AMD's upcoming Zen CPUs and new AM4 socket are closer and closer to being revealed, and now we have leaked shots of the PGA (Pin Grid Array) design which looks to have 1331 pins - just 6 short of that awesome 1337 number, AMD.
The current AM3+ socket from AMD is running out of steam, with the upcoming AM4 socket delivering native PCIe 3.0 support as well as native USB 3.0 and USB 3.1 functionality. AM4 will support dual-channel DDR4 RAM with up to 4 x DDR4 DIMMs at 3200MHz. The new socket will support up to 24 PCIe lanes, depending on the chipset - bringing AMD to an equal playing field against Intel's mainstream LGA 1151 socket.
AM4, just like previous sockets from AMD, is a PGA socket with zero insertion force. The same locking mechanism is in play, with the same 40x40mm space used on AM4, just like AM3+ before it. AMD has changed the pin count up on the AM4 however, with 1331 pins over the 942 of AM3+ and even more than Intel's current LGA 1151 socket - a first for AMD.
AMD will be launching its next-gen Zen architecture in February 2017, alongside a new high-end X370 series chipset. The news is coming from Benchlife, teasing the Bristol Ridge series of APUs based on the Excavator CPU architecture, and GCN 3.0-based graphics architecture.
The upcoming AM4 socket will be split into three different chipsets, with the high-end X370 leading the way, and behind it we have the B350 and A320 chipsets. The Zen FX-based CPUs will be part of the upcoming Summit Ridge family of processors, including PCIe 3.0 support, dual channel DDR4 memory controllers, lots of L3 cache, updated storage features including USB 3.1 and NVMe, and more. We can also expect to see CPUs arrive with between 4/8 cores with 8/16 threads, and consuming between 65-95W of power.
Motherboard makers are already reportedly building their inventories of X370-based motherboards, with companies preparing to unveil some of their AM4 motherboards in October. AMD will most likely show off their Zen CPUs and new X370 chipset during CES 2017 in January.
Intel had previously pegged its upcoming shift to the 7nm node for 2020, but according to the latest rumors it has been delayed through to 2020.
The company has been looking for a processor designer to work in their new Microarchitecture Research Lab that's based in Bangalore, India. The new processor designer would join a team of engineers to "spearhead the research and advanced development" of both processor cores and graphics processors that will be deployed in the "2020 and beyond timeframe" using Intel's "futuristic" 7nm manufacturing tech.
Intel has recently updated the job advertisement, changing the date of its 7nm node from 2020 to 2022, with the job noting: "The India Lab specifically, in collaboration with MRL-US and Intel product architecture teams worldwide, will spearhead the research and advanced development of Microprocessor Cores in the 2022 and beyond timeframe. By conceiving of and prototyping radical approaches, the Lab will aim to deliver much greater CPU power and area efficiency while still delivering industry-leading performance. The microarchitecture and design of these advanced CPUs will be aggressively co-optimized with Intel's sub-10nm technology nodes deep into the next decade".
AMD's upcoming Zen architecture continues to build hype, with the new Zen-based Naples processor having benchmarks leaked onto the web over the weekend.
The Geekbench database had a new addition with the codename for the platform teased as 2S1451A4VIHE4_29/14_N, which looks to be a dual-socket platform that can take two of AMD's next-gen Naples CPUs. Each CPU has 32 cores and 64 threads, meaning we will see 128-threaded systems in 2017 from AMD, running at 1.44GHz base, and 2.9GHz boost (at least for now, these clocks could improve with time).
Each CPU complex has 4 cores with 8MB of L3 cache, which means the 32-core variant will feature up to 64MB of L3 cache, which means Intel might have something to worry about in 2017. The motherboard used was codenamed "AMD Corporation Diesel", which was powered by a huge 128GB of DDR4 memory, leaving the Geekbench results at 1141 points for single-threaded operation, and 15,620 for multi-threaded performance.
AMD will have their 32-core/64-thread Naples CPU out in Q2 2017, ready for the workstation market with dual-socket goodness.
AMD has been making strides in its CPU division, but the next big leap looks like it'll come from something called 'Starship'. What is Starship? According to our friends at Fudzilla, Starship rocks 48 physical CPU cores, and 96 threads in total - yeah, it's a beast.
Starship is a concept project right now, with AMD looking to land the CPU sometime in 2018 or beyond. AMD will tap the 7nm process, skipping the 10nm node for Starship, using GlobalFoundries as their semiconductor manufacturer.
Fudzilla reports - and yes we know, this report is from June - that AMD's next-gen Starship will arrive in TDPs of 35W and 180W, so we're sure the 180W part will be the 96-threaded beast.
Why it matters: AMD is getting into the headlines for all the right reasons, and a 96-threaded server CPU is something I want to see. Now I want to get my hands-on some server-grade hardware to prepare to salivate all over Starship and its 96-threaded CPU goodness.
