NVIDIA’s forthcoming GeForce RTX 4090 and GeForce RTX 4070 graphics playing cards based on the Ada Lovelace GPU architecture have been further thorough by Kopite7kimi in his most up-to-date tweet. These graphics playing cards will rock some significantly insane specs and efficiency in contrast to the present Ampere variants.
NVIDIA Rumored To Equip GeForce RTX 4090 With 24 GB / 21 Gbps Memory at 600W, RTX 4070 With 12 GB / 18 Gbps Memory at 300W
In his tweet, Kopite7kimi focuses on two NVIDIA ‘Ada Lovelace’ graphics cards, the GeForce RTX 4090 and the GeForce RTX 4070. The leaker has already uncovered the alleged specs of other cards such as the RTX 4090 Ti and the RTX 4080 but it seems like he was able to dig further more information with regards to the two playing cards pointed out previously mentioned. The major variance is that the GeForce RTX 4090 will be dependent on an optimized AD102 GPU main although the RTX 4070 will employ an optimized AD104 GPU core.
RTX 4090, PG137/139-SKU330, AD102-300, 21Gbps 24G GDDR6X, 600W
RTX 4070, PG141-SKU341, AD104-400, 18Gbps 12G GDDR6, 300W
— kopite7kimi (@kopite7kimi) Might 10, 2022
NVIDIA Ada Lovelace ‘GeForce RTX 40’ GPU Configurations
|GPU Title||GPCs / TPCs||SMs For each TPC / Overall||CUDA Cores||L2 Cache||Memory Bus|
|AD102||twelve / 6||2 / a hundred and forty four||18432||96 MB||384-bit|
|AD103||7 / 6||two / 84||10752||64 MB||256-bit|
|AD104||five / 6||2 / 60||7680||forty eight MB||192-bit|
|AD106||3 / 6||2 / 36||4608||32 MB||128-bit|
|AD107||3 / four||2 / 24||3072||32 MB||128-little bit|
NVIDIA GeForce RTX 4090 / AD102 GPU But Optimized
The NVIDIA GeForce RTX 4090 will also rock the exact AD102 GPU as the ‘Ti’ variant but in a slash-down configuration dependent on the rumor. The AD102 GPU will make use of 144 SMs, a 71% increase over the existing GA102 GPU and home a substantial 18,432 CUDA core rely. In addition to the SM counts, the Ada Lovelace GPUs will also function elevated L2 cache dimensions. Setting up with the AD102 GPU, the flagship would be outfitted with up to ninety six MB of L2 cache, an insane 16x maximize above the 6 MB L2 cache featured on GA102.
The graphics card will in addition rock 24 GB of GDDR6X memory clocking in at 21 Gbps throughout a 384-little bit bus interface and delivering above one. TB/s bandwidth. From the TGP figures we have noticed, the card will have a TBP of 600W at reference and that is going to go up with the non-reference patterns.
As we noticed with the GeForce RTX 3090 Ti, there are specified models that go past the 450W TBP (reference) and up to 516W (customized). The identical is going to be the circumstance with Ada Lovelace-based cards. And whilst this will involve plenty of cooling, it seems to be like AIBs are very well and all set. We are probable heading to see the use of additional hybrid-styled GPU coolers rather than common air-coolers at the time we go above the 600W TBP restrict. This will mark a 33% raise in TBP in excess of the RTX 3090 Ti.
NVIDIA GeForce RTX 4070 / AD104 at 300W
Last but not least, there are the NVIDIA GeForce RTX 4070 class GPUs which will utilize the AD104 GPUs. The AD104 GPUs will feature 60 SM models for a whole of 7680 CUDA cores & a overall of forty eight MB of L2 cache. The RTX 4070 collection in specific will characteristic twelve GB of GDDR6 memory managing throughout a 192-little bit bus interface and feature a TGP of 300W. Thinking about that these are GDDR6 chips, we are probable going to get 16-18 Gbps speeds which should offer a bandwidth of all around four hundred GB/s. This card is explained to make the most of the PG141-SKU341 PCB.
Now thinking of the 300W TDP, the GeForce RTX 3070 Ti presently attributes a reference TDP of 290W so you are hunting at a measly 10W enhance but at the exact same time, the functionality is claimed to be on par with an RTX 3080 Ti-course graphics card which is rated at 350W. The 12 GB memory also appears incredibly desirable considering that it truly is the identical quantity highlighted on the RTX 3080 Ti. It will all rely on pricing of this card as the 70 course is one particular of the most preferred among the superior-conclude graphics card and for consumers to get 3080-Ti class efficiency ($1199 US) at all around $500-$600 would be a massive upgrade.
The NVIDIA Ada Lovelace GPU relatives is predicted to deliver a generational jump identical to Maxwell to Pascal. It is envisioned to launch in the 2nd 50 % of 2022 but expects provide and pricing to be identical to current playing cards despite NVIDIA spending billions of dollars to acquire individuals great great TSMC 5nm (4N) wafers.
NVIDIA GeForce RTX 40 Series Graphics Card Lineup (Rumored):
|Graphics Card||GPU||PCB Variant||SM Units / Cores||Memory / Bus||Memory Clock / Bandwidth||TGP||Electric power Connectors|
|NVIDIA GeForce RTX 4090 Ti||AD102-350?||TBD||a hundred and forty four / 18432?||48 GB / 384-little bit||24 Gbps / 1.15 TB/s||~900W||2x sixteen-pin|
|NVIDIA GeForce RTX 4090||AD102-three hundred?||PG137/139 SKU330||>144 / 18432?||24 GB / 384-little bit||21 Gbps / 1.00 TB/s||~600W||1x sixteen-pin|
|NVIDIA GeForce RTX 4080||AD103-three hundred?||TBD||>84 / 10752?||sixteen GB / 256-little bit||21 Gbps / 672 GB/s||~350W||1x 16-pin|
|NVIDIA GeForce RTX 4070||AD104-four hundred?||PG141-310 SKU341||>60 / 7680||twelve GB / 192-little bit||eighteen Gbps / 432 GB/s||~300W||1x sixteen-pin|