Full NVIDIA AD102 GPU Allegedly Runs Control At Over 160 FPS In 4K

A recent leak about a full-fat NVIDIA AD102 Ada Lovelace GPU suggests that the top-tier GPU can and will run rings around the current king-of-the-hill, the GeForce RTX 3090. How fast is the card, you ask? Apparently, it can run Remedy’s Control at 4K resolution and above the 100 fps mark.

Now, to be clear, this rumour about the AD102 isn’t to be confused with the non-Ti variant of the supposed RTX 4090 that has been making its rounds online. By comparison, the full-fat AD102 GPU ships out with as many as 18432 CUDA cores, which is approximately 12% more CUDA cores than what the RTX 4090 is expected to ship out with.

Getting back on point, and according to a Twitter user that goes by the handle AGF (@XpeaGPU), the AD102 GPU was able to run the game Control at 4K resolution, with the framerates sustaining themselves beyond the 160 fps mark. Further, those framerates were also achieved with ray-tracing and DLSS enabled, but not without its own set of drawbacks. In order to achieves those numbers, the GPU was also reported to have pulled more than 800W off the wall.

By comparison, the RTX 3090 at the same resolution and ray-tracing parametres turned on would yield approximately half the number of frames. With our RTX 3090 Founders Edition, our framerate in-game hovered between 60 and 74 fps.

The arrival of this AD102 rumour is interesting, given that just yesterday, we posted a report about another rumour regarding the RTX 4090 and how it is expected to be nearly 70% faster than the RTX 3090. However, those performance numbers were supposedly pulled off an as-of-yet unconfirmed 3DMark Time Spy Extreme benchmark and as such, should be treated with the usual consumption of salt.

For another matter, there are other missing variable that can render claims about the AD102’s prowess unreliable at this point. For starters, the source doesn’t specify the exact game settings the GPU was running Control with, nor did they say if the Lovelace GPU is actually using a new generation of Tensor Cores. Much like Ampere and its 2nd generation Tensor Cores.

Second, there aren’t any additional details about the AD102’s specs, not even its TDP or clockspeeds.

ARM Announces New Immortalis Mobile GPU With Ray-Tracing Support

Arm recently announced its latest flagship GPU, the Immortalis-G715. The new mobile GPU is the first of its kind to implement hardware-based ray tracing for mobile.

The Immortalis-G715 signals ARM’s shift in bringing its products into the mainstream, in the same way that Sony and Microsoft brought ray-tracing to their respective consoles. Specs-wise, the new mobile GPU is built on top of Mali and designed to accommodate between 10 and 16-cores, with a promised performance boost of up to 15% over previous generation Mali GPUs. This is alongside double the architectural machine learning (ML) improvements. 

With the presence of hardware accelerated ray-tracing, Andy Craigen, director of Product Management at ARM. says that the Immortalis-G715 will use approximately 4% of power per the shader core area while delivering more than 300% performance improvement. The trade-off here is that the Immortalis can and will use significant power, energy, and area within the SoC it is attached to.

The company has not provided any more particular details on what devices the Immortalis-G715 will be featured in, but you may expect that the first phones that have them will arrive at the beginning of 2023. 

Moving on, Arm also announced the Mali-G715 and Mali-G615 as premium GPUs, which are aimed at devices with lesser pricing, but with fewer cores and no ray tracing support. The Mali-G615 has 6 cores or less while the Mali-G715 has 7 to 9 cores respectively, but unlike the Immortalis-G715, they will only have Variable Rate Shading that provides up to 40% on frames per second (FPS) and Execution Engine.

ARM also gave a glimpse of its roadmap, which said that it plans on following up Immortalis other flagship mobile GPU next year, codenamed “Titan”, while it plans on bringing “Krake” in 2024. The chipmaker did not indicate if it intends to extend the same hardware-based ray-tracing to the other two.

Intel Goes On Hiring Spree Of Game Developers For ARC Alchemist GPU

Intel appears to have jumped head-on into a hiring frenzy of late. The chipmaker has reportedly managed to fill four positions within the company, with each new member having a background in game development, in one form or another.

It should be noted that Intel’s new hires are being brought in to work on “gaming and graphics”, making it obvious that the company is recruiting hard for the development of its ARC Alchemist GPU. As mentioned, the list of individuals who have a gaming background but more specifically, they made their bones over at companies like Electronic Arts and AMD.

