CPU, APU & Chipsets - Page 146

Computex 2018 - AMD is pushing for a strong Computex this year, with an exciting unveiling of Vega 20 on 7nm with 32GB of HBM2, but also their new project: Fenghuang 15FF, a new graphics chip for APUs,

Fenghuang 15FF is the internal codename for the graphics chip, with the current codename of DG02SRTBP4MFA applied. This chip should be part of the Zen+ APU family with 4C/8T of CPU power, 3GHz clock speed, and 2GB of HBM2 clocked at 1200MHz (2.4GHz effective).

AMD's new chip sports 28 CUs and 1792 SPs (up from 24 CUs and 1536 SPs on Vega 24) while there's 2GB of HBM2 at 2.4Gbps offering 307.2GB/sec memory bandwidth, which is up from the 1.6Gbps and 204.8GB/sec offered on Vega 24.

AMD really raised the bar with multi-core processors with the successful launches of Ryzen and Threadripper, and while Intel sales are barely affected by it, it shouldn't stop AMD from continuing to push out new generations of these CPUs with many more cores.

The latest rumor is from WCCFTech so please take it with a grain of salt, that AMD will making the AM4 socket detect CPUs with greater than 8 cores, ushering in support for possibly 12- and 16-core CPUs. WCCFTech adds "according to private conversations we've had with incredibly knowledgeable people in the industry the actual core count figure might be closer to 12 cores" and not the 16C/32T that we're hearing with the latest rumors.

Ryzen 7 3700/X or Ryzen 7 3800/X with 12C/24T at over 4GHz would be quite the chip, quite the chip indeed. Having that type of power at the top of the Ryzen 7 mainstream lineup opens the flood gates to Threadripper ramping up to 20C/40T, 24C/48T, and even 32C/64T which will meet EPYC. EPYC can be opened up further with next-gen nodes and Zen architectures, where we will see beyond 32C/64T which will be very impressive, time will tell though.

It looks like Intel is set for a huge Computex this year with the launch of their beefed up Z390 chipset, but the tease of a new 8C/16T variant of its Coffee Lake CPU, a new processor that would succeed the already kick ass Core i7-8700K.

The Core i7-8700K is a 6C/12T gaming beast in its own right, but Intel will want to crush the 8C/16T space that AMD is shifting on with the Ryzen 7 2700. There's been a new listing on SiSoft's benchmark database, with the unnamed chip running at 2.6GHz with 8 cores in tow.

We should expect Intel to announce the new 8C/16T processor that should arrive during Computex, but there's one thing I'm concerned over: could Intel lock the new 8C/16T chip to only work on the new Z390 chipset? Or will they open up support for the new CFL processor on their other chipsets, too?

Intel is very close to the launch of its new Z390 chipset that has been rumored since last year, and will succeed the current Z370 chipset while keeping most of its features and expanding on some.

Intel is building Wireless-AC 802.11 AC and Bluetooth 5.0 into their new Z390 chipset, while also adding in 6 x USB 3.1 Gen 2 ports. Motherboard manufacturers might offer this connectivity on their Z370-based motherboards, but they rely on third-party controllers which increase the BOM of a motherboard.

Intel has rolled these features into the Z390 chipset, which is a win-win for Intel and motherboard partners. Until now, the Z370 featured integrated USB 3.1 Gen 1 and Thunderbolt 3 (Alpine Ridge) connectivity, while Z390 adds the integrated Wireless-AC and BT 5.0 controllers, as well as an integrated SDXC 3.0 controller and updated Thunderbolt 3.0 (Titan Ridge) with DP 1.4 connectivity.

Intel has been having serious troubles with its 10nm process, but it looks like we might see one of their new 10nm Cannonlake CPUs in the very near future.

According to a listing on a Chinese retail outlet, Lenovo is preparing their new Ideapad 330 that will be the first notebook to market with an Intel 10nm Cannonlake CPU. Lenovo's upcoming Ideapad 330 packs the Cannonlake-powered Core i3-8121U processor, AMD Radeon RX 540 2GB discrete graphics card, and up to 8GB of DDR4.

