An unreleased NVIDIA GeForce RTX 3090 SUPER Founders Edition graphics card was last spotted just over a year ago. A fortunate member of the Chinese NGA discussion board provided a close-up shot of a shroud bearing "super." A new leak gives us a full view of the RTX 3090 SUPER FE with prominent branding—KittyYYuko declared: "WTF, I have indeed heard of this leak before" upon posting this discovery to social media.

According to ITHome, the example from last year appeared to be a publicly released variant of "an unpackaged GeForce RTX 3090 Ti," and the latest finding seems to be identical. A seller, tbNick_dn86z, has created an entry for his GeForce RTX 3090 SUPER Founders Edition card with a value of 9999 RMB (~$1370) on Xianyu (Taobao's second hand market)—it is advertised as being "original and not modified, with a pure black casing." When confronted about identifying any apparent differences between the SUPER and officially launched Ti version, tbNick_dn86z confirmed that they are largely the same (minus external branding)—a matching device ID is shared across both variants.

An unusual MSI RTX 3080 Ti SUPRIM X graphics card is up for sale, second hand, on Facebook Marketplace—the Sydney, Australia-based seller is advertising this component as a truly custom model with a non-standard allocation of VRAM: "Yes this is 20 GB not 12 GB." The used item is said to be in "good condition" with its product description elaborating on a bit of history: "There are some scuff marks from the previous owner, but the card works fine. It is an extremely rare collector's item, due to NVIDIA cancelling these variants a month before release. This is not an engineering sample card—this was a finished OEM product that got cancelled, unfortunately." The seller is seeking AU$1100 (~$740 USD), after a reduction from the original asking price of AU$1,300 (~$870 USD).

MSI and Gigabyte were reportedly on the verge of launching GeForce RTX 3080 Ti 20 GB variants two years ago, but NVIDIA had a change of heart (probably due to concerns about costs and production volumes) and decided to stick with a public release of the standard 12 GB GPU. Affected AIBs chose to not destroy their stock of 20 GB cards—these were instead sold to crypto miners and shady retailers. Wccftech points out that mining-oriented units have identifying marks on their I/O ports.

NVIDIA's H100 Hopper GPU is a device designed for pure AI and other compute workloads, with the least amount of consideration for gaming workloads that involve graphics processing. However, it is still interesting to see how this 30,000 USD GPU fairs in comparison to other gaming GPUs and whether it is even possible to run games on it. It turns out that it is technically feasible but not making much sense, as the Chinese YouTube channel Geekerwan notes. Based on the GH100 GPU SKU with 14,592 CUDA, the H100 PCIe version tested here can achieve 204.9 TeraFLOPS at FP16, 51.22 TeraFLOPS at FP32, and 25.61 TeraFLOPS at FP64, with its natural power laying in accelerating AI workloads.

However, how does it fare in gaming benchmarks? Not very well, as the testing shows. It scored 2681 points in 3DMark Time Spy, which is lower than AMD's integrated Radeon 680M, which managed to score 2710 points. Interestingly, the GH100 has only 24 ROPs (render output units), while the gaming-oriented GA102 (highest-end gaming GPU SKU) has 112 ROPs. This is self-explanatory and provides a clear picture as to why the H100 GPU is used for computing only. Since it doesn't have any display outputs, the system needed another regular GPU to provide the picture, while the computation happened on the H100 GPU.

TechPowerUp today released the latest version of TechPowerUp GPU-Z, the popular PC graphics information, monitoring, and diagnostics utility. Version 2.52.0 adds support for AMD Radeon RX 7900 XTX, RX 7900 XT, RX 6300 OEM; NVIDIA GeForce RTX 4070 Ti, and a few rare "Ampere" based GPUs in circulation these days, including the RTX 3080 Ti 20 GB, RTX 3070 Ti based on GA102 silicon, RTX 3050 based on GA107, and the PCIe AIC version of the A800 80 GB accelerator. Detection is improved for the Xe LP-based iGPU of Intel Core "Raptor Lake" processors. NVIDIA GPUs with ECC memory now have ECC status reported in the Advanced panel. On GPUs where the boost frequency can't be read, the base frequency will be used to calculate fillrates. Clock speed detection for Intel Arc "Alchemist" GPUs has been improved. Vendor detection has been added for several new graphics card brands such as Corsair (gaming notebooks), Maxsun, and Wingtech.

