How Does Intel’s Atom Z3580 Hold Up Against Qualcomm’s Snapdragon 805?

In the article Intel Corporation's Atom Z3580 Is Another Big Step Forward, we talked through a set of Intel (NASDAQ: INTC  ) marketing claims touting the performance of its Z3580. Intel's chip competes well on CPU and GPU performance against the Qualcomm (NASDAQ: QCOM  ) Snapdragon 801 found in flagship phones such as the Samsung (NASDAQOTH: SSNLF  ) Galaxy S5 and the LG G3, losing only in camera/imaging performance and integration. However, a reader noted in the comments section of that piece that a more appropriate comparison would be to Qualcomm's recently announced Snapdragon 805.

Putting the pieces together
We now have preliminary Atom Z3580 benchmark numbers -- run on Intel's 4.5-inch reference design -- and we now have performance numbers from Qualcomm's Snapdragon 805, courtesy of a benchmarking session done by AnandTech. Note that the numbers from Qualcomm are from the company's 10.6-inch Mobile Development Platform, so the Qualcomm platform probably can stretch its legs a bit with respect to power consumption.

That said, let's take the numbers that Intel gave for GFXBench 3.0 and 3D Mark Unlimited Ice Storm and see how these two parts actually stack up. First, GFXBench 3.0 T-Rex (these results are all at the same resolution):

Source: AnandTech, Intel.

We can see here that the Snapdragon 805 is quite a bit zippier than the Atom Z3580 on the order of 32%, but Z3580 slightly edges out the Snapdragon 801.

Source: AnandTech, Intel

In 3D Mark Icestorm 1.2 Unlimited's graphics test, the Atom Z3580 and the Snapdragon 805 are neck and neck, with the difference between them working out to less than 1%. It's hard to forget, however, that the Snapdragon 805 gave the Z3580 a beating in GFXBench 3.0, so at least for now, it would seem that the Snapdragon 805 in the reference tablet offers better graphics performance than the Z3580 found in Intel's reference smartphone. 

CPU performance -- Intel's the winner
When it comes to general purpose CPU performance, it's very hard to find a good benchmark that will give "accurate" results. Synthetic benchmarks can be useful, but are very easy to manipulate, and web-browser-based benchmarks are usually just as dependent on the optimization of the JavaScript (or similar) engine as they are on the actual CPU. Intel likes to promote Mobile XPRT as a good test of "real-world" performance, but those also seem to be very software/browser dependent, and could obscure the underlying CPU performance. 

Let's turn to the 3D Mark Unlimited Physics test. This is a test that performs plenty of complex calculations in what is probably the most demanding CPU task that a user is likely to perform on a mobile device -- physics simulation in a game. This is a test that actually exposed a pretty significant glass jaw in Apple's A7, and is, at the very least, better than most of the synthetic CPU benchmarks available on the Google Play store.

Source: AnandTech, Intel. 

Intel was kind enough to provide me with a breakdown of the sub-scores for the Atom Z3580 reference smartphone, and I have included the physics results here for the Z3580. It's no contest -- the Atom Z3580 in a 4.5-inch smartphone pulls nicely ahead (about 20% more performance) of the Qualcomm chips in this test. In terms of CPU performance, Intel's Silvermont does very well in real-world complex workloads.  

Foolish bottom line
The Snapdragon 805, as tested in Qualcomm's reference tablet, appears to offer lower CPU performance than the Z3580 in the reference smartphone than while offering higher graphics performance. However, what we don't know is how the Snapdragon 805 will perform in a 4.5-5.0-inch handset or how well Z3580 will perform in a real-world implementation. Further, even if Qualcomm has been able to pack this much more graphics performance into the Snapdragon 805 over the 801, it's unclear if the company dramatically increased graphics area (and therefore increased costs) to do so.

At any rate, Qualcomm appears to have the more competitive chip assuming that it can deliver this kind of performance in a handset and sell the chip at a reasonable cost structure. Further, Qualcomm's imaging subsystem is far ahead of what Intel currently has in its mobile platforms, which is very important as megapixel count is a selling-point for higher end devices. Intel has made a lot of progress (and is delivering leadership CPU performance and did a good implementation of the PowerVR G6430), but as noted in the previous article, it is still not quite in a position to capture a leadership position for hero devices as the rest of the system-on-chip needs to catch up with what the company has done on CPU and graphics.

Leaked: Apple's next smart device (warning, it may shock you)
Apple recently recruited a secret-development "dream team" to guarantee its newest smart device was kept hidden from the public for as long as possible. But the secret is out, and some early viewers are claiming its everyday impact could trump the iPod, iPhone, and the iPad. In fact, ABI Research predicts 485 million of this type of device will be sold per year. But one small company makes Apple's gadget possible. And its stock price has nearly unlimited room to run for early in-the-know investors. To be one of them, and see Apple's newest smart gizmo, just click here!