There will be up to 1TB 5400RPM mechanical HDD and 256GB SSD that will be included in the highest-end Ideapad 330, which will arrive for around $450. Lenovo's new Ideapad 330 features a 15.6-inch TN panel with a native resolution of 1366 x 768, so it won't be busting any records, that's for sure.

Intel is expected to launch its refreshed Z390 chipset in the next couple of months, where I'm sure Chipzilla will show off the Z390 during Computex 2018 next week.

We're now being exposed to some of the secrets behind the Z390 chipset, with SuperMicro's upcoming and always confusingly named C9Z390-CG-IW. The board packed a Core i7-8700T processor, a low-power 6C/12T chip that is already available, as well as the Core i7-8700K which is an 8C/16T chip.

This means that the new Z390 chipset will support all current-gen Coffee Lake S (CFS) parts, as it uses the same LGA1151 socket found in current-gen 300-series motherboards. We should expect Intel to show off more about the Z390 next month at Computex, and we'll be there bringing you the news as it happens.

Intel didn't really have a good time with their Kaby Lake-X line of processors, with Chipzilla announcing that they are discontinuing the CPUs after only 11 months on the market.

Originally, Kaby Lake-X were designed and marketed as an enthusiast HEDT platform to give quad-core CPUs on the 14nm+ architecture, but after a lack of sales and enthusiasm around them, Intel has put Kaby Lake-X into EOL (end of life).

At the time, Intel announced the X299 platform and two quad-core SKUs of KLX that were based on the enhanced 14nm+ node, but had less PCIe lanes, dual-channel DDR4 support, and next to no performance gains over the Skylake-X family of CPUs at the time on the much cheaper Z270 platform.

AMD will be continuing their onslaught of CPU releases throughout 2018, starting the year out with a bang with the Ryzen 2000 series processors led by the flagship Ryzen 7 2700X.

But now we're hearing about new Ryzen desktop CPUs with the Ryzen 2100, 2300X, and 2500X. These processors will join the Ryzen 2200 and 2400 processors, so we should expect the Ryzen 5 2500X to be a 4C/8T chip for mainstream consumers and gamers.

Moving onto the mobile offerings with Ryzen Mobile 2000U, 2600U, and 2800U. The interesting note here is the Ryzen 7 2800U that is a full Ryzen 7 flagship offering in mobile designs, very cool to see, but will it be very hot under full load? That's what I want to see out of it all.

Intel has delayed its 10nm node yet again, but during AMD's recent Q1 2018 earnings call the company teased that they have not just their 7nm Radeon GPUs being tested in their labs, but upcoming 7nm server CPUs as well.

AMD President and CEO Lisa Su explained: "We have 7nm GPU based on Vega that we'll sample later this year. We have a 7nm server CPU that we'll sample later this year. And then, obviously, we have a number of products that are planned for 2019 as well. So it's a very, very busy product season for us. But we're pleased with the sort of the execution on the product roadmap".

Where we normally see Intel nailing the process node game, AMD is ahead here with what seems to be Globalfoundries' mature 7nm LP node that will power desktop Ryzen 3000 series CPUs in 2019.

Intel has delayed its upcoming 10nm processors yet again, with Cannon Lake stuck back in the firing line with mass production delayed until 2019, instead of late-2018 like Intel promised.

The company promised 10nm back in 2015, but delayed it into 2017 and then again into 2018, so this new delay marks the fourth delay in a row for Chipzilla. Instead, the company has been reworking its seventh-gen 14nm+ node (Kaby Lake R) and their newer 14nm++ (Coffee Lake). We were meant to have eighth-gen parts in 10nm Cannon Lake node, but yeah that won't be happening until 2019 now.

Intel revealed the fourth delay to its 10nm node in their recent earnings report, and while they nailed it with record Q1 revenue of $16.8 billion (up 16% year-over-year), high-volume 10nm production has been delayed until sometime in 2019, with no specifics provided at all.