[zhrooms]

WORK-IN-PROGRESS

Last Updated: March 29, 2022

Note: This content is licensed under Creative Commons 3.0. This means that you are free to copy and redistribute this material, but only if the following criteria are met: 1) You must give appropriate credit by linking back to this thread. 2) You may not use this material for commercial purposes or place this on a for-profit website with ads. 3) You cannot create derivative work based on this material.

NVIDIA GeForce® RTX 3090 Ti

⠀⠀⠀⠀⠀RTX 3070 / Ti Owner's Club
⠀⠀RTX 3080 Owner's Club
⠀⠀⠀⠀RTX 3080 Ti Owner's Club
⠀⠀RTX 3090 Owner's Club
⠀⠀→ RTX 3090 Ti Owner's Club

Click here to join the discussion on Discord or join directly through the Discord app with the code kkuFR3d

Source: NVIDIA

SPECS (Click Spoiler)

Spoiler

 
   Architecture Ampere
   Chip GA102-350
   Transistors 28,300 million
   Die Size 628 mm²
   Manufacturing Process Samsung 8nm

   CUDA Cores 5376 (10752)
   TMUs 336
   ROPs 112
   SM Count 84
   Tensor Cores 336
   GigaRays -- GR/s

   Core Clock 1560 MHz
   Boost Clock 1860 MHz
   Memory 24GB GDDR6X
   Memory Bus 384-bit
   Memory Clock 1313 MHz / 21008 MHz
   Memory Bandwidth 1008 GB/s
   External Power Supply 12-Pin
   TDP 450W

   DirectX 12.2 Ultimate
   OpenGL 4.6
   OpenCL 2.0
   Vulkan 1.2
   CUDA 8.6

   Interface PCIe 4.0 x16
   Connectors 1x HDMI 2.1, 3x DisplayPort 1.4a
   Dimensions 313 x 138mm (3-Slot)

   Price $1999 US

   Release Date March 29, 2022

RTX 3090 Ti | GA102-350 |  8nm | 628mm² | 28.3 BT | 10752 CCs | 336 TMUs | 112 ROPs | 84 SMs | 1860 MHz |  24GB | 2048MB x 12 | GDDR6X | 384-bit | 1008 GB/s | 450W⠀⠀
RTX 3090    | GA102-300 |  8nm | 628mm² | 28.3 BT | 10496 CCs | 328 TMUs | 112 ROPs | 82 SMs | 1695 MHz |  24GB | 1024MB x 24 | GDDR6X | 384-bit | ⠀936 GB/s | 350W⠀⠀
RTX 3080 Ti | GA102-XXX |  8nm | 628mm² | 28.3 BT | 10240 CCs | 320 TMUs | 112 ROPs | 80 SMs | 1665 MHz |  12GB | 1024MB x 12 | GDDR6X | 384-bit | ⠀912 GB/s | 320W⠀⠀
RTX 3080    | GA102-200 |  8nm | 628mm² | 28.3 BT |⠀ 8704 CCs | 272 TMUs |  96 ROPs | 68 SMs | 1710 MHz |  10GB | 1024MB x 10 | GDDR6X | 320-bit | ⠀760 GB/s | 320W

Note: Ampere gaming performance does not scale linearly with CUDA core count when compared with previous generations

