yitit
Home
/
Hardware
/
NVIDIA Ampere GA102F GPU Support Added To AIDA64, Is This The Chip Powering The GeForce RTX 3090 Ti?
NVIDIA Ampere GA102F GPU Support Added To AIDA64, Is This The Chip Powering The GeForce RTX 3090 Ti?-December 2024
Dec 27, 2024 11:48 AM

A mysterious new NVIDIA Ampere GPU, the GA102F, has been added support within the latest version of AIDA64. There's currently no new graphics card that utilizes such a chip but it is very likely that this could be the one powering NVIDIA's upcoming flagship, the GeForce RTX 3090 Ti.

AIDA64 Adds Support For NVIDIA Ampere GA102F GPU, Is This The Chip That Powers GeForce RTX 3090 Ti?

The Ampere GA102 GPU is currently the biggest gaming silicon that NVIDIA ships. It powers a range of graphics cards including the RTX 3090, RTX 3080 Ti, RTX 3080, RTX A6000 & the RTX A5000. There is going to be a brand new SKU for the GeForce RTX 3090 Ti which will also be based on the GA102 silicon but it is going to be the fastest that the company has ever built. Sporting higher clocks & the full core configuration, the RTX 3090 Ti's GA102 GPU has been a difficult chip to manufacture & one of the reported causes for the delay of the flagship card.

AIDA64 has now added support for an entirely new GA102 GPU SKU, the GA102F & first guess will be that this is the chip that might power the GeForce RTX 3090 Ti graphics card. We can't say for sure what the 'F' stands for, it might be 'Full' since the RTX 3090 Ti rocks the full GA102 core config. The other guess is that this could be a specialized variant of the GA102 for the data center or cloud computing segment. Once again, I wouldn't say we know this for sure until we get to see the chip in action somewhere.

NVIDIA GeForce RTX 3090 Ti 24 GB Graphics Card Specifications

At the heart of the NVIDIA GeForce RTX 3090 Ti graphics card lies the GA102 GPU. The GA102 is the flagship gaming GPU and also the fastest gaming GPU that NVIDIA has produced. The GPU is based on Samsung's 8nm custom process node designed specifically for NVIDIA and features a total of 28 Billion transistors. It measures 628mm2 which makes it the 2nd biggest gaming GPU ever produced right below the Turing TU102 GPU which powered the RTX 2080 Ti and Titan RTX.

nvidia-geforce-rtx-30-series-graphics-cards_announcement_geforce-rtx-3090_rtx-3080_rtx-3070_4

nvidia-geforce-rtx-30-series-graphics-cards_announcement_geforce-rtx-3090_rtx-3080_rtx-3070_5

nvidia-geforce-rtx-30-series-graphics-cards_announcement_geforce-rtx-3090_rtx-3080_rtx-3070_2

nvidia-geforce-rtx-30-series-graphics-cards_announcement_geforce-rtx-3090_rtx-3080_rtx-3070_3

nvidia-geforce-rtx-30-series-graphics-cards_announcement_geforce-rtx-3090_rtx-3080_rtx-3070_6

nvidia-geforce-rtx-30-series-graphics-cards_announcement_geforce-rtx-3090_rtx-3080_rtx-3070_7

2 of 9

For the GeForce RTX 3090, NVIDIA has enabled a total of 84 SM units on its flagship which results in a total of 10,752 CUDA cores (vs 82 SM / 10496 cores on RTX 3090 Non-Ti). In addition to the CUDA cores, NVIDIA's GeForce RTX 3090 Ti also comes packed with next-generation RT (Ray-Tracing) cores, Tensor cores, and brand new SM or streaming multi-processor units. The GPU runs at a base clock speed of 1560 MHz and a boost clock speed of 1860 MHz. The card has a TDP of 450W (a 100 Watt increase over the RTX 3090).

In terms of memory, the GeForce RTX 3090 Ti comes packed with 24 GB of memory and that too the next-generation GDDR6X design. With Micron's latest and greatest graphics memory dies, the RTX 3090 Ti can deliver GDDR6X memory speeds of 21 Gbps. That along with a bus interface of 384-bit will deliver a cumulative bandwidth of 1008 Gbps.

The NVIDIA GeForce RTX 3090 Ti graphics card is also rumored to be the first PCIe Gen 5.0 compliant graphics card, rocking a single 16-pin power connector that can supply up to 600 Watts of power to the card.

Currently, it's an ongoing debate internally at several AIBs whether to feature the PCIe Gen 5.0 connector like NVIDIA's Founder Edition or go with standard 8-pin connectors. The Founders Edition is also going to utilize what seems to be an updated revision of the PG136 PCB board known as PG136C.

As for its feature set, the NVIDIA GeForce RTX 3090 Ti 24 GB graphics card rocks all the modern NV feature set such as the latest NVENC Encoder and NVCDEC Decoder, support for the latest APIs, 2nd Generation ray-tracing cores, 3rd Gen Tensor cores. It packs all the modern RTX features such as DLSS, Reflex, Broadcast, Resizable-BAR, Freestyle, Ansel, Highlights, Shadowplay, and G-SYNC support too. The graphics card is expected to launch on 29th of March for a massive premium over the existing $1499 MSRP of the non-Ti variant.