DOWNLOAD: TechPowerUp GPU-Z v2.52.0

NVIDIA manufactured a heap of large "GA102" Ampere silicon to cater to demand from the crypto-mining boom; only to see that demand vanish. With next-gen RTX 40-series awaiting ramp; the company has to digest these GA102 chips somehow, and is apparently letting its partners use them on performance-segment SKUs such as the GeForce RTX 3070 Ti. The RTX 3070 Ti is normally based on the GA104 silicon, which it maxes out, enabling all 6,144 CUDA cores, 48 RT cores, 192 Tensor cores, 192 TMUs, and 96 ROPs, besides the chip's full 256-bit wide GDDR6X memory interface. This SKU is now being carved out on the larger GA102, by enabling 48 out of 84 streaming multiprocessors (just 57% of the CUDA cores enabled); and narrowing the memory bus from its normal 384-bit, down to 256-bit.

The memory size remains at 8 GB, memory type at GDDR6X, and memory speed at 19 Gbps, working out to 608 GB/s of bandwidth. The most interesting aspect of carving the RTX 3070 Ti out of the GA102 has to be board power; with a ZOTAC-branded card listing it at 320 W, higher than the 290 W of GA104-based cards from the company. Sadly, this is a China-only SKU. Every custom-design graphics card, especially from a reputed AIC such as ZOTAC, has to go through qualification with NVIDIA; which means NVIDIA is not only aware of GA102-based RTX 3070 Ti cards, but is behind fusing the SMs to carve out the SKU, and developing the video BIOS and driver support. ZOTAC is kind enough to list the ASIC code on its website, and for this SKU it is "GA102-150-xx."

With NVIDIA cracking the whip on the supply chain to pave the way for the next-generation RTX 40-series, we're seeing some very oddball graphics card hit the market as official/retail products. One such card in China is the MSI GeForce RTX 3080 Ventus 3X with 20 GB of memory. The card is neither based on the GA102-200 ASIC powering the original RTX 3080 with 10 GB of memory, nor is it based on the GA102-220 powering the RTX 3080 12 GB, but rather the "GA102-250," or at least the board description sticker says so. The card features 20 GB of memory across a 320-bit wid memory bus. Its core-config remains unknown.

The SKU positioned above the RTX 3080 12 GB, the RTX 3080 Ti, is powered by the GA102-225. VideoCardz discovered that the card lacks the LHR hardware-limitation for blockchain hashrates, which could be evidence that MSI had a special stock of graphics cards that it was directly selling to miners even while LHR-enabled cards were being pushed to the retail market. The card yields a hash-rate of 97.48 MH/s in the ETHASH test, proving the lack of the LHR block. The card also features display connectors. The leaker behind pictures of this card says that there may be as many as a 100 of these cards in circulation, priced around RMB 3xxx (USD $432 to $600).

This is probably the only picture of a GeForce RTX 3090 SUPER Founders Edition graphics card. NVIDIA allegedly decided against this branding, in favor of the RTX 3090 Ti, to designate its maxed-out GA102-based graphics card. With no other "SUPER" SKUs in the RTX 30-series and plenty of "Ti," the company probably thought it wasn't worth the trouble to leave the odd SUPER SKU sticking out at the top. The picture surfaced on the NGA Forums, showing the card with prominent "RTX 3090 SUPER" branding etched along the top frame of the cooler. This card has the same device ID as the RTX 3090 Ti, so GPU-Z detects it as such. It also has identical specs to an RTX 3090 Ti Founders Edition.

