MediaTek Helio X30

Ailuros

Epsilon plus three
Legend
Supporter
It's just very unfortunate that Helio X30 won't likely meet sales expectations :(
 
What makes you say that?

http://www.anandtech.com/show/11166/mediatek-helio-x30-10-cores-on-10nm

In a short presentation at its MWC booth Monday, Executive Vice President & Co-COO Jeffrey Ju stated that MediaTek expects only a limited number of phones to use the Helio X30, perhaps less than ten. He also mentioned that a low yield rate for TSMC’s 10nm process has delayed the X30’s availability.

http://www.fudzilla.com/news/mobile/42903-mediatek-s-helio-x30-10nm-demand-is-lower

Irrelevant of what many think of the author itself or the site, Xiaomi going for its own SoCs doesn't seem to be wrong amongst others.
 
If that number of phones sells well maybe is not that bad. But Qualcomm is getting a lot of the mid-high end phones from China.
 
Irrelevant of what many think of the author itself or the site, Xiaomi going for its own SoCs doesn't seem to be wrong amongst others.

Xiaomi's Surge S1 is a 28nm 4+4 A53 little.LITTLE SoC with a Mali T860MP4.
It's a competitor to Mediatek's P20 and Snapdragon 625, not to X30.

This is most probably going into their Redmi line. I'm a big fan of the Redmis, I always recommend them to people on a 150-200€ budget, but that line of phones would never get a X30.
 
Xiaomi's Surge S1 is a 28nm 4+4 A53 little.LITTLE SoC with a Mali T860MP4.
It's a competitor to Mediatek's P20 and Snapdragon 625, not to X30.

This is most probably going into their Redmi line. I'm a big fan of the Redmis, I always recommend them to people on a 150-200€ budget, but that line of phones would never get a X30.

I obviously didn't mean the V670 but the V970 which comes a bit later: http://www.gsmarena.com/two_xiaomi_...cortexa73_and_malig71_included-news-23560.php

As for the X30 if I'm going for it, I'd rather have a LeEco/LeTV device.
 
A 28nm 8 cores A53 is disappointing for a 2017 SOC. When there are cheap Finfet process nodes and new IPs like Cortex A73 or new Mali G-51/71 that brings a noticeable bump in performance and efficiency without increasing the cost/area.
 
A 28nm 8 cores A53 is disappointing for a 2017 SOC. When there are cheap Finfet process nodes and new IPs like Cortex A73 or new Mali G-51/71 that brings a noticeable bump in performance and efficiency without increasing the cost/area.
Yeah, even helio x20 with 20nm is already a power hog.

I'm baffled why xiaomi chose 28nm
 
Overwhelmingly the reason will be cost. 28 is much much cheaper than 16/14/12/10, due to tech and maturity.
 
Overwhelmingly the reason will be cost. 28 is much much cheaper than 16/14/12/10, due to tech and maturity.
The tiny Qualcomm SD625 is manufactured at 16nmFF+ ( i think it was launch before the more cheap Compact version was available ) and it is being sold in devices around 200$. I just think it is too little too late. It is not like they are doing this starting from zero they partnership with Leadcore ( if i am not wrong ) which has some experience designing SOCs. And if we go to details like memory controller, image signal processor, modem and other stuff... They have a long way.
 
Qualcomm have different economic scales in play to justify advanced tech for something like 625. Pine Cone are just getting started, so playing it safe with something easy to make and integrate, with established IP, on a cheap and very mainstream process, is a good way to test the waters with low risk. It's their first chip after all.
 
Mediatek doesn't manufacture its mainstream (careful mainstream for Mediatek's standards) in anything better than 28HPm right now unless I'm missing something. One might say that MTK's next batch of SoCs will be for N manufacturing process, but by that time Pinecone V670 won't be a mainstream SoC in the grander scheme of things anymore either.

The tiny Qualcomm SD625 is manufactured at 16nmFF+ ( i think it was launch before the more cheap Compact version was available ) and it is being sold in devices around 200$. I just think it is too little too late. It is not like they are doing this starting from zero they partnership with Leadcore ( if i am not wrong ) which has some experience designing SOCs. And if we go to details like memory controller, image signal processor, modem and other stuff... They have a long way.

In line with the above: http://www.anandtech.com/show/10758/qualcomm-adds-new-snapdragon-socs-include-support-for-8gb-ram

No idea if QCOM is dual sourcing the 625, but if not it's being manufactured on 14LPP Samsung. Besides the point. The real point is that a 28nm process even today is by far not unreasonable for anything mainstream or lower. For the given case unless Xiaomi hasn't announced all its SoC plans yet, it smells more like that they'll use a high end Snapdragon above their V970 which looks like their mainstream offering of the neareast future.

Yeah, even helio x20 with 20nm is already a power hog.

The Note Pro 3 seems to have a better battery life at 28nm than the Note 4 here despite the 20SoC for the latter http://www.anandtech.com/show/11137...port-xiaomi-redmi-note-4-mediatek-helio-x20/4
 
Last edited:
I don't get why Snapdragon 650 isn't used a lot more widely, as it does so much better than any of the little.LITTLE solutions out there.
Maybe it's a lot more expensive but damn, the 650 really is a sheep in wolf's clothing.
they did give it a weaker gpu that the qc 8xx socs
 
they did give it a weaker gpu that the qc 8xx socs
Which 8xx?
The GPU is practically on par with the Snapdragon 808. In fact, I'd say the 650 is pretty much a 808 "done right".
Qualcomm could probably get away with calling it a Snapdragon 809 or 812 or something.

It's such a shame they haven't introduced a 14FF shrink of Snapdragon 650, like they've done with the 615 -> 625.
Maybe they haven't because doing so could turn the 820 into a harder sale.
 
I don't get why Snapdragon 650 isn't used a lot more widely, as it does so much better than any of the little.LITTLE solutions out there.
Maybe it's a lot more expensive but damn, the 650 really is a sheep in wolf's clothing.

I can also guess it's primarily a cost issue and that mostly for their partners. Freely invented numbers: if you can get the job done for mainstream smartphone X with a $10 SoC, why go for a $15 solution instead?

To turn that around a connect it back to Helio X30 I don't know any details, but I would imagine that the SoC is neither small, nor is manufacturing at 10FF TSMC at this stage cheap (we have the bad yields for 10FF tidbit from official MTK lips....). So far MTK's "high end" SoCs are more or less somewhere between QCOM's high end and mainstream SoCs all factors counted. In order for the X30 to stand its ground it would have to be significantly cheaper than QCOM's high end 10FF SoCs. Because faster than the average mainstream but quite a bit more expensive isn't going to save the X30's day/sales either.
 
Back
Top