Tons of reports and rumors have gone flying around concerning the upcoming AMD Ryzen 7 9800X3D processor, including release dates, clock speeds, and more. But thanks to a new leak from the motherboard manufacturer, we may now have a good idea of the specs it’s packing.
Maxsun, which manufactures the motherboard for the processor, has seemed to reveal some juicy information regarding the chip. Interestingly enough, it’s referred to as the AMD Ryzen 9 9800X3D in the leak but most likely it’s the Ryzen 7. According to the listing, it has a TDP of 120W and 96MB of cache, including 32MB of L3 cache and 64MB of 3D V-Cache. This means the cache size is still the same as in previous models, which honestly isn’t great.
The leaked document also reveals that the base clock speed is 4.7 GHz — 900 MHz faster than the 9700X. Most likely this is due to the TDP being higher in the 9800X3D. The leak also states that it’ll have a boosted clock speed of 5.2 GHz, which would oddly be 300MHz slower than the 9700X. This all lines up with an earlier leak that confirms the same information. As for single-core clock speed, that hasn’t been revealed at this time but it’s assumed to be 5.4/5.5GHz.
If past reports are to be trusted, then we should have official confirmation of the 9800X3D’s specs on October 25, 2024. Though most likely the actual launch date would be sometime in November 2024.
Disappointing rumored performanceThough base clock speeds seem to be promising enough (while boosted not so much), a recent report from a German tech site seems to point to rather unimpressive benchmark scores. In gaming comparisons, the Ryzen 9800X3D is 11% faster than the 7800X3D in Far Cry 6. Shadow of the Tomb Raider sees an uptick of only 4% and Black Myth: Wukong a dismal 2%.
Cinebench does look a bit more promising in this regard, with a huge performance increase of around 18% and 28% faster than its predecessor for single-core and multi-core, respectively. But still, considering this is supposed to be a next-gen chip, the overall results are rather lackluster.
You might also likeLooking to beef up server capacity? The Inventec 96 DIMM CXL box could be what you’re looking for. Unveiled at OCP Summit 2024 alongside Astera Labs, the CXL expansion box allows users to connect up to 96 DDR5 DIMMS to a single server, providing enormous memory capacities ranging up to the tens of terabytes per memory server.
As reported by ServeTheHome, the expansion box can connect to the upcoming eight-way Intel Xeon 6 Granite Rapids-SP server, offering an even bigger punch.
Combined with the Intel Xeon 6 server, which boasts a total of 128 DDR DIMM slots, this means users could harness a massive 244 DIMM slots for a single server, providing significant memory capacity.
What you can expect from the 96 DIMM CXL expansion boxSo what does this announcement mean exactly? Well, we’re talking about significant memory capacity boosts for servers.
The DDR5 DIMMS in question here are DDR5-4800, and the expansion box uses the Astera Labs Leo. Elsewhere, users are able to draw up 24 CDFP ports, each of which boasts a PCIe Gen5 x16 connection linking the CXL boat and the servers themselves.
As Servethehome noted, the launch shows users can draw upon a sizable 20TB of memory in a single server, which has massive long-term implications for server capacities and marks a huge development.
Underpinning all of this is the growing potential of compute express link (CXL) technology.
In 2023, Meta announced a partnership with AMD to showcase a memory type that can be added onto servers and capable of recycling petabytes worth of RAM.
As part of the collaboration, the two firms showed CXL can be used to markedly improve memory efficiency, in turn reducing costs and enhancing performance.
The demo board showcased by AMD and Meta included an AMD EPYC 9004 Genoa, boasted four dual in-line memory module (DIMM) slots, and a heat sink and fan.
Astera Labs’ involvement here marks the second major announcement from the semiconductor company so far this year. In April 2024, Astera Labs showcased its Aries 6 PCIe retimer board, which TechRadar Pro reported at the time could unlock significant benefits for hyperscalers ramping up AI development.
This retimer board can help maintain data signal integrity over the PCIe interface.
These high-speed data transfers typically suffer from degradation over long distances or due to interference.
The Aries 6 retimers marked the first in Astera Labs’ PCIe 6.x portfolio, and are aimed specifically at enhancing networking capabilities for next-generation GPUs, accelerators, CPUs, and CXL memory controllers.
More from TechRadar ProPicture the scene: you jump on an important work call, fresh-faced and ready to confidently deliver your trademark “nothing from me” line, only to discover that the light (or lack thereof) in your work-from-home office has left you looking less like a serious businessperson and more like a seedy Omegle user from 2010. Don’t worry – we’ve all been there.
