Print 49 comment(s) - last by BillyBatson.. on Mar 13 at 7:10 PM

New 28 nm chip is more than twice as powerful as anything on the market

Qualcomm, Inc. (QCOM) impressed at the 2012 Consumer Electronic Show in January.  At the show it talked about its transition down to the 28 nm node (via Taiwan Semiconductor Manufacturing Comp., Ltd.'s (TPE:2330new processes) and pushing Snapdragon 4 (S4) system-on-a-chip designs onto the market.

Developers' first taste of S4 will be delivered via the MSM8960 Mobile Development Platform (MDP), which will be distributed at the 2012 Mobile World Congress, which starts Feb. 27 in Barcelona, Spain.  But AnandTech scored an early unit and has been busy benchmarking it.

How does the fourth generation Snapdragon fare? To quote AnandTech:

The [CPU] performance advantage... is insane.

The super-chip's performance comes, in part, thanks to the new Krait core, which is built on a licensed ARMv7 instruction set from ARM Holdings plc. (LON:ARM).  The new core is similar, in some ways, to the ARM Cortex-A15 intellectual property core from ARM Holdings.

The configuration tested by AnandTech was the MSM 8960 MDP, a dual-core 1.5 GHz design.  Quad-core and single-core variants will also be available.  The cores onboard pack twice the L2 cache (1 MB) of their predecessor, a deeper 11-stage pipeline a 50% wider instruction decoder (3-wide), and twice the execution ports (7).  The cores support out-of-order execution and 128-bit NEON instructions.

In certain math-heavy (e.g. Linpack) or cache-heavy benchmarks, the S4 more than doubles the scores of its closest competitors -- the Motorola Droid RAZR (1.2 GHz dual-core OMAP4430 from Texas Instruments, Inc. (TXN)) and the Galaxy S II (also OMAP4430), and the Galaxy Nexus (1.2 GHz dual-core OMAP4460).  In browser benchmarks, the new S4 chip is 20-35 percent faster.

AnandTech summarizes:

These results as a whole simply quantify what we've felt during our use of the MSM8960 MDP: this is the absolute smoothest we've ever seen Ice Cream Sandwich run.

The on-die Adreno 225 GPU also receives a nice bump.  While it does not dominate like the CPU, the transition to 28 nm allowed Qualcomm to squeeze Direct3D 9_3 and bump the clock speed to 400 MHz, up substantially from the 266 MHz Adreno 220.

The improvements allow the Adreno 225 to trade blows with the PowerVR SGX543 MP2 (Imagination Technologies Group plc. (LON:IMG)) in Apple, Inc.'s (AAPL) iPhone 4S and ARM Holdings' Mali 400 in the Galaxy S II.  In both cases, the Adreno 225 proves its mettle, offering performance that can match or beat the competition in some cases.

In its power benchmarks, AnandTech used the platform's built in power-measuring tools, the Trepn Profiler software.  The results showed that in periods of modest activity (e.g. web browsing) a single Krait core was used, with the second core and GPU only occasionally becoming activity.  Power performance hovered around 400-750 milliwatts, with a peak consumption of about 750 milliwatts per core.  Idle periods showed about 150 milliwatts.

The Snapdragon S4 looks to be the most powerful smartphone CPU when it launches later this year in handsets. [Image Source: Jason Mick/DailyTech]

During demanding GPU activity like games, the platform could suck down 800-1200 milliwatts.

The average Android cell phone has been a 1600 and 2000 milliwatt-hour battery, so this indicates that the phone is good for about 3-4 hours of web browsing and 1.5-2 hours of gaming.  Again, this lower battery life is thanks to the development platform's larger 4-inch display, much as the iPhone's strong battery life is thanks to its small 3.5-inch display.

The development model was running Android 4.0.3.

Overall, the S4 is likely to see heavy pickup in Windows Phone models.  It also could potentially win business from Samsung Electronics Comp., Ltd. (KS:005930), who experimented with the S3 in some of its Galaxy S II models.

Source: Anandtech

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

RE: Always the idiot comment
By messele on 2/21/2012 4:49:44 PM , Rating: -1
Yup, somehow that marginal difference in surface area (Mick probably calls it "girth") is more responsible for an Android's power sucking ways than poor software engineering and a myriad of unnecessary, cruddy processes sucking all those electrons. After all, 3.5" is tiny compared to 4" as any hooker will tell you.

Still, shame on your for pointing it out. I hope you are downrated to oblivion by the Andrilluminati for this. Oh... wait...

RE: Always the idiot comment
By Kwiker on 2/21/2012 5:08:09 PM , Rating: 5
3.5" display = 1.94"x2.92" @1:1.5 = 5.66"sq
4" display = 2.22"x3.33" @1:1.5 = 7.39"sq

that .5" accounts for a 23.4% increase in the amount of surface area that needs to be illuminated.

