Processors - Page 84

As of right now, this is very much a rumor, so I want to put that out front. However, this lends far more credence to the claim as it has come from several sources over time and seems to be more of a beginning to a realization. Sometimes it is simply not possible, and I have to assume as of the time of writing that is the case.

Word has been coming around over the past month or so, and it's not great for anyone looking at a new platform on the Threadripper front. Especially if you are looking at a new or better motherboard for your existing Threadripper or looking to put a newer chip in your current X399 motherboard.

The new TRX40, which is one of the claimed new chipsets/board monikers for the latest 3rd generation of Ryzen Threadripper from AMD. While AMD always pushes to make a socket or platform last a long time, we have seen issues with X570 where 1st gen Ryzen support was cut in many cases. However, in the case of AM4, older boards could still support the newer chips, at least mostly with a few exceptions across the line.

Intel's next-gen Cascade Lake-X family of CPUs was unveiled lat week, with the new competitors to AMD's Ryzen Threadripper family of CPUs set to be a big one for Intel.

So big that the flagship Core i9-10980XE processor will reportedly rock a huge 5.1GHz all-core boost when cooled with liquid cooling. Intel's new Core i9-10980XE processor has a base CPU clock of 3GHz, Turbo Boost 2.0 clock of 4.6GHz boost, and an all-core boost of 3.8GHz boost under regular cooling.

Intel's EMEA Technical PR Manager, Mark Walton, told PCGamesN: "You can overclock the heck out of these and get some really interesting results. For example, we've had the 10980XE, the eighteen-core processor, up as high as 5.1GHz in the lab using standard liquid cooling. And that, I believe, is all cores".

Last week we revealed a leak about Intel's upcoming 10th generation HEDT processors and their improvements and additions, but there was more to the story than that. Intel also briefed us on upcoming revamped Xeon-W 2200 series CPUs as well as improved pricing on some of their mainstream 9000 series processors.

The new features of the new CPUs include up to 4.8GHz Turbo speeds, up to 1TB of DRAM, which is double the amount of the previous generation, higher DRAM speeds up to 2933MHz, new deep learning ISA, support for Intel's 2.5G LAN controller, and support for Intel's new WIFI 6 controller. The new CPUs will work with the same motherboards the previous generation used.

In regards to performance increases, we see huge improvements especially when it comes to deep learning. Intel's new DL Boost ISA and improved frequencies basically double performance when it comes to AI inference. When it comes to other workloads we see large boosts in editing, rendering, and compiling.

AMD has teased the world with some more news on its next-gen EPYC processors, with the new chips to be based on its next-gen Zen 3 and Zen 4 cores. The news was unleashed at the HPC-AI Advisory Council UK conference.

The new road map shows the current EPYC Rome CPUs going into production in the middle of 2019, which they have -- and are now being installed into datacenters and servers worldwide. The next-gen EPYC 'Milan' CPUs taped out in Q2 2019, and will go into production this time next year (Q3 2020).

EPYC Milan will use Zen 3 cores and be on the 7nm+ node with up to 64 cores, and TDPs of between 120-225W depending on the processor. But it's the next-gen Genoa that has us excited, with the new EPYC 'Genoa' chips arriving sometime in 2021 and are in the "definition phase".

At about 2:45PM yesterday, Intel's price slashing was leaked. We were actually at the briefing when it happened, so we will clarify some of the new details with Intel's official slides. For starters, prices are being slashed by up to 50%, which is huge news. However, if you have had experience with Intel's top of the line CPUs over the past decade, their Extreme Edition flagships have always been priced around $1000. It's refreshing to see Intel react to market demand and bring the price down to more normal levels.

The 10th Generation 14-nm i9-10980XE will be their flagship CPU with 18 cores, and all the new CPUs are Cascade Lake-X series. All the CPUs get considerable boosts to their maximum Turbo frequencies, with the 18, 14, and 12 core variants going up to 4.8GHz and the 10-core CPU going to 4.7GHz. When we inquired to the turbo tables Intel told us they would have to get back with us.

We did notice that Intel has dropped their 16-core variant, and we believe this was done due to low demand. Total platform PCI-E lanes have been increased to 72, which is due to the addition of four PCI-E lanes, so the CPUs all have 48 PCI-E lanes instead of 44.

AMD is expected to launch its next-gen third-generation Ryzen Threadripper in November, powered by the new Zen 2 architecture and offering more cores and threads than ever before. But what about X399 compatibility? According to the new rumors: that's a big NOPE.

1usmus is the creator of the DRAM Calculator for AMD Ryzen CPUs (his real name is Yuri Bubily) and his post on Overclock.net he said that AMD has removed X399 compatibility from its upcoming third-gen Ryzen Threadripper CPUs. There are lots of changes with the new Threadripper CPUs, which seem to have forced AMD into new chipsets for the new TR processors.