Thankfully, Qualcomm knows we’ve all been there, which is why the American semiconductor giant has seen fit to equip its newly announced Snapdragon 8 Elite mobile chipset with an AI-powered, real-time lighting tool that could banish terrible video call lighting to the annals of internet history.
AI Portrait Video Re-lighting lets you add a virtual, moveable light source that maps to the contours of your face to better illuminate it in heavily backlit environments. This lighting can be adjusted manually, or you can let your phone’s AI do the work for you, and the entire process takes place on-device, with no added latency. Wait a minute, did Qualcomm just kill the ring light?
Qualcomm debuted the new tool at Snapdragon Summit 2024, demonstrating its potential in a nameless dummy phone, though we could conceivably see AI Portrait Video Re-lighting feature (albeit with a different name) in some of the best Android phones of 2025 – think the much rumored Samsung Galaxy S25, OnePlus 13, Xiaomi 15 Ultra, and so on.
That’s because all of those upcoming flagship phones are expected to be powered by the Snapdragon 8 Elite, meaning their respective OEMs (original equipment manufacturers) will have the option of implementing this new AI technology as they see fit.
Indeed, as Qualcomm’s Senior Director of Product Management, Judd Heape, confirmed to TechRadar in a roundtable interview at Snapdragon Summit (with reference to Limitless Segmentation, another Snapdragon 8 Elite-specific camera feature): “Limitless Segmentation is embedded into the [Snapdragon 8 Elite] camera framework – it’s shipping with the baseline camera. So, it will be up to OEMs as to how they integrate it [and other AI features] into the main products.”
Limitless Segmentation segments an image into over 250 layers, optimizing and enhancing each layer (including faces, hair, clothing, objects, and backgrounds) in the process, so presumably Qualcomm’s AI Portrait Video Re-lighting technology could be applied to still portraits, too. Might we see a “Portraits by Qualcomm” label added to images taken with the best Samsung phones in the near future? It’s certainly a possibility.
In any case, we’re glad to see Qualcomm using the power of its Snapdragon 8 Elite chipset to develop practical generative AI tools like this one. Sure, bespoke emoji creators are great – but they’re not going to help you save face when your video call lighting just ain’t it.
You might also likeMicrosoft might have debuted its major Windows 11 24H2 update only recently, but it’s already becoming infamous due to its habit of appearing in headlines for causing user issues. This time, there have been reports of a bug that causes devices to crash and show the blue screen of death (BSoD). This comes after previous reports of problems that were serious enough that Microsoft ended up blocking 24H2 from being installed on affected devices.
Today, we can add another glitch to the list - a system crash that apparently occurs on PCs with the Voicemeeter app installed. Voicemeeter is a popular app that helps people manage their PC’s audio, and according to Neowin, Microsoft has published a note in Windows 11 24H2’s official documentation describing the problem. Unfortunately, it seems like the driver software (which enables the app to “talk” to your system) is incompatible with Windows 11’s memory manager.
This incompatibility results in many users’ systems becoming unresponsive, leading to the dreaded blue screen with the MEMORY_MANAGEMENT error code - a rather nasty bug, as it could mean people’s unsaved work is lost if the PC suddenly reboots.
To prevent this from happening, Microsoft has put a hold on rolling out update 24H2 to PCs that are running Voicemeeter. If you want to continue using the app in Windows 11, you won’t even be able to see the option to update to 24H2 in Windows Update when you check for new updates.
(Image credit: Future) What Windows 11 Voicemeeter users can expect nextIt does seem like Microsoft is aware of the problem and, as reported by Neowin, it advises users not to force their devices to update to 24H2 using methods like the Installation Assistant or the Media Creation tool. I’m fairly sure this means that Microsoft is working on an update that delivers a fix that will allow users to update to 24H2 and continue using Voicemeeter trouble-free.
If you can give up Voicemeeter, then apparently you can remove the app altogether, and the 24H2 update will become available again for you to download after around 24 hours. So far, it seems like this compatibility issue only impacts PCs running 24H2, and earlier versions of Windows 10 and 11 aren’t affected.
This is yet another potential issue to consider if you’re looking to upgrade to Windows 11 24H2, and you can read some of the other problems users have been bumping into after trying to update here.
A BSoD isn’t fatal, but it can be incredibly frustrating and annoying. Along with all of the other errors and problems, 24H2 is shaping up to be pretty disruptive to some people. It also brings back bad memories of Windows 10 suffering from numerous update problems a few years ago.
This draws attention away from any actual quality and functionality improvements that Microsoft has included in 24H2. Still, it’s worth staying updated on issues that arise so you can know how to deal with them. It shows how difficult it can be when trying to roll out major updates across a wide range of software and hardware – and how Microsoft is still struggling to overcome those difficulties, even after decades of experience.
