Factoring attacks against the RSA algorithm were generally considered infeasible, but that may change dramatically if the findings in a paper from UC San Diego (California) and Inria Nancy (France) prove correct after peer review. The paper outlines a new attack that breaks RSA in record time. That’s concerning because RSA-based encryption was in use for quite some time, and although it’s been generally deprecated in favor of ECC and post-quantum algorithms, it’s still employed today in a substantial portion of services.
Go deeper with TH Premium: Silicon Photonics
(Image credit: Getty Images / Bloomberg)
Assuming the research holds up and the attack conditions are met, the number of necessary operations for cracking an RSA-encrypted key (and thus, inspecting network traffic encrypted with it) drops dramatically. (From 280, 2112, 2128, and 2144 down to 265, 290, 2105, and 2119 — for 1024, 2048, 3072-, and 4096-bit key lengths, respectively.)
Even for the 1024-bit keys, performing 265 calculations is still a lot, but it starts getting into the range where it’s feasible for an entity with large resources at their disposal, like a number of state actors. Crucially, the paper indicates that the algorithms the team used are far from optimized and could hypothetically be significantly sped up after optimization passes, potentially with the help of AI tools and by leveraging GPUs for the task.
Latest Videos FromTom’s Hardware
The attack only works against the textbook (or “raw”) variant of RSA, which isn’t widely used for commonplace computing like website certificates and remote access, but is nevertheless present in services like Cloudflare Privacy Pass, iCloud Private Relay, and Private Cloud Compute. Equipment and services using PKCS #11, like many smart cards, USB security tokens, code-signing pipelines, hardware security (HSM) and trusted platform (TPM) modules also use this variation. But fulfilling the attack conditions is much easier said than done.
In order to reach a point where an attacker can decipher data, they first must perform an exceedingly high number of queries against the key they’re attacking, collecting enough data points from the key’s oracle. In practical terms, this often means repeatedly poking a liveserver that’s using that key to encrypt traffic, billions of times over, though it can also be done against a standalone hardware device.
In the practical context of an encryption algorithm, an oracle takes a number in, runs it through a computation involving a private key, and spits it back out, without revealing its key. The textbook variant of RSA under attack here lets the oracle be queried with any number, thus allowing the attacker to collect enough data points to perform the mathematical wizardry described in the paper. They can then proceed to throw a lot of computing power to ultimately inspect whichever traffic they captured that was using that same encryption key.
As network administrators have certainly figured out by now, running billions of queries against a live service is almost guaranteed to earn them an IP ban, among other measures. It’s a reasonable assumption that entities like Cloudflare and Apple have those kinds of controls in place. Still, not everyone does, especially in the case of key management services that aren’t closely monitored, including hardware-based ones. The scientists worked their magic on Thales Luna HSMs, in both hardware and online service form.
Get Tom’s Hardware’s best news and in-depth reviews, straight to your inbox.
Although the bar for a practical application of this attack remains quite high, the key finding is that it may be significantly lowered. This opens the possibility that oracle data collected today may be used to decrypt saved traffic in a matter of weeks or months, especially in the near future as hardware advances. It’s also possible that other scientists can take this idea and expand on it, finding additional flaws.
Nvidia has updated its RTX Mega Geometry SDK to 2.0, adding streaming support for “continuous level-of-detail clusters” to handle high-density meshes, according to thedeveloper blog post. Nvidia says it works even for scenes too big for the VRAM budget. The technology is in addition to its RTX Kit, arriving alongside RTX Kit 2026.3. The announcement comes close to the release of Gears of War: E-Day, which makes use of RTX Mega Geometry technology. Nvidia has not explicitly said what version of Mega Geometry the game uses.