RTX 2080 Ti | TU102-300 | 12nm | 754mm² | 18.6 BT |⠀ 4352 CCs  | 272 TMUs |  88 ROPs | 68 SMs | 1635 MHz |  11GB | 1024MB x 11 | GDDR6  | 352-bit | ⠀616 GB/s | 250W
RTX 2080 S  | TU104-450 | 12nm | 545mm² | 13.6 BT |⠀ 3072 CCs  | 192 TMUs |  64 ROPs | 48 SMs | 1815 MHz |   8GB | 1024MB x 8  | GDDR6  | 256-bit | ⠀496 GB/s | 250W
RTX 2080    | TU104-400 | 12nm | 545mm² | 13.6 BT |⠀ 2944 CCs  | 184 TMUs |  64 ROPs | 46 SMs | 1710 MHz |   8GB | 1024MB x 8  | GDDR6  | 256-bit | ⠀448 GB/s | 215W
GTX 1080 Ti | GP102-350 | 16nm | 471mm² | 12.0 BT |⠀ 3584 CCs  | 224 TMUs |  88 ROPs | 28 SMs | 1582 MHz |  11GB | 1024MB x 11 | GDDR5X | 352-bit | ⠀484 GB/s | 250W
GTX 1080    | GP104-400 | 16nm | 314mm² |  7.2 BT | ⠀2560 CCs  | 160 TMUs |  64 ROPs | 20 SMs | 1733 MHz |   8GB | 1024MB x 8  | GDDR5X | 256-bit | ⠀320 GB/s | 180W
GTX 980 Ti  | GM200-310 | 28nm | 601mm² |  8.0 BT |⠀ 2816 CCs  | 172 TMUs |  96 ROPs | 22 SMs | 1076 MHz |   6GB |  512MB x 12 | GDDR5  | 384-bit | ⠀336 GB/s | 250W
GTX 980     | GM204-400 | 28nm | 398mm² |  5.2 BT |⠀ 2048 CCs  | 128 TMUs |  64 ROPs | 16 SMs | 1216 MHz |   4GB |  512MB x 8  | GDDR5  | 256-bit | ⠀224 GB/s | 165W
GTX 780 Ti  | GK110-425 | 28nm | 551mm² |  7.1 BT |⠀ 2880 CCs  | 240 TMUs |  48 ROPs | 15 SMs |  928 MHz |   3GB |  256MB x 12 | GDDR5  | 384-bit |⠀ 336 GB/s | 250W
GTX 780     | GK110-300 | 28nm | 551mm² |  7.1 BT | ⠀2304 CCs  | 192 TMUs |  48 ROPs | 12 SMs |  900 MHz |   3GB |  256MB x 12 | GDDR5  | 384-bit |⠀ 288 GB/s | 250W
GTX 680     | GK104-400 | 28nm | 294mm² |  3.5 BT |⠀ 1536 CCs  | 128 TMUs |  32 ROPs |  8 SMs | 1058 MHz |   2GB |  256MB x 8  | GDDR5  | 256-bit | ⠀192 GB/s | 200W
GTX 580     | GF110-375 | 40nm | 520mm² |  3.0 BT |  ⠀512 CCs  |  64 TMUs |  48 ROPs | 16 SMs |  772 MHz | 1.5GB |  128MB x 12 | GDDR5  | 384-bit | ⠀192 GB/s | 250W

TECHPOWERUP | GPU-Z

Download TechPowerUp GPU-Z

NVIDIA | NVFLASH

Download NVIDIA NVFlash

BIOS | ROM

TechPowerUp BIOS Collection < Verified

TechPowerUp BIOS Collection < Unverified

FLASH | GUIDE (Click Spoiler)