First of is Andre Bremer, who joins Intel as the Vice President (VP) and General Manager (GM) of Gaming and Graphics Workload Engineering, and previously worked EA, LucasArts, and Amazon. Next on the list is Michael Heilemman, who comes in as the Senior Director of Game Developer Tools and Technologies.

The third hire is Ritche Corpus, who now heads Intel’s Game Ecosystem Business Development and Developer Relations as Vice President and General manager. Prior to his latest appointment, Corpus was a member of the PC Gaming Alliance, working as its Treasurer.

Last on the list is Steve Bell, who joins Intel directly from AMD as the former’s new Senior Director of Gaming Developer Relations. Bell reportedly spent more than 13 years at AMD, working in exactly the same capacity as his current position.

NVIDIA GeForce RTX 3090 Super GPU With 21Gbps GDDR6X Memory Reportedly In The Works

As NVIDIA reportedly preps its next-generation GPU architecture, Ada Lovelace, rumours have surfaced that the GPU brand plans on sending off its current Ampere lineup with one last hurray. That hurray, by the way, is rumoured to come in the form of a GeForce RTX 3090 Super.

Rumours of the RTX 3090 Super first came to light via prominent leakster Greymon55 (@greymon55), who said that the card was going to ship out with 10753 CUDA cores, which is 256 CUDA cores more than the current RTX 3090. Additionally, they also said that the card would require upwards of 400W of power to run, which clearly exceeds the current 350W TGP of the RTX 3090.

Less than a day after that rumour, another popular leakster, kopite7kimi (@kopite7kimi), posted more information about the RTX 3090 Super, saying that the card TGP was higher at 450W. Further, the card also would not have an NVLink unlike the RTX 3090, because SLI is pretty much dead at this point. On that note, it doesn’t appear that the card’s 24GB GDDR6X memory is getting increased, but until its launch, that is still up for change.

Another interesting plot point is that the RTX 3090 Super is expected to run on a new GA102-350-A1 architecture, along with GDDR6X memory running at 21Gbps. Again, for context, the GDDR6X on the currently existing RTX 3090 run at 19.5Gbps, which is already extremely fast. Even more interesting, though, is that kopite7kimi says that NVIDIA plans on launching the RTX 3090 Super before the year is out.

As exciting as this news may sound to us, it should also be noted that these rumours do not mention or affirm if NVIDIA also plans on releasing other Super variants of other RTX 30 series GPUs, including the RTX 3060, RTX 3070, and RTX 3080. That said, the chances of that happening for the latter two models are exceedingly slim, seeing how the company just recently launched the “Ti” versions of said cards.

If these rumours actually do become reality, it is also likely that NVIDIA’s AiB partners will have no choice but to fit their own custom-cooled versions of the GPU with no less than three 8-PIN PCIe power ports in order to run the card. That, or adopt the 12-pin power connector that NVIDIA uses on its Founders Edition cards.

On that note and as always, NVIDIA has neither confirmed nor denied the existence of such a card, as is the brand’s policy. So, take this news with a liberal sprinkling of salt.

Alleged MSI Delta 15 Laptop With AMD Radeon RX 6700M GPU Leaks

Leaked information of an MSI Delta 15 with an AMD Radeon RX 6700M discrete GPU have reportedly made their way on to the internet. The leaks first appeared on the Chinese news site, Zhihu and seems to detail the synthetic performance of the AMD discrete GPU, as well as its expected specifications.

As per Videocardz, the 6700M will be a water-down version of AMD’s Navi 22 GPU, have 2304 Stream Processors, 10 GB GDDR6 running through a 160-bit memory bus, and possibly be available in three TGP variants: 80W, 110W, and 135W. In this instance, the GPU in question seems to be the 110W version. Additionally, the GPU seems to be paired with an AMD Ryzen 9 5900HX.

Moving on, the Zhihu post also seems to be benching the 6700M-powered Delta 15 against a Razer Blade 14 Ryzen, fitted with an 80W NVIDIA GeForce RTX 3070 GPU. Overall, the Radeon GPU appears to outperform its Team Green rival by an average of 9%, with the highest gains in performance seen with Hitman 2.