Go deeper with TH Premium: GPUs
(Image credit: Noctua)
Mega Geometry “organizes detailed meshes into clusters so ray-traced scenes can adapt geometric detail efficiently,” according to Nvidia. The SDK appeared at version 0.9.0 beta in 2025 and found its way into Alan Wake 2’s title update 1.2.8 early that same year. Ourtest on an RTX 4090 found about 1GB of VRAM saved with a performance boost of 13%, at native 4K and with DLSS Quality. On an 8GB graphics card, that 1GB is an eighth of the memory.
The SDK is at version 2.0.0 on GitHub as of this week. The technology makes dense scenes less expensive to ray trace, with 2.0 specifically targeting geometry that doesn’t fit in VRAM. This means that “a scene’s source geometry no longer has to fit in VRAM [with detail] bounded by a memory budget rather than by mesh count,” according to Nvidia’s changelog.
Latest Videos FromTom’s Hardware
When a scene goes over budget, it settles at lower detail instead of constantly swapping geometry in and out of VRAM. By default, this is based on 2GB of VRAM for streamed geometry, another 2GB for ray-tracing acceleration structures, and 4GB of material textures, but these are adjustable. The existing tessellation path remains.
In the case of Nvidia’s sample Zorah, there are 1.6 billion unique triangles in a 70GB download, with a first load that can require up to 64GB of RAM at peak. Zorah uses 31GB of mesh data, but the courtyard sample view only has 1.5GB of mesh data in VRAM, using an RTX 5090 at 4K with DLSS Quality.
Image 1 of 2
(Image credit: Nvidia)
(Image credit: Nvidia)
The sample requires an RTX GPU with 10GB of VRAM or more, a Game Ready driver 570 or later, Microsoft’s DXR 1.1 or later, and Windows 10 at a minimum. For now, this looks more like a tool for developers than players. It’s an SDK with a reference sample, and currently no game has been confirmed to use it.
The rest of RTX Kit 2026.3 includes RTX Character Rendering 1.4, RTX Dynamic Illumination 3.1, RTX Neural Texture Compression (NTC) 0.10 beta, RTX Neural Shading 1.4, and RTX Texture Filtering 1.3. NTC has been at v0.10.0 beta on GitHub since Aug. 4, with the DX12 LinAlg path only for testing purposes. The DX12 LinAlg path requires preview driver 620.12, the preview Agility SDK, and Windows Developer Mode. That leaves Vulkan as the only way for released games to use NTC with Tensor Core acceleration today.
Get Tom’s Hardware’s best news and in-depth reviews, straight to your inbox.
An existing problem, Nvidia said in August, is that under ray tracing, Nanite elements “switch to a single, static low-detail mesh” because it’s “prohibitively expensive” to build acceleration structures for them with “existing DXR 1.2 and Vulkan APIs.” It has taken six years of research and development to pioneer the new technology, with the innovations debuting with Gears of War: E-Day on Oct. 6. These will later be part of Microsoft’s DirectX Raytracing 2.0 standard, according to Nvidia.
On Sept. 22, in a Game Ready driver post, Nvidia stated that Mega Geometry enables “accurate ray tracing of Nanite geometry for the very first time.” Nvidia’s Game Ready driver 617.14 for the game is already out. Early access on Oct. 1 is the first chance for players to see Nvidia’s Nanite ray tracing technology in action.
A newly published patent shows that Nvidia has been working on an AI chatbot that could make it quicker and easier for developers to optimize their PC games. The chatbot would be capable of receiving a plain English request for help before running diagnostics based on the specific issue at hand.
Go deeper with TH Premium: AI and data centers
(Image credit: Microsoft)
PC gamers are well aware of the impact of poor optimization on games. And while there are tools available to developers that help them identify issues and then address them, they require specialist knowledge to use. They also take time, something that developers rarely have — especially as they get closer to a release date.
The patent, first spotted by Patentlyze, could help. It depicts a chatbot-like interface in which developers would be able to outline the issue that they are seeing. They would be able to do so using normal language, allowing even relatively inexperienced developers to use the tool.