NVIDIA has reportedly resumed production of its GA102-based RTX 3080 12 GB graphics cards, according to a Tweet from GPU leaker Zed__Wang. The reason cited has to do with oversupply of the company's GA102 chips, which powers the company's high-end lineup from the RTX 3080 through the RTX 3090 Ti (in all, there are six RTX 30-series cards powered by this chip, alongside the CMP 90HX mining-specific card, datacenter and AI inferencing accelerators A10, A10G, and A40, as well as the company's RTX A4500, A5000, A5500, and A6000 series for a total of 14 SKUs).

Oversupply, in this case, has more to do with contracting demand - not only is NVIDIA's next-gen RTX 40-series right around the corner, but the already-announced death of Ethereum's Proof of Work mining has flooded the market with second-hand RTX 30-series cards. This, alongside the already long-winded shelf-life of the RTX 30-series - which hit the market back in September 2020 - has led to contracting demand for NVIDIA's GPUs. Rampant inflation and general macroeconomic indicators also do little to instill confidence in the purchase of non-essential products.

NVIDIA's upcoming GeForce RTX 40 series Ada Lovelace graphics card lineup is slowly shaping up to be a significant performance uplift compared to the previous generation. Today, according to a well-known hardware leaker kopite7kimi, we are speculating that a mid-range AD104 SKU could match the performance of the last-generation flagship GeForce RTX 3090 Ti graphics card. The full AD104 SKU is set to feature 7680 FP32 CUDA cores, paired with 12 GB of 21 Gbps GDDR6X memory running on a 192-bit bus. Coming with a large TGP of 400 Watts, it should have a performance of the GA102-350-A1 SKU found in GeForce RTX 3090 Ti.

Regarding naming this complete AD104 SKU, it should end up as a GeForce RTX 4070 Ti model. Of course, we must wait and see what NVIDIA decides to do with the lineup and what the final models will look like.

ASUS looks set to release a second graphics card featuring a Noctua cooling solution nearly a year after the two companies announced the ASUS x Noctua GeForce RTX 3070. This new card features the GA102-based 10 GB version of the RTX 3080 instead of the newer 12 GB version launched in January. The quad-slot cooling solution present on the card is very similar to the existing configuration with dual 120 mm Noctua fans and a slightly modified backplate. The card features dual 8-pin power connectors in addition to non-standard display outputs with 3x DisplayPort 1.4 and 2x HDMI 2.1. The ASUS x Noctua GeForce RTX 3080 is expected to launch soon and retail for 1150 EUR which is 250 EUR more then the existing ASUS x Noctua GeForce RTX 3070.

Extreme overclocking is an enthusiast discipline where overclockers try to push their hardware to extreme limits. Combining powerful cooling solutions like liquid nitrogen (LN2), which reaches sub-zero temperatures alongside modified hardware, the silicon can output tremendous power. Today, we are witnessing a custom XOC (eXtreme OverClocking) BIOS for the NVIDIA GeForce RTX 3090 Ti graphics card that can push the GA102 SKU to impressive 890 Watts of power, representing almost a two-fold increase to the stock TDP. Enthusiasts pursuing large frequencies with their RTX 3090 Ti are likely users of this XOC BIOS. However, most likely, we will see GALAX HOF or EVGA KINGPIN cards with dual 16-pin power connectors utilize this.

As shown below, MEGAsizeGPU, the creator of this BIOS, managed to push his ASUS GeForce RTX 3090 Ti TUF with XOC BIOS to 615 Watts, so KINGPIN and HOF designs will have to be used to draw all the possible heat. The XOC BIOS was uploaded to our VGA BIOS database, however, caution is advised as this can break your graphics card.

With the HEATKILLER V eBC - active backplate for RTX 3080/3090 EVGA FTW3 graphics cards, the German-based manufacturer of high-quality liquid cooling presents its interpretation of an active backplate. The backplate's solid copper design is a unique selling point. The large surface area guarantees an excellent performance especially for the hot spot temperatures in the areas of the graphics card processor, VRAM and the voltage converters. The cooling channels were arranged to provide active cooling for all relevant components.