Of course, one important note to bear in mind about the real-world benchmarks is that only a handful of these games are DXR-capable and judging from the framerates, it is likely that these numbers were obtained without the feature turned on. As such, it is unclear how the 6700M will perform with the feature turned on.

As these things usually are, MSI has not released any official statement about the Delta 15, nor has AMD actively confirmed the existence of the Radeon RX 6700M GPU. So, take this news with a grain of salt.

NVIDIA To Cease GPU Driver Support For Windows 7 And 8 By October 2021

Despite being several years old, NVIDIA is still releasing GPU driver updates for older Microsoft operating systems including Windows 7, 8 and 8.1. However, this will no longer be the case once October 2021 rolls in, as announced by the company via a newly published support page on its website.

“Effective October 2021, Game Ready Driver upgrades, including performance enhancements, new features, and bug fixes, will be exclusively available for systems utilizing Windows 10 as their operating system,” NVIDIA wrote. “Critical security updates will be available on systems utilizing Windows 7, Windows 8, or Windows 8.1 through September 2024.”

Seeing that current games and applications are only compatible with the current generation Windows systems, NVIDIA’s (and even Microsoft’s) decision to cease support for the older platforms shouldn’t be an issue to most users. That is, provided that they have already upgraded to the popular Windows 10 system, which will soon hit its six-year-old mark late next month.

AMD Radeon RX 6000M Series Mobile GPU Now Official; Features Up To 12GB GDDR6 Memory

Beside its latest Zen 3-powered Ryzen APUs, today’s keynote also saw the official announcement of AMD’s new Radeon RX 6000M mobile graphics series for laptops. Much like its desktop counterparts, the new mobile GPU series are based on the brand’s current 7nm, RDNA2 architectures.

In total, the RX 6000M comprises three different GPUs; sitting pretty at the top is the RX 6800M, while the RX 6700M dominates the middle ground, and last but not least, we have the RX 6600M covering base for the entry-level.

As you’ve no doubt guess, the RX 6800M is the most powerful of the trio, sporting 12GB GDDR6 graphics memory, 40 Compute Units (CUs) and Ray Accelerators, a game clock of 2.3GHz at full whack, and a 96MB Infinity Cache. Again, as it is an AMD product , the GPU also supports both Smart Access Memory (SAM) and Smartshift when paired with an AMD Ryzen 5000 series mobile CPU.

In terms of raw gaming performance, the 6800M isn’t quite set up to dominate the 4K realm, but to AMD’s credit, it is setting realistic expectations, aiming for max graphics settings at 1440p with this mobile GPU, which is more than plenty for us.

As to which laptop will be the first to showcase the 6800M, that honour goes to ASUS’ ROG Strix G15 “AAA” Edition gaming laptop. In addition to the 6800M, the laptop is also fitted with AMD’s Rzyen 9 5900HX CPU, both of which are also kept cooled through the use of Liquid Metal and vapour chamber technology.

Moving on, the RX 6700M serves as the halfway mark of the three mobile GPUs. Specs-wise, the mobile GPU has 10GB GDDR6 memory, 36 CUs and Ray Accelerators, an 80MB Infinity Cache, and a 2.3GHz game clock. Whereby the 6800M is designed to enable gamers to play their favourite triple-A title at 1440p resolution at their highest graphics settings, the 6700M unsurprisingly takes it down a notch. By allowing gamers to still game at 1440p, but at either the second or third highest graphics settings of each respective games.

Last on the list is the entry-level RX 6600M. On paper, this mobile GPU ships out with 8GB GDDR6 memory, 28 CUs and Ray Accelerators, a 32MB Infinity Cache, and a game clock of 2.18GHz. Whereas the 6800M is designed to enable laptops to maintain high framerates at 1440p and with the graphics settings maxed out, the 6600M is targeted at the gamer that is perfectly content with gaming at Full HD resolution. While still maintaining max graphical settings for their favourite title.

The AMD Radeon RX 6800M and RX 6600M are available starting today. While the 6800M is available via the ASUS Strix G15 “AAA” Edition, the RX 6600M will first be available via the HP Omen 16. At the time of writing, there is still no word regarding the availability of the 6700M.

Intel Confirms Specs For Xe DG2 GPU; Will Have Up To 16GB GDDR6 And 512EU

Intel recently confirmed the specifications of its Xe DG2 GPUs by furnishing its official, repository website with new details about its discrete graphics for laptops. According to Videocardz, details from the repository is usually restricted to individuals and entities that are under NDA, yet for some reason, the information was easily found online.

