Low-cost emerging market SoC/phone discussion

[Deleted member 13524]

Like Tegra 3, it seems the new MTK6588 will use single-channel LPDDR2/1.

The great thing about MTK chipsets is that they seem to be better at supporting dual-sim handsets.


BTW, looking at the results from MTK6577, it looks like that SGX531 is clocked at ~400MHz or so.
That's not bad at all, especially for a low-cost SoC.

 

[Lazy8s]

Unless I'm reading incorrectly, the leak shows that MediaTek's new multicore CPUs are A7 based, not A9.

 

[Laurent06]

Unless I'm reading incorrectly, the leak shows that MediaTek's new multicore CPUs are A7 based, not A9.

The slide for MT6588 on gizmochina indeed says Cortex-A7.

 

[Deleted member 13524]

Unless I'm reading incorrectly, the leak shows that MediaTek's new multicore CPUs are A7 based, not A9.

You're right.
Furthermore, they don't say the clock speeds yet.

At 28nm, a Cortex A7 is either going to consume very little or be clocked very high.

 

[wco81]

Heard ZTE or Huawei saw a big drop in sales, profits.

Low cost Android may not be taking over yet.

 

[Laurent06]

You're right.
Furthermore, they don't say the clock speeds yet.

At 28nm, a Cortex A7 is either going to consume very little or be clocked very high.

Mediatek says 1.X GHz and ARM says >1 GHz at 28nm

 

[Zaphod]

Heard ZTE or Huawei saw a big drop in sales, profits.

Heard where? According to IDC, at least ZTE grew their smartphone sales 300% YOY in Q2 2012.



Edit: Huawei is increasing their shipments too, according to iSuppli. However, they seem to be growing at a (significantly) slower rate than previously forecast.

 

[wco81]

http://www.mobilebusinessbriefing.c...ed/24593?elq=b12671aac0b14b8c93fc60ead445975f

 

[ams]

That article doesn't mention anything about a drop in actual units sold, revenue, market share, etc. All it says is that profit has dropped, largely due to lower investment income earned compared to prior reporting period. Also note that ZTE does not have extravagent dollar amount of profit in the first place, so a big swing in investment income for the period could materially affect profitability.

 

[Deleted member 13524]

That article doesn't mention anything about a drop in actual units sold, revenue, market share, etc. All it says is that profit has dropped, largely due to lower investment income earned compared to prior reporting period. Also note that ZTE does not have extravagent dollar amount of profit in the first place, so a big swing in investment income for the period could materially affect profitability.

Even then, firing 12000 people is big. It's Nokia big.

 

[tangey]

Not sure how practical it is, but mediatek are said to be launching the worlds first 8-core arm based smartphone/tablet chip in Q1 2013.

http://www.itproportal.com/2012/09/28/mediatek-to-launch-8-core-mt6599-soc-with-lte-in-q1-2013/

Although the following article quotes a more realistic q4 2013
http://techideki.com/mediatek-mt6599-8-core-with-lte-slated-q4-next-year/#more-1054

Also sen some news indicating that the quad core chip will clock the A7s @ 1.2ghz, and Lenovo will be first to market with a phone.

 

[redtailboas]

Not sure how practical it is, but mediatek are said to be launching the worlds first 8-core arm based smartphone/tablet chip in Q1 2013.

http://www.itproportal.com/2012/09/28/mediatek-to-launch-8-core-mt6599-soc-with-lte-in-q1-2013/

Although the following article quotes a more realistic q4 2013
http://techideki.com/mediatek-mt6599-8-core-with-lte-slated-q4-next-year/#more-1054

Also sen some news indicating that the quad core chip will clock the A7s @ 1.2ghz, and Lenovo will be first to market with a phone.

itportal maybe made a mistake or didn't research properly. the mtk6588 is coming out in q1 2013...not mtk6599.

gizbeat.com (http://gizbeat.com/781/breaking-mtk...at-mtk-mt6588-mtk6588-will-be-coming-q4-2012/) shows it coming out sometime in 2013.

i wouldn't be surprised to see it q3 of 2013.

the dual cored mtk6577 just started having phones in middle of this year, and already the 4 core mt6588 will come out end of this year or very beginning of 2013.

