• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

NVIDIA RTX 4090 Doesn't Max-Out AD102, Ample Room Left for Future RTX 4090 Ti

btarunr

Editor & Senior Moderator
Staff member
Joined
Oct 9, 2007
Messages
47,244 (7.55/day)
Location
Hyderabad, India
System Name RBMK-1000
Processor AMD Ryzen 7 5700G
Motherboard ASUS ROG Strix B450-E Gaming
Cooling DeepCool Gammax L240 V2
Memory 2x 8GB G.Skill Sniper X
Video Card(s) Palit GeForce RTX 2080 SUPER GameRock
Storage Western Digital Black NVMe 512GB
Display(s) BenQ 1440p 60 Hz 27-inch
Case Corsair Carbide 100R
Audio Device(s) ASUS SupremeFX S1220A
Power Supply Cooler Master MWE Gold 650W
Mouse ASUS ROG Strix Impact
Keyboard Gamdias Hermes E2
Software Windows 11 Pro
The AD102 silicon on which NVIDIA's new flagship graphics card, the GeForce RTX 4090, is based, is a marvel of semiconductor engineering. Built on the 4 nm EUV (TSMC 4N) silicon fabrication process, the chip has a gargantuan transistor-count of 76.3 billion, a nearly 170% increase over the previous GA102, and a die-size of 608 mm², which is in fact smaller than the 628 mm² die-area of the GA102. This is thanks to TSMC 4N offering nearly thrice the transistor-density of the Samsung 8LPP node on which the GA102 is based.

The AD102 physically features 18,432 CUDA cores, 568 fourth-generation Tensor cores, and 142 third-generation RT cores. The streaming multiprocessors (SM) come with special components that enable the Shader Execution Reordering optimization, which has a significant performance impact on both raster- and ray traced graphics rendering performance. The silicon supports up to 24 GB of GDDR6X or up to 48 GB of GDDR6+ECC memory (the latter will be seen in the RTX Ada professional-visualization card), across a 384-bit wide memory bus. There are 568 TMUs, and a mammoth 192 ROPs on the silicon.



The RTX 4090 is carved out of this silicon by enabling 16,384 out of 18,432 CUDA cores. 512 out of 568 Tensor cores, 512 out of 568 TMUs, 128 out of 142 RT cores, and unless NVIDIA has touched the ROP count, it could remain at 192. The memory bus is maxed out, with 24 GB of 21 Gbps GDDR6X memory across the 384-bit bus-width. In creating the RTX 4090, NVIDIA has given itself a 10% headroom in the number-crunching machinery, from which to carve out future SKUs such as the possible RTX 4090 Ti. Until that SKU is needed in the product-stack, NVIDIA will use this 10% margin toward harvesting the AD102 silicon.

View at TechPowerUp Main Site
 
Joined
Sep 17, 2014
Messages
22,452 (6.03/day)
Location
The Washing Machine
Processor 7800X3D
Motherboard MSI MAG Mortar b650m wifi
Cooling Thermalright Peerless Assassin
Memory 32GB Corsair Vengeance 30CL6000
Video Card(s) ASRock RX7900XT Phantom Gaming
Storage Lexar NM790 4TB + Samsung 850 EVO 1TB + Samsung 980 1TB + Crucial BX100 250GB
Display(s) Gigabyte G34QWC (3440x1440)
Case Lian Li A3 mATX White
Audio Device(s) Harman Kardon AVR137 + 2.1
Power Supply EVGA Supernova G2 750W
Mouse Steelseries Aerox 5
Keyboard Lenovo Thinkpad Trackpoint II
Software W11 IoT Enterprise LTSC
Benchmark Scores Over 9000
'Samsung 8 was a good node'.
Mhm, but TSMC's are much better.
 
Joined
Nov 11, 2020
Messages
460 (0.31/day)
Location
Earth, Solar System, Milky Way Galaxy, Local Group
Processor AMD Ryzen 7 5700X
Motherboard Asus TUF Gaming B550M-Plus (Wi-Fi)
Cooling Thermalright PA120 SE; Arctic P12, F12
Memory Crucial BL8G32C16U4W.M8FE1 ×2
Video Card(s) Sapphire Nitro+ RX 6600 XT
Storage Kingston SKC3000D/2048G; Samsung MZVLB1T0HBLR-000L2; Seagate ST1000DM010-2EP102
Display(s) AOC 24G2W1G4
Case Sama MiCube
Audio Device(s) Somic G923
Power Supply EVGA 650 GD
Mouse Logitech G102
Keyboard Logitech K845 TTC Brown
Software Windows 10 Pro 1903, Dism++, CCleaner
Benchmark Scores CPU-Z 17.01.64: 3700X @ 4.6 GHz 1.3375 V scoring 557/6206; 760K @ 5 GHz 1.5 V scoring 292/964
RTX 4090 doesn't max-out Max TBP, ample room left for future 800 W power.
Good job, nVIDIA.
 