NVIDIA GeForce RTX 30 Series Graphics Card Specifications

Graphics Card NameNVIDIA GeForce RTX 3090 TiNVIDIA GeForce RTX 3090NVIDIA GeForce RTX 3080 TiNVIDIA GeForce RTX 3080 12 GBNVIDIA GeForce RTX 3080NVIDIA GeForce RTX 3070 TiNVIDIA GeForce RTX 3070NVIDIA GeForce RTX 3060 TiNVIDIA GeForce RTX 3060 TiNVIDIA GeForce RTX 3060NVIDIA GeForce RTX 3060NVIDIA GeForce RTX 3050NVIDIA GeForce RTX 3050 6 GB
GPU NameAmpere GA102-350?Ampere GA102-300Ampere GA102-225Ampere GA102-220Ampere GA102-200Ampere GA104-400Ampere GA104-300Ampere GA104Ampere GA104-200
Ampere GA103-200
Ampere GA106-300Ampere GA106Ampere GA106-150Ampere GA106-325
Process NodeSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nm
Die Size628.4mm2628.4mm2628.4mm2628.4mm2628.4mm2395.2mm2395.2mm2395.2mm2395.2mm2 (GA104)276mm2276mm2276mm2 (GA106)200mm2 (GA107)
Transistors28 Billion28 Billion28 Billion28 Billion28 Billion17.4 Billion17.4 Billion17.4 Billion17.4 Billion (GA104)13.2 Billion13.2 Billion13.2 Billion (GA106)8.7 Billion (GA107)
CUDA Cores1075210496102408960870461445888486448643584358425602048
TMUs / ROPs336 / 112328 / 112320 / 112280 / 104272 / 96184 / 96184 / 96152 / 80152 / 80112 / 64112 / 6480 / 32TBD
Tensor / RT Cores336 / 84328 / 82320 / 80280 / 70272 / 68184 / 46184 / 46152 / 38152 / 38112 / 28112 / 2880 / 20TBD
Base Clock1560 MHz1400 MHz1365 MHzTBA1440 MHz1575 MHz1500 MHz1410 MHz1410 MHz1320 MHz1320 MHz1552 MHzTBD
Boost Clock1860 MHz1700 MHz1665 MHzTBA1710 MHz1770 MHz1730 MHz1665 MHz1665 MHz1780 MHz1780 MHz1777 MHz1470 MHz
FP32 Compute40 TFLOPs36 TFLOPs34 TFLOPsTBA30 TFLOPs22 TFLOPs20 TFLOPs16 TFLOPs16 TFLOPs13 TFLOPs13 TFLOPs9.1 TFLOPsTBD
RT TFLOPs74 RFLOPs69 TFLOPs67 TFLOPsTBA58 TFLOPs44 TFLOPs40 TFLOPs32 TFLOPs32 TFLOPs25 TFLOPs25 TFLOPs18.2 TFLOPsTBD
Tensor-TOPsTBA285 TOPs273 TOPsTBA238 TOPs183 TOPs163 TOPs192 TOPs192 TOPs101 TOPs101 TOPs72.8 TOPsTBD
Memory Capacity24 GB GDDR6X24 GB GDDR6X12 GB GDDR6X12 GB GDDR6X10 GB GDDR6X8 GB GDDR6X8 GB GDDR68 GB GDDR6X8 GB GDDR612 GB GDDR68 GB GDDR68 GB GDDR66 GB GDDR6
Memory Bus384-bit384-bit384-bit384-bit320-bit256-bit256-bit256-bit256-bit192-bit128-bit128-bit96-bit
Memory Speed21 Gbps19.5 Gbps19 Gbps19 Gbps19 Gbps19 Gbps14 Gbps19 Gbps14 Gbps16 Gbps15 Gbps14 Gbps14 Gbps
Bandwidth1008 GB/s936 GB/s912 Gbps912 Gbps760 GB/s608 GB/s448 GB/s608 GB/s448 GB/s384 GB/s240 GB/s224 GB/s168 GB/s
TGP450W350W350W350W320W290W220W175W175W170W150W130W (GA106)
115W (GA107)
70W
Price (MSRP / FE)TBD$1499 US$1199$999 US?$699 US$599 US$499 US$399 US$399 US$329 USTBD$249 USTBD
Launch (Availability)29th March 2022?24th September 20203rd June 202111th January 202217th September 202010th June, 202129th October 2020October 20222nd December 202025th February 2021October 202227th January 20222024

Comments
Welcome to yitit comments! Please keep conversations courteous and on-topic. To fosterproductive and respectful conversations, you may see comments from our Community Managers.
Sign up to post
Sort by
Show More Comments
Hardware
Recent News
Copyright 2023-2024 - www.yitit.com All Rights Reserved