 

[shyam]

I didn't see anyone mentioning about upcoming XBurst JZ4780.
It's the successor of JZ4770.
It's a MIPS32 dual core clocked at 1.5GHz with 64bit DDR3-1066 and sgx540 manufactured on 40 nm.
Some preliminary bench-es here.

Supposedly very cheap and low on power.

 

[tangey]

Pretty sure its targetting tablets as opposed to Smartphones, but none the less, the Allwinner A31 Soc seems to be highly impressive.

quad core A7, and wait for it, EIGHT core SGX544. Depending on clock speed, this might well out-perform the ipad4 graphically.
I wonder if the spec is correct ? the 3Gpixel/s fill rate suggests around 185Mhz clock.

It also supports 4K video out.

http://www.jsxltech.com/blogs/allwi...ner-a31-tablet-pc-designed-and-made-by-ployer

http://www.jsxltech.com/blogs/allwi...-first-allwinner-a31-quad-core-tablet-by-icoo

 

[Ailuros]

Hmmm from those links:

The overall performance is 2 times faster as that of the Quad Core Tegra 3, 1.5 times faster than iPad 2,and rival that of the New iPad.

More like iPad3 performance? For iPad4 GPU performance there are quite a few GFLOPs missing if the frequency is truly around 185MHz.

Quite a surprise!

 

[Deleted member 13524]

In other news by the same site regarding another Allwinner A31 tablet, it's pretty explicit that it's a SG544MP2:

I guess that by saying "8-core" they're counting the amount of "shader units", nvidia-style.
But the clocks seem high, though. 375-400MHz? It borders on the performance of an ipad3, which is excellent.
The Allwinner A31 is also going with dual 32-bit DDR3 @ 1600MHz-ish, going for the same ~12GB/s bandwidth as the A5X.
And the Wii U.
Lawlz.

 

[Ailuros]

Good job detecting that slide!

It's not 375MHz then either, since fillrate is most likely quoted with depth complexity (probably a factor of 2.5x encounted).

SGX544MP2, 2 TMUs/core = 4 TMUs * 300MHz * 2.5x overdraw = 3000.

***edit: not that the marketing from Allwinner doesn't stink regarding the GPU, but if they'd really quote ALU lanes NV style, they would had quoted 32 cores [2*(4*Vec4)] ALUs.

 

[tangey]

yeah, well worked out both of you. First time I've seen PowerVr stated in terms of the number of ALUs.

Still 544MP2@300MHZ is pretty decent, probably not a match for the 543MP4@250 of the ipad3, but none the less, quite competitive amd with a better compliance set.

Allwinner (?) seem to have done a decent job in Soc design time. IMG licensed to Allwinner sometime between May '11 and April '12
(was mentioned in the full year results to April 2012 as a new licencee), so looks like they went from licence to end silicon in 18 months max.

Also, Is this the first tablet boasting 4K video out ? , wonder are they implementing some PowerVR video decode ?

 

[Ailuros]

http://www.glbenchmark.com/phonedet...ro25&D=Alps+mt8389_tablet_s3&testgroup=system

This one smells like the Mediatek SoC with a quad A7@1.2GHz + SGX544MP1@320MHz.

And Qualcomm announces mainstream SoCs with quad A7@1.2GHz just in time:

http://www.anandtech.com/show/6498/...y-msm8x26-and-wtr2605-quad-core-arm-cortex-a7

For the Allwinner stuff above I'd love to know how good sw integration is and the according MSRPs.

 

[Deleted member 13524]

With such high-performance mobile GPUs going into these low-cost SoCs (which already started with Rockchips getting a Mali 400MP4 into the low-cost RK3066 for tablets), these little guys are putting an enormous pressure on Intel.

Now that TI will be pretty much off the tablet market (will we ever see an OMAP5 in handhelds?), Intel will remain with the lowest-performing GPU of all tablet SoCs.
CloverTrail is set for the whole year of 2013 and has a 500MHz SGX545, and although 2014's ValleyView-T claims to have a Gen7 GPU, it only predicts a 3x faster performance than CloverTrail, which is.. dirt-slow for a 2014 tablet.