on a prototype fresh out of the fabricator, and we have our doubts that these benchmark apps are optimized for the new platform — so don’t give up on ‘ol x86 just yet.

Honeycomb on Oak Trail gets benchmarked on prototype Compal tablet, numbers ensue originally appeared on Engadget on Fri, 03 Jun 2011 09:26:00 EDT. Please see our terms for use of feeds.

Permalink   |  tweakers.net  | Email this | Comments

A very strange thing popped up on mobile graphic benchmarking site NenaMark the other day — an entry for the Droid Bionic. Now, it would be very easy to fake this test, and you’d be right to be skeptical given the incomplete score and the fact that it’s reporting PowerVR’s SGX 540 GPU, instead of the Tegra 2 we saw at CES. But, let’s not be too hasty — we heard back in April that NVIDIA’s mobile chip wasn’t playing nice with Verizon’s LTE. Perhaps when Motorola said it was delaying the Bionic to incorporate “several enhancements” it really meant “rebuilding the phone with a more LTE friendly CPU.” Both Samsung and Texas Instruments use the SGX 540, and Moto has previously turned to TI’s OMAP for the Droid, Droid 2, and Droid X. Then again, a single, suspiciously low benchmark score isn’t the most convincing basis for a rumor.

Droid Bionic benchmark reports PowerVR GPU, new SOC inside? originally appeared on Engadget on Sun, 29 May 2011 05:05:00 EDT. Please see our terms for use of feeds.

Permalink Droid Life  |  NenaMark  | Email this | Comments

A lot of portable PC power cells last for only four or five hours, after which you’ll find yourself chained to a wall socket. Good thing there are external batteries to keep us in current when a plug’s nowhere to be found, and Tom’s Hardware has done some benchmarking on a slew of such devices so you’ll know which one’s suited for you. A Dell Vostro 3300 and an Inspiron Mini 10 running Windows 7 were used to put packs from Amstron, Brunton, Digipower, Electrovaya, Energizer, Lenmar, PowerTraveller and Tekkeon through some real-world paces — we’re talking word processing and web surfing, not fragging and film editing. So if you’re in the market for a mobile power unit, hit the source link and get the down and dirty on which external battery’s best.

External batteries benchmarked, the portable juice is loose originally appeared on Engadget on Wed, 13 Apr 2011 09:59:00 EDT. Please see our terms for use of feeds.

Permalink   |  Tom’s Hardware  | Email this | Comments

iFixit may have physically uncovered Apple’s latest silicon, but it’s the processor gurus that have discovered what’s truly inside — using software benchmarks, they’ve unearthed the speeds and feeds of the Apple A5. As you’ll no doubt be aware having read our headline above, there actually isn’t a 1GHz CPU at the helm, as AnandTech and IOSnoops report the dual-core ARM chip is dynamically clocked around 900MHz, likely in search of reduced power consumption. Perhaps more interestingly for all you gamers in the audience, the iPad 2 reports that it has a dual-core PowerVR SGX543MP2 GPU on the die as originally foretold — and, spoiler alert — it mops the floor with both the original iPad and the Motorola Xoom. Though the new chip didn’t quite demonstrate 9X the graphical prowess of its predecessor, it rendered 57.6 frames per second in a GLBenchmark test where the (admittedly higher-res) Tegra 2 tablet managed only 26.7fps, and last year’s iPad pulled only 17.6fps. That’s some serious Tai Chi. Hit up our source links to see the difference it can make in games like Infinity Blade.

Update: Though it sure sounds like there’s a dual-core ARM Cortex A9 in there, that’s not yet a proven fact — we only know that it’s a dual-core ARM v7 chip which performs relatively similarly in non-graphical tests. [Thanks, Jim]

[Thanks to everyone who sent this in]

iPad 2 specs discerned, 900MHz dual-core ARM CPU and PowerVR SGX543MP2 GPU blow away graphical benchmarks originally appeared on Engadget on Sat, 12 Mar 2011 20:01:00 EDT. Please see our terms for use of feeds.

Permalink   |  AnandTech (1), (2)  | Email this | Comments