HW Information – NVIDIA Crypto Self-Personal, Intel Xe HPG GPU Tease, Silicon Scarcity Timeline, & Google Lawsuit | GamersNexus

01:17 | Intel Posts Gaming GPU Decoder Problem

Intel lately tweeted a teaser to its Xe HPG gaming structure, supposed to be used in discrete GPUs within the close to future. Whereas all of us look ahead to Intel to save lots of us from silicon shortages, the corporate has given indicators that there could also be a cypher problem (no, not the rap form) to decrypt its messages and study extra concerning the Xe HPG {hardware}. NVIDIA considerably famously had a cryptographic set of packages and webpages again for the GTX 10 sequence launch (the “Order of 10”), and if we’re fortunate, Intel will do one thing equally enjoyable.

The teaser video had a binary string hidden within the backside nook. In the event you don’t wish to hear what that string decoded to (in an effort to remedy it your self), skip to the following story.

The string was an 8×4-digit string and decoded to 35 (.) 160 (.) 237 (.) 208, which, when entered as an IP deal with into an internet browser, directs to an Intel server and divulges an internet deal with of “xehpg.intel.com,” which says “welcome to the Xe HPG Scavenger Hunt” and has a launch date of March 26, 2021. Now, to be clear, that’s the launch date of the scavenger hunt. We’ll hold you posted on what occurs with it. We’ve participated in a few of these prior to now and, when not too corny and achieved cryptically sufficient, they could be a lot of enjoyable. Right here’s hoping Intel has one thing promising to ship to the present GPU panorama. They are going to by no means have a greater likelihood than proper now.

Supply: https://twitter.com/IntelGraphics/standing/1372594993781694465

Supply: https://xehpg.intel.com/

05:09 | Silicon Scarcity Seems to be Higher in Q3

In talking with trade sources at main board companions and PC half producers, we’ve acquired phrase that the present expectation of easing provide shortages isn’t till the distant future: Just a few main board companions reported to GN that their steerage from NVIDIA (particularly) has indicated extra potential to fulfill demand in Q3 of 2021. Present steerage means that AMD’s CPU silicon has already turn out to be extra prevalent (as we reported final week, through PC World), and that the higher-end CPUs ought to turn out to be extra available into mid-year this 12 months. The 5800X and 5600X have already been out there for a couple of days at a time these days. No phrase on AMD GPU shortages but, from our contacts, but it surely does sound like Q3 is the present goal for extra fixed provide of NVIDIA GPUs. That’s additional than we’d like, and consider it’s simply trade steerage, however that’s the present information.

Supply: GN

07:15 | NVIDIA’s Crypto Self-Personal

NVIDIA needs to rejoice. NVIDIA was cleared by a Federal Choose of the crypto lawsuit final week, whereupon traders claimed NVIDIA underreported its mining income and overreported gaming, and so the corporate is celebrating by hacking its personal drivers.

NVIDIA’s RTX 3060 launched with each firmware and driver locks in place, forming what the corporate described as a “handshake” between the 2 to hinder Ethereum mining hash charges. The RTX 3060s had been nonetheless capable of mine different cryptocurrencies on the unobstructed charge, and a few experiences urged that this meant that the block had been bypassed, however in truth, Ethereum was nonetheless restricted.

Not less than, till NVIDIA hacked itself.

Bored with ready for reddit to crack the answer, NVIDIA launched beta driver model 470.05 forward of the game-ready launch, and in so doing, by chance totally unlocked Ethereum mining efficiency. We’d assume that the corporate had a code department in some unspecified time in the future simply earlier than launch to fork safety into the drivers and firmware, and maybe it wasn’t grafted into the beta branches. Both manner, PC Watch, Laptop Base, and {Hardware} Luxx have confirmed that the beta driver did, in truth, take away the block. As can be apparent, this screw-up can’t be unscrewed: Miners might simply decide to run older drivers completely.

Main supply: https://computer.watch.impress.co.jp/docs/information/yajiuma/1312/085/index.html

Secondary supply: https://www.pcworld.com/article/3611422/nvidia-cracks-its-own-geforce-rtx-3060-anti-mining-protection.html

09:57 | AMD Reveals Epyc Milan 7003-Sequence CPUs

As promised final week, AMD took the lid off of its newest Epyc Milan CPUs. AMD’s new Epyc 7003-series will probably be topped by the 64C/128T Epyc 7763, which additionally runs a base frequency of two.45GHz and a lift clock of three.50GHz. All in all, the brand new Epyc Milan 7003 lineup will encompass 19 SKUs and can goal AMD’s traditional 1P and 2P server clients. 

