Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Wednesday March 01 2017, @11:18AM   Printer-friendly [Skip to comment(s)]
from the a-core-USED-to-refer-to-a-single-bit-of-memory dept.

MediaTek has released more details of an upcoming 10-core SoC:

MediaTek first unveiled the Helio X30—its next-generation high-end SoC—last fall, but today at Mobile World Congress the Taiwanese company announced its commercial availability. The Helio X30 is entering mass production and should make its debut inside a mobile device sometime in Q2 2017.

The Helio X30, like the Helio X20 family before it, incorporates 10 CPU cores arranged in a Max.Mid.Min tri-cluster configuration. Two of ARM's latest A73 CPU cores replace the two A72s in the Max cluster, which should improve performance and reduce power consumption. The Mid cluster still uses 4 A53 cores, but they receive a 10% frequency boost relative to the top-of-the-line Helio X27. In the X30's Min cluster we find the first implementation of ARM's most-efficient A-series core. The A35 consumes 32% less power than the A53 it replaces (same process/frequency), while delivering 80%-100% of the performance, according to ARM. With a higher peak frequency of 1.9GHz, the X30's A35 cores should deliver about the same or better performance than the X20's A53 cores and still consume less power.

Also at Tom's Hardware, entitled "The 10nm Helio X30 May Be MediaTek's First Truly Competitive High-End Chip".

While some smartphone SoCs like the X30 are a bit of an exception due to cluster configurations, there are going to be many CPUs with 8+ cores sold in 2017. Some examples that come to mind: AMD's Ryzen 7 desktop CPUs, the AMD APUs in the Xbox One, PS4, and PS4 Pro (with 7 cores usable in these consoles), and other smartphone SoCs like the Exynos 7 Octa 7880, which uses equivalent cores rather than clusters. Will games and popular applications be able to exploit this newfound glut of cores?

Related: Samsung's Exynos 8895 to be the First 10nm Chip on the Market


Original Submission

Related Stories

Moore's Law: Not Dead? Intel Says its 10nm Chips Will Beat Samsung's 15 comments

Intel is talking about improvements it has made to transistor scaling for the 10nm process node, and claims that its version of 10nm will increase transistor density by 2.7x rather than doubling it.

On the face of it, three years between process shrinks, rather than the traditional two years, would appear to end Moore's Law. But Intel claims that's not so. The company says that the 14nm and 10nm process shrinks in particular more than doubled the transistor density. At 10nm, for example, the company names a couple of techniques that are enabling this "hyperscaling." Each logic cell (an arrangement of transistors to form a specific logic gate, such as a NAND gate or a flip flop) is surrounded by dummy gates: spacers to isolate one cell from its neighbor. Traditionally, two dummy gates have been used at the boundary of each cell; at 10nm, Intel is reducing this to a single dummy gate, thereby reducing the space occupied by each cell and allowing them to be packed more tightly.

Each gate has a number of contacts used to join them to the metal layers of the chip. Traditionally, the contact was offset from the gate. At 10nm, Intel is stacking the contacts on top of the gates, which it calls "contact over active gate." Again, this reduces the space each gate takes, increasing the transistor density.

Samsung's Exynos 8895 to be the First 10nm Chip on the Market 7 comments

Samsung will be the first company to sell a 10nm chip:

Samsung announced its next-generation mobile application processor, the Exynos 9 Series 8895, and said it's the first 10nm processor built so far. This means Samsung beat Intel and TSMC to the next-gen process node, but Qualcomm's Snapdragon 835 could soon follow on the same 10nm process.

The 10nm FinFET process brings an improved 3D transistor structure that allows for up to 27% higher performance or up to 40% power consumption when compared to Samsung's previous 14nm FinFET process.

Samsung said the Exynos 8895 is also the first chip to embed a gigabit LTE modem that supports five carrier aggregation (5CA). This allows wireless operators to combine multiple fragments of a spectrum to deliver higher data throughput. The modem can achieve up to 1Gbps (Cat. 16) downlink with 5CA, and 150 Mbps uplink with 2CA.