The new chipsets in question would be the purported TRX40 and TRX80 (really, AMD? Mix that around and it spells RTX... your competitors graphics cards) that would support the next-gen Ryzen Threadripper CPUs.

Intel will soon be launching its new Core i9-9900KS offering 8C/16T with all-core 5GHz boost, but we don't have a firm price -- the only indicator to price we have is a new listing that sees it dropping in at around $600.

According to a new listing with Australian e-tailer Mwave, which had it listed for AUD$899 which converts to around $605 (remember Australia has a 10% GST so this price will vary).

Don't go thinking this is some next-gen CPU at all, as it is a binned 9900K that offers a higher base CPU clock (4GHz versus 3.6GHz) and all-core Turbo at 5GHz (versus 4.7GHz). The single-core boost is identical to the 9900K, with the new 9900KS rolling out with 5GHz single-core boost.

Prepare the salt shakers because this rumor is a doozy -- AMD's next-gen Zen 3 architecture reportedly features massive changes to its SMT capabilities that would enable 4 threads per core.

This means Zen 3 processors with 64 cores could have up to 256 threads, something that would be quite the leap from AMD compared to the current EPYC Rome CPUs with 64C/128T max. The 4 threads per core processors would be perfect for the data center and HPC markets, and would pave the way for gigantic performance improvements from all those threads.

4 cores per thread isn't exactly new, with IBM supporting it with its Power CPU architecture supporting a huge 8 execution threads per core. AMD would, however, be the first with 4 cores per thread on an x86 micro architecture.

We've been wondering if it would even be unveiled, but yes -- AMD is indeed working on the third-gen Ryzen Threadripper range of CPUs, something that will be unveiled in November. The news comes after the delay of its R zen 9 3950X processor.

AMD has been shifting the goal posts on the release of the world's first 16-core gaming CPU, with the upcoming Ryzen 9 3950X expected this month but has been delayed into November.

The company said in its announcement on the delay: "We are focusing on meeting the strong demand for our 3rd generation AMD Ryzen processors in the market and now plan to launch both the AMD Ryzen 9 3950X and initial members of the 3rd Gen AMD Ryzen Threadripper processor family in volume this November".

AMD has just announced the latest and fastest-ever EPYC processor yet, the new EPYC 7H12 -- which beats out the previous flagship EPYC 7742 processor. The new continues to fill out the game-changing EPYC Rome CPU family.

The new EPYC 7H12 has been designed specifically for liquid-cooled servers and still maintains its industry-leading 64C/128T of CPU grunt, with 256MB of L3 cache, and 128 PCIe 4.0 lanes.

The big difference here with the new EPYC 7H12 is that it has a base CPU clock of 2.6GHz (the highest on any EPYC processor) and a boost CPU clock of 3.3GHz -- 100MHz short of the EPYC 7742. Power wise it is a much bigger beast, as it'll chew down a 280W TDP compared to the 225W TDP on the EPYC 7742.

AMD broke the world record for Cinebench R20 with its EPYC 7742 processor, and now the 64C/128T beast has another win in powering the world's first real-time 8K HEVC encoder.

Beamr Imaging says it reached a world first with its new software powered by the AMD EPYC 7742 processor, in which it can handle 8K live encoding. Beamr's latest software when powered by the EPYC 7742 can handle real-time 8K video in 10-bit with HDR at a huge 79FPS, which is damn impressive.

The company says the new software will be ready for live linear broadcast level streaming, premium VOD entertainment, and cloud gaming content streaming. Beamr has some of the biggest companies in the world as its customers and partners, with the likes of Netflix, Intel, IBM, Microsoft, Sony, Xilinx and Yahoo.

AMD is having a big show this year at IBC 2019, with their new EPYC 'Rome' CPU demolishing some world records in V-Ray and Cinebench.

The big surprise here would definitely have to be Cinebench R20, with dual AMD EPYC 7742 processors hitting a new world record with 31,913. A single EPYC 7742 has 64C/128T so the dual EPYC 7742 processor pack a huge 128C/256T.

Compared to the Xeon Platinum 8168 processor and its 48C/96T with its 16,536 -- the dual EPYC 7742 rig rips it to shreds with 31,913. Hell, the single AMD EPYC 7742 beats it with 20,645. If you look down the list, right down the list, an Intel Core i7-7700K scores 2420 to give you some perspective.

Intel is expected to launch its next-gen Tiger Lake CPU architecture sometime in late 2020 or 2021, which will feature Gen12 graphics and a total overhaul of the Execution Unit.

The revamped Execution Unit is expected to provide the largest GPU performance increase in over 10 years, something that was revealed in a new release of Linux patches from Intel. The post talks about a new Display State Buffer (DBS) engine that will take care of batch submit display register programming.

The documentation states that DBS reduces "loading time and CPU activity, thereby making the context switch faster. DSB Support added from Gen12 Intel graphics based platform".