Latest Videos FromTom’s Hardware
Example questions provided by the patent include “What is causing the micro-stutter during camera cuts?” and “Why is frame time spiking when ray-traced reflections enter the viewport?”.
Once the question has been posed, the AI would then write a diagnostic specific to the issue before automatically running it on the GPU and then providing the results for interpretation. The developer would then be able to implement any fixes and re-run any testing that is still required.
The patent doesn’t make it clear whether there will be any validation or testing of the diagnostic code this tool would create, something that may give some pause. Test results are only as strong as the test that was used to get them, and developers will need to trust that Nvidia’s chatbot is up to the job before relying on it.
However, it’s important to remember that companies like Nvidia apply for countless patents every year. Not all of those patents turn into shipping products or features, and there is no guarantee that this one will ever see the light of day. That being said, it does seem similar to G-Assist, a chatbot-like tool that allows gamers to get AI-powered assistance for their GeForce RTX PC. Nvidia has also sought to bring its AI chops to bear on other aspects of gaming, not least DLSS 5 and its multiple AI modes for different levels of detail — whether gamers want it or not.
Get Tom’s Hardware’s best news and in-depth reviews, straight to your inbox.
For gamers, the hope will be that this patent will result in a tool that makes it easier for game developers to optimize their wares ahead of release.
PlayStation 5 emulation has witnessed massive growth ever since Sony confirmed its plans to end PC ports for its first-party, narrative-driven single-player games. The latest name joining the list is the open source project AnyPS5, which is claimed to run PS5 games natively on PC without requiring any emulation. The official GitHub page describes it as a tool for automatically porting executables to Linux and Windows.
Since the PS5 is powered by AMD’s Zen 2 hardware, it uses the x86-64 architecture, thus allowing AnyPS5 to work as a compatibility translation layer somewhat similar to Proton or Wine. It takes the original PS5 executable files and converts them into Windows and Linux formats, while the shaders are recompiled into SPIR-V for Vulkan. By avoiding emulation altogether, AnyPS5 can potentially eliminate some of the computational overhead, resulting in a smoother experience.
That said, development is still in the early stages, with test builds reportedly managing to boot select games as far as the main menu, with some even reaching gameplay with audio. However, the tool still requires a lot of work, particularly when it comes to system libraries and proprietary APIs unique to the PlayStation 5’s operating system. While the project is currently active on GitHub, a working public build is yet to be released.
Latest Videos FromTom’s Hardware
PS5 emulation is still far from being practical, but if you are interested or have been following the topic, make sure to check out KytyPS5, one of the most prominent and popular PlayStation 5 emulators in active development. It recently managed to run the PS5 version of Grand Theft Auto V as high as 40 to 60 FPS using high-end PC hardware. Unfortunately, this one is also in an early experimental phase and is not stable enough to maintain playable frame rates.
There is also SharpEmu, which focuses on accuracy and the underlying infrastructure needed to improve compatibility with PS5 games. The project has managed to successfully run Astro’s Playroom on a Steam Deck and was shown booting on the handheld at around 0.5 to 1 FPS. The game remains unplayable at those frame rates and does not progress beyond the introductory sequence, but getting PS5 emulation to run on the Steam Deck represents a major step forward for the project.
Microsoft launched the Copilot+ brand in 2024 to differentiate its offerings with local AI capabilities from other devices that do not have NPUs or fail to meet the minimum 40+ TOPS requirement. But a little over two years after the disastrous launch and poor sales, the company has quietly moved away from the brand. Microsoft Surface CVP Brett Ostrum confirmed this toWindows Central in an interview on the sidelines of Snapdragon Summit 2026.
Go deeper with TH Premium: AI and data centers
(Image credit: Microsoft)
“These are not called Copilot+ PCs,” Ostrum said. “They do meet all the requirements of our previous bar for what Copilot+ devices are. We still lean into the narrative around AI on the edge and being able to have a hybrid out there.” Even other devices that would’ve been eligible under the branding, including the upcoming Nvidia RTX Spark N1X and Microsoft’s own Project Zenith, are missing the Copilot+ badges. The company also said that it’s scaling back Copilot’s presence in Windows 11, promising to improve the operating system’s performance, reliability, and updates. Microsoft also vowed to optimize it to run smoothly on just 8GB of RAM amid the AI-driven memory shortage.
