Processors - Page 72

Intel has announced that it will be updating us on their process and packaging roadmaps, with Intel CEO Pat Gelsinger set to provide more details on July 26 during the just-announced Intel Accelerated webcast.

The upcoming Intel Accelerated webcast will have Intel CEO Pat Gelsinger joined by Senior Vice President and General Manager of Technology Development, Dr. Ann Kelleher, where we'll find out about Intel's process and packaging roadmaps. Intel will have its next-gen 12th Gen Core "Alder Lake" CPUs on its in-house 10nm Enhanced SuperFin technology, but we'll know more details on this on July 26.

Intel's first 7nm CPU will be its next-gen Meteor Lake, which is the 14th Gen Core CPU -- we're at the 11th Gen Core CPU right now (with the flagship Core i9-11900K) while the Alder Lake architecture drops with the 12th Gen Core and then 13th Gen Core "Raptor Lake" will drop before the 14th Gen Core "Meteor Lake" CPUs on 7nm.

AMD launched the flagship Ryzen Threadripper PRO 3995WX processor earlier this year, with its 64 cores and 128 threads of CPU power costing $5489 -- but there's the new Ryzen Threadripper 5000 series "Monster Truck" CPUs now launching in August 2021 -- just weeks away.

The latest rumor teases AMD will launch the Ryzen Threadripper 5000 series codenamed "Chagall" CPUs in August, based on the new Zen 3 (possibly Zen 3+) architecture and some tweaks to its xGMI (Inter-Chip Global Memory Interconnect) speeds from 16GT/s to 18GT/s.

Other than that, we're still looking at the same CPU core count, the same N7 node over at TSMC, the same DDR4 memory support and TDP (280W max).

I don't know why you'd try to bother doing this thinking no one would see the 308 processors physically strapped to your body, but a Chinese citizen has been caught trying to smuggle Intel CPUs out of the country.

In total, the man had 308 processor strapped to him -- Intel's own 10th Gen Core i7-10700 and Intel Core i9-10900K processors. Chinese customs officials caught the person smuggling Intel CPUs over the Hong Kong-Macau border, with a suspicious-looking driver busted.

Chinese custom officials have had one driver attempt to smuggle 256 processors strapped to their legs and chest, while two weeks ago a separate CPU smuggling attempt had 52 hidden processors between seats -- this person was arrested. The two-week operation has at least 308 smuggled CPUs recovered, worth over $123,000+ but it could be much bigger than this -- we just don't know yet.

AMD is now powering more of the world's fastest supercomputers than Intel processors, with EPYC overtaking Xeon in a huge way.

During the ISC High Performance 2021 event, it was revealed that 3 out of the top 10 fastest supercomputers in the world are powered by AMD EPYC processors. AMD EPYC chips are inside of the Perlmutter (NERSC Lawrence Berkeley National Laboratory), Selene (NVIDIA), and JUWELS Booster Module (Forschungzentrum Julich FZH).

AMD EPYC processors are used in custom designs in all 3 of these supercomputers, while Intel Xeon processors are in just 2 of the top 10 fastest supercomputers.

We thought AMD's interesting 4700S desktop kit was based on the Xbox Series X APU, but it's not -- it seems it's based on the PlayStation 5 SoC instead.

Thanks to new photos by Bodnara, the new AMD 4700S desktop it is based on the Sony PlayStation 5 SoC which packs a CPU based on the AMD Zen 2 architecture with 8 cores and 16 threads of CPU power. There's also Hynix GDDR6 memory that is physically soldered onto the back of the Mini-ITX motherboard.

The AMD 4700S APU itself is on the AMD A77E Fusion Controller Hub (FCH) which is also used on the AMD A9-9820 system which is based on the Xbox One APU codenamed "Durango". VideoCardz explains that this is the reason everyone thought the AMD 4700S was based on the Xbox Series X APU.

We've been hearing more and more about Intel's new Sapphire Rapids platform, which was destined for beyond the consumer markets -- but not anymore, according to the latest rumors.

VideoCardz is reporting that Intel's new Sapphire Rapids HEDT platform will be succeeding Cascade Lake on the Glacial Falls platform, with the new Sapphire Rapids HEDT platform dropping somewhere in late 2022. But the more interesting fact here is that Intel is dumping the "X" chipset codenames with the new "W" series chipset.

Intel could launch its next-gen Sapphire Rapids HEDT platform for consumers alongside the release of the Raptor Lake CPU family, which will be dropping in the second half of 2022. The new W790 chipset could be a monster, but we're going to have to wait and see how that goes in at least 12 months from now.

It looks like AMD could be working on an absolute stomping monster of a CPU with rumors that its next-gen codename Bergamo CPU rocking a huge 128 cores and 256 threads of Zen 4 processing power.

We did hear whispers of a 128-core, 256-thread EPYC CPU not too long ago now, up from the 96 cores and 192 threads expected on AMD's next-gen EPYC "Genoa" CPU. In that article, I explained: "new Zen 4-powered EPYC or Threadripper CPU with 128C/256T will also support up to 12-channel DDR5-5200 memory".