Last edited:

ir_cow

Staff member
Joined
Sep 4, 2008
Messages
4,483 (0.76/day)
Location
USA
Is anyone surpised by this? At one point the Ti model was a max core GPU. Than it shifted to the Titan Model. Now the Titan is gone. The Ti is once again the highest core count.
 
Joined
Aug 10, 2008
Messages
294 (0.05/day)
Location
Saigon city
System Name Kurise PC
Processor i7 5820k 4,7ghz / Ryzen 1700x 4ghz / 8700k
Motherboard Asus X99 deluxe / MSI x370 gaming pro carbon / z370i strix
Cooling EK evo, xspc slim 360 rad, D5 pump, dual alpha cool GPU mono block, dual xspc 240 radiator, DDC 18w
Memory Crucial sport white 16gb x 8 128gb 2666mhz/ Crucial sport white 16gb x 4 64gb 2933 / ddr4 chinese 32
Video Card(s) GTX 1080Ti SLI / HP gtx 1080 SLI 1850/1520 / 2080ti ref
Storage Lite on 512GB x 3 / Plextor m2 256gb / samsung 970 evo
Display(s) AOC I2769Vm, AOC U3477PQU, AOC I2769Vm / Koios 40''/ eizo EV2730QFX 1:1
Case Xigmatek Elysium / Corsair 750D / Bitfenix prodigy M
Audio Device(s) creative blaster ZX / Blaster ZXR / Blaster x7 lmt + burson v5i upgraded
Power Supply Be Quiet 1200 / Thermaltake toughpower 1200w / chinese 750w sfx PSU
Mouse Asus Echelon/ steelseries black ops II/ james donkey
Keyboard Cm storm quickfire pro / Fire rose steampunk kb/ corsair k70
Software Windows 10
I would want the rtx4000 to be at 3000 series with same performance and cosume 1/2 to 1/3 power
 
Joined
Feb 15, 2019
Messages
1,659 (0.78/day)
System Name Personal Gaming Rig
Processor Ryzen 7800X3D
Motherboard MSI X670E Carbon
Cooling MO-RA 3 420
Memory 32GB 6000MHz
Video Card(s) RTX 4090 ICHILL FROSTBITE ULTRA
Storage 4x 2TB Nvme
Display(s) Samsung G8 OLED
Case Silverstone FT04
1000W just isn't enough

season 5 episode 10 GIF by SpongeBob SquarePants
 
Joined
Aug 3, 2022
Messages
133 (0.16/day)
Processor i7-7700k @5ghz
Motherboard Asus strix Z270-F
Cooling EK AIO 240mm
Memory Hyper-X ( 16 GB - XMP )
Video Card(s) RTX 2080 super OC
Storage 512GB - WD(Nvme) + 1TB WD SDD
Display(s) Acer Nitro 165Hz OC
Case Deepcool Mesh 55
Audio Device(s) Razer Karken X
Power Supply Asus TUF gaming 650W brozen
Mouse Razer Mamba Wireless & Glorious Model D Wireless
Keyboard Cooler Master K70
Software Win 10
'Samsung 8 was a good node'.
Mhm, but TSMC's are much better.
Agreed

RTX 4090 doesn't max-out Max TBP, ample room left for future 800 W power.
Good job, nVIDIA.
Damn dude - EVGA made 3090Ti liquid cooler then they didn't had any room left for future 4090ti :D - Thats why they said bye bye Nvidia
 
Joined
Mar 28, 2020
Messages
1,755 (1.03/day)
I feel even with a fully enabled chip, its not going to result in a significant improvement over a slightly gimped one. Increasing number of cores will end up with diminishing returns. And knowing Nvidia, they will likely increase power limit in their mid cycle refresh, plus a further price increase.

'Samsung 8 was a good node'.
Mhm, but TSMC's are much better.
I am not sure if its a good node, I.e. Samsung 8nm which is essentially 10nm. Compared to AMD, Nvidia seems to be doing well with a node disadvantage. But frankly, this may be attributed to better architecture than RDNA2. Considering the huge jump in specs and clocks peed on TSMC 4nm (5nm) I feel the Samsung node actually was holding Ampere's performance back.
 