You Might Also Like...Some cloud storage providers offering end-to-end encryption (E2EE) are largely operating a broken ecosystem which could, in very realistic theory, allow threat actors to tamper with the files in a way that should not be possible, experts have claimed.
In an in-depth analysis, recently published on the brokencloudstorage.info website, cybersecurity researchers Jonas Hofmann and Kien Tuong Truong from ETF Zurich noted if a threat actor compromises a company server, they can “inject files, tamper with file data, and even gain direct access to plaintext."
During their research, the two experts analyzed five major providers in the field - Sync, pCloud, Icedrive, Seafile, and Tresorit, concluding, “many of our attacks affect multiple providers in the same way, revealing common failure patterns in independent cryptographic designs."
Nation-state targetsOn Sync and pCloud, a compromised server could be abused to break the confidentiality of uploaded files, inject files, and tamper with their content, while for Seafile such a server could be used to speed-up brute-force attacks, inject files, and tamper with the content.
For Icedrive, hackers could use a compromised server to break the integrity of uploaded files, inject files, and tamper with their content, while for Tresorid, a broken server could be used to present non-authentic keys, when sharing files. Crooks would also be able to tamper with some metadata in the storage.
The researchers stress that this doesn’t mean that the service providers are malicious, but that these flaws make them a major target for threat actors. More importantly - nation-state threat actors. They also added that compromising a server belonging to an E2EE cloud storage provider isn’t as far-fetched as it may seem at first.
In fact, they argue it’s the “most realistic” adversary model for E2EE cloud storage.
The majority of the service providers mentioned in the report - Sync, Seafile, and Tresorit, were said to have acknowledged the report. Icedrive is yet to address the issue, while there are no reports for pCloud just yet.
Via The Hacker News
More from TechRadar ProApple Intelligence is set to release next week, as Apple has confirmed that iOS 18.1 and AirPods Pro 2' hearing health features will begin rolling out next week at some point.
Apple confirmed in a press release that the AirPods Pro 2 hearing aid upgrade, which requires iOS 18.1, iPadOS 18.1, and macOS Sequoia 15.1, will be available to the public beginning next week. Putting two and two together means we'll be ushering in the Apple Intelligence era for iPhone and iPad very shortly.
You'll be able to access the first batch of Apple Intelligence features in these software updates, including Writing Tools for proofreading and rewriting, Smart Replies to quickly reply to messages, Notification Summaries, Clean Up in Photos, and a redesign of Siri.
Plus, if you have AirPods Pro 2, the earbuds can perform a hearing test and even act as FDA-approved over-the-counter hearing aids for those with mild to moderate hearing loss. Of course, you'll need to install that firmware update for AirPods alongside iOS 18.1 on your iPhone.
Apple Intelligence will work on some of the best iPhones, including the iPhone 16 lineup and the iPhone 15 Pro / Pro Max. Unfortunately, iPhone users of devices as recent as the iPhone 15 will not get access to any of the features Apple Intelligence offers.
Apple had previously announced that Apple Intelligence would become available in October, but for owners of the iPhone 16, iPhone 16 Plus, iPhone 16 Pro, and iPhone 16 Pro Max, waiting over a month to get the features on their new iPhones might still sting a little.
That said, now that we've got a better idea of when to expect Apple Intelligence, we can start to get excited about everything Apple's AI has to offer.
A new era Siri's redesign on an iPhone 16 Pro Max (Image credit: Future / Apple )If you're waiting for some of the more exciting Apple Intelligence features like ChatGPT integration in Siri and the voice assistant's major overhaul with on-screen awareness and personal context, then you'll need to stay patient as we won't be using those for a while.
Genmoji, Image Playground, ChatGPT integration, and Visual Intelligence are set to arrive as part of iOS 18.2, which should enter beta in the next month or so. Apple has said all of these features are coming "later this year," so considering there are only about two months left in 2024, we should get our hands on these AI tools soon. Siri's overhaul, on the other hand, won't be available until 2025.
Next week's release date is only for US English, with other parts of the world needing to wait until December to get their hands on Apple Intelligence. While you can easily change your device's language, you might find Apple Intelligence will struggle – trust me, it doesn't cope well with Scottish slang! As for Apple users in the EU, we still have no information on whether you'll be getting AI features.
Check out our iOS 18.1 Apple Intelligence hands-on to get a rundown of everything you'll be able to use at some point next week, as well as our deep dive into the AirPods Pro 2's hearing health features here. If you're looking for a new Apple Intelligence-compatible iPhone, check out our iPhone 16 review.
You might also like...