"The current mockups of the AMD Genoa CPU have it with 12 CCDs, so a tweaked chip (and real, not the mockup) with room for 16 CCDs would unleash the 128 cores and 256 threads, up from the 96 cores and 192 threads with the 12 CCDs mockup of Genoa".

What was once a rumor has now been confirmed: Intel will be adopting HBM (High Bandwidth Memory) and will be deploying it with its next-gen Sapphire Rapids CPU.

Intel's upcoming Sapphire Rapids CPUs will be deployed in two variants: one without HBM, and other variants with HBM. The new HBM-powered Sapphire Rapids CPUs are codenamed SPR-HBM, and will not be launching at the same time as the Sapphire Rapids CPUs without HBM.

Intel confirmed the news to AnandTech, where the company said: "Intel has stated to AnandTech that they are committed to making HBM-enabled Sapphire Rapids available to all enterprise customers and server vendors as well. These versions will come out after the main Sapphire Rapids launch, and entertain some interesting configurations. We understand that this means SPR-HBM will be available in a socketed configuration".

It seems like Intel has just kicked off the Xe-HPG graphics card marketing campaign, also confirming that it has begun shipping its next-gen DG2 family of GPUs.

The company has started inviting Odyssey Cardholders to register, where they might get their hands-on some Intel Xe-HPG related swag. Intel kicking off its journey and asking for gamers to join the company on its Odyssey, with the company saying they're "heading toward a milestone moment, the pending release of the Xe HPG microarchitecture from Intel".

Intel explains: "Thank you for joining Intel on this Odyssey. We are soon heading toward a milestone moment, the pending release of the Xe HPG microarchitecture from Intel. Some of you received an Odyssey card and were the first to engage with Intel on this journey".

Can you feel that? We're getting closer to Intel catching up to where AMD was a few years ago, with Intel's first-ever 7nm CPU not too far away now with Meteor Lake being teased again.

The latest on Intel's new Meteor Lake CPU is from the LGA18xx socket it will use, with @momomo_us sharing a new LGA17xx/18xx CPU socket. Intel's new 12th Gen Core "Alder Lake" and 13th Gen Core "Raptor Lake" CPUs will be using the new LGA1700 socket, while Intel's next-gen 14th Gen Core "Meteor Lake" will move to the LGA18xx socket.

We heard about the new LGA1800 socket from Igor's Lab not too long ago, with Intel's new CPU socket introducing a new mounting system. We could see Intel not forcing consumers into that rabbit hole, and instead the new LGA1800 socket is for another series of CPU... we just don't know any of it for sure just yet.

If you want to stress test and benchmark your new CPU inside of 3DMark, you can now finally do just that with UL Benchmarks' latest update to 3DMark.

The new 3DMark CPU Profile will test your CPU performance, either in single or multi-threaded benchmarks where it will provide a score at the end and show you how your CPU performance will scale with the amount of cores and threads being used.

There's a full benchmark run included, but if you want to test individual CPU cores and threads then the other benchmark modes are there for you: the 16 threads test, 8 threads test, 4 threads and 2 threads test, and finally the single threads test. Each of these will test CPU performance to measure how good your CPU is for older games, normal performance, new games, and the computational intensive tasks (content creation, 3D editing, high-end gaming, etc).

AMD will have its RDNA 2 graphics architecture inside of Samsung's next-gen Exynos SoC, which is said to be revealed next month.

The new Samsung Exynos processor with RDNA 2 inside was meant to be unveiled this month but has been delayed according to leaker Ice Universe on Twitter. Samsung's next-gen Exynos processor will be offering ray tracing support as well as variable rate shading (VRS).

Samsung isn't so great on the CPU side of things, which is why the South Korean giant has reportedly been hiring architecture engineers from both Apple and AMD -- but with AMD providing its RDNA 2 GPU architecture to Samsung for the GPU side of its next-gen Exynos, this is going to be interesting.

The big silicon fight is only just beginning, with the likes of Apple, Samsung, Intel, Qualcomm, NVIDIA, AMD, and many others fighting for supremacy -- but Samsung is reportedly making some big, and very interesting (but expected) news.

The latest is that ARM's next-gen Cortex X series CPU cores are reportedly not meeting Samsung's high expectations, with Samsung and Qualcomm "turning their plans to a custom architecture development". As it stands, Samsung and Qualcomm license the CPU cores designed by ARM -- but both companies would be silly to not design their own CPU architectures.

Samsung did precious have an in-house team designing custom CPU cores, this was codenamed Mongoose which I wrote about back in August 2015. Samsung shut this division down not long after that, licensing CPU cores designed by ARM again for all of its phones including the flagship Galaxy S21 Ultra.

Intel won't be launching its next-gen Alder Lake-S platform and new flagship Z690 chipset until later this year in Q4 2021, and then after that, we have Raptor Lake-S with the even more next-gen Z790 chipset.

Both of Intel's new Z690 and Z790 chipsets will have the same LGA 1700 socket, with the new Z790 chipset launching alongside the Intel 13th Gen Raptor Lake-S CPUs dropping in Q3 2022. The new rumors are coming from PJ, an editor and reviewer with Uniko's Hardware.