Certainly one of AMD’s greatest strengths with Epyc is its simplified and non-price-segmented function set. As in, every thing AMD has to supply with Epyc is on faucet: all SKUs provide 128 PCIe 4.0 lanes, 8-channel reminiscence help, and DDR4-3200 help. Based mostly on Zen 3, Epyc Milan may also convey a tangible IPC uplift and improved boosting algorithms to extract larger frequencies from the silicon. 

As with different Zen 3 CPUs, Epyc Milan is manufactured on TSMC’s N7 node and can use AMD’s tried and true chiplet topology, permitting for as much as 9 chiplets (CCDs) to be stitched collectively. As traditional, Zen 3 CCDs span as much as eight cores, and are afforded entry to 32MB of L3 cache. Milan will use the SP3 socket, and will probably be backwards appropriate with Rome and Naples. 

AMD famous that its companions like AWS, Dell, HPE, Google Cloud, Microsoft Azure, and extra will all be adopting Third-gen Epyc for their very own Epyc 7003-based designs.        

Supply: https://www.amd.com/en/press-releases/2021-03-15-amd-epyc-7003-series-cpus-set-new-standard-highest-performance-server

12:19 | System76 Thelio Mira Linux Desktop

We’ve talked about System76 right here earlier than, particularly for the truth that it’s a producer specializing in Linux-based programs. System76 presents plenty of laptops, desktops, mini-PCs, and servers all operating a Linux distro. System76 lately trotted-out its newest desktop design, the Thelio Mira desktop.

Like the remainder of its catalog, System76’s Thelio Mira ships with Linux — both Ubuntu 20.04 LTS or System76’s personal Ubuntu-based Pop!_OS 20.10 or Pop!_OS 20.04 LTS. The Thelio Mira is geared toward customers on the lookout for a compact workstation, and may be configured with AMD’s Ryzen 3000 or 5000 CPUs and Nvidia GPUs. System76 is providing a GT 1030 on the bottom-end, and as much as two Quadro RTX playing cards on the high-end, with choose  RTX 30-series playing cards being supplied in between.

Elsewhere, Thelio Mira will probably be configurable with as much as 128GB of dual-channel ECC DRR4 reminiscence and as much as 36TB of storage, with a number of M.2 PCIe Gen4 NVMe slots and a pair of.5” bays on faucet. 

Thelio Mira additionally comes with System76’s Thelio IO daughterboard, which is an open-source I/O board that strikes chassis and fan controls from the motherboard to the daughterboard, which aligns with System76’s purpose of shifting performance away from proprietary and closed-source {hardware}. System76 additionally contends that this enables for extra granular management over fan, storage, and energy. The board makes use of an embedded Microchip ATmega32U4 for the controller, and the design information are hosted on Github, so anybody can iterate or enhance on them. 

Supply: https://system76.com/desktops/thelio-mira

14:19 | NZXT May Be Increasing Into Shows

A current job posting means that NZXT is trying to get its ft moist within the ever-crowded show market. The itemizing was caught by OC3D.web, and requires a “Show/Monitor Engineering Lead.” Moreover, in accordance with the itemizing, anybody assuming this place “will probably be liable for the event and qualification of Flat Panel Shows for desktop screens and moveable merchandise.”

The itemizing is in search of candidates which have expertise with a number of show applied sciences, a MS or BS diploma in Electrical Engineering, 15+ years expertise in creating show merchandise, and confirmed expertise in taking 5 or extra merchandise from idea to manufacturing.

NZXT is barely the most recent pc {hardware} firm trying to get into shows. MSI and Cooler Grasp have already taken the plunge, and Corsair posted an identical job itemizing on Certainly calling for a “Merchandise Supervisor” for shows again in 2018.  

Supply: https://overclock3d.web/information/gpu_displays/nzxt_plans_enter_the_gaming_monitor_market_-_job_listing_confirms/1

15:59 | Rockstar Listens, Will Implement Consumer-Created GTA:On-line Repair

We lately reported on a programmer and GTA On-line participant, going by the identify of t0st, bettering the sport’s sluggish loading occasions with a customized DLL. On the time, there was hypothesis as to how severe Rockstar would take the findings, however surprisingly, Rockstar has confirmed that what t0st discovered is correct. 