Joined
Sep 17, 2014
Messages
22,452 (6.03/day)
Location
The Washing Machine
Processor 7800X3D
Motherboard MSI MAG Mortar b650m wifi
Cooling Thermalright Peerless Assassin
Memory 32GB Corsair Vengeance 30CL6000
Video Card(s) ASRock RX7900XT Phantom Gaming
Storage Lexar NM790 4TB + Samsung 850 EVO 1TB + Samsung 980 1TB + Crucial BX100 250GB
Display(s) Gigabyte G34QWC (3440x1440)
Case Lian Li A3 mATX White
Audio Device(s) Harman Kardon AVR137 + 2.1
Power Supply EVGA Supernova G2 750W
Mouse Steelseries Aerox 5
Keyboard Lenovo Thinkpad Trackpoint II
Software W11 IoT Enterprise LTSC
Benchmark Scores Over 9000
I feel the Samsung node actually was holding Ampere's performance back.
Of course, every half wit knows this, but there is a strong following of GPU owners that is adamant Samsung's nodes 'are not bad at all'. The latest argument in favor of that was pushing the blame to GDDR6X for the monumental power consumption, never mind the fact clocks were lower than in 2016 on... TSMC...16nm ;)

And here we are seeing the same GDDR6X on TSMC with more memory alongside a smaller die with many more transistors at a relatively small increase of power budget :)
 
Joined
Jul 15, 2020
Messages
1,021 (0.64/day)
System Name Dirt Sheep | Silent Sheep
Processor i5-2400 | 13900K (-0.02mV offset)
Motherboard Asus P8H67-M LE | Gigabyte AERO Z690-G, bios F29e Intel baseline
Cooling Scythe Katana Type 1 | Noctua NH-U12A chromax.black
Memory G-skill 2*8GB DDR3 | Corsair Vengeance 4*32GB DDR5 5200Mhz C40 @4000MHz
Video Card(s) Gigabyte 970GTX Mini | NV 1080TI FE (cap at 50%, 800mV)
Storage 2*SN850 1TB, 230S 4TB, 840EVO 128GB, WD green 2TB HDD, IronWolf 6TB, 2*HC550 18TB in RAID1
Display(s) LG 21` FHD W2261VP | Lenovo 27` 4K Qreator 27
Case Thermaltake V3 Black|Define 7 Solid, stock 3*14 fans+ 2*12 front&buttom+ out 1*8 (on expansion slot)
Audio Device(s) Beyerdynamic DT 990 (or the screen speakers when I'm too lazy)
Power Supply Enermax Pro82+ 525W | Corsair RM650x (2021)
Mouse Logitech Master 3
Keyboard Roccat Isku FX
VR HMD Nop.
Software WIN 10 | WIN 11
Benchmark Scores CB23 SC: i5-2400=641 | i9-13900k=2325-2281 MC: i5-2400=i9 13900k SC | i9-13900k=37240-35500
try to look surprised..
 
Joined
Dec 1, 2020
Messages
468 (0.32/day)
Processor Ryzen 5 7600X
Motherboard ASRock B650M PG Riptide
Cooling Noctua NH-D15
Memory DDR5 6000Mhz CL28 32GB
Video Card(s) Nvidia Geforce RTX 3070 Palit GamingPro OC
Storage Corsair MP600 Force Series Gen.4 1TB
Looking at this 4080 12gb, which is slower or on par in rasterization compared to 3090 with 285W vs 350W makes me think that the 8nm samsung was not that bad, but just nvidia can't produce effective card
 
Joined
May 20, 2020
Messages
1,372 (0.83/day)
Looking at this 4080 12gb, which is slower or on par in rasterization compared to 3090 with 285W vs 350W makes me think that the 8nm samsung was not that bad, but just nvidia can't produce effective card
I believe greed is the key word here, as in ngreedia as we've seen nvidiot_central being called in the past. Let's see the reviews...
 
Joined
May 31, 2016
Messages
4,437 (1.43/day)
Location
Currently Norway
System Name Bro2
Processor Ryzen 5800X
Motherboard Gigabyte X570 Aorus Elite
Cooling Corsair h115i pro rgb
Memory 32GB G.Skill Flare X 3200 CL14 @3800Mhz CL16
Video Card(s) Powercolor 6900 XT Red Devil 1.1v@2400Mhz
Storage M.2 Samsung 970 Evo Plus 500MB/ Samsung 860 Evo 1TB
Display(s) LG 27UD69 UHD / LG 27GN950
Case Fractal Design G
Audio Device(s) Realtec 5.1
Power Supply Seasonic 750W GOLD
Mouse Logitech G402
Keyboard Logitech slim
Software Windows 10 64 bit
Looking at this 4080 12gb, which is slower or on par in rasterization compared to 3090 with 285W vs 350W makes me think that the 8nm samsung was not that bad, but just nvidia can't produce effective card
maybe the node was not bad at all. The proper question is, does the architecture is good? Maybe what was not so great was the architecture itself and the node change is not gonna change it either.
 
