Processors - Page 71
CPU and SoC news from Intel, AMD, Apple Silicon, ARM, and Qualcomm - launches, benchmarks, and architecture updates from TweakTown. - Page 71
Stay Updated
Follow TweakTown for breaking tech news, reviews, and daily updates.
As an Amazon Associate, we earn from qualifying purchases. TweakTown may also earn commissions from other affiliate partners at no extra cost to you.
Intel's new ARC series graphics card teased with dual-fan cooler
Intel has now formally introduced its new ARC series graphics cards -- at least by naming scheme only, with a nice marketing trick teasing the ARC GPU with 1000 drones all synched together. Check it out:
The design of the new Intel ARC series graphics cards were teased by drones, and looks virtually identical to the leaked photos that Tom from Moore's Law is Dead shared months ago, back in April 2021. It appears he was right, again. The upcoming Intel ARC series GPU had 9 blades in the leak... and 9 blades shown by the drones as well.
Intel's upcoming ARC "Alchemist" will be based on the Intel DG2 GPU family, powered by the Xe-HPG microarchitecture. We should expect Intel's flagship ARC graphics card to feature 512 Execution Units, and 16GB of GDDR6 memory.
Continue reading: Intel's new ARC series graphics card teased with dual-fan cooler (full post)
Intel's new Core i9-12900K appears in BaseMark benchmark database
Intel's new Core i9-12900K processor is still in the oven and months away, but it has appeared in the CPU black market -- and now on the BaseMark benchmark database.
The new Alder Lake-S flagship processor has 12 cores and a CPU clock of 3.2GHz on the BaseMark database, but we've been hearing it'll be a 16-core, 24-thread CPU. This will see 8 high-performance cores (with HT) and 8 high-efficiency cores for 24 threads in total on the Core i9-12900K processor.
Intel's Core i9-12900K processor was reportedly run on an "Acer Z69H6-AM" which going by the Z69 part of the product name could mean a new Acer Z690 motherboard which would be capable of taking a new 12th Gen Core "Alder Lake" CPU.
Continue reading: Intel's new Core i9-12900K appears in BaseMark benchmark database (full post)
Intel rumored to secure most of TSMC's new 3nm production capacity
We're hearing new rumors that Intel has secured most of TSMC's upcoming 3nm production capacity, underscoring rumors from all the way back in January 2021.
Now we have Chinese publication UDN reporting that mass production of Intel's new CPUs on TSMC's new 3nm node will begin in Q2 2022, with production capacity expected to hit 4000 wafers in May 2022, and in the months after that, it will flow out 10,000 wafers per month.
The new report says that Intel will be using TSMC's new 3nm node for 3 x GPUs, and 1 x GPU -- we should expect the 3nm CPUs to be next-gen Xeon processors for next-gen servers and data centers -- not the upcoming Raptor Lake CPUs that will be launching in the second half of 2022.
Continue reading: Intel rumored to secure most of TSMC's new 3nm production capacity (full post)
AMD Ryzen 3 5300G APU overclocked: insane 5.5GHz under LN2 cooling
AMD's new Ryzen 3 5300G processor isn't the one you'd think of when it comes to gaming your days and nights away on, but for an overclocker... well, holy sh*t -- the APU has just been OC'd all the way up to an insane record-breaking 5.5GHz.
The Ryzen 3 5300G is a quad-core APU that you can buy for around $110-$130 in the US, with a Vega-based integrated GPU for gamers that don't have deeper pockets for a higher-end CPU and GPU combination. Well, during an overclocking marathon by Safedick, who used an ASUS ROG Crosshair VIII Extreme motherboard and 16 GB of G.Skill Ripjaws V DDR4-2066 (CL14-14-14-28-1T) memory.
With that combo, the AMD Ryzen 3 5300G processor was cranked up to 5.5GHz under LN2 cooling, so don't be thinking you'd be pushing that OC under a normal air cooler or AIO liquid cooler.