The connection terminal has been modified to connect the graphics card cooler directly to the backplate. Not only does this look cleaner, but it also allows a parallel inflow, which increases the flow rate while reducing the flow resistance. To achieve the best possible cooling performance, the flow ratio between the active backplate and the graphics card cooler was optimized during development by performing countless tests. The assembly is done in two simple steps. The active backplate and the graphics card cooler are mounted first and then both are connected by the new connection terminal.

NVIDIA released the RTX A5500 professional graphics card in the PCI-Express AIC form-factor, and the RTX A5500 Laptop GPU. Based on the "Ampere" graphics architecture, the RTX A5500 AIC features NVIDIA's biggest graphics silicon, the "GA102." It features 10,240 CUDA cores, 320 Tensor cores, 80 RT cores, and 24 GB of ECC GDDR6 memory across a 384-bit wide memory interface. The card uses some aggressive power management, with the typical board power rated at just 230 W. You also get an NVLink interface, and four DisplayPort 1.4 output connectors. The A5500 is being positioned a notch below the company's flagship pro-vis product, the RTX A6000, which comes with 48 GB of memory.

The RTX A5500 Laptop GPU is targeted at mobile workstations, and is based on the "GA103" silicon, with 7,424 CUDA cores, 232 Tensor cores, 58 RT cores, and 16 GB of ECC GDDR6 memory across a 256-bit wide memory interface. NVIDIA has given this product some Max-Q power and thermal optimization, with typical graphics power being rated in the range of 80 W to 160 W. This is NVIDIA's most powerful mobile pro-vis solution, being slotted higher than the RTX A4500. It's worth noting here, that both these products use ECC GDDR6 memory instead of GDDR6X.

A few days ago, we found out that NVIDIA corporation has been hacked and that attackers managed to steal around 1 TB of sensitive data from the company. This includes various kinds of files like GPU driver and GPU firmware source codes and something a bit more interesting. The LAPSUS$ hacking group responsible for the attack is now threatening to "help mining and gaming community" by releasing a bypass solution for the Lite Hash Rate (LHR) GPU hash rate limiter. As the group notes, the full LHR V2 workaround for anything between GA102-GA104 is on sale and is ready for further spreading.

Additionally, the hacking group is making blackmailing claims that the company should remove the LHR from its software or share details of the "hw folder," presumably a hardware folder with various confidential schematics and hardware information. NVIDIA did not respond to these claims and had no official statement regarding the situation other than acknowledging that they are investigating an incident.Update 01:01 UTC: The hackers have released part of their files to the public. It's a 18.8 GB RAR file, which uncompresses to over 400,000 (!) files occupying 75 GB, it's mostly source code.

When NVIDIA announced the appearance of the GeForce RTX 3080 Ti mobile graphics card, we were left with a desire to see just what the GA103 silicon powering the GPU looks like. And thanks to the Chinese YouTuber Geekerwan, we have the first pictures of the GPU. Pictured below is GA103S/GA103M SKU with GN20-E8-A1 labeling. It features 58 SMs that make up for 7424 CUDA cores in total. The number of Tensor cores for this SKU is set to 232, while there are 58 RT cores. NVIDIA has decided to pair this GPU with a 256-bit memory bus and 16 GB GDDR6 memory.

As it turns out, the full GA103 silicon has a total of 7680 CUDA cores and a 320-bit memory bus, so this mobile version is a slightly cut-down variant. It sits perfectly between GA104 and GA102 SKUs, providing a significant improvement to the core count of GA104 silicon. Power consumption of the GA103 SKU for GeForce RTX 3080 Ti mobile is set to a variable 80-150 Watt range, which can be adjusted according to the system's cooling capacity. An interesting thing to point out is a die size of 496 mm², which is a quarter more significant compared to GA104, for a quarter higher number of CUDA cores.