Joined
Feb 25, 2016
Messages
396 (0.12/day)
System Name 06/2023
Processor R7 7800X3D
Motherboard ROG STRIX B650E-I GAMING WIFI
Cooling Custom 240mm cooling (for CPU) with noctua nfa12x25 and Phantek T30
Memory 32gb Gskill 6000 CL30
Video Card(s) RTX 4070 dual asus deshrouded with 120mm NF-A12x25
Storage 2tb samsung 990 pro + 4tb samsung 870 evo
Display(s) Asus 27" Oled PG27AQDM + Asus 27" IPS PG279QM
Case Ncase M1 v6.1
Audio Device(s) Steelseries arctis pro wireless + Shure SM7b with Steinberg UR
Power Supply Corsair SF750 Platinum
Mouse Corsair scimitar pro (this mouse need an overall guys pls) + Logitech G Pro wireless with powerplay
Keyboard Sharkoon purewriter
Software windows 11
Benchmark Scores Over 9000 !
The 3090 didnt max out GA102 either and we didn't get a 3090ti using a bigger die.
 
Joined
May 31, 2016
Messages
4,437 (1.43/day)
Location
Currently Norway
System Name Bro2
Processor Ryzen 5800X
Motherboard Gigabyte X570 Aorus Elite
Cooling Corsair h115i pro rgb
Memory 32GB G.Skill Flare X 3200 CL14 @3800Mhz CL16
Video Card(s) Powercolor 6900 XT Red Devil 1.1v@2400Mhz
Storage M.2 Samsung 970 Evo Plus 500MB/ Samsung 860 Evo 1TB
Display(s) LG 27UD69 UHD / LG 27GN950
Case Fractal Design G
Audio Device(s) Realtec 5.1
Power Supply Seasonic 750W GOLD
Mouse Logitech G402
Keyboard Logitech slim
Software Windows 10 64 bit
The 3090 didnt max out GA102 either and we didn't get a 3090ti using a bigger die.
It is not about bigger die since that would mean a different chip. 3090 has 2 Sm units disabled. 3090 has 82 SM units vs 84 for 3090ti. It is not about bigger die since that one is the same. You have more resources.
 
Joined
Dec 12, 2016
Messages
1,852 (0.64/day)
I meant to post this comment here:

There seems to be some ambiguity around the ROP count. Is there an official number yet?
 
Joined
Jun 14, 2020
Messages
3,474 (2.13/day)
System Name Mean machine
Processor 12900k
Motherboard MSI Unify X
Cooling Noctua U12A
Memory 7600c34
Video Card(s) 4090 Gamerock oc
Storage 980 pro 2tb
Display(s) Samsung crg90
Case Fractal Torent
Audio Device(s) Hifiman Arya / a30 - d30 pro stack
Power Supply Be quiet dark power pro 1200
Mouse Viper ultimate
Keyboard Blackwidow 65%
Releasing the 4090ti a year from now, mere months before a 5xxx launch is just.....I don't know. I'd never buy an xx90ti if it doesn't come out near the launch day of the current gen
 
Joined
Dec 12, 2016
Messages
1,852 (0.64/day)
Releasing the 4090ti a year from now, mere months before a 5xxx launch is just.....I don't know. I'd never buy an xx90ti if it doesn't come out near the launch day of the current gen
Don’t get bogged down in model letters and numbers. Currently Ti cards are the year out refresh that graphics manufacturers have been doing for years. In the past, different model letters and numbers have been used such as Super and xx50 XT.

Refreshes require more mature manufacturing nodes and a build up of harvested dies that yield more or less silicon to be activated. It’s a way to sell as many chips as possible given the reality of defects and poor yields near the beginning of new product series.

As an aside, its also easier to make one complete chip and then lock otherwise functioning parts to create lower SKUs. This only works up to a point when the ‘dead’ or ‘locked’ silicon exceeds the portion of working parts of the chip at which point you manufacture a smaller ‘native’ chip.

Edit: Oh and sometimes later SKU refreshes are just added in response to competition product releases. A company might even save such responses from the beginning on purpose to see how the competition reacts.
 