Spoiler

└ Step 01 of 27 - Download NVFlash ┘

└ Step 02 of 27 - Downloads Folder ┘

└ Step 03 of 27 - Open Zip File ┘

└ Step 04 of 27 - Copy Files ┘

└ Step 05 of 27 - Create New Folder ┘

└ Step 06 of 27 - Name Folder ┘

└ Step 07 of 27 - Paste Files ┘

└ Step 08 of 27 - Installation Successful ┘

└ Step 09 of 27 - Find BIOS ┘

└ Step 10 of 27 - Download BIOS ┘

└ Step 11 of 27 - Name BIOS ┘

└ Step 12 of 27 - Copy or Cut BIOS ┘

└ Step 13 of 27 - Paste BIOS ┘

└ Step 14 of 27 - Download Successful ┘

└ Step 15 of 27 - Before Flash ┘

└ Step 16 of 27 - Maximum Power Limit (330W) ┘

└ Step 17 of 27 - Starting Command Prompt (Administrator) ┘

chdir C:\nvflash

└ Step 18 of 27 - Changing Directory ┘

nvflash64 --protectoff

└ Step 19 of 27 - Disable Flash Protection ┘

nvflash64 --save Partner2080TiModel.rom

└ Step 20 of 27 - (Optional) BIOS Backup ┘

└ Step 21 of 27 - BIOS Saved ┘

nvflash64 -6 Partner2080TiModel.rom

Y

└ Step 22 of 27 - Flash BIOS ┘

Y

└ Step 23 of 27 - Confirm Update ┘

exit

└ Step 24 of 27 - Flash Completed ┘

└ Step 25 of 27 - After Flash ┘

└ Step 26 of 27 - Maximum Power Limit (380W) ┘

chdir C:\nvflash

nvflash64 --protecton

exit

└ Step 27 of 27 - (Optional) Enable Flash Protection ┘

OVERCLOCKING | TOOLS

Download ASUS GPUTweak II

Download EVGA Precision X1

Download Gainward EXPERTool

Download Galax/KFA2 Xtreme Tuner Plus

Download Gigabyte AORUS Engine

Download Inno3D TuneIT

Download MSI Afterburner

Download Palit ThunderMaster

Download PNY Velocity X

Download Zotac FireStorm​

 

[Panchovix]

Interesting, there are already some confirmed VBIOS out there, and the max power limit seems to be 516W from the Strix LC.

I guess you could just flash that VBIOS in the other cards, right? I think all 3090Ti use the PCI-E 5.0 connector

 

[zhrooms]

I guess you could just flash that VBIOS in the other cards, right? I think all 3090Ti use the PCI-E 5.0 connector

Good point, all of the cards use the same MP86957 controller for the VRM, and all of them use the same power connector as well. So I would assume you could flash any BIOS you wanted, including any XOC ones.

 

[Awsan]

Good point, all of the cards use the same MP86957 controller for the VRM, and all of them use the same power connector as well. So I would assume you could flash any BIOS you wanted, including any XOC ones.

The FTW one in Steve's video had an extra connector on the side, just in case you want another 600 watts pumped into your GPU

 

[chibi]

So.... who's getting the 3090 Ti? This will be a cool thread to follow

 

[zhrooms]

The FTW one in Steve's video had an extra connector on the side, just in case you want another 600 watts pumped into your GPU

Interesting, noticed it now on the PCB shots by TPU, the EVGA and Zotac card has an option for a second connector (not installed).

Strix LC​

FTW3 Ultra​

Suprim X​

AMP Extreme Holo​

The Kingpin card is confirmed to feature two connectors (installed). Might cause issues with the XOC BIOS then, time will tell.

 

[Afferin]

I grabbed a FTW3 Ultra Gaming while I could, when I really wanted the Suprim X. The Suprim X was listed on Newegg Canada about 10 minutes after my FTW3 shipped out. Sucks.

Whatever, this will be a fun experiment.

 

[yzonker]

Interesting, noticed it now on the PCB shots by TPU, the EVGA and Zotac card has an option for a second connector (not installed).

Strix LC​

FTW3 Ultra​

Suprim X​

AMP Extreme Holo​

The Kingpin card is confirmed to feature two connectors (installed). Might cause issues with the XOC BIOS then, time will tell.

That's a good point. Very likely will screw that up. EVGA didn't want that bios shared on the 3090, so they may have a fix for that. Galax has 2 also. That only leaves Asus as a possible xoc bios to flash.

 

[HyperMatrix]

Watching this thread with curiosity. From the OC results in reviews, it seems these cards are all hitting 2150-2200MHz which I didn't expect considering slightly more cores and also higher heat levels to manage. TechPowerUp was able to hit 24Gbps on memory OC as well if I remember correctly. In memory bandwidth heavy games like Horizon Zero Dawn, this could result in a 10% performance increase even over an overclocked 3090. For most games I'm expecting 5%. If 4090 weren't right around the corner I'd be tempted. Put in a notify for the Kingpin model just in case.

 

[BattlePhenom]

I've got a FTW3 card shipped. Really wanted the ASUS Strix LC, but it was gone in seconds.

The FTW one in Steve's video had an extra connector on the side, just in case you want another 600 watts pumped into your GPU

Pretty sure it only has one 16 pin, but EVGA elected to put it in the front.

 

[cfranko]

I've got a FTW3 card shipped. Really wanted the ASUS Strix LC, but it was gone in seconds.



Pretty sure it only has one 16 pin, but EVGA elected to put it in the front.

Why do these cards have that 16 pin thing? Why didn’t manufacturers put the standard 3x8 pin layout? The 16 pin and the adapter looks super ugly and I can’t understand the point behind it.