NVIDIA today sneaked in a major update to its high-end GeForce RTX-30 "Ampere" series with the new RTX 3080 12 GB. Based on the same 8 nm "GA102" silicon as the original RTX 3080 (10 GB), the RTX 3080 Ti, the RTX 3090, and the upcoming RTX 3090 Ti, this SKU maxes out the 384-bit wide GDDR6X memory interface of the silicon, giving it 12 GB of 19.5 Gbps GDDR6X memory, resulting in a memory bandwidth of 912 GB/s, compared to 760 GB/s of the RTX 3080 (10 GB), and its 320-bit memory bus.

Memory isn't the only upgrade, the RTX 3080 12 GB gets a few more CUDA cores. With 70 out of 84 streaming multiprocessors (SM) enabled, the GPU gets 8,960 CUDA cores. In comparison, the RTX 3080 (10 GB) has 68 SM and 8,704 CUDA cores. This results in a Tensor core and TMU count of 280, and RT core count of 70. NVIDIA is positioning this SKU in between the RTX 3080 and the RTX 3080 Ti, and real-world prices of the card can be as high as $1,700, if not higher. TechPowerUp has several RTX 3080 12 GB graphics cards, but our editor and graphics card reviewer, W1zzard, is on a much-needed skiing holiday in the Alps, since we got no heads-up on this launch, and no marketing materials to help us understand the product. Hopefully NVIDIA puts out a public GeForce driver update later today, and we'll use it to test the cards we have. Expect our reviews to go live next week.

During the CES 2022 keynote, we have witnessed NVIDIA update its GeForce RTX 30 series family with GeForce RTX 3050 and RTX 3090 Ti. However, this is not an end to NVIDIA's updates to the Ampere generation, as we now hear industry sources from Wccftech suggest that we could see a GeForce RTX 3080 GPU with 12 GB of GDDR6X VRAM enabled, launched as a separate product. Compared to the regular RTX 3080 that carries only 10 GB of GDDR6X, the new 12 GB version is supposed to bring a slight bump up to the specification list. The GA102-220 GPU SKU found inside the 12 GB variant will feature 70 SMs with 8960 CUDA, 70 RT cores, and 280 TMUs.

This represents a minor improvement over the regular GA102-200 silicon inside the 8 GB model. However, the significant difference is the memory organization. With the new 12 GB model, we have a 384-bit memory bus allowing GDDR6X modules to achieve a bandwidth of 912 GB/s, all while running at 19 Gbps speeds. The overall TDP will also receive a bump to 350 Watts, compared to 320 Watts of the regular RTX 3080 model. For more information regarding final clock speeds and pricing, we have to wait for the alleged launch date - January 11th.

NVIDIA GeForce RTX 3090 Ti is real, and coming to gamers miners within Q1 2022. The new SKU maxes out NVIDIA's largest silicon based on the GeForce "Ampere" graphics architecture, the GA102, and pairs it with even faster 21 Gbps GDDR6X memory; but these changes come at a significant impact on power, with the typical board power for a stock RTX 3090 Ti reportedly being rated at 450 W, compared to the 350 W value of the RTX 3090. For enthusiast-class custom-design boards such as the EVGA KINGPIN, this only means even more elaborate setups, as QuasarZone forums found out.

While the current RTX 3090 KINGPIN comes with three 8-pin PCIe power connectors, a configuration capable of delivering up to 525 W (including the PCIe slot), the new RTX 3090 Ti KINGPIN ropes in the small but mighty 12-pin Molex MicroFit 3.0 connector, and comes with not one, but two of them! NVIDIA debuted the 12-pin connector in the consumer space with its RTX 30-series Founders Edition graphics cards. The connector is capable of delivering 450 W of power. With two of these on the RTX 3090 Ti, you're looking at a maximum power input capability of 975 W! It's only now, that PSUs are shipping with 12-pin cables, and even the high-Wattage ones we've come across, only pack one such connector. The EVGA card could hence include several dual- or triple-8-pin to 12-pin adapter cables. Pictured below is the RTX 3090 KINGPIN.