“After an intensive investigation, we will verify that participant t0st did, in truth, reveal a facet of the sport code associated to load occasions for the PC model of GTA On-line that might be improved. Because of these investigations, we’ve got made some modifications that will probably be applied in a forthcoming title replace,” mentioned Rockstar.

As of March 16, Rockstar has rolled out a patch that implements the findings of t0st, and if the GTA On-line subreddit is something to go by, the outcomes are vital. Moreover, t0st confirmed that additionally they acquired a $10K award by way of Rockstar’s bug bounty program. In accordance with t0st, Rockstar made a particular exception with the award, as its bug bounties are normally just for safety associated bugs. 

Supply: https://www.techspot.com/information/88932-rockstar-implementing-fan-discovered-fix-gta-online-boost.html


17:47 | Qualcomm Completes Nuvia Acquisition

As anticipated, Qualcomm has closed its acquisition of silicon/CPU design firm Nuvia. Qualcomm shelled out $1.4B to accumulate Nuvia, and plainly will afford it the chance to tackle Apple and Arm, and even take one other shot on the server market. Qualcomm’s Snapdragon SoCs have lived within the shadow of Apple’s A-series SoCs for years now, and Qualcomm’s customized designs have by no means strayed too removed from vanilla Arm designs.

Now, Qualcomm has wasted no time in stating its fast intent to deal with high-performance laptops. “The primary Qualcomm Snapdragon platforms to function Qualcomm Applied sciences’ new internally designed CPUs are anticipated to pattern within the second half of 2022 and will probably be designed for prime efficiency ultraportable laptops,” mentioned Qualcomm in its press launch. 

Moreover, Qualcomm will leverage its Nuvia acquisition to design in-house silicon for “flagship smartphones, laptops, and digital cockpits, in addition to Superior Driver Help Programs, prolonged actuality, and infrastructure networking options.”

As a reminder, Nuvia was based by former Apple A-series architects John Bruno, Manu Gulati, and Gerard Williams III. Nuvia by no means really launched a product earlier than getting Qualcomm’s consideration, however the firm was engaged on customized silicon for servers and HPC purposes. It was little question the engineering pedigree of Nuvia’s founders that obtained Qualcomm’s curiosity, and as famous by Qualcomm’s press launch, Gerard Williams is now serving as Qualcomm’s SVP of Engineering.    

Supply: https://www.qualcomm.com/information/releases/2021/03/16/qualcomm-completes-acquisition-nuvia


19:41 | Google Lawsuit Over Incognito Mode Continues

In June of 2020, we mentioned a $5B class-action lawsuit being slapped in opposition to Google for the alleged monitoring of customers whereas in Google’s “non-public” incognito searching mode. The lawsuit is leaning on Google violating the Federal Wiretap Act and deliberately deceiving customers. 

For its half, Google tried to have the case dismissed on the idea that Chrome help paperwork make it clear that “exercise would possibly nonetheless be seen.” In accordance with Bloomberg and ArsTechnica, Google was unable to persuade the courts and the lawsuit will transfer ahead. 

“The courtroom concludes that Google didn’t notify customers that Google engages within the alleged information assortment whereas the person is in non-public searching mode,” in accordance with US District Choose Lucy Koh. As traditional, Google disputes this declare, and has vowed to “vigorously” defend itself. The lawsuit seeks no less than $5 billion in damages, as much as $5,000 per affected person. The go well with is retroactively concentrating on any person that has used Chrome Incognito mode since June 1, 2016.

Supply: https://arstechnica.com/devices/2021/03/judge-rules-5-billion-google-chrome-incognito-mode-lawsuit-can-go-forward/


21:09 | Micron Exits The 3D Xpoint Enterprise

Just a few years in the past, Micron and Intel introduced a three way partnership to develop 3d Xpoint reminiscence, which was concepted as one thing of an evolution of NAND Flash and DRAM, with theoretical larger endurance, efficiency, and density. Intel branded its model of 3D Xpoint as Optane, whereas Micron went on to model its personal 3D Xpoint merchandise as QuantX. Nevertheless, none of Micron’s QuantX merchandise ever shipped, apart from Micron’s X100 SSD that shipped to pick datacenter clients. 

Within the intervening years, Intel and Micron dissolved their 3D Xpoint partnership, and Intel has even scaled again its personal Optane improvement and ambitions, primarily specializing in datacenter and enterprise purposes. Now, Micron has determined to desert the 3D Xpoint enterprise fully. 