Continue reading: AMD Ryzen 3 5300G APU overclocked: insane 5.5GHz under LN2 cooling (full post)
Google's new Tensor chip in Pixel 6 will be made on Samsung's 5nm node
We are receiving drops and drops of information on Google's next-gen ultra-high-end Pixel 6 and Pixel 6 Pro smartphones, where they will be manufactured using Samsung's next-gen 5nm process node.
Google is using a custom in-house Tensor chip inside of its next-gen Pixel 6 and Pixel 6 Pro smartphones, with new rumors suggesting that Samsung will be making the chip using its new 5nm process technology. The news is coming from sources that go against Nikkei, that reported TSMC was going to be building the Tensor chips for Google.
The new Tensor core is going to built on Samsung's new 5nm process according to the latest reports, but of course -- Google and Samsung are refusing to comment.
Continue reading: Google's new Tensor chip in Pixel 6 will be made on Samsung's 5nm node (full post)
ASUS rumor teams with Noctua on wicked new GeForce RTX 30 series cards
Imagine in a few month's time you'll be able to maybe buy an ASUS graphics card with a wicked Noctua cooler on it, it looks like it'll be happening and I couldn't be happier.
According to a new filing with the Eurasian Economic Commission, it seems ASUS is teaming up with Noctua on some new cooling technology for GeForce RTX 30 series graphics cards. Noctua is well known for their beige/brown combo cooling solutions, which don't just look great, they're some of the best-performing coolers in the industry.
The reason for all this hype is a model that was slipped in called "RTX3070-8G-NOCTUA" which kinda gives it away, doesn't it -- a GeForce RTX 3070 graphics card with 8GB of GDDR6 memory and a Noctua branded cooler. I guess you'd have to be blind to not see it.
Continue reading: ASUS rumor teams with Noctua on wicked new GeForce RTX 30 series cards (full post)
Intel next-gen CPUs: Arrow Lake, Lunar Lake, Nova Lake in 2023 to 2025
You'll need a bucket of salt for this one, but Intel's next-gen series of CPUs has been leaked out on Reddit... and quickly deleted... but the internet being the internet means nothing is ever deleted.
Anyway, we're now hearing about Intel's next-gen Arrow Lake, Lunar Lake, and Nova Lake CPU architectures that will drop in 2023, 2024, and 2025, respectively. Arrow Lake and Lunar Lake will reportedly use Lion Cove high-performance "big" cores and Skymont high-efficiency "small" cores.
Arrow Lake CPUs should have 8 big cores and 32 small cores, for a total of 40 cores... doubling the core count over Raptor Lake, which will have double the core count of Alder Lake (8+8 with Alder Lake, 8+16 with Raptor Lake, and 8+32 with Arrow Lake).
Continue reading: Intel next-gen CPUs: Arrow Lake, Lunar Lake, Nova Lake in 2023 to 2025 (full post)
Hot Chips 33: AMD, Intel and TSMC will tease 3D packaging technology
It's that time of the year, with the Hot Chips 33 technology symposium kicking off on August 22 for a huge 3-day online event. We'll hear all about delicious new technology, with a highlight on discussions with advanced
The organizers of Hot Chips 33 have confirmed an exciting schedule, where we will receive some juicy details on "Advanced Packaging" technology. Intel and TSMC will be discussing their "packaging technologies for chiplets and 3D" while AMD will only be discussing their 3D packaging technology.
Intel will be talking about its next-gen Alder Lake and Sapphire Rapids CPU architectures on day two of Hot Chips 33, which I'm sure is going to be a different deep dive than traditional ones where we hear about specs, performance, and all that jazz. The packaging technology side of things is probably going to get a total nerd-out, which is going to be awesome.
Continue reading: Hot Chips 33: AMD, Intel and TSMC will tease 3D packaging technology (full post)
Here's our best look yet at AMD's next-gen AM5 socket for Zen 4 CPUs
The last we heard on AMD's next-gen AM5 socket was from a mockup of the new LGA1718 socket, but now we have a proper render that looks fantastic, thanks to ExecutibleFix on Twitter. Check it out:
The new render shows how AMD's new AM5 socket locking mechanism would work, which looks similar to Intel's own HEDT LGA sockets -- there are no screws here like the LGA socket-based EPYC processors use. The pins being moved from the CPU to the motherboard is a big change for AMD with Zen 4 + AM5 socket, just like it was for Intel all those years ago.