However, that doesn’t mean Microsoft is backing away completely from Copilot. Despite dropping the hardware branding, it’s still working on numerous AI features for its product lineup. The company recently revamped Copilot and added new features like Home, Code, and Autopilot. It has also tested Copilot+ AI features that run on discrete GPUs, possibly expanding the feature set to PCs that use processors without a built-in NPU. In fact, one AI consultant mapped out Microsoft Copilot products and discovered more than 80 unique offerings, with many more possibly unaccounted for.
Latest Videos FromTom’s Hardware
Microsoft did not give a specific reason for dropping the branding, but as many modern mainstream and high-end processors now come with NPUs that can hit 40 TOPS or more, it seems that it no longer makes sense to offer that differentiation. Aside from that, the current memory chip shortage has made the minimum 16GB requirement rather expensive, with Microsoft even purging its 32GB RAM recommendation from its website and releasing brand-new Surface for Business laptops with as little as 8GB of RAM.
It’s still unclear if the quiet sunset of the Copilot+ brand means that every Windows 11 PC will now get all the AI features that were once locked behind the name. But even if it will no longer exist on the hardware side, we still expect the Copilot branding to stay inside Windows and other Microsoft products for better or for worse.
The Russian military has started striking data centers within Ukraine in its latest escalation against the country. According to the BBC, a Russian drone hit the data center, which was situated in a central business district in Kyiv, damaging it and killing four people in the area. Multiple reports also suggest that other data centers and network infrastructure have been hit across the country as well, with President Volodymyr Zelensky saying that Moscow wants to maximize disruption for the common people.
“All this affects people’s ability to stay connected, study, work,” Zelensky said. He also wrote in a social media post, “The Russians are constantly expanding their escalation operation: American and other businesses, data centers, internet providers — for Russia, all ordinary life is simply a target. It is important that the world responds to all of this and that Russia feels the response to its terror.” Russia, on the other hand, confirmed that it attacked the Datagroup site, claiming that Ukrainian military intelligence used it, and that it’s also targeting other data and telecommunications centers owned by New-Telco, United DC, Kyivstar, and Parkovyi.
These attacks have left about 100,000 households in Ukraine without internet connectivity, and some companies have begun migrating their data across the border to avoid disruption. Nevertheless, one government official said that Ukraine’s internet network is highly decentralized, but people should still expect some local disruptions. However, Ukraine Foreign Minister Andrii Sybiha said that Russia’s targeting of data centers would affect the flow of “life-saving information.” “Rapid alerts about missile and drone threats are essential — they save lives. This is crucial civilian infrastructure… It is not a military target,” Sybiha said. “It is essential to keeping everyday life functioning.”
While gamers wait for the release of Grand Theft Auto VI, whether on their consoles or gaming PCs with one of the best graphics cards, they can relive one of the franchise’s most memorable titles. Modder gebdag has unveiled GTA2 RTX Remix, a mod that harnesses Nvidia’s RTX Remix technology to add full path tracing and dynamic time-of-day cycles to the 27-year-old classic, Grand Theft Auto 2.
Go deeper with TH Premium: GPUs
(Image credit: Noctua)
Grand Theft Auto 2, released in 1999, used older graphics APIs such as DirectDraw and an earlier version of Direct3D. The problem is that these legacy APIs create a technical barrier to modern technologies like RTX Remix. Nvidia’s platform lacks native support for DirectDraw and only plays well with Direct3D 9. It requires a wrapper to interface with previous versions of Direct3D, which is where the GTA2 RTX Remix comes in.