Back in January Microsoft wrote up a controversial edict that tried to shorten Skylake's lifecycle on legacy Windows OS and lock newer CPU hardware--including Kaby Lake and Zen--exclusively to Windows 10. After tons of pressure from the PC crowd, Redmond relented on Skylake...but what about Kaby Lake and Zen? Both processors are on the horizon, and PC enthusiasts need to know if their OS will be compatible.
Microsoft has now confirmed to PC World that its decision concerning Kaby Lake and Zen is final: Windows 10 only. "As new silicon generations are introduced, they will require the latest Windows platform at that time for support," a Microsoft spokeswoman told PC World, reiterating the initial proclamation. "This enables us to focus on deep integration between Windows and the silicon, while maintaining maximum reliability and compatibility with previous generations of platform and silicon."
Earlier this year Tesla showed their intent to beef up their chip expertise with the hiring of former AMD and Apple CPU designer Jim Keller. That move indicated that Tesla might be considering building their own chips for their cars which previously seemed far-fetched since they had stuck with NVIDIA and Mobileye for so long.
Then, following a high-profile accident which resulted in a death of a customer Tesla dropped Mobileye as a chip supplier which seemed sudden but may have made more sense since they were acquiring chip experts like Jim Keller.
Now, Tesla appears to be expanding their hiring to GPU engineers with the hiring of Anand Mandapati who was most recently employed by Apple according to his LinkedIn.
Just as we're recovering from the massive $1650 price of the Core i7-6950X, we're now seeing leaked benchmarks of Intel's upcoming Kaby Lake-based Core i7-7700K processor.
Some CPU benchmark results have been spotted on SiSoft SANDRA 2015, with the Intel Core i7-7700K processor rocking a default clock speed of 4.2GHz, and a Turbo Boost frequency of 4.5GHz. The Core i7-7700K looks like it might feature 8MB of L3 cache, and of course - HyperThreading technology.
We can expect a refreshed Z270 chipset that will co-exist with the current Z170 platform, meaning that LGA1151-based boards might support the new Kaby Lake processors with a simple BIOS update.
As for the performance of the Core i7-7700K, we're looking at just under 10% more performance over the older Skylake-based Core i7-6700K. The Core i7-7700K scores 151.94 FOPS in SiSoft SANDRA's "Processor Arithmetic" performance, while the older Core i7-6700K scores 140.88 GOPS. Comparing its "Processor Multimedia" score, the Kaby Lake-based Core i7-7700K is capable of 379.8 Mpix/s, while the Core i7-6700K scores 353.8 Mpix/s.
Is the upgrade to the Core i7-7700K worth it for existing Core i7-6700K owners? Probably not, unless there are some changes that we don't know about in the upcoming Z270 chipset.
Researchers at Princeton University have created an open-source 25-core processor that can be scaled in an array to create "a single system containing millions of cores."
Princeton's new computer chip is called Piton, and it's a many-core open-source research processor aimed at revolutionizing data-center and enterprise-grade cloud-based solutions that power the internet--from email to Facebook and Twitter--with a cheap yet extremely efficient new scalable CPU architecture. The Princeton Piton Processor is a 25-core customized CPU designed by Professor David Wentzlaff and his team, featuring 25 modified OpenSPARC T1 cores that operate at a 1GHz per-core clock frequency. Piton is also outfitted with 460 million transistors, and was taped-out in IBM's 32m SOI process. Piton's main advantage is that it's scalable, meaning an array of some thousand Piton chips could be combined in an array to "enable thousands of cores on a single chip."
"The Princeton researchers designed their chip specifically for massive computing systems. Piton could substantially increase processing speed while slashing energy usage. The chip architecture is scalable - designs can be built that go from a dozen to several thousand cores, which are the independent processors that carry out the instructions in a computer program. Also, the architecture enables thousands of chips to be connected into a single system containing millions of cores," reads a Princeton report.
You've got to hand it to AMD, they have been having a great past 12 months or so, riding the success of their new Polaris architecture, and the tease of Zen has investors pleased. So much so, that AMD stock has increased by a massive 330% in the last year, rising another 8% today on the back of the positive news of Zen from IDF 2016.
AMD has had a few years of financial grief, but it looks like the company is getting past this, returning to profitability in Q2 and should hopefully continue that with the sales of the Radeon RX 480, RX 470 and RX 460 throughout the rest of 2016, and into 2017. But the last week has been all about Zen, AMD's next-gen CPU architecture, which has been impressing us for quite a while now. AMD's new Zen architecture looks to have the company on an even playing field with Intel once again, who has had a monopoly on the market for what feels like forever.
Zen won't see the light of day for consumers and gamers until early 2017, but it will be delivering a 40% improvement of IPC (Instructions Per Clock), which is beyond impressive for something that has probably only reached testing phases now. AMD will continue to improve its Zen craft, with next-gen motherboards also featuring new connectivity like M.2 SSD support, USB Type-C and support for the likes of DDR4 RAM, and more.