 

[BattlePhenom]

Why do these cards have that 16 pin thing? Why didn’t manufacturers put the standard 3x8 pin layout? The 16 pin and the adapter looks super ugly and I can’t understand the point behind it.

One theory is that with one power connector you have more room on the board for other components. This will likely be implemented with 40 series as well.

 

[yzonker]

Watching this thread with curiosity. From the OC results in reviews, it seems these cards are all hitting 2150-2200MHz which I didn't expect considering slightly more cores and also higher heat levels to manage. TechPowerUp was able to hit 24Gbps on memory OC as well if I remember correctly. In memory bandwidth heavy games like Horizon Zero Dawn, this could result in a 10% performance increase even over an overclocked 3090. For most games I'm expecting 5%. If 4090 weren't right around the corner I'd be tempted. Put in a notify for the Kingpin model just in case.

It's really hard to assess the true increase in performance since the manufacturers play games with the power limit. Looks like 5-10% though depending on the game. Reviews showing 10%+ are mostly comparing to a base 350w card. You would think at least one reviewer might equalize power as part of the review, but I guess it's expecting too much for them to write a review for the enthusiast community.

 

[yzonker]

One theory is that with one power connector you have more room on the board for other components. This will likely be implemented with 40 series as well.

No power balancing issues other than with the PCIE slot. Also once PSUs have this plug, it will reduce cable clutter. And yes it's really for 40 series.

 

[Maximization]

hoping to get a water blocked one

 

[yzonker]

If anyone has or is getting one of these cards, please post a screenshot of the voltage/frequency curve from Afterburner if you can. I'm curious what the base curve looks like. Take the screenshot with the card cooled down at idle as it shifts with temperatures.

 

[ZealotKi11er]

I have lost interest with RTX-30 series after having 3080, 3080 Ti and now 3070 Ti Mobile. Part of reason was lack of games apart from just doing 4K to really enjoy these cards.

 

[Falkentyne]

It's really hard to assess the true increase in performance since the manufacturers play games with the power limit. Looks like 5-10% though depending on the game. Reviews showing 10%+ are mostly comparing to a base 350w card. You would think at least one reviewer might equalize power as part of the review, but I guess it's expecting too much for them to write a review for the enthusiast community.

An overclocked 3090 FE limited to 520-580W from internal rail limits (shunt mod) already exceeds a stock 3090 Ti And my 3090 FE didn't win the VRAM lottery at all (+600 max) but the core can do +150 it seems. There. 3090 Ti with fewer cores!

Anyway these cards are an epic cash grab. Once you pass 450W, the performance/watt skyrockets massively in the watts favor just for increasing nothing but power limit, and heat is hard to control. You only get a real increase if you have a card that can clock the core sky high, e.g. Asus with Elmor device or Kingpin with voltage dip switches enabled/tool.

I wonder what the custom loop people will get on these cards though.

So who here is actually buying a Ti? They went for sale today and were in stock for over 2 hours....yet some people on the ATC discord seem to be picking some up anyway....

 

[BehindTimes]

So who here is actually buying a Ti? They went for sale today and were in stock for over 2 hours....yet some people on the ATC discord seem to be picking some up anyway....

I went ahead and purchased one. I initially purchased one online, and went to grab another one at MicroCenter, but the card is way too big for my bottom PCI slot, which only has room for 3 slots, and this exceeded that. So I have to return the online one.

NVIDIA GeForce RTX 3090 Ti video card benchmark result - Intel Core i9-12900K Processor,Micro-Star International Co., Ltd. MEG Z690 ACE (MS-7D27) (3dmark.com)

Yeah, I have the slowest score right now, but I haven't applied overclocking yet, and I need the RAM, so I ended up with 4 slots, thus limiting that speed. I.e. I spent way too much on stuff when I could have gotten it for significantly cheaper.

 

[neteng101]

I.e. I spent way too much on stuff when I could have gotten it for significantly cheaper.

This might depress you then but hopefully its inspiration instead, time to start overclocking... I have 4 sticks too, SR DDR4. Really cheap RAM sticks, they were originally 3000-16 before OC. And a 3080Ti on air.

I scored 20 820 in Time Spy

Intel Core i7-12700K Processor, NVIDIA GeForce RTX 3080 Ti x 1, 65536 MB, 64-bit Windows 11}

www.3dmark.com