The new Z790 chipset will support both the 12th Gen Core and 13th Gen Core CPUs, so if you buy a CPU later this year you'll be able to put it into a next-gen motherboard next year.

Up until now, we've been expecting AMD's next-gen Zen 4 architecture to have CPU cores topping out at the 96-core mark, but according to new rumors that's not the top -- 128 cores is the peak for Zen 4.

AMD's next-gen Zen 4 architecture will reportedly offer up to 128 cores and 256 threads, with the next-gen AMD EPYC "Genoa" CPUs offering up to 96 cores and 192 threads on the Zen 4 architecture. Genoa will still have 96 cores, but whatever this new CPU is, it won't be Genoa with its 128 cores and 256 threads.

A new Zen 4-powered EPYC or Threadripper CPU with 128C/256T will also support up to 12-channel DDR5-5200 memory. The current mockups of the AMD Genoa CPU have it with 12 CCDs, so a tweaked chip (and real, not the mockup) with room for 16 CCDs would unleash the 128 cores and 256 threads, up from the 96 cores and 192 threads with the 12 CCDs mockup of Genoa.

Intel's next-gen Alder Lake-S processors will reportedly be launching on October 25, 2021 according to the latest rumors.

The new "S1" processors will be launching first, with the Intel 12th Gen Core "Alder Lake-S" processors launching in full desktop form first -- with the new LGA 1700 socket, and 125W K-series CPUs. Most of these high-end Alder Lake-S processors will consume most of the volume "if not all" says Tom from Moore's Law is Dead, until 2022.

The flagship Core i9 processor will have 8 Golden Cove (GLC) and 8 Gracemont (GRT) cores for a total of 16 threads. The 8 Golden Cove cores are the high-performance CPU cores, while the 8 Gracemont cores are the high-efficiency cores.

Intel is working on its 12th Gen Core "Alder Lake" CPU family for later this year, ushering in next-gneeration technologies like DDR5 and PCIe 5.0 standards... right at the same time as Microsoft is hyping its next-gen Windows 11 operating system.

In the middle of these is Tom from Moore's Law is Dead, who has tweeted that Microsoft and Intel will be launching their new OS and CPUs "around a spooky time of the year" hinting at Halloween... which is October 31. The two companies are working closely together as they always do, with optimizations for Alder Lake built into the next-gen Windows 11.

Windows 11 will have huge scheduling upgrades, as well as many more tricks under its revised hood, that will be perfect for Intel's new Alder Lake CPUs. Intel's 12th Gen CPUs see the company moving into a new hybrid design, which will have two different architectures on the same package.

Intel's next-gen Alder Lake-S processors will arrive on the new LGA1700 socket, which is codenamed Socket V, with the new CPU package and the socket being thinner.

The new information is coming from Igor Wallossek who posted the schematics and drawings of Intel's new cooling for Socket V (LGA1700). Intel looks to be changing the z-height by at least 1mm for Socket V versus Socket H (LGA1200). The lowered z-height on Socket V will see lower loads compared to Socket H, but the new LGA1700 socket will have a new hole pattern.

There's also a tease of preliminary schematics of high-performance thermoelectric cooling, where we should expect similar for LGA1700. Intel worked with Cooler Master on the launch of its MasterLiquid ML360 Sub-Zero cooling, which uses a Peltier effect to create a heat flux between two types of materials.

Intel has just confirmed that its next-next-gen 14th Gen Core "Meteor Lake" compute tile has taped in, meaning that the design is finalized and can be completed to tape out the whole chip.

Intel confirmed the news with Executive Vice President & GM of Intel's Client Computing Group, Gregory M Bryant, tweeting out that it was a "Great way to start the week! We are taping in our 7nm Meteor Lake compute tile right now. A well-deserved celebration by the team on this milestone".

The new Meteor Lake architecture should be powering Intel's next-next-gen 14th Gen Core CPUs which are expected in 2023, they will be following up the next-gen 12th Gen "Alder Lake" CPUs coming in 2022 with DDR5 and PCIe 5.0 support and the 13th Gen Core "Raptor Lake" CPUs after that.

AMD has been ahead of the curve for a while now, with its chiplet-based Zen CPU architecture making waves through the industry... but the industry moves so fast it's always time for change. Well, that change comes in the form of "X3D" packaging, and it's coming to codename Milan-X.

Patrick Schur tweeted out today that AMD is "working on a new CPU" that is codenamed Milan-X that will use stacked dies. ExecutiveFix backed up the leak, saying that codename Milan-X will use X3D stacking technology and will feature the Genesis-IO die, which is the codename of the current-gen Zen 3-based EPYC architecture.

This won't be inside of your new Ryzen CPU, but rather it will be what AMD will use to fight its competitors in the data center market. AMD recently teased a simplified diagram during its Financial Analyst Day that packed four compute tiles in a 2 x 2 pattern, with stacked memory surrounding the chip. This hybrid approach is called X3D, as it combines Hybrid 2.5 and 3D stacking technologies.