AMD is preparing to launch its new Ryzen 9 3950X -- its flagship 16C/32T processor and the world's first 16-core gaming CPU that will boost up to 4.7GHz and offer 72MB of cache.

Now we have German retailer Digitec listing the chip with a release date of September 30, and what appears to be a placeholder price of 999 euros which works out to $1102 or so -- much more than the MSRP of $749.

AMD will have quite the gaming and multi-tasking chip on its hands with the Ryzen 9 3950X, offering a huge 16C/32T of CPU performance at 3.5GHz base clock and up to 4.7GHz boost -- the fastest for a Ryzen 3000 series CPU.

We all knew that AMD had been doing some damage to Intel over the last year or two with its excellent new Ryzen CPUs, but now Intel itself has confirmed the news -- it has lost CPU market share to AMD, and it is going to fight harder from now on.

Jason Grebe, Corporate Vice President, General manager Cloud Platforms and Technology Group, speaking to an analyst at the Citi Global Tech Conference said that Intel lost CPU market share to AMD and that the company will have to fight more aggressively to continue its dominance in the market.

While his comments don't acknowledge AMD by name, he says Intel lost "some channel desktop share" to who we can only presume is AMD.

Intel teased its all-core 5GHz beast a couple of months ago, but now we know that the new Core i9-9900KS will launch next month. The news was announced by ex-media and PCPer founder Ryan Shrout, who is now the new chief performance strategist of Intel.

Firstly, the new Core i9-9900KS was first unveiled at Computex 2019 and will offer up 8 cores and 16 threads at 4GHz, but will boost with all cores at 5GHz. Shrout used the new 9900KS to push the boot into AMD's neck with its current Ryzen 3000 series processors, saying: "The point is, we're not taking this sitting down, we see the competition, we see the landscape as it is".

He continued: "We're adjusting because we take these customers very seriously. And we want to give them the best product possible".

AMD had the entire world in its hands leading into the Zen 2 architecture, resulting in the great Ryzen 3000 series CPUs -- but now there's some serious drama with bold claims of false advertising being thrown around. Check out Der8auer's video below for a great recap:

Once the new Ryzen 3000 series CPUs began getting into consumer' hands, it was quickly noticed that the boost clocks AMD had advertised weren't being met -- and that boost speeds were locked to a single CPU core. AMD didn't confirm this before the launch, but later confirmed it -- going as far as updating the Ryzen product pages to clarify the slip up.

But then Der8auer used a survey in which 2700 people responded to, building data to show that just 5.6% of the reported Ryzen 9 3900X processors in his survey hit the advertised maximum boost clocks. The mid-range Ryzen 5 3600 was much better with 49.8% of users saying they hit max boost clocks... but this is a big issue -- a big, big issue.

AMD's kick ass new EPYC Rome CPUs are here and boy are they throwing Intel around like a rag doll, with ServeTheHome setting a new world record on a dual-CPU rig using two AMD EPYC 7742 processors.

STH compared the dual AMD EPYC 7742 chips against four Intel Xeon Platinum 8180M processors, which battles out the 64C/128T processor from AMD against Intel's best 28C/56T offering with some very, very surprising results. The AMD system is not only much cheaper, but it is also faster.

This means that STH ran benchmarks with 128C/256T of power from AMD and its EPYC 7742 processors, while the quad-CPU rig from Intel packed just 112C/224T in comparison, but cost over 3x as much.

Cerebras Systems just unveiled the largest chip ever at Hot Chips 31, which is a A Symposium on High Performance Chips hosted yearly with the biggest companies in the tech space involved.

The new Cerebras Wafer Scale Engine chip from Cerebras has a huge 1.2 trillion (1,200,000,000,000) transistors compared to the largest GPU in NVIDIA's GV100 and its 21.1 billion (21,100,000,000). NVIDIA requires just 815 square millimeters of die space for its GV100 and 21.1 billion transistors, while Cerebras Systems requires a huge 46,225 square millimeters of silicon.

Cerebras' new chip is so big that in comparison photos, the company compares its new 1.2T transistor chip side-by-side with a freaking keyboard. Yes, this new chip is as big as a desktop keyboard.

AMD released its game-changing EPYC Rome CPUs and in the hours afterwards, AMD shares spiked by a huge 16% in between Google and Twitter announcing they're using the new EPYC Rome chips in their servers.

AMD shares closed at $33.92 on Thursday, up by a huge 84% this year alone after riding the success of Zen 2 powering the new Ryzen 3000 series CPUs, and the new Navi-powered Radeon RX 5700 series graphics cards.

Patrick Moorhead, Founder, President and Principal Analyst of Moors Insights & Strategy had some comments on AMD's new EPYC Rome CPUs, where he said: "AMD took a big step forward today in the datacenter with its launch of the 2nd Gen EPYC processor and platform. It is a bigger leap forward than I had expected".