The SoC - OMAP3 3430

Before we dive into the phones, let's take a look at the SoC at the core of both. We talked about the OMAP 3430 back in 2009 (well maybe not way back, but it sure feels like it) in the context of the Palm Pre, which uses the OMAP 3430. It turns out that Texas Instruments' OMAP 3430 SoC is quite popular if not a microcosm of the generation of smartphones we're talking about; it's common to the N900, Motorola Droid, and Palm Pre.

 
OMAP 3430 Block Diagram
 

The Texas Instruments OMAP 3430 is designed for a 65nm manufacturing process, in fact, all of the OMAP34x series are designed for 65-nm process, whereas OMAP36 is intended for 45-nm processes. OMAP 3430 supports up to 12 megapixel cameras on its onboard image-signal processor, and packs an onboard IVA 2+ image, video and audio DSP accelerator clocked at 430 MHz, and a PowerVR SGX 530 GPU. Those clocks are recommended by TI, but in practice handset OEMs set them through a dialog with carriers to meet appropriate performance and battery life targets. OMAP3430's IVA 2+ supports MPEG-4 and H.264 encode/decode of 30 fps video at 720x480, and WMV9 decode at the same resolution and framerate. OMAP3430 can drive displays up to WXGA (1280x800) resolution at 24-bits. Note that the OMAP3 series SoCs don't pack internal cellular modems, meaning OEMs shop around for their own and connect over internal USB or multi channel buffered serial (McBSP).

At the heart of the OMAP 3430 SoC is the 600 MHz ARM Cortex A8 CPU. Remember that Cortex-A8 is an ARMv7 design with a 13 stage pipeline, compared to the 8 stage ARM11 (ARMv6) design common to OMAP2 and the CPU in the iPhone 3G. In practice, the performance delta between ARMv6 (ARM11) and ARMv7 (Cortex-A8) is between 2x and 3x.

 

Birds of a feather: N900 and Motorola Droid Physical Comparison and OMAP 3430 Continued
POST A COMMENT

68 Comments

View All Comments

  • Wadzii22 - Friday, June 11, 2010 - link

    Out of curiosity I ran linpack and Benchmark pi on my droid that's oc'd to 1ghz

    my benchmark pi score was 1280 and linpack gives me 17.24 mflops
    Reply
  • strikeback03 - Friday, June 11, 2010 - link

    Did you run them stock? As those numbers seem to be a ~4x improvement over what is shown here, which seems odd given the ~2x increase in clockspeed. Reply
  • Wadzii22 - Wednesday, June 16, 2010 - link

    with the phone completely stock my scores were basically the same as whats in the original article. Reply
  • jamyryals - Friday, June 11, 2010 - link

    Please continue this type of in depth comparison with current and future hardware. PC hardware is all well and good, but it's all so fast now the mobile space is a much more interesting battle. Not to mention with how fast things are evolving there is the opportunity for a lot of content. Reply
  • Ratman6161 - Friday, June 11, 2010 - link

    For eample, my droid purchased in early April came out of the box running at 600 MHz (though now it actually runs at up to 900 Mhz). My wife got hers in early June and hers is 600 MHz too and also came out of the box with Android 2.1 already on it. Reply
  • Wadzii22 - Friday, June 11, 2010 - link

    For whatever reason, setcpu always sees a stock droid's max at 600, but they do run at 550. I just got a new one yesterday after bricking my old droid, it showed the same thing. Reply
  • CharonPDX - Friday, June 11, 2010 - link

    Nokia was the originator of the "sell unsubsidized smartphones direct" model, years before Apple or Google. You could get a Nokia N80 at CompUSA completely unlocked for $800 in 2006, a year before the unsubsidized iPhone. Reply
  • Stas - Friday, June 11, 2010 - link

    0.1 build with fixed WiFi and maps.

    LinPack - 12.2 (twelve point two)MFLOPS
    Engadget.com loads in 20 sec (default browser)

    'nuff said.
    Reply
  • Stas - Friday, June 11, 2010 - link

    forgot to mention. the CPU is at 800Mhz. I've had it up at 900Mhz with bare Android build (leaked 2.2) and the performance seems no different, but no numbers, sorry. Reply
  • milli - Saturday, June 12, 2010 - link

    That's pretty wrong what you're saying there.
    Qualcomm didn't even license the A8 (nor will they ever).
    What they did license is the ARMv7 instruction set (and that's a huge difference). With that they made a custom implementation of the ARMv7 architecture. (BTW Qualcomm already stated in 2005 that they're an architectural licensee for ARM’s ARMv7 instruction set)

    There are many differences between Scorpion and A8.
    I'll quote from a certain article since i can't say it better:
    'Although Scorpion and Cortex-A8 have many similarities, based on the information released by Qualcomm, the two cores differ in a number of interesting ways. For example, while the Scorpion and Cortex-A8 NEON implementations execute the same SIMD-style instructions, Scorpion’s implementation can process128 bits of data in parallel, compared to 64 bits on Cortex-A8. Half of Scorpion’s SIMD data path can be shut down to conserve power. Scorpion’s pipeline is deeper: It has a 13-stage load/store pipeline and two integer pipelines—one of which is 10 stages and can perform simple arithmetic operations (such as adds and subtracts) while the other is 12 stages and can perform both simple and more complex arithmetic, like MACs. Scorpion also has a 23-stage floating-point/SIMD pipeline, and unlike on Cortex-A8, VFPv3 operations are pipelined. Scorpion uses a number of other microarchitectural tweaks that are intended to either boost speed or reduce power consumption. (Scorpion’s architects previously designed low-power, high-performance processors for IBM.) The core supports multiple clock and voltage domains to enable additional power savings."

    "Qualcomm claims that Scorpion will have power consumption of roughly 200 mW at 600 MHz (this figure includes leakage current, though its contribution is typically minimal in low-power processes). In comparison, ARM reports on its website that a Cortex-A8 in a 65 nm LP process consumes .59 mW/MHz (excluding leakage), which translates into about 350 mW at 600 MHz."

    With that said, i don't understand where the misconception about the Scorpion being an A8 started. Even Qualcomm states clearly on their own website that Scorpion is not licensed from ARM. They also state that they invested hundred of millions in creating their own core based on the ARMv7 instruction set.
    I hope now all the staff from Anand will stop saying that there's an A8 inside of Snapdragon. Or maybe you should even clarify that with a small article.
    Reply

Log in

Don't have an account? Sign up now