In the past few months, we have heard rumors of NVIDIA launching an upgraded version of the GA102 silicon called GeForce RTX 3090 Ti. The upgraded version is supposed to max out the chip and bring additional performance to the table. According to anonymous sources of VideoCardz, MSI, one of NVIDIA's add-in board (AIB) partners, is preparing to update its SUPRIM X lineup of graphics cards with the MSI GeForce RTX 3090 Ti SUPRIM X GPU, scheduled for January 27th launch date. This suggests that the official NDA lifts for these RTX 3090 Ti GPUs on January 27th, meaning that we could see AIBs teasing their models very soon.

As a general reminder, the GeForce RTX 3090 Ti graphics card should use a GA102-350 silicon SKU with 84 SMs, 10752 CUDA cores, 336 TMUs, 24 GB of GDDR6X memory running on a 384-bit bus at 21 Gbps speed with 1008 GB/s bandwidth, and a TBP of a whopping 450 Watts. If these specifications remain valid, the GPU could become the top contender in the market, however, with a massive drawback of pulling nearly half a KiloWatt of power.

It looks like GeForce RTX 3090 Ti is indeed the name of the maxed out "Ampere" GA102 silicon, and NVIDIA did not go with "RTX 3090 SUPER" for its naming. A picture emerged of an ASUS TUF Gaming GeForce RTX 3090 Ti graphics card that confirms the naming. The board design of the card looks similar to the RTX 3090 TUF Gaming, except that the Axial-Tech fans have changed, with more blades on the impellers.

The GeForce RTX 3090 Ti is expected to max out the GA102 silicon, featuring all 10,752 CUDA cores, 84 RT cores, and 336 Tensor cores, physically present on the silicon. The memory size is unchanged over the RTX 3090, with 24 GB of GDDR6X memory. What's new is that NVIDIA is reportedly using faster 21 Gbps-rated memory chips, compared to 19.5 Gbps on the RTX 3090. The typical board power is rated at 450 W, compared to 350 W on the RTX 3090. NVIDIA is expected to announce this card at its January 4 press event along the sidelines of the 2022 International CES.

NVIDIA has reportedly pushed midlife refreshes to its GeForce RTX 30-series "Ampere" graphics card family to beyond January, according to a report by Igor's Lab. The company was supposed to launch as many as three high-end graphics card SKUs in early-Q1, which include a 16 GB variant of the RTX 3070 Ti, which maxes out the GA104 silicon, a refreshed RTX 3080 12 GB variant, which keeps the CUDA core count of the original RTX 3080, but maxes out the 384-bit memory bus width of the GA102; and the RTX 3090 Ti, which maxes out both memory and CUDA cores on the GA102.

It's being reported that while the RTX 3090 Ti launch is on schedule, with announcements slated for NVIDIA's 2022 International CES presentation; the RTX 3070 Ti 16 GB and RTX 3080 12 GB could be launched only after the Chinese New Year (at least mid-February). The three SKUs, we predict, are designed to calibrate NVIDIA's lineup against a possible refresh of AMD's RDNA2 graphics architecture on the TSMC N6 (6 nm) node, which could allow the company to dial up engine clocks across the board, along with faster 18 Gbps GDDR6 memory, to better compete against the RTX 30-series. AMD is expected to announce these 6 nm GPUs in its CES presentation.

NVIDIA's upcoming flagship graphics card, the GeForce RTX 3090 Ti (also known as RTX 3090 SUPER in some rumor circles); could feature the company's fastest memory solution for GeForce, according to a Uniko's Hardware report. The card maxes out not just the 384-bit bus width of the "GA102" silicon, but also uses Micron-sourced "MT61K512M32KPA-21U" GDDR6X memory chips, which are rated for 21 Gbps date-rates. This results in a roughly 7.7% increase in memory bandwidth over the RTX 3090, with a whopping 1008 GB/s on tap. The title of the highest memory bandwidth on a client-segment graphics card would still go to the Radeon VII, which uses 4096-bit HBM2 to achieve 1024 GB/s. The RTX 3090 Ti also maxes out the "GA102" silicon, by enabling all 84 streaming multiprocessors (SM).