Last edited:
Joined
May 13, 2015
Messages
632 (0.18/day)
Processor AMD Ryzen 3800X / AMD 8350
Motherboard ASRock X570 Phantom Gaming X / Gigabyte 990FXA-UD5 Revision 3.0
Cooling Stock / Corsair H100
Memory 32GB / 24GB
Video Card(s) Sapphire RX 6800 / AMD Radeon 290X (Toggling until 6950XT)
Storage C:\ 1TB SSD, D:\ RAID-1 1TB SSD, 2x4TB-RAID-1
Display(s) Samsung U32E850R
Case be quiet! Dark Base Pro 900 Black rev. 2 / Fractal Design
Audio Device(s) Creative Sound Blaster X-Fi
Power Supply EVGA Supernova 1300G2 / EVGA Supernova 850G+
Mouse Logitech M-U0007
Keyboard Logitech G110 / Logitech G110
It's 5nm "marketed" as "4nm". The people writing articles should be filtering the BS, not echoing it.
 
D

Deleted member 185088

Guest
They need another card to milk us. The worst part is we used to have Titans that at least brought professional things to the mainstream.
 
Joined
Nov 30, 2021
Messages
135 (0.12/day)
Location
USA
System Name Star Killer
Processor Intel 13700K
Motherboard ASUS RO STRIX Z790-H
Cooling Corsair 360mm H150 LCD Radiator
Memory 64GB Corsair Vengence DDR5 5600mhz
Video Card(s) MSI RTX 3080 12GB Gaming Trio
Storage 1TB Samsung 980 x 1 | 1TB Crucial Gen 4 SSD x 1 | 2TB Samsung 990 Pro x 1
Display(s) 32inch ASUS ROG STRIX 1440p 170hz WQHD x 1, 24inch ASUS 165hz 1080p x 1
Case Lian Li O11D White
Audio Device(s) Creative T100 Speakers , Razer Blackshark V2 Pro wireless
Power Supply EVGA 1000watt G6 Gold
Mouse Razer Viper V2 Wireless with dock
Keyboard ASUS ROG AZOTH
Software Windows 11 pro
If performance is 10-20 percent higher than anything that AMD has, then it will be branded as a Titan GPU. The reason the 30 series didn't have one is partially due to how close AMD was in performance. They will not risk the headline, "Titan loses"
 
Joined
Jan 5, 2017
Messages
307 (0.11/day)
System Name Main
Processor 8700K
Motherboard Maximus Hero X
Cooling EVGA 280 CLC w/ Noctua silent fans
Memory 2x8GB 3600/16
Video Card(s) EVGA 2080TI Hybrid
Datacenters will get all the fully functional dies, gamers get the broken scraps.
 
Joined
Feb 20, 2019
Messages
8,284 (3.93/day)
System Name Bragging Rights
Processor Atom Z3735F 1.33GHz
Motherboard It has no markings but it's green
Cooling No, it's a 2.2W processor
Memory 2GB DDR3L-1333
Video Card(s) Gen7 Intel HD (4EU @ 311MHz)
Storage 32GB eMMC and 128GB Sandisk Extreme U3
Display(s) 10" IPS 1280x800 60Hz
Case Veddha T2
Audio Device(s) Apparently, yes
Power Supply Samsung 18W 5V fast-charger
Mouse MX Anywhere 2
Keyboard Logitech MX Keys (not Cherry MX at all)
VR HMD Samsung Oddyssey, not that I'd plug it into this though....
Software W10 21H1, barely
Benchmark Scores I once clocked a Celeron-300A to 564MHz on an Abit BE6 and it scored over 9000.
Yields on something that big must be horrible.
 
Joined
Aug 20, 2007
Messages
21,469 (3.40/day)
System Name Pioneer
Processor Ryzen R9 9950X
Motherboard GIGABYTE Aorus Elite X670 AX
Cooling Noctua NH-D15 + A whole lotta Sunon and Corsair Maglev blower fans...
Memory 64GB (4x 16GB) G.Skill Flare X5 @ DDR5-6000 CL30
Video Card(s) XFX RX 7900 XTX Speedster Merc 310
Storage Intel 905p Optane 960GB boot, +2x Crucial P5 Plus 2TB PCIe 4.0 NVMe SSDs
Display(s) 55" LG 55" B9 OLED 4K Display
Case Thermaltake Core X31
Audio Device(s) TOSLINK->Schiit Modi MB->Asgard 2 DAC Amp->AKG Pro K712 Headphones or HDMI->B9 OLED
Power Supply FSP Hydro Ti Pro 850W
Mouse Logitech G305 Lightspeed Wireless
Keyboard WASD Code v3 with Cherry Green keyswitches + PBT DS keycaps
Software Gentoo Linux x64 / Windows 11 Enterprise IoT 2024
but there is a strong following of GPU owners that is adamant Samsung's nodes 'are not bad at all'.
They weren't bad at all...

...on launch day.

This is how tech advancement works.
 
Top