Replies to post #113125 on Intel Corp (INTC)

[wbmw]

ARM have produced a processor that looks too hot for phones at 32nm at least in dual-core form which is probably why they have this big.little combination lined up as a single-core A15 combined with a single-core A7 should be acceptable.

I think they'll be a bit more aggressive than that. We've already seen quad core A9 in the latest Nexus 4, as well as the very same dual core Exynos5 A15 in the latest Nexus 10.

Phones usually operate <1W, but they can peak at about 2-3W. That's similar to how Medfield runs at 1.3GHz, but can peak at 2GHz when needed. Tablets, however, are larger and can sustain probably 2W of power, while peaking at 4-5W. And if you add a fan... well, you've seen Ivy Bridge tablets, right?

So I think we'll be seeing dual core A15, even down into phones, eventually - but they'll be the superphones and phablets, at 4.5-5+ inches in display. However, in order to do that, they'll need to voltage scale down from where the Chromebook is. I don't know if they've announced Exynos 5 operating range inside the Nexus 10, but I bet it's lower than 1.7GHz. In a phone, it will probably be closer to 1.0-1.2GHz - and by losing so much of its frequency, it will probably lose a lot of its performance advantage relative to Clover Trail and CT+.

[Colonel Sanders]

I'm kind of surprised that the revamped OoO Atom core on a process advantage wouldn't be more dominant than that.

[wbmw]

Nexus 10 Performance Below Medfield

http://www.anandtech.com/print/6425

Some results (Nexus 10 = 1.0 baseline in relative performance):

Sunspider Javascript 0.9.1 (lower = better)
- Motorola RAZR-i: 1087 (1.3x)
- Nexus 4 (Qualcomm Quad Core Krait): 1857 (0.75x)
- Nexus 10 (Samsung Dual Core A15): 1384 (1.0x)

BrowserMark
- Motorola RAZR-i: 150,991 (1.1x)
- Nexus 4 (Qualcomm Quad Core Krait): 86,550 (0.62x)
- Nexus 10 (Samsung Dual Core A15): 139,561 (1.0x)

Google Octane Benchmark v1
- Motorola RAZR-i: 2048 (0.75x)
- Nexus 4 (Qualcomm Quad Core Krait): 1124 (0.41x)
- Nexus 10 (Samsung Dual Core A15): 2748 (1.0x)

Mozilla Kraken Benchmark (lower = better)
- Motorola RAZR-i: 14,455 (0.77x)
- Nexus 4 (Qualcomm Quad Core Krait): 23,191 (0.48x)
- Nexus 10 (Samsung Dual Core A15): 11,146 (1.0x)

In these, since Nexus 10 was not tested, Nexus 4 = 1.0x baseline

Battery Life: AT Smartphone Bench: Web browsing over 3G/4G (lower = better)
- Motorola RAZR-i: 6.47 hours (1.5x)
- Nexus 4 (Qualcomm Quad Core Krait): 4.18 hours (1.0x)

Battery Life: AT Smartphone Bench: Web browsing over WiFi (lower = better)
- Motorola RAZR-i: 7.88 hours (1.3x)
- Nexus 4 (Qualcomm Quad Core Krait): 6.27 hours (1.0x)

I find these results to be totally amazing. Intel pretty much destroys the Nexus 4 - but for that matter most other dual core phones destroy the Nexus 4 - including those with Qualcomm dual cores in them. That pretty much tells me that quad core is nearly useless under the current Android benchmarks, and yet LG probably needed to downclock and undervolt the S4-Pro processor in order to fit it in a phone, which hurts the single threaded performance.

Ah, the irony! People thought that Intel would be too short on cores against the onslaught of ARM competition, and yet it seems like going fast to quad core has more handicapped the ARM based ecosystem, rather than benefitted it. Medfield with just one core is blazing through the benchmarks and running circles around the other phones.

As for the Nexus 10 with dual core A15, many thought that would run circles around Medfield, but it's interesting to see Medfield put up a good fight - and as an unfair fight of a phone going up against a tablet. The Exynos 5 processor has more thermal envelope in the Nexus 10 than the Qualcomm S4-Pro has in the Nexus 4. We'll have to see how well A15 does in a phone form factor - but for now the results are highly encouraging!

Of course, props do go to Samsung and ARM for the Mali-T604 graphics core, currently outperforming both the Adreno 320 as well as the SGX543-MP3 core in the Apple A6, based on the Egypt 1080p offscreen test. Of course, thanks also go to the higher power dissipation of the Nexus 10 tablet, compared to the phone chips. I highly expect Apple to take back the crown with the A6X enabled iPad.

But at least in terms of phones, the Nexus 4 had a rather poor showing in performance, even relative to the other Qualcomm S4-Pro based LG phone, the Optimus G. I think we'll need to see more data, but so far I am fairly bullish that Intel will have a leadership product next year with CT+.

Edit: Interesting note from Anand on the Optimus G, which also uses the Adreno 320:

Normalize resolution and the Mali-T604 actually does very well here, setting a new performance record. Despite being based on the same hardware, the Optimus G is able to post a much higher score here than the Nexus 4. The explanation is simple: the Optimus G can't complete a single, continuous run of GLBenchmark 2.5 - the app will run out of texture memory and crash if you try to run through the entire suite in a single setting. The outcome is that the Optimus G avoids some otherwise nasty throttling. The Nexus 4 on the other hand manages to complete everything, but likely quickly throttles its clocks down due to thermal constraints. The Nexus 4 was really hot by the end of our GLBenchmark run, which does point to some thermal throttling going on here. I do wonder if the Snapdragon S4 Pro is a bit too much for a smartphone, and is better suited for a tablet at 28nm.