There's plenty of room at the bottom???


Original Submission

MediaTek Dimensity 1000 SoC Supports 5G Speeds, 16 GB RAM, and AV1 Video Decoding 13 comments

MediaTek Dimensity 1000 octa-core SoC promises 5G for the masses when it launches in 2020

The 5G SoC will support 90 Hz QHD displays, up to 16 GB of quad-channel LPDDR4x RAM, Wi-Fi 6, Bluetooth 5.1, hexa-core AI processor, download/upload speeds of up to 4.7/2.5 Gbps, and more with the promise of faster performance than the current Qualcomm Snapdragon 855.

While a handful of 5G smartphones are already available today, they are all prohibitively expensive. The Samsung Galaxy S10 5G, for example, currently retails for $1300 unlocked. MediaTek hopes to be the catalyst for 5G adoption next year by launching an all-in-one SoC solution that integrates an octa-core CPU, octa-core Mali-G77 MC9 GPU, hexa-core AI APU, and a 5G modem for more affordable smartphones.

Called the Dimensity 1000, the SoC will be the first in a series of SoCs with integrated support for 2G, 3G, 4G, and sub-6 GHz 5G networks. MediaTek is also claiming it to be the world's first SoC to support 5G dual-SIM for better worldwide appeal and versatility. While single-SIM smartphones are still prevalent in the U.S., most smartphones overseas tend to carry two SIM slots.

MediaTek's presentation shows that the SoC will support AOMedia Video 1 (AV1) hardware decoding at up to 4K60:

In addition to hardware video encoding and decoding at 4K 60FPS, the MediaTek Dimensity 1000 is the world's 1st mobile SoC with AV1 format support.

Also at AnandTech.

Related: MediaTek Announces 10-Core SoC for Phones and Tablets
MediaTek Helio X30: 10 Cores on a 10nm Process
Qualcomm's Snapdragon 855 SoC Will Optionally Enable 5G Connections with Added X50 Modem
Realtek RTD2983 SoC for 8K TVs: Supports AV1 Codec
Huawei: ARM Cortex-A77 Cores Would Shorten Battery Life (Dimensity 1000 includes 4x Arm Cortex-A77 cores)


Original Submission

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 0) by Anonymous Coward on Wednesday March 01 2017, @05:11PM (2 children)

    by Anonymous Coward on Wednesday March 01 2017, @05:11PM (#473397)

    I've never understood why these companies announce their products to the public; it's not like the public can purchase them for their own projects, or even just program them in the corporate devices that do incorporate them.

    Why do any of you care about a tool you'll never wield?

    • (Score: 2) by Aiwendil on Wednesday March 01 2017, @05:52PM

      by Aiwendil (531) on Wednesday March 01 2017, @05:52PM (#473417) Journal

      Probably same reason as why they read about jetplanes, sportcars and swimsuit models - with the exception that the crowd here on soylent has higher likelyhood of actually have use for knowing roughly what is out there.

      Heck - prior to the RPi I never tough I would mess around that much with ARM-based machines (excepting heavily locked diwn stuff (smartphones, tablets)

    • (Score: 0) by Anonymous Coward on Wednesday March 01 2017, @08:33PM

      by Anonymous Coward on Wednesday March 01 2017, @08:33PM (#473496)

      To increase demand for their products. Why else would they announce to the public?

      Why do you think Intel used to* constantly advertise their products, on the whole they don't sell a lot of individual chips to enthusiasts, but it increases demand for products containing their chips, which does increase their sales to OEMs.

      *Maybe they still do, I don't pay much attention to adverts these days.

  • (Score: 2) by TheRaven on Wednesday March 01 2017, @07:37PM (2 children)

    by TheRaven (270) on Wednesday March 01 2017, @07:37PM (#473459) Journal
    Scheduling with big.LITTLE is still really hard and nothing does it well. These big.Medium.LITTLE ones make the problems even harder. A53s are pretty anaemic (64-bit version of the Cortex-A7), so the A35s are pretty useless as a third performance point.
    --
    sudo mod me up
    • (Score: 2) by takyon on Wednesday March 01 2017, @08:28PM

      by takyon (881) <takyonNO@SPAMsoylentnews.org> on Wednesday March 01 2017, @08:28PM (#473493) Journal

      So what about Exynos 7 Octa 7880 or other octo-core SoCs with equivalent cores?