NVIDIA has been preparing a new flagship graphics card with their RTX 3090 SUPER featuring a fully unlocked GA102 GPU. The RTX 3090 SUPER will include 10,752 CUDA cores, 336 Tensor cores, and 84 RT cores paired with 24 GB of GDDR6X memory on a 384-bit wide memory bus. These speed improvements are set to increase the cards TDP to 400 W compared to 350 W for the base RTX 3090. We have recently seen a new rumor from @hongxing2020 that claims NVIDIA will be launching the RTX 3090 SUPER, a refreshed RTX 3070 Ti with 16 GB memory, and the RTX 2060 12 GB in January 2022. The RTX 3090 SUPER and RTX 2060 12 GB launches have been expected for some time but this is the first mention we have seen of an RTX 3070 TI 16 GB.

NVIDIA is readying the GeForce RTX 3090 SUPER, the first "SUPER" series model from the RTX 30-series, following a recent round of "Ti" refreshes for its product stack. According to kopite7kimi and Greymon55, who each have a high strike-rate with NVIDIA rumors, the RTX 3090 SUPER could finally max-out the 8 nm "GA102" silicon on which nearly all high-end models from this NVIDIA GeForce generation are based. A fully unlocked GA102 comes with 10,752 CUDA cores, 336 Tensor cores, 84 RT cores, 336 TMUs, and 112 ROPs. The RTX 3090 stops short of maxing this out, with its 10,496 CUDA cores.

NVIDIA's strategy with the alleged RTX 3090 SUPER will be to not only max out the GA102 silicon, with its 10,752 CUDA cores, but also equip it with the fastest possible GDDR6X memory variant, which ticks at 21 Gbps data-rate, compared to 19.5 Gbps on the RTX 3090, and 19 Gbps on the RTX 3080 and RTX 3080 Ti. At this speed, across the chip's 384-bit wide memory bus, the RTX 3090 SUPER will enjoy 1 TB/s of memory bandwidth. Besides more CUDA cores, it's possible that the GPU Boost frequency could be increased. All this comes at a cost, though, with Greymon55 predicting a total graphics power (TGP) of at least 400 W, compared to 350 W of the RTX 3090. A product launch is expected within 2021.

ZOTAC may have inadvertently leaked the GeForce RTX 3090 Ti. The latest version of its FireStorm utility mentions support for the RTX 3090 Ti. This would indicate that NVIDIA has been working on a new top-of-the-line graphics card that replaces the RTX 3090 as its most premium consumer graphics offering. Until now, it was expected that NVIDIA would hold onto the RTX 3090 as its top client product, with the gap between it and the RTX 3080 being filled up by the RTX 3080 Ti, to help it better compete with the AMD Radeon RX 6900 XT. AMD's introduction of the new RX 6900 XT (XTXH silicon), and more surprisingly, the introduction yielding a 10% clock-speed increase, has changed the competitive outlook of the very top of NVIDIA's product-stack.

There are no specifications out there, but in all likelihood, the GeForce RTX 3090 Ti maxes out the 8 nm "GA102" silicon. The RTX 3090 enables all but one of the 42 TPCs physically present on the silicon, and it's likely that this disabled TPC, amounting to an additional 256 CUDA cores, could be unlocked. This would put its CUDA core count at 10,752, compared to 10.496 on the RTX 3090. The only other area NVIDIA could squeeze out performance is GPU clock speeds—an approach similar to AMD's to come up with the RX 6900 XT (XTXH). The highest bins of GA102 could go into building the RTX 3090 Ti. The RTX 3090 already maxes out the 384-bit GDDR6X memory interface, uses the fastest 19.5 Gbps memory chips available, and offers a massive 24 GB of video memory, so it remains to be seen what other specs NVIDIA could tinker with to create the RTX 3090 Ti.