AMD's next-gen Raphael CPUs are teased here in the render, with the Zen 4-based processors expected to drop in the second half of 2022 -- but between now and then we can expect Vermeer-X3D which is a Vermeer refresh, but it'll rock AMD's new 3D V-Cache technology.
Continue reading: Here's our best look yet at AMD's next-gen AM5 socket for Zen 4 CPUs (full post)
Don't worry, Intel's new Z690 chipset mobos keep 24-pin connector
Intel will be launching its new 12th Gen Core "Alder Lake" CPUs in October 2021, just a few months away from now -- and we now know that they should be retaining that old faithful 24-in power connector.
The reason? Well, motherboard makers don't want to adopt the new ATX12VO standard that Intel was pushing a few months ago, but motherboard makers don't want to use it on the upcoming 600-series motherboards. Not even the new flagship Z690 chipset and its custom motherboards will have the new ATX12VO standard, we're sticking with 24-pin power connectors for a little while longer yet.
It was actually kind of expected that Intel would push the new ATX12VO standard with their new 12th Gen Core "Alder Lake" platform... but it seems they won't. It would've been nice, as we'd shift away from the bulky 24-pin connector to a much, much smaller 10-pin connector. This would be akin to the big change to the 12-pin connector on NVIDIA's new Ampere-based GeForce RTX 30 series Founders Edition graphics cards.
Continue reading: Don't worry, Intel's new Z690 chipset mobos keep 24-pin connector (full post)
Intel retires +++ scheme: Intel 7, Intel 4, Intel 3, Intel 20A nodes
Intel is doing some much-needed changes to its node naming scheme, shifting away from its just-introduced 10nm Enhanced SuperFin nomenclature with the introduction of Intel 7.
Intel 7 was previously referred to as Enhanced SuperFin, and will have 10-15% performance-per-watt improvements, FinFET transistor optimizations, and is currently in volume production. The new Intel 7 node will arrive with consumer-focused 12th Gen "Alder Lake" CPUs as well as data center-bound Sapphire Rapids CPUs.
During its Intel Accelerated event, the company kinda confirmed that its next-gen Alder Lake CPUs will have a hybrid design. This means we're looking at 8 high-performance cores (Golden Cove) and 8 efficiency optimized cores (Gracemont). We expect Intel to launch its next-gen 12th Gen Alder Lake CPUs with Microsoft's launch of its next-gen Windows 11 operating system... but it looks like Intel will be launching its next-gen CPU on October 27.
Continue reading: Intel retires +++ scheme: Intel 7, Intel 4, Intel 3, Intel 20A nodes (full post)
China's new Loongson CPU is nearly as good as first-gen AMD Ryzen
I still remember writing about Chinese chipmaker Loongson all the way back in 2012, that they were working on a new custom CPU design ready to take on Intel, AMD, and others in the tech industry... well, here it is.
Loongson's new 3A5000 CPUs are based on the Chinese chipmaker's own in-house proprietary microarchitecture, offering a quad-core design with CPU clocks between 2.3GHz and 2.5GHz with up to 50% more performance and lower power consumption over their previous-gen designs.
The CPU cores are superscalar, with each packing 4 general-purpose ALUs and 2 x 256-bit vector operation units, with the cores based on the LoongArch GS464V instruction set architecture. LoongArch supports binary conversion extension instructions (LBT), vector processing extension instructions (LSX), advanced vector processing extension instructions (LASX), and virtualization extension instructions.
Continue reading: China's new Loongson CPU is nearly as good as first-gen AMD Ryzen (full post)
Intel Core i9-12900K up to 25% faster than Ryzen 9 5950X in new leaks
You can already buy Intel's next-gen flagship Core i9-12900K processor on the black market for $1000+ but what about benchmarks? Another day has passed, and more news on Intel's next-gen "Alder Lake-S" CPUs are here this time in the form of benchmarks.