      --
      [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
    • (Score: 2) by bob_super on Wednesday March 01 2017, @09:52PM

      by bob_super (1357) on Wednesday March 01 2017, @09:52PM (#473553)

      Actually, the A35 are the most useful for menial OS tasks and HW management. The A73s give you the good benchmarks numbers people crave, at the cost of your battery life, while the A53 are perfectly capable of running most apps smoothly...

      But I agree that a smart scheduler is hard. I'd like to know how much of a hit you get when you have to switch between those cores. I sure hope they have some shared cache.

  • (Score: 2) by bzipitidoo on Wednesday March 01 2017, @10:58PM (1 child)

    by bzipitidoo (4388) Subscriber Badge on Wednesday March 01 2017, @10:58PM (#473592) Journal

    This Helio chip is good news for smartphones. That it is on a 10nm die may be bigger news than anything about the chip itself. Ought to help push desktop CPU manufacturers to shrink more. Though, 14nm is a big leap from 22nm and larger.

    I could use updated desktops and laptops. x86 stuff I have is pushing 10 years of age. No SSE4, which wasn't a problem until I ran into a game binary that requires it. I've been looking at what's available and thinking about it. The question is, wait a bit longer for the AMD Ryzen and 10nm chips? Also have to figure out where I want to be on the performance vs low power curve, and am favoring low power, but not too low.

    I really like not having fans droning away all day long, however cheap stick computers with those Atom X5-Z8300 CPUs like the Asus Vivostick are a mite too sluggish to suit me. I got one for under $100, and found it able to play videos on Youtube and run high end games at very low graphics settings. Had to cut the screen size down, 1920x1080 was pushing it too much. It wasn't the performance, so much as a lack of RAM. But at that price, and at a peak power usage of only 6W, lack of performance is totally forgivable. Works just fine for checking email. The NUC is more muscular but gets pricey in a hurry. Or there are these Zotac PCs. There's this Taiwanese manufacturer named Giada that has some interesting offerings I can get through NewEgg. Also on NewEgg is this Kangaroo Mobile Desktop. Another company I watch is Polywell. I have a Soekris I use as a web server, but those computers aren't suitable for desktops. Tried a Beagleboard a while back, and while I was able to get Ubuntu running on it, it was unstable, probably because the USB power supply I'd gotten for it wasn't powerful enough.

    Or I can just get a plain old laptop. It's really tough to beat a commodity laptop for value, with it by definition including a screen. Computer retailers have used the old "monitor sold separately" trick since the 1980s to make systems seem less expensive, and the Zotacs, Polywells, Kangaroos, Giadas and others all employ that.

    • (Score: 2) by takyon on Thursday March 02 2017, @03:57AM

      by takyon (881) <takyonNO@SPAMsoylentnews.org> on Thursday March 02 2017, @03:57AM (#473720) Journal

      Ought to help push desktop CPU manufacturers to shrink more.

      Not necessarily. AMD's fortunes are dependent on the fabrication entity they spun off in 2008, GlobalFoundries. Only Intel is independent. The only other "desktop chip" I can think of is POWER9, and that's also made by GlobalFoundries.

      Now we know that Intel is sticking to 14nm for roughly 4 generations of chips (you could say 3.5 if they move some chips to 10nm early). They killed Tick-Tock. Did they do that because AMD didn't represent serious competition, or because their 10nm process is not mature and yields aren't ideal? I'd say the latter. They also compete with Nvidia (Xeon Phi) and IBM/ARM/etc.

      We are nearing the end of the line for "easy" shrinking. 10nm works, 7nm seems to be feasible. 5nm and below? Unknown. Carbon nanotubes or other materials might need to be thrown into the equation. And it's an economic equation [wikipedia.org].

      --
      [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(1)