“Micron has now decided that there’s inadequate market validation to justify the continuing excessive ranges of investments required to efficiently commercialize 3D XPoint at scale to deal with the evolving reminiscence and storage wants of its clients,” says Micron. “In keeping with this new strategic focus, Micron is engaged in discussions for the sale of its Lehi, Utah fab at present devoted to 3D XPoint manufacturing. The corporate goals to achieve a sale settlement inside calendar 12 months 2021.”

Micron additionally made it clear that going ahead, it can give attention to creating reminiscence and storage merchandise that help the rising Compute Specific Hyperlink (CXL) interface. That is one thing that Intel most likely isn’t eager on getting onboard with, as Optane is proprietary and unique to Intel’s CPUs. 

With Micron exiting the 3D Xpoint enterprise, there was some hypothesis on what that might imply for Intel, as Micron has been producing the reminiscence for Intel’s Optane merchandise. Tom’s {Hardware} spoke to each Micron and Intel, and acquired the next statements: 

“Micron is dedicated to finishing our obligations underneath the prevailing wafer provide settlement with Intel, which at present goes by way of the top of CY21, and Micron intends to retain all of its mental property related to 3D XPoint,” says Micron (through Tom’s {Hardware}). 

“Micron’s announcement would not change our technique for Intel Optane or our potential to provide Intel Optane merchandise to our clients,” says Intel (through Tom’s {Hardware}).

Supply: https://traders.micron.com/news-releases/news-release-details/micron-updates-data-center-p



23:01 | AMD to Develop into TSMC’s Second-Largest Buyer

New experiences detailing the capital expenditure (CapEx) spending of Samsung and TSMC recommend that Samsung won’t be displacing TSMC because the trade chief anytime quickly, although Samsung does seem like closing the hole in total manufacturing prowess and expertise. 

Again in November of 2020, Bloomberg reported that Samsung deliberate to spend $116 billion to bolster its manufacturing capabilities in an effort to catch as much as TSMC by 2022. This consists of increasing the event of GAAFET transistors, which we’ve been discussing lots these days. As we’ve mentioned beforehand, Samsung intends to make use of GAAFETs, or its trademarked Multi-Bridge Channel FETs (MBCFET) at 3nm, whereas TSMC will nonetheless be utilizing FinFETs. On the Institute of Electrical and Electronics Engineer’s current Worldwide Stable-State Circuits Convention (that might be the IEEE’s ISSCC for acronym fanatics), Samsung lately demonstrated the efficiency and energy effectivity enhancements that it might accomplish with GAAFETs and SRAM.

Looking for Alpha experiences that TSMC at present holds a bonus over Samsung in each manufacturing capability and income, in addition to a broader buyer base for contract chip making. TSMC has an total 2.3x larger capability than Samsung. Samsung has been slowly closing the hole, in 2016 TSMC’s capability benefit was 3.3 occasions what Samsung might produce. Smaller nodes present a good larger benefit for TSMC. In 2021, TSMC’s capability for 7nm is projected to be between 5 and 6 occasions that of Samsung, and for 5nm, TSMC is projected to have 4 occasions the capability of Samsung. TSMC can be trying to start 3nm manufacturing this 12 months, one thing Samsung just isn’t ready for but. 

Moreover, TSMC at present boasts a broader shopper base than Samsung. Business estimates state that as a lot as 60% of Samsung’s capability is allotted for its personal gadgets, like Exynos SoCs for Samsung smartphones. In the meantime, TSMC serves 460 clients, together with Apple, Hello-Silicon, AMD, Nvidia, Broadcom, Qualcomm, and Intel, amongst others. 

Talking of consumers, one spotlight of the experiences is that AMD is on observe to turn out to be TSMC’s second largest buyer for 2021, accounting for an estimated 9.2% of TSMC’s 2021 income. AMD will nonetheless be far and away from TSMC’s prime buyer in Apple, which can account for 25.4% of TSMC’s 2021 income. Nevertheless, AMD will probably be forward of different outstanding clients like Nvidia, Qualcomm, and Broadcom.

Supply: https://seekingalpha.com/article/4414346-samsung-won-t-soon-overtake-taiwan-semi-despite-massive-capex-spend


Supply hyperlink

Leave a comment

Your email address will not be published. Required fields are marked *