The new Core i9-12900K in its Qualification Sample (QS) form is already faster than AMD's current-gen Ryzen 9 5950X processor, with early benchmark results comparing Alder Lake vs Zen 3 in Cinebench R20. In the single-threaded Cinebench R20 benchmark, the new Core i9-12900K processor is 26% faster than the Ryzen 9 5950X, while in multi-threaded tests it's 11% faster.
This is super, super early so don't go judging the performance from these leaked benchmarks of a Qualification Sample CPU that is months away from being released. Intel's new Core i9-12900K will feature 8 high-performance "Golden Cove" cores and 8 high-efficiency "Gracemont" cores -- for a total of 16 cores and 24 threads. The smaller high-efficiency cores do not have Hyper-Threading (HT).
Continue reading: Intel Core i9-12900K up to 25% faster than Ryzen 9 5950X in new leaks (full post)
AMD's Zen 4 EPYC could get HBM upgrade to fight Intel Sapphire Rapids
Intel has confirmed that its next-gen "Sapphire Rapids" Xeon CPUs would have HBM, and now we're hearing that AMD will be upgrading its Zen 4-based EPYC "Genoa" CPUs with HBM.
The news is coming from Inpact-Hardware, which is reporting that they've got sources that tell them an HBM-powered version of AMD's next-gen EPYC Genoa CPUs is being planned. Inpact-Hardware reports: "There is indeed a recurring question of an HBM version of Zen 4 among partners, but for the moment nothing seems definitively decided on this subject. The manufacturer could indeed reserve such a solution for certain customers or ultimately prefer a variation with 3D V-Cache".
We're to expect Intel's next-gen Sapphire Rapids CPUs in 2022 with HBM variants in volume sometime in 2023, with AMD partners asking the company if there would be an EPYC CPU with HBM on-board. It seems they have been pressured enough from their partners, and now Intel with HBM on Sapphire Rapids, for Zen 4-based EPYC "Genoa" CPUs to have HBM.
Continue reading: AMD's Zen 4 EPYC could get HBM upgrade to fight Intel Sapphire Rapids (full post)
Intel's next-gen Core i9-12900K CPU on the black market from $1064
Intel is still a few months away from unleashing its next-gen 12th Gen Core CPUs with the new Alder Lake-S platform coming in October 2021 according to the latest reports -- with qualifying samples leaking out onto the CPU black market for $1064.
The new Intel Core i9-12900K processor can be purchased for anywhere between $1064 and $1157 on the "Chinese black market". VideoCardz reports that it has confirmed these Core i9-12900K processors are indeed Qualification Samples and the major reason why sellers wouldn't want to post pictures of their super-secret CPUs.
The new Intel Core i9-12900K QS processor clocks in at up to 5.3GHz on its big cores and 3.7GHz on the small cores, with power limits of 125W for PL2 and up to 228W for PL2. Inside, the new Core i9-12900K will have 8 high-performance Golden Cove cores and 8 high-efficiency Gracemont mini cores.
Continue reading: Intel's next-gen Core i9-12900K CPU on the black market from $1064 (full post)
AMD Ryzen Threadripper 5995X: 64 cores on Zen 3 coming in 2021
AMD is reportedly going to unleash its new codename Chagall CPU later this year, a new Zen 3-based range of Ryzen Threadripper CPUs with up to 64 cores and 128 threads.
We've been hearing more and more on the new AMD codename Chagall lately, but now Tom from Moore's Law is Dead is providing some more detailed information on the new CPUs. Tom says that the new Chagall CPU will have up to 64 cores with the Ryzen Threadripper 5995X processor, and will launch in November 2021.
The new Ryzen Threadripper 5995X (and other Threadripper 5000 series CPUs) will have quad-channel DDR4-3200 RAM support, 64 PCIe 4.0 lanes in total, and should be more expensive than the Zen 2-based Ryzen Threadripper 3000 series CPUs but Tom points out that "exact pricing isn't finalized yet".