Read/Post Comments (7) | Recommend This Article (4)

Comments from our Foolish Readers

Help us keep this a respectfully Foolish area! This is a place for our readers to discuss, debate, and learn more about the Foolish investing topic you read about above. Help us keep it clean and safe. If you believe a comment is abusive or otherwise violates our Fool's Rules, please report it via the Report this Comment Report this Comment icon found on every comment.

  • Report this Comment On June 06, 2014, at 6:39 AM, maguro01 wrote:

    Power consumption is a very important parameter in such a comparison since we are talking portable devices. So would be a brief discussion of it .

    The Intel x86 cpu architecture is inherently more power consuming than the Qualcom ARM, but Intel has still competed through fab technology. Also with all the graphics and other functionality on the chip perhaps the cpu consumes a smaller portion of the power overall ameliorating the problem..

  • Report this Comment On June 06, 2014, at 8:14 AM, TMFAeassa wrote:

    "The Intel x86 cpu architecture is inherently more power consuming than the Qualcom ARM"

    This is not necessarily true. The X86 decode block that people attribute to making X86 inherently less efficient is a pretty negligible part of performance/power story at these levels.

    The micro-architecture, physical implementation, and of course the manufacturing technology all go into the power efficiency of a given CPU core, and Silvermont looks like it does a very good job here.

    That said, we will see Silvermont on TSMC 28nm soon enough, so we can strip out Intel's process lead and do apples-to-apples comparisons next year. Exciting, eh?

  • Report this Comment On June 06, 2014, at 9:13 AM, SemiWiki wrote:

    The Intel chip is 22nm FinFet and QCOM is 28nm planar?

    If so I would give the architecture design win to QCOM.

    Chip cost should also be a benchmark, for phones certainly. QCOM wins that one as well.

  • Report this Comment On June 06, 2014, at 9:57 AM, TMFAeassa wrote:

    Hi SemiWiki

    Thank you for taking the time to comment!

    Regards,

    AE

  • Report this Comment On June 06, 2014, at 1:41 PM, raghu78 wrote:

    Ashraf

    "This is not necessarily true. The X86 decode block that people attribute to making X86 inherently less efficient is a pretty negligible part of performance/power story at these levels. "

    wrong. Jim Keller categorically says ARM has inherent efficiency advantage over x86 by spending more transistors driving performance rather than spend more transistors dealing with the decoding complexity of x86. In fact he calls it a straightforward proposition.

    http://www.youtube.com/watch?v=eTVnxaXCLg0

    video at 54:50

    Sorry but your word means nothing in this context.

    "The micro-architecture, physical implementation, and of course the manufacturing technology all go into the power efficiency of a given CPU core, and Silvermont looks like it does a very good job here."

    While talking about x86 vs ARM its clearly advantage ARM. Now when you bring Intel's manufacturing process lead then only do things get close in terms of efficiency and perf/watt. Silvermont on TSMC 28HPM vs Krait or A15 on the same process would be a no contest and ARM will win easily. You could wait for the confirmation till next year. :-)

  • Report this Comment On June 06, 2014, at 2:27 PM, TMFAeassa wrote:

    raghu78

    "wrong. Jim Keller categorically says ARM has inherent efficiency advantage over x86 by spending more transistors driving performance rather than spend more transistors dealing with the decoding complexity of x86. In fact he calls it a straightforward proposition."

    That's funny...especially since AMD's Fred Weber indicated the following:

    "Fred said that the overhead of maintaining x86 compatibility was negligible, at the time around 10% of the die was the x86 decoder and that percentage would only shrink over time. We're now at around 8x the transistor count of the K8 processor that Fred was talking about back then and the cost of maintaining x86 backwards compatibility has shrunk to a very small number. But the benefit of backwards compatibility is huge."

    http://www.anandtech.com/show/2493/3

    AMD even published a paper proving that the overhead was negligible. I'll try to dig it up :-)

    AMD is simply talking its current book, and right now it needs to convince the world that it has something great and magical with a custom ARM Core. ARM is the hot buzzword, and AMD is trying to capitalize.

    p.s. Silvermont on 28nm HPM is coming with SoFIA. We'll be able to test your claim them :-)

  • Report this Comment On August 19, 2014, at 11:22 PM, HibikiTaisuna wrote:

    Wow, didn't expect that my Tegra K1 is so much better than these SoCs. 50% more graphics performance and even 10% to 20% more CPU Performance (the week part of the Tegra K1).

Add your comment.

Sponsored Links

Leaked: Apple's Next Smart Device
(Warning, it may shock you)
The secret is out... experts are predicting 458 million of these types of devices will be sold per year. 1 hyper-growth company stands to rake in maximum profit - and it's NOT Apple. Show me Apple's new smart gizmo!

DocumentId: 2984393, ~/Articles/ArticleHandler.aspx, 9/1/2014 8:19:37 AM

Report This Comment

Use this area to report a comment that you believe is in violation of the community guidelines. Our team will review the entry and take any appropriate action.

Sending report...


Advertisement