Modder gebdag ingeniously developed a custom Direct3D 9 renderer for Grand Theft Auto 2. It acts as a generational bridge between the original game and Nvidia’s RTX Remix. To do so, the mod relies on two DLLs (gta2dx9.dll and gta2dx9_vid.dll) that serve as translators, converting the game’s legacy graphics pipeline into a modern Direct3D 9 renderer that RTX Remix can understand. As a result, RTX Remix can interpret Grand Theft Auto 2’s visuals in real time and do its magic.
Latest Videos FromTom’s Hardware
As you can see in the embedded video below, the mod preserves the core gameplay fans will appreciate while adding a dramatic graphic overhaul to the classic top-down action title. Anywhere City now looks like a living metropolis, with realistic lighting and shadow effects throughout. The dynamic day-night cycle is a great addition that lets you see the sun set and the moon rise. Things like explosions, vehicle fires, or neon signs now have their own light sources with glows and reflections. Press F4 to open the menu and tweak the lighting to your taste.
Here comes a GTA2 RTX Remix mod with full path tracing + dynamic time of day! From modder gabdeg 🤯 pic.twitter.com/dZ0eMyIfmsSeptember 25, 2026
The graphics revamp is just one of the mod’s improvements. Beyond the eye-candy upgrade, it also adds other features to bring Grand Theft Auto 2 up to modern standards. While the original was limited to a 4:3 native aspect ratio, the mod supports widescreen 16:9 displays, giving gamers a more immersive view of Anywhere City. The mod also improves gameplay by adding support for frame rates up to 60 FPS through Frame Generation technology. It also fixes the nuisance of cars and pedestrians popping in at the screen edges.
GTA2 RTX Remix is an open-source project available on GitHub. To use the mod, you will need a copy of Grand Theft Auto 2. The modder tested it specifically with version 9.6, which Rockstar Games distributed for free in 2004. You will also need a 64-bit version of Windows 10 or Windows 11, along with the required RTX Remix runtime files.
RTX Remix has opened the floodgates to a plethora of mods from studios and the modding community that transform some of the most beloved games from our childhood. Now, Grand Theft Auto 2 has joined this notable list, which also includes Need For Speed Carbon, Tomb Raider, Max Payne, and The Elder Scrolls III: Morrowind, to name just a few.
Get Tom’s Hardware’s best news and in-depth reviews, straight to your inbox.
Our expert reviewers spend hours testing and comparing products and services so you can choose the best for you. Find out more about how we test.
The good news is that Sandisk is still putting out SSDs for consumers, and these are not just cheap replacements. The Optimus GX Pro 850P is a bona fide high-end Gen 4 SSD that puts Sandisk’s name on theWD Black SN850X’s legendary design. This means DRAM, TLC flash, and good performance for any role.
Like theBlack SN850P, which is a Black SN850X made with a heatsink and for the PS5, the 850P is made for the console. Its heatsink is excellent, and it would fit right into your desktop PC, too. We wouldn’t recommend shucking it for use in a laptop, but theoretically it could work there as well. Our main concern would be the pricing, which is a bit steep – this is a nice drive, but you’ll pay a premium for it.
The Sandisk Optimus GX Pro 850P is available at the same capacities as itsWD Black SN850P forebear: 1TB, 2TB, 4TB, and 8TB. This also matches the originalWD Black SN850X and its later8TB SKU. Prices at the time of review for the Optimus GX Pro 850P are $309.99, $609.99, $1,139.99, and $2,249.99. These prices are well above the $234.99, $494.99, and $1,019.00 that the Black SN850P is currently demanding. We think the prices will come down on the Sandisk drive, but be aware of the comparable alternatives when making a purchase.
Latest Videos FromTom’s Hardware
The drive can reach up to 7,300 / 6,600 MB/s for sequential reads and writes and up to 1.2 million random read/write IOPS. This performance is at the top for a Gen 4 drive and, frankly, is capable of delivering an excellent experience even for heavier workloads. The drive’s warranty is standard at five years with 600TB of writes for every TB of capacity. The Optimus GX Pro 850P supports data-at-rest encryption through TCG Opal v2.01, but this is not explicitly listed for the SN850P, likely as this model is designed for use in the PS5.