Continue reading: AMD Ryzen Threadripper 5995X: 64 cores on Zen 3 coming in 2021 (full post)
TSMC eyes continued US expansion, could build new fab in Japan
Taiwan Semiconductor Manufacturing Company (TSMC) isn't stopping, with the latest news teasing the semiconductor giant would build even more factories in the US -- and a new fab in Japan.
TSMC confirmed it would be expanding its production capacity in China, but also teased that it could kick off a "second phase" expansion of its US operations, a new factory worth $12 billion in Arizona, USA. TSMC is also looking at the possibility of opening up a specialty technology wafter plant in Japan, which comes at a very interesting time.
China has said that it would take Taiwan by force if it needed to, and more recently the CCP military warned that it would drop nuclear bombs continuously on Japan if they interfered with China in any way, shape, or form. China could simply invade Taiwan and gain control of TSMC, something that was being discussed just a couple of months ago.
Continue reading: TSMC eyes continued US expansion, could build new fab in Japan (full post)
Intel's new Xeon CPU: 38C/76T chip still loses to AMD, so, so badly
Intel is reportedly preparing a new family of Xeon CPUs with the new Xeon W-3300 series, with its flagship model packing 38 cores and 76 threads to better fight AMD's chiplet onslaught with the Ryzen Threadripper PRO family of processors.
The new Intel Ice Lake Xeon workstation CPUs would pack more CPU cores and threads, more cache, a new architecture that's based on the 10nm process node, more I/O and memory support, and more. We have the LGA 4189 socket, with Xeon W-3300 series CPUs coming in at up to 270W TDP.
We have 64 PCIe 4.0 lanes so you can throw in multiple Sabrent Rocket 4 Plus Destroyer SSDs and a couple of high-end GPUs and still have PCIe 4.0 lanes to spare. There's also 8-channel DDR4-3200 memory support, which will offer support for up to 4TB of system RAM.
Continue reading: Intel's new Xeon CPU: 38C/76T chip still loses to AMD, so, so badly (full post)
AMD RDNA 2 GPU codename 'Voyager' inside of Samsung Exynos 2200 on 4nm
Samsung's next-gen Exynos 2200 mobile processor is in the works, and according to the latest leaks, we're to expect the next-gen SoC to use AMD's next-gen codename Voyager GPU based on the RDNA 2 architecture.
The new rumor is that Samsung's next-gen Pamir SoC will be built on the new 4nm process, with the GPU side of things in the form of the Voyager GPU built with Samsung and AMD working together on the mobile GPU based on the RDNA 2 architecture.
We're to expect Samsung's next-gen SoC with the mobile RDNA 2 architecture to debut in early 2021, which will compete against Qualcomm's next-gen Snapdragon 895 which is expected to use the ARM-based Adreno 730 GPU, and also made on 4nm node from Samsung.
Continue reading: AMD RDNA 2 GPU codename 'Voyager' inside of Samsung Exynos 2200 on 4nm (full post)
AMD's next-gen Zen 4 desktop CPUs will offer 16 cores max
AMD won't be offering more CPU cores on its next-gen Zen 4 desktop CPUs according to the latest leaks, with the flagship Zen 4 desktop processor having 16 cores max.
The current-gen Zen 3-based Ryzen 9 5950X processor has 16 cores and 32 threads, with previous rumors teasing a flagship Zen 4 desktop processor would have 24 cores and 48 threads. Patrick Schur on Twitter tweeting that Raphael would have up to 16 cores, up to 170W TDP, and made on 5nm.
AMD's next-gen Zen 4 processors will be made on TSMC's new 5nm fabrication node, with future Ryzen 6000 series CPUs support next-gen DDR5 memory as well as the first with the new IHS design for the LGA1718 socket codenamed AM5.
Continue reading: AMD's next-gen Zen 4 desktop CPUs will offer 16 cores max (full post)





