That aside, the repository initially lists down a total of five SKUs for the Xe DG2. Firstly, the variants will come with 512 Execution Units (EU), 384EU, and 128EU. Each EU will have eight shading units, so the total number of shaders for the 512EU comes up to 4096 shading units, or ALUs. The odd thing is that the same source also lists down two other EU variants, which are 256EU and 96EU.

Additionally, there is reason to believe that Intel is preparing four memory configurations for the Xe DG2: 4GB, 6GB, 8GB, and 16GB, the latter obviously being the flagship SKU. Moreover, the laptop GPUs are expected to support memory speeds between 14Gbps and 18Gbps, which is also running on the GDDR6 memory format. As promised earlier by Intel.

A lot of this information seems to align with earlier reports of the Xe DG2; back in January this year, leaks suggested that one of Intel’s upcoming Xe-based laptop GPU would come with 8GB GDDR6 graphics memory, as well as 512EU. Moreover, the GPU was supposed to be paired with Intel’s Tiger Lake-H series, which is the chipmaker’s high performance lineup of mobile CPUs.

As for when we can expect to see this GPUs, it is rumoured that we could see them sometime during the second quarter of this year. Intel has yet to make an official announcement, though, so take this bit of news with a grain of salt.

GPU Miners Bypass NVIDIA GeForce RTX 3060 Hashrate Limiter With Cheap HDMI Dummy Plug

Back in February, NVIDIA made two announcements in relation to the renewed cryptomining craze. The first was the launch of its Cryptocurrency Mining Processor (CMP) GPU, made specifically for cryptomining. The second was that it was limiting the hashrate of Ethereum mining for the RTX 3060, a pledge that has already been thwarted by GPU miners.

Unfortunately for NVIDIA, its anti-mining algorithm for the RTX 3060 are still actively being rubbished by GPU miners, with their most workaround the GPU brand’s Ethereum mining limiter being an HDMI dummy plug that can easily be procured from online stores.

For the uninitiated, NVIDIA’s anti-mining algorithm is designed to nerf the hashrate for Ethereum mining for all RTX 3060 cards. However, and for reasons still unknown, the company later released a GeForce beta driver that disabled the limiter within the algorithm, albeit still with some restrictions in place, in the form of parameters.

First, the motherboard that holds the RTX 3060 must have a PCIe 3.0 x8 interface as a minimum, which is pretty much any motherboard within the last six to seven years. The second parameter requires a monitor to be connected RTX 3060 via the onboard HDMI or DisplayPort output.

Bypassing the first parameter is clearly simple enough, and as we mentioned earlier, it didn’t take long for the GPU miners to realise that, while the driver is able to detect when a monitor is connected via the ports, it is unable to differentiate the difference between a real monitor or the dummy plug. To add insult to injury, the plugs themselves are cheap and go for around US$6 (~RM24) per unit.

As shown by one user from Quasar Zone, the workaround enabled him to run a four-card RTX 3060 setup, with each card able to mine Ethereum at an average of 48MH/s, bringing the total hashrate to 192MH/s. In terms of profitability, that’s a monthly average of US$562 (~RM2318).

GPU Mining Farms Are Now The Leading Cause Of Power Outages In Iran

Since the crypto mining frenzy that started in 2013, there are periods in time when the GPU miners have inconvenienced others in the lust for the digital coin. In Iran, the country has reportedly been investigating power outages caused by – you guessed it – GPU mining farms.

The punchline here is that all these mining farms aren’t locally owned, but are in fact, systems owned by a Chinese company named RHY. According to a report by El Chapuzas Informatico, the company shifted its cryptomining operations from the mainland into Iran. Why, you ask? Because in Iran, the cost of electricity consumption is four times lower than in China; Iran is one of the several countries that can boast having one of the lowest cost for electricity consumption at US$0.027 (~RM0.11) per kilowatt.

At the time of writing, Iran is home to no less than 14 large mining farms running continuously, consuming a total of 300 megawatts of power. For context, that’s the power consumption of a city with an average of 100000 people living in it.

Despite the power outages, Iranian authorities have said that the company that owns the mining farms weren’t doing anything wrong and had obtained all the necessary authorisation to operate the facilities in the area.