Sandisk Optimus GX Pro 850P Software and Accessories
Sandiskfully supports its drives with downloadable software. For the Optimus GX Pro 850P this includes the Sandisk Dashboard, which is Sandisk’s version of WD’s Dashboard, and Acronis True Image for Sandisk. The former is an SSD toolbox with all of the expected features: drive health and SMART monitoring, firmware updating, performance testing, etc. Acronis True Image is OEM-licensed software used for backing up files and imaging drives.
Sandisk Optimus GX Pro 850P: A Closer Look
Image 1 of 2
(Image credit: Tom’s Hardware)
(Image credit: Tom’s Hardware)
The Optimus GX Pro 850P is designed to specifically fit into the PS5 with an effective heatsink. The bare drive underneath is single-sided at 1TB and 2TB but double-sided at 4TB and 8TB. For that reason, the rated height is larger for the latter two SKUs. In general, this should not be an issue for fit in desktop PCs. The heatsink itself is finned for extra surface area for better heat dissipation and should keep the drive cool even under harsh conditions. It does not have RGB, unlike the WD Black SN850’s heatsinked SKU.
The drive label gives some additional information, such as a rating for 3.3V / 2.8A. This is ~9W, which matches the drive’s highest reported power state. This is certainly higher than the ~5W we see for many DRAM-less drives, which have half the flash channels on top of not having DRAM to worry about, but it is not enough to be a problem with this heatsink.
Get Tom’s Hardware’s best news and in-depth reviews, straight to your inbox.
Image 1 of 3
(Image credit: Tom’s Hardware)
(Image credit: Tom’s Hardware)
(Image credit: Tom’s Hardware)
The Optimus GX Pro 850P has a relatively standard SSD layout with an SSD controller, a DRAM module, power management circuitry, and two NAND flash packages. The larger, double-sided versions of the drive will have more NAND flash packages, with two additional packages on the back. We’ve covered the controller and its iterations in the past. The DRAM, for what it’s worth, is DDR4, which is serviceable for the role volatile memory plays on an SSD: mostly as a metadata cache. The one caveat here is that our 2TB drive only has 1GB of DRAM, which is one-half the usual and likewise one-half of what we had on our 2TB WD Black SN850X sample. In practice, this should have no impact on intended workload performance and could potentially reduce cost a small amount.
The flash is potentially more interesting, as it could be the expected BiCS5, or it could have been updated. The8TB WD Black SN850X uses BiCS6, and it can make sense to cycle in newer flash later on in a product’s lifetime. The Optimus GX Pro 850P is essentially in that line. What we can say for sure is that the flash on here was manufactured more recently than the flash on the Black SN850X samples we’ve reviewed at both capacities – newer in production date, not necessarily in generation – but the performance is close enough that there would be little perceptible difference even if it were generationally different. This drive is at the edge of what you can get out of this controller technology and Sandisk could also choose to tune it specifically for this performance range, so at the end of the day one could say this is, effectively, equivalent to the WD Black SN850X with BiCS5 TLC flash.
We believe our sample is still using BiCS5 anyway – the coding’s sixth character “C” implies that – but that statement applies to any potential BiCS6 swap. In this case, moving to newer BiCS5 means a higher I/O rate, 1,200 MT/s to 1,600 MT/s. The former is plenty to saturate a PCIe 4.0 link with an eight-channel controller like the 850P’s, but putting in newer flash makes sense from a supply perspective. Higher MT/s flash can have lower latency and potentially better power efficiency when run at a lower MT/s, but we don’t believe either is a motivation here. Real-world impacts are, again, minimal, with minor exceptions that might not exist if we were comparing to a current Black SN850X.