I don't use either platform, I just like correct math :)

RE: Always the idiot comment
By messele on 2/21/12, Rating: -1
RE: Always the idiot comment
By BSMonitor on 2/23/2012 2:28:24 PM , Rating: 1
BSMonitor sucks at mathematics but somehow managed to actually gain 2 points in his rating in the time that Kwiker (top of the class) pointed that out. Mick will be happy that new improved 'droidagra makes you 23.4% less flaccid.

7/6 is ~16%. The math was correct. Donkey.

It's called an estimate of the dimensions. 16%, 25%, 35%.. None of them is close to saying Large vs. Small.

You my friend are a moron.

RE: Always the idiot comment
By Jeremy87 on 2/21/2012 6:47:53 PM , Rating: 3
7.39/5.66 = 30.6% increase, not 23.4.

I just like correcting corrected math.

RE: Always the idiot comment
By Solice55 on 2/22/2012 10:19:41 AM , Rating: 2
5.66/7.39 = 0.7659, or 76.6%

If 7.39 is 100%, then this would be a change of 23.4% (100 - 76.6).

RE: Always the idiot comment
By GmTrix on 2/22/2012 11:39:15 AM , Rating: 2
7.39/5.66 = 1.305 or (an increase of 30.5%)

5.66/7.39 = 0.765 or (a decrease of 76.5%)

It's not accurate to say that (100 - 76.5)% is the percent increase because that 76.5% is a percent of 7.39 not 5.66.

To make it clear:

"What percentage do you times by 5.66 to get 7.39?", i.e.
5.66*increase% = 7.39 ==> 1.305 ==> 30.5% increase

"What percentage do you times by 7.39 to get 5.66?", i.e.
7.39*decrease% = 5.66 ==> 0.765 ==> 76.5% decrease

RE: Always the idiot comment
By theapparition on 2/22/2012 12:02:22 PM , Rating: 2
Not to nitpick, as I agree with most everything you have listed, but when talking about decrease, it's 1-the percentage.

So it would be more correct to say it's 76.5% of the brightness or a 23.5% decrease.

76.5% decrease implies that something is only 23.5% as much.

RE: Always the idiot comment
By GmTrix on 2/22/2012 12:53:47 PM , Rating: 2
Yep you're right. My bad.

RE: Always the idiot comment
By theapparition on 2/22/2012 10:08:08 AM , Rating: 3
Need to correct some things a bit:
You correctly listed Apple screen as a 1:1.5 ratio.

3.5" display @ 1:1.5 = 1.94"x2.92" = 5.66"sq

But all 4" displays are not 1:1.5 since they are on Android devices, the aspect ratio is closer to 1.6667 (480x800), or qHD and HD are 1.7777 (960x540 or 1280x720).

So replacing some numbers with real life android equivalents on some models.
4.0" display @ 1:1.6667 = 2.06"x3.43" = 7.07"sq
4.3" display @ 1:1.7778 = 2.11"x3.75" = 7.91"sq
4.65"display @ 1:1.7778 = 2.28"x4.05" = 9.24"sq

So that would be 24.9%, 39.8% or a whopping 63.3% increase of lighted area.

Considering that a full 720 HD screen is also pushing 50% more pixels than the iPhone 4/4S, then you can see where some extra power is going.

RE: Always the idiot comment
By Kwiker on 2/22/2012 10:45:13 AM , Rating: 3
very nice :) sorry I was too lazy to dig up exact ratios. I was mostly just trying to point out that the .5" was far less of a "trivial" increase as was being claimed.

RE: Always the idiot comment
By BSMonitor on 2/23/2012 2:25:09 PM , Rating: 1
Donkey. Missed point entirely.

Large compared to small is not 25%. Grow up

RE: Always the idiot comment
By BSMonitor on 2/23/2012 2:22:55 PM , Rating: 1
Well, it's called quick estimating. I was off 7%. I was not lame enough to look up the exact dimensions down to the 100ths of an inch.

7/6 = 16% more. Grow up donkey. Math was correct.

23% or 16% my point is the same.

If Large is 1.25 and Small is 1.00. Large is then defined as pathetic. As the difference doesn't even make up a quarter of small.

Or, the terms do not apply correctly. Hence JMicks BS use of the English language.

"Death Is Very Likely The Single Best Invention Of Life" -- Steve Jobs

Latest Headlines
Inspiron Laptops & 2-in-1 PCs
September 25, 2016, 9:00 AM
The Samsung Galaxy S7
September 14, 2016, 6:00 AM
Apple Watch 2 – Coming September 7th
September 3, 2016, 6:30 AM
Apple says “See you on the 7th.”
September 1, 2016, 6:30 AM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki