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

Will a RTX 4070 TI super bottleneck a Ryzen 9 7950X3D?

Joined
Mar 23, 2024
Messages
24 (0.59/day)
Location
Almelo, Netherlands
System Name my system
Processor AMD Ryzen 7 5800X3D
Motherboard Asus Rog strix b550 gaming wifi II
Cooling noctua nh-u12s se-am4
Memory corsair vengeance 32gb (2x16) 3600mhz
Video Card(s) Gigabyte RTX 2070 Windforce 8g (planning to upgrade to an asus TUF 4070 TI super soon)
Storage 1x 240gb kingston ssd (win) 1x500gb kingston m.2 1x1tb 870 evo ssd 2x1tb hdd (storage)
Display(s) gigabyte aorus cv27f
Case Sharkoon Rebel c50 RGB
Audio Device(s) Creative soundblaster x-ae5 plus
Power Supply Seasonic Focus GX-850 ATX 3.0
Mouse Logitech g502x
Keyboard Microsoft ernomic buisness
VR HMD none
Software win 11 64 bit home
Benchmark Scores To be tested, so none yet
So what resolution is he going to play at? That is the most important question. If 2560x1440, the 4070ti Super is an excellent choice. If 3440x1440 ultra wide, it is still a good choice, though a 4080 Super will of course be better. If going to 4K, I would certainly want a 4080 Super.


Generally a CPU will bottleneck a GPU if your GPU is too fast and the CPU is slow and can't keep the minimum frame rate stable. For example, you may be getting 50 frame rates and you upgrade your GPU. You now get 80 frame rates average, but there are still large frame rate drops occasionally in situations where the CPU is limited. So the overall minimum frame rate will not be as stable. Average is important but you also want consistency and a high minimum frame rate. Your CPU will not be limiting your GPU. A 7800XD3 is better for gaming than the CPU selected, though that has more cores so might be better in productivity. The difference isn't that big though and it is an excellent CPU.

For example, you can view my benchmarks with a 7800X3D and a slower 4070 Super at 2560x1440. You can view the Cyberpunk settings here.


View attachment 343903


Dying Light 2. I have ray tracing enabled as well as DLSS.

View attachment 343904


If you are going to use ray tracing, you will need to use DLSS and possibly frame generation.
Thank you for this!

With his new screen hes gonna play at 2560x1440, since it doesnt support higher then that.
He wanted to grab a philips evnia screen at first, but then found out of the asus rog strix monitor, and decided that is good enough for what he wants, and not getting the evnia.

The evnia had 4k though, but he realised he'd need at least a 4080, or maybe even a 4090 to crank the settings up and still get enjoyable frame rates.

So all in all, its a step up for him nevertheless, with the 4070 ti super and the asus rog strix 1440p monitor.
He now has a simple samsung 24 inch screen that even lacks a DP connection lol. So, no HDR, no g-sync, no high refresh, no local dimming, nothing.

His new asus rog strix is 27 inch, has hdr400 (not the best, but still better then nothing) nvidia certified g-sync (so no compatible with g-sync), 175hz, and local dimming (although it isnt that great, but still better then nothing)

I myself am going to check out how games and stuff look at his screen, and then i might switch myself over to a samsung odyssey g7, which also has local dimming and even hdr600.

Nevertheless, i think my question is solved, so thanks a lot for everything!

This whole bottleneck drama needs to stop. Buy the card you want and leave the damn thing alone.

Honestly, if you think its drama, then i don't agree with it. I think its a fair question, cause were new to the table here, and don't know everything ourself yet.
But as the question is solved, probally fewer replies, or none whatsoever comes here anymore, so your wish is fulfilled i think ;)
 

Toothless

Tech, Games, and TPU!
Supporter
Joined
Mar 26, 2014
Messages
9,292 (2.52/day)
Location
Washington, USA
System Name Veral
Processor 5950x
Motherboard MSI MEG x570 Ace
Cooling Corsair H150i RGB Elite
Memory 4x16GB G.Skill TridentZ
Video Card(s) Powercolor 7900XTX Red Devil
Storage Crucial P5 Plus 1TB, Samsung 980 1TB, Teamgroup MP34 4TB
Display(s) Acer Nitro XZ342CK Pbmiiphx + 2x AOC 2425W
Case Fractal Design Meshify Lite 2
Audio Device(s) Blue Yeti + SteelSeries Arctis 5 / Samsung HW-T550
Power Supply Corsair HX850
Mouse Corsair Nightsword
Keyboard Corsair K55
VR HMD HP Reverb G2
Software Windows 11 Professional
Benchmark Scores PEBCAK
Honestly, if you think its drama, then i don't agree with it. I think its a fair question, cause were new to the table here, and don't know everything ourself yet.
But as the question is solved, probally fewer replies, or none whatsoever comes here anymore, so your wish is fulfilled i think ;)
A lot of people asking about bottlenecks is due to Youtube/Tiktok/whatever else videos and "bottleneck" websites making people second, third, fourth guess their hardware. It's all made a bigger deal than it actually is and yes it is drama.
 
Joined
May 18, 2011
Messages
1,070 (0.23/day)
Processor 7800X3D
Motherboard ASRock X670E Lightening PG
Cooling Thermalright Phantom Spirit 120SE CPU cooler, 3x 140mm, 1x 120mm case fan
Memory 32GB G.SKILL Flare X5 DDR5 6000 (PC5 48000) F5-6000J3038F16GX2-FX5
Video Card(s) ASUS RTX 4070
Storage 2TB Sandisk SSD, 2TB P31 SK Hynix, 4TB WD SN850X, WD Black 6TB, WD Red Plus 12TB
Case Fractal Design Definse S
Power Supply Seasonic Focus 750
Mouse Logitech Pro
Keyboard Corsair Strafe Cherry MX Silent w/red LED
VR HMD HTC Vive
Software Win 10 Pro
Thank you for this!

With his new screen hes gonna play at 2560x1440, since it doesnt support higher then that.

If that is the case I would get a 4070ti Super. It has 16GB VRAM, is almost as fast as a regular RTX 4080. The way I see it, that card will max out all games with 60+ frame rates if you use DLSS with ray tracing. With a 4080 Super it might be a bit faster and in some cases you can avoid using DLSS, but I think you'll still end up using DLSS regardless.



More examples here.

Sure, 60-62 frame rates is better than 50-52, but most likely they'll end up turning on DLSS. I don't like 60 frame rates and try to keep it above 90. So while a 4080 Super will be faster, it isn't fast enough to be worthwhile at that resolution. Without ray tracing a 4070ti Super is enough, with ray tracing a 4080 Super isn't enough without using DLSS.

Of course, you can always buy the fastest GPU you can afford and it depends on local prices. Personally I would save that money and put it towards something else even if it means saving it to upgrade to the next generation of GPUs next year.
 
Joined
Mar 23, 2024
Messages
24 (0.59/day)
Location
Almelo, Netherlands
System Name my system
Processor AMD Ryzen 7 5800X3D
Motherboard Asus Rog strix b550 gaming wifi II
Cooling noctua nh-u12s se-am4
Memory corsair vengeance 32gb (2x16) 3600mhz
Video Card(s) Gigabyte RTX 2070 Windforce 8g (planning to upgrade to an asus TUF 4070 TI super soon)
Storage 1x 240gb kingston ssd (win) 1x500gb kingston m.2 1x1tb 870 evo ssd 2x1tb hdd (storage)
Display(s) gigabyte aorus cv27f
Case Sharkoon Rebel c50 RGB
Audio Device(s) Creative soundblaster x-ae5 plus
Power Supply Seasonic Focus GX-850 ATX 3.0
Mouse Logitech g502x
Keyboard Microsoft ernomic buisness
VR HMD none
Software win 11 64 bit home
Benchmark Scores To be tested, so none yet
If that is the case I would get a 4070ti Super. It has 16GB VRAM, is almost as fast as a regular RTX 4080. The way I see it, that card will max out all games with 60+ frame rates if you use DLSS with ray tracing. With a 4080 Super it might be a bit faster and in some cases you can avoid using DLSS, but I think you'll still end up using DLSS regardless.



More examples here.

Sure, 60-62 frame rates is better than 50-52, but most likely they'll end up turning on DLSS. I don't like 60 frame rates and try to keep it above 90. So while a 4080 Super will be faster, it isn't fast enough to be worthwhile at that resolution. Without ray tracing a 4070ti Super is enough, with ray tracing a 4080 Super isn't enough without using DLSS.

Of course, you can always buy the fastest GPU you can afford and it depends on local prices. Personally I would save that money and put it towards something else even if it means saving it to upgrade to the next generation of GPUs next year.
Mind that he comes from a 3060, and i myself from a 2070, that are more or less equall in performance, and that we did upgrade the rest of our rig already in december last year.

He came from a ryzen 7 2700x, 32gb 2400mhz ddr4, and i came from a ryzen 7 1800x with the same ram.
So all in all with the system we both have now - he a 7950x3d, and 64gb 5600mhz ddr5 ram, me an 5800x3d with 32gb ddr4 3600mhz ram (planning to upgrade to 64 btw) - we really do need a better gpu to run all these new games.

Really the only game that i could run pretty well with raytracing, and my 1800x was cyberpunk back when it first launced. I got quite good fps with all ray tracing on that was available back then, while also having dlss on quality, but all of this was on 1080p ofcourse, as my screen doesnt support higher.

Havent tried yet even the new version, but honestly, that old cypberpunk was the only game that ran well with raytracing.
Maybe control ran well as well, but didnt played it that long honestly.

But later games ran worse and worse when applying ray tracing. But it is something that is of nowadays, so ye, if you want to have it, you gotta bring a good gpu to the table.

But.. As i said, the 4070ti, he (and me too) will get, and that will suffice for some years from now. I mean, I myself have always skipped at least one generation, which is true now again, going from a 20xx to a 40xx. Same goes for him, as the 3060 is - like i said before - almost equal to the 2070.

In any case, still thanks for your reply. As like i said before, more info is always welcome!
 
Joined
Oct 2, 2020
Messages
667 (0.51/day)
System Name ASUS TUF F15
Processor Intel Core i5-10300H
Motherboard ASUS FX506LHB
Cooling Laptop built-in cooling lol
Memory 24GB @ 2933 Dual Channel
Video Card(s) Intel UHD & Nvidia GTX 1650 Mobile
Storage WD Black SN770 NVMe 1TB PCIe 4.0
Display(s) Dell 27 4K Monitor S2721QS; Samsung Odyssey G55 Curved 2K 144 Hz LC27G55TQWRXEN
Audio Device(s) LOGITECH 2.1-channel
Power Supply ASUS 180W PSU (from more powerful ASUS TUF DASH F15 lol)
Mouse Logitech G604
Keyboard SteelSeries Apex 7 TKL
Software Windows 11 Pro
Hello!

I ask this question for a friend.

His pc is the following:

Ryzen 9 7950X3d
Asus Rog strix B650-A Gaming
64gb Corsair vengeance ddr5 5600mhz
Windows 11 pro 64 bit installed on a samsung evo 860

And currently he has a gtx 3060 12gb as a gpu.

He wants to upgrade to a new videocard in about a month from now, and he is wondering, if a 4070 ti super would still bottleneck his cpu.

I myself dont think so. But i'm not really sure about it, so therefore this question.

He also have the option to buy a 4080 super, but that one is more expensive, so its only an option if the 4070 ti super is guaranteed to bottleneck his 7950x3d.

So, can anyone provide some info on this matter?

Thank you very much on forehand from myself and my friend.

Greetings from Martijn

ps. He also believes that his ram might be bottlenecking his cpu, and asked me to type this in this post as well. But honestly, cause of it being 64gb and ddr5, I dont think thats an issue that big.
lol, usually GPU are "bottlenecked" by CPU. In this case, all is good, spend the amount that is comfortable for him. It's 18% difference in performance, but:
- TDP of 4080S is 320 W - umm, crazy.:kookoo:
- price difference is OBVIOUSLY more than 18%.:rolleyes::D
 

Toothless

Tech, Games, and TPU!
Supporter
Joined
Mar 26, 2014
Messages
9,292 (2.52/day)
Location
Washington, USA
System Name Veral
Processor 5950x
Motherboard MSI MEG x570 Ace
Cooling Corsair H150i RGB Elite
Memory 4x16GB G.Skill TridentZ
Video Card(s) Powercolor 7900XTX Red Devil
Storage Crucial P5 Plus 1TB, Samsung 980 1TB, Teamgroup MP34 4TB
Display(s) Acer Nitro XZ342CK Pbmiiphx + 2x AOC 2425W
Case Fractal Design Meshify Lite 2
Audio Device(s) Blue Yeti + SteelSeries Arctis 5 / Samsung HW-T550
Power Supply Corsair HX850
Mouse Corsair Nightsword
Keyboard Corsair K55
VR HMD HP Reverb G2
Software Windows 11 Professional
Benchmark Scores PEBCAK
lol, usually GPU are "bottlenecked" by CPU. In this case, all is good, spend the amount that is comfortable for him. It's 18% difference in performance, but:
- TDP of 4080S is 320 W - umm, crazy.:kookoo:
- price difference is OBVIOUSLY more than 18%.:rolleyes::D
I'm pretty sure the people buying high end cards can afford the power bill and power supply used. 320w isn't unheard of when we've had 250w for years and years.
 
Joined
Feb 23, 2019
Messages
5,644 (2.98/day)
Location
Poland
Processor Ryzen 7 5800X3D
Motherboard Gigabyte X570 Aorus Elite
Cooling Thermalright Phantom Spirit 120 SE
Memory 2x16 GB Crucial Ballistix 3600 CL16 Rev E @ 3800 CL16
Video Card(s) RTX3080 Ti FE
Storage SX8200 Pro 1 TB, Plextor M6Pro 256 GB, WD Blue 2TB
Display(s) LG 34GN850P-B
Case SilverStone Primera PM01 RGB
Audio Device(s) SoundBlaster G6 | Fidelio X2 | Sennheiser 6XX
Power Supply SeaSonic Focus Plus Gold 750W
Mouse Endgame Gear XM1R
Keyboard Wooting Two HE
Don't worry about bottlenecks on GPU side or else you'll always end up buying flagship cards. 4070 Ti is perfectly fine card for 1440p gaming and with things like DLSS you can cheat your way out of super high GPU loads.
 
Joined
Dec 16, 2021
Messages
121 (0.14/day)
Location
Denmark
Processor AMD Ryzen 7 3800X
Motherboard ASUS Prime X470-Pro
Cooling bequiet! Dark Rock Slim
Memory 64 GB ECC DDR4 2666 MHz (Samsung M391A2K43BB1-CTD)
Video Card(s) eVGA GTX 1080 SC Gaming, 8 GB
Storage 1 TB Samsung 970 EVO Plus, 1 TB Samsung 850 EVO, 4 TB Lexar NM790, 12 TB WD HDDs
Display(s) Acer Predator XB271HU
Case Corsair Obsidian 550D
Audio Device(s) Creative X-Fi Fatal1ty
Power Supply Seasonic X-Series 560W
Mouse Logitech G502
Keyboard Glorious GMMK
In my extremely humble opinion a 4070 Ti Super is pretty much the same as a 4080 which is pretty much the same as a 4080 Super. Just go with what is the cheapest of the three, which presumably is a 4070 Ti Super. Unless your friend is made of money. In that case the 4090 is the obvious choice.
 
Joined
Aug 18, 2022
Messages
308 (0.49/day)
In my extremely humble opinion a 4070 Ti Super is pretty much the same as a 4080 which is pretty much the same as a 4080 Super.
They may look the same on paper - the only real differences are 8448 cores for the 4070 Ti Super against 10240 for the 4080 Super that also has slightly faster memory. But it does make a measurable difference at 4K in games like The Last of Us Part 1. Yes it does come down to what you can afford but I think most of the buyers of these cards are planning to keep them for a while. So there is always the temptation to spend quite a bit more to get just a bit more performance.
 

dgianstefani

TPU Proofreader
Staff member
Joined
Dec 29, 2017
Messages
4,371 (1.89/day)
Location
Swansea, Wales
System Name Silent
Processor Ryzen 7800X3D @ 5.15ghz BCLK OC, TG AM5 High Performance Heatspreader
Motherboard ASUS ROG Strix X670E-I, chipset fans removed
Cooling Optimus AMD Raw Copper/Plexi, HWLABS Copper 240/40+240/30, D5, 4x Noctua A12x25, Mayhems Ultra Pure
Memory 32 GB Dominator Platinum 6150 MHz 26-36-36-48, 56.6ns AIDA, 2050 FLCK, 160 ns TRFC
Video Card(s) RTX 3080 Ti Founders Edition, Conductonaut Extreme, 18 W/mK MinusPad Extreme, Corsair XG7 Waterblock
Storage Intel Optane DC P1600X 118 GB, Samsung 990 Pro 2 TB
Display(s) 32" 240 Hz 1440p Samsung G7, 31.5" 165 Hz 1440p LG NanoIPS Ultragear
Case Sliger SM570 CNC Aluminium 13-Litre, 3D printed feet, custom front panel with pump/res combo
Audio Device(s) Audeze Maxwell Ultraviolet, Razer Nommo Pro
Power Supply SF750 Plat, transparent full custom cables, Sentinel Pro 1500 Online Double Conversion UPS w/Noctua
Mouse Razer Viper Pro V2 Mercury White w/Tiger Ice Skates & Pulsar Supergrip tape
Keyboard Wooting 60HE+ module, TOFU Redux Burgundy w/brass weight, Prismcaps White & Jellykey, lubed/modded
Software Windows 10 IoT Enterprise LTSC 19053.3803
Benchmark Scores Legendary
They may look the same on paper - the only real differences are 8448 cores for the 4070 Ti Super against 10240 for the 4080 Super that also has slightly faster memory. But it does make a measurable difference at 4K in games like The Last of Us Part 1. Yes it does come down to what you can afford but I think most of the buyers of these cards are planning to keep them for a while. So there is always the temptation to spend quite a bit more to get just a bit more performance.
4080S has the full cache, 64 MB, not the 48 MB of the 4070 Ti S.

Helps at 4K.
 
Joined
May 18, 2011
Messages
1,070 (0.23/day)
Processor 7800X3D
Motherboard ASRock X670E Lightening PG
Cooling Thermalright Phantom Spirit 120SE CPU cooler, 3x 140mm, 1x 120mm case fan
Memory 32GB G.SKILL Flare X5 DDR5 6000 (PC5 48000) F5-6000J3038F16GX2-FX5
Video Card(s) ASUS RTX 4070
Storage 2TB Sandisk SSD, 2TB P31 SK Hynix, 4TB WD SN850X, WD Black 6TB, WD Red Plus 12TB
Case Fractal Design Definse S
Power Supply Seasonic Focus 750
Mouse Logitech Pro
Keyboard Corsair Strafe Cherry MX Silent w/red LED
VR HMD HTC Vive
Software Win 10 Pro
Mind that he comes from a 3060, and i myself from a 2070, that are more or less equall in performance, and that we did upgrade the rest of our rig already in december last year.

He came from a ryzen 7 2700x, 32gb 2400mhz ddr4, and i came from a ryzen 7 1800x with the same ram.
So all in all with the system we both have now - he a 7950x3d, and 64gb 5600mhz ddr5 ram, me an 5800x3d with 32gb ddr4 3600mhz ram (planning to upgrade to 64 btw) - we really do need a better gpu to run all these new games.

Really the only game that i could run pretty well with raytracing, and my 1800x was cyberpunk back when it first launced. I got quite good fps with all ray tracing on that was available back then, while also having dlss on quality, but all of this was on 1080p ofcourse, as my screen doesnt support higher.

Havent tried yet even the new version, but honestly, that old cypberpunk was the only game that ran well with raytracing.
Maybe control ran well as well, but didnt played it that long honestly.

But later games ran worse and worse when applying ray tracing. But it is something that is of nowadays, so ye, if you want to have it, you gotta bring a good gpu to the table.

But.. As i said, the 4070ti, he (and me too) will get, and that will suffice for some years from now. I mean, I myself have always skipped at least one generation, which is true now again, going from a 20xx to a 40xx. Same goes for him, as the 3060 is - like i said before - almost equal to the 2070.

In any case, still thanks for your reply. As like i said before, more info is always welcome!

If you're only playing games 64GB of RAM is useless. Put that money into the GPU. Or save it and put it towards something else. A 4070ti Super will be a big jump over those GPUs. At 2560x1440 I personally wouldn't think going to a higher end GPU is worthwhile. Again, put that aside for your next GPU upgrade.

I think you'll be hard pressed to find a game that uses over 16GB of RAM, though it isn't that uncommon either. Mainly niche games like Flight Simulator and whatnot. Maybe some of the latest games will benefit from more than 16GB, which is why 32GB is recommended. But more than that is useless for gaming.
 
Joined
Dec 16, 2021
Messages
121 (0.14/day)
Location
Denmark
Processor AMD Ryzen 7 3800X
Motherboard ASUS Prime X470-Pro
Cooling bequiet! Dark Rock Slim
Memory 64 GB ECC DDR4 2666 MHz (Samsung M391A2K43BB1-CTD)
Video Card(s) eVGA GTX 1080 SC Gaming, 8 GB
Storage 1 TB Samsung 970 EVO Plus, 1 TB Samsung 850 EVO, 4 TB Lexar NM790, 12 TB WD HDDs
Display(s) Acer Predator XB271HU
Case Corsair Obsidian 550D
Audio Device(s) Creative X-Fi Fatal1ty
Power Supply Seasonic X-Series 560W
Mouse Logitech G502
Keyboard Glorious GMMK
They may look the same on paper - the only real differences are 8448 cores for the 4070 Ti Super against 10240 for the 4080 Super that also has slightly faster memory. But it does make a measurable difference at 4K in games like The Last of Us Part 1. Yes it does come down to what you can afford but I think most of the buyers of these cards are planning to keep them for a while. So there is always the temptation to spend quite a bit more to get just a bit more performance.
Of course my statement was explicitly oversimplified, but it still comes down to not overly massive differences in performance. I got a 4080 "cheap" (around 1050€) at the beginning of the year and I plan to keep it for the next many years. The 1080 currently in my system I got in 2018, so it has served me for 6 years. I fully expect the 4080 to do likewise.
 
Joined
Sep 3, 2019
Messages
2,987 (1.75/day)
Location
Thessaloniki, Greece
System Name PC on since Aug 2019, 1st CPU R5 3600 + ASUS ROG RX580 8GB >> MSI Gaming X RX5700XT (Jan 2020)
Processor Ryzen 9 5900X (July 2022), 150W PPT limit, 79C temp limit, CO -9~14
Motherboard Gigabyte X570 Aorus Pro (Rev1.0), BIOS F37h, AGESA V2 1.2.0.B
Cooling Arctic Liquid Freezer II 420mm Rev7 with off center mount for Ryzen, TIM: Kryonaut
Memory 2x16GB G.Skill Trident Z Neo GTZN (July 2022) 3600MHz 1.42V CL16-16-16-16-32-48 1T, tRFC:288, B-die
Video Card(s) Sapphire Nitro+ RX 7900XTX (Dec 2023) 314~465W (390W current) PowerLimit, 1060mV, Adrenalin v24.3.1
Storage Samsung NVMe: 980Pro 1TB(OS 2022), 970Pro 512GB(2019) / SATA-III: 850Pro 1TB(2015) 860Evo 1TB(2020)
Display(s) Dell Alienware AW3423DW 34" QD-OLED curved (1800R), 3440x1440 144Hz (max 175Hz) HDR1000, VRR on
Case None... naked on desk
Audio Device(s) Astro A50 headset
Power Supply Corsair HX750i, 80+ Platinum, 93% (250~700W), modular, single/dual rail (switch)
Mouse Logitech MX Master (Gen1)
Keyboard Logitech G15 (Gen2) w/ LCDSirReal applet
Software Windows 11 Home 64bit (v23H2, OSB 22631.3155)
Avg 3060/12GB relative performance from top 10 GPUs in non RT gaming. You just choose whatever your wallet can buy and go one and live your life.
Especially when the system runs one of the fastest CPUs out there.
Everything else is subject for consumption on YT content.

IMG_7968.jpeg
 
Joined
Sep 17, 2014
Messages
20,968 (5.96/day)
Location
The Washing Machine
Processor i7 8700k 4.6Ghz @ 1.24V
Motherboard AsRock Fatal1ty K6 Z370
Cooling beQuiet! Dark Rock Pro 3
Memory 16GB Corsair Vengeance LPX 3200/C16
Video Card(s) ASRock RX7900XT Phantom Gaming
Storage Samsung 850 EVO 1TB + Samsung 830 256GB + Crucial BX100 250GB + Toshiba 1TB HDD
Display(s) Gigabyte G34QWC (3440x1440)
Case Fractal Design Define R5
Audio Device(s) Harman Kardon AVR137 + 2.1
Power Supply EVGA Supernova G2 750W
Mouse XTRFY M42
Keyboard Lenovo Thinkpad Trackpoint II
Software W10 x64
This whole bottleneck drama needs to stop. Buy the card you want and leave the damn thing alone.
Yeah, if you're gonna spend money on a gaming rig, start off with a real gaming CPU, simple as, and add the best GPU you can pay, done. Bottlenecks are then self imposed. Every system has a bottleneck, its called peak performance. Even with the best CPU some games won't utilize 100% of GPU. Fact of life.

In my extremely humble opinion a 4070 Ti Super is pretty much the same as a 4080 which is pretty much the same as a 4080 Super. Just go with what is the cheapest of the three, which presumably is a 4070 Ti Super. Unless your friend is made of money. In that case the 4090 is the obvious choice.
This. We're late in gen as well, don't pay for top end GPUs anymore. Its a monumental waste of money.

4080S has the full cache, 64 MB, not the 48 MB of the 4070 Ti S.

Helps at 4K.
Aye but it really doesn't pay off relative to the cost difference.
 
Joined
May 18, 2011
Messages
1,070 (0.23/day)
Processor 7800X3D
Motherboard ASRock X670E Lightening PG
Cooling Thermalright Phantom Spirit 120SE CPU cooler, 3x 140mm, 1x 120mm case fan
Memory 32GB G.SKILL Flare X5 DDR5 6000 (PC5 48000) F5-6000J3038F16GX2-FX5
Video Card(s) ASUS RTX 4070
Storage 2TB Sandisk SSD, 2TB P31 SK Hynix, 4TB WD SN850X, WD Black 6TB, WD Red Plus 12TB
Case Fractal Design Definse S
Power Supply Seasonic Focus 750
Mouse Logitech Pro
Keyboard Corsair Strafe Cherry MX Silent w/red LED
VR HMD HTC Vive
Software Win 10 Pro
Aye but it really doesn't pay off relative to the cost difference.



4080 Super is around 3% faster than a 4080, so around 15% faster on average than a 4070ti Super. Though it will come down mostly to what games they play. Overall, an average of 15% more performance for 20% more cost. It isn't that bad when it comes to the performance/price ratio, but I would still recommend the 4070ti Super for 2560x1440. That 12-15% performance increase will generally be unnoticeable as frame rates are high enough in raster, or in ray tracing, will only give maybe 2-3 extra frame rates which isn't quite worth $200.



Of course there will be some outliers.




TLDR: OP, get a 4070ti Super. Skip the 64GB of RAM. You can probably save $350-400. If you find the 4070ti Super is not fast enough, a year from now you can put that $400 towards an RTX 5070 which will likely be faster and more efficient than a 4080 Super.
 
Joined
Oct 2, 2020
Messages
667 (0.51/day)
System Name ASUS TUF F15
Processor Intel Core i5-10300H
Motherboard ASUS FX506LHB
Cooling Laptop built-in cooling lol
Memory 24GB @ 2933 Dual Channel
Video Card(s) Intel UHD & Nvidia GTX 1650 Mobile
Storage WD Black SN770 NVMe 1TB PCIe 4.0
Display(s) Dell 27 4K Monitor S2721QS; Samsung Odyssey G55 Curved 2K 144 Hz LC27G55TQWRXEN
Audio Device(s) LOGITECH 2.1-channel
Power Supply ASUS 180W PSU (from more powerful ASUS TUF DASH F15 lol)
Mouse Logitech G604
Keyboard SteelSeries Apex 7 TKL
Software Windows 11 Pro
I'm pretty sure the people buying high end cards can afford the power bill and power supply used. 320w isn't unheard of when we've had 250w for years and years.
the problem isn't in "bills" or "psus". The problem is just the design. High number of power draw, monstrous cooler, and all of them doesn't give you 200+ fps in 4K ultra with RT in Cyberpunk, unless you turn on the scaling magic... LOL

If you're only playing games 64GB of RAM is useless. Put that money into the GPU. Or save it and put it towards something else. A 4070ti Super will be a big jump over those GPUs. At 2560x1440 I personally wouldn't think going to a higher end GPU is worthwhile. Again, put that aside for your next GPU upgrade.

I think you'll be hard pressed to find a game that uses over 16GB of RAM, though it isn't that uncommon either. Mainly niche games like Flight Simulator and whatnot. Maybe some of the latest games will benefit from more than 16GB, which is why 32GB is recommended. But more than that is useless for gaming.
LMFAO, you compare CONSOLE to PC or what? WINDOWS, my friend, likes RAM too, and I, personally, find it "OKAY" to open the game when I have browsers or other crap left OPENED too. So, that's WHY I recommend and will recommend 24-32 GB RAM for builds even since like 2020. One could play even with 8 GB RAM some games, closing everything in background and cutting Windows services lol
 
Joined
May 18, 2011
Messages
1,070 (0.23/day)
Processor 7800X3D
Motherboard ASRock X670E Lightening PG
Cooling Thermalright Phantom Spirit 120SE CPU cooler, 3x 140mm, 1x 120mm case fan
Memory 32GB G.SKILL Flare X5 DDR5 6000 (PC5 48000) F5-6000J3038F16GX2-FX5
Video Card(s) ASUS RTX 4070
Storage 2TB Sandisk SSD, 2TB P31 SK Hynix, 4TB WD SN850X, WD Black 6TB, WD Red Plus 12TB
Case Fractal Design Definse S
Power Supply Seasonic Focus 750
Mouse Logitech Pro
Keyboard Corsair Strafe Cherry MX Silent w/red LED
VR HMD HTC Vive
Software Win 10 Pro
LMFAO, you compare CONSOLE to PC or what? WINDOWS, my friend, likes RAM too, and I, personally, find it "OKAY" to open the game when I have browsers or other crap left OPENED too. So, that's WHY I recommend and will recommend 24-32 GB RAM for builds even since like 2020. One could play even with 8 GB RAM some games, closing everything in background and cutting Windows services lol

32GB is still plenty for that. If you have programs open that use so much RAM that 32GB is not enough you'll have having a CPU and/or GPU performance impact most likely. Example, if you're running UE5 or Blender in the background you'll probably loose some frame rates. Generally I find games crash more with constant minimizing as well. Some games work fine, others crash or have frame rate issues if constantly minimizing and maximizing. I can't recall which games it was but I recently played a game that would get frame rate issues if minimized 3-4 times. Restarting the game would of course fix it.

So I'd always recommend shutting down programs unless necessary. Leaving your browser, Discord or whatever on is fine of course. But if your background programs are using 10-20GB of RAM I assume they're quite demanding and not just in RAM.
 

Toothless

Tech, Games, and TPU!
Supporter
Joined
Mar 26, 2014
Messages
9,292 (2.52/day)
Location
Washington, USA
System Name Veral
Processor 5950x
Motherboard MSI MEG x570 Ace
Cooling Corsair H150i RGB Elite
Memory 4x16GB G.Skill TridentZ
Video Card(s) Powercolor 7900XTX Red Devil
Storage Crucial P5 Plus 1TB, Samsung 980 1TB, Teamgroup MP34 4TB
Display(s) Acer Nitro XZ342CK Pbmiiphx + 2x AOC 2425W
Case Fractal Design Meshify Lite 2
Audio Device(s) Blue Yeti + SteelSeries Arctis 5 / Samsung HW-T550
Power Supply Corsair HX850
Mouse Corsair Nightsword
Keyboard Corsair K55
VR HMD HP Reverb G2
Software Windows 11 Professional
Benchmark Scores PEBCAK
32GB is still plenty for that. If you have programs open that use so much RAM that 32GB is not enough you'll have having a CPU and/or GPU performance impact most likely. Example, if you're running UE5 or Blender in the background you'll probably loose some frame rates. Generally I find games crash more with constant minimizing as well. Some games work fine, others crash or have frame rate issues if constantly minimizing and maximizing. I can't recall which games it was but I recently played a game that would get frame rate issues if minimized 3-4 times. Restarting the game would of course fix it.

So I'd always recommend shutting down programs unless necessary. Leaving your browser, Discord or whatever on is fine of course. But if your background programs are using 10-20GB of RAM I assume they're quite demanding and not just in RAM.
That depends but applies on nearly all systems that aren't high core systems. I can sit idle around 20GB used but only like, 20% CPU usage at most. Memory amount is always based on user needs but overall yes closing background tasks will help and 32GB is plenty for majority users.
 
Joined
Sep 17, 2014
Messages
20,968 (5.96/day)
Location
The Washing Machine
Processor i7 8700k 4.6Ghz @ 1.24V
Motherboard AsRock Fatal1ty K6 Z370
Cooling beQuiet! Dark Rock Pro 3
Memory 16GB Corsair Vengeance LPX 3200/C16
Video Card(s) ASRock RX7900XT Phantom Gaming
Storage Samsung 850 EVO 1TB + Samsung 830 256GB + Crucial BX100 250GB + Toshiba 1TB HDD
Display(s) Gigabyte G34QWC (3440x1440)
Case Fractal Design Define R5
Audio Device(s) Harman Kardon AVR137 + 2.1
Power Supply EVGA Supernova G2 750W
Mouse XTRFY M42
Keyboard Lenovo Thinkpad Trackpoint II
Software W10 x64
That depends but applies on nearly all systems that aren't high core systems. I can sit idle around 20GB used but only like, 20% CPU usage at most. Memory amount is always based on user needs but overall yes closing background tasks will help and 32GB is plenty for majority users.
20% CPU usage is substantial and will likely impact game performance won't it?

I'm still here 'getting by' on 16GB... can't say I've ever noticed a detrimental impact from having too little, but then again, I'm on an ancient CPU as well.
 

Toothless

Tech, Games, and TPU!
Supporter
Joined
Mar 26, 2014
Messages
9,292 (2.52/day)
Location
Washington, USA
System Name Veral
Processor 5950x
Motherboard MSI MEG x570 Ace
Cooling Corsair H150i RGB Elite
Memory 4x16GB G.Skill TridentZ
Video Card(s) Powercolor 7900XTX Red Devil
Storage Crucial P5 Plus 1TB, Samsung 980 1TB, Teamgroup MP34 4TB
Display(s) Acer Nitro XZ342CK Pbmiiphx + 2x AOC 2425W
Case Fractal Design Meshify Lite 2
Audio Device(s) Blue Yeti + SteelSeries Arctis 5 / Samsung HW-T550
Power Supply Corsair HX850
Mouse Corsair Nightsword
Keyboard Corsair K55
VR HMD HP Reverb G2
Software Windows 11 Professional
Benchmark Scores PEBCAK
20% CPU usage is substantial and will likely impact game performance won't it?

I'm still here 'getting by' on 16GB... can't say I've ever noticed a detrimental impact from having too little, but then again, I'm on an ancient CPU as well.
Not for the high core count chips like my 5950x.
 
Joined
Sep 17, 2014
Messages
20,968 (5.96/day)
Location
The Washing Machine
Processor i7 8700k 4.6Ghz @ 1.24V
Motherboard AsRock Fatal1ty K6 Z370
Cooling beQuiet! Dark Rock Pro 3
Memory 16GB Corsair Vengeance LPX 3200/C16
Video Card(s) ASRock RX7900XT Phantom Gaming
Storage Samsung 850 EVO 1TB + Samsung 830 256GB + Crucial BX100 250GB + Toshiba 1TB HDD
Display(s) Gigabyte G34QWC (3440x1440)
Case Fractal Design Define R5
Audio Device(s) Harman Kardon AVR137 + 2.1
Power Supply EVGA Supernova G2 750W
Mouse XTRFY M42
Keyboard Lenovo Thinkpad Trackpoint II
Software W10 x64
Not for the high core count chips like my 5950x.
Well I reckon you have the cores available. But isn't RAM exactly the bottleneck here if you're moving that much data around? Not capacity wise, but in bandwidth/transfer speed? Especially on DDR4...?
 
Joined
Dec 25, 2020
Messages
4,697 (3.83/day)
Location
São Paulo, Brazil
System Name Project Kairi Mk. IV "Eternal Thunder"
Processor 13th Gen Intel Core i9-13900KS Special Edition
Motherboard MSI MEG Z690 ACE (MS-7D27) BIOS 1G
Cooling Noctua NH-D15S + NF-F12 industrialPPC-3000 w/ Thermalright BCF and NT-H1
Memory G.SKILL Trident Z5 RGB 32GB DDR5-6800 F5-6800J3445G16GX2-TZ5RK @ 6400 MT/s 30-38-38-38-70-2
Video Card(s) ASUS ROG Strix GeForce RTX™ 4080 16GB GDDR6X White OC Edition
Storage 1x WD Black SN750 500 GB NVMe + 4x WD VelociRaptor HLFS 300 GB HDDs
Display(s) 55-inch LG G3 OLED
Case Cooler Master MasterFrame 700
Audio Device(s) EVGA Nu Audio (classic) + Sony MDR-V7 cans
Power Supply EVGA 1300 G2 1.3kW 80+ Gold
Mouse Razer DeathAdder Essential Mercury White
Keyboard Redragon Shiva Lunar White
Software Windows 10 Enterprise 22H2
Benchmark Scores "Speed isn't life, it just makes it go faster."
lol, usually GPU are "bottlenecked" by CPU. In this case, all is good, spend the amount that is comfortable for him. It's 18% difference in performance, but:
- TDP of 4080S is 320 W - umm, crazy.:kookoo:
- price difference is OBVIOUSLY more than 18%.:rolleyes::D

320 W is a standard TDP for a GPU of this segment, it is lower than the RX 7900 XTX's, and adding to that, the RTX 4080 does not reach ~300 W for the vast majority of workloads, and I'm gaming at 4K. The 4080S is no different.

The 4080/S are, in fact, the most power efficient GPUs currently available from a raw performance to power draw perspective.
 

Toothless

Tech, Games, and TPU!
Supporter
Joined
Mar 26, 2014
Messages
9,292 (2.52/day)
Location
Washington, USA
System Name Veral
Processor 5950x
Motherboard MSI MEG x570 Ace
Cooling Corsair H150i RGB Elite
Memory 4x16GB G.Skill TridentZ
Video Card(s) Powercolor 7900XTX Red Devil
Storage Crucial P5 Plus 1TB, Samsung 980 1TB, Teamgroup MP34 4TB
Display(s) Acer Nitro XZ342CK Pbmiiphx + 2x AOC 2425W
Case Fractal Design Meshify Lite 2
Audio Device(s) Blue Yeti + SteelSeries Arctis 5 / Samsung HW-T550
Power Supply Corsair HX850
Mouse Corsair Nightsword
Keyboard Corsair K55
VR HMD HP Reverb G2
Software Windows 11 Professional
Benchmark Scores PEBCAK
Well I reckon you have the cores available. But isn't RAM exactly the bottleneck here if you're moving that much data around? Not capacity wise, but in bandwidth/transfer speed? Especially on DDR4...?
I haven't personally seen ram bandwidth being an issue but it could be something I've missed. I currently have three games, video stream, Discord, etc up with zero issues or any "laggy" feelings.
 

dgianstefani

TPU Proofreader
Staff member
Joined
Dec 29, 2017
Messages
4,371 (1.89/day)
Location
Swansea, Wales
System Name Silent
Processor Ryzen 7800X3D @ 5.15ghz BCLK OC, TG AM5 High Performance Heatspreader
Motherboard ASUS ROG Strix X670E-I, chipset fans removed
Cooling Optimus AMD Raw Copper/Plexi, HWLABS Copper 240/40+240/30, D5, 4x Noctua A12x25, Mayhems Ultra Pure
Memory 32 GB Dominator Platinum 6150 MHz 26-36-36-48, 56.6ns AIDA, 2050 FLCK, 160 ns TRFC
Video Card(s) RTX 3080 Ti Founders Edition, Conductonaut Extreme, 18 W/mK MinusPad Extreme, Corsair XG7 Waterblock
Storage Intel Optane DC P1600X 118 GB, Samsung 990 Pro 2 TB
Display(s) 32" 240 Hz 1440p Samsung G7, 31.5" 165 Hz 1440p LG NanoIPS Ultragear
Case Sliger SM570 CNC Aluminium 13-Litre, 3D printed feet, custom front panel with pump/res combo
Audio Device(s) Audeze Maxwell Ultraviolet, Razer Nommo Pro
Power Supply SF750 Plat, transparent full custom cables, Sentinel Pro 1500 Online Double Conversion UPS w/Noctua
Mouse Razer Viper Pro V2 Mercury White w/Tiger Ice Skates & Pulsar Supergrip tape
Keyboard Wooting 60HE+ module, TOFU Redux Burgundy w/brass weight, Prismcaps White & Jellykey, lubed/modded
Software Windows 10 IoT Enterprise LTSC 19053.3803
Benchmark Scores Legendary
I haven't personally seen ram bandwidth being an issue but it could be something I've missed. I currently have three games, video stream, Discord, etc up with zero issues or any "laggy" feelings.
Most games only use a few GB/s of bandwidth.

It's latency that causes performance improvements for games. Latency improves with higher MT at the same timings, so faster kits are generally lower latency, unless it's a particularly egregriously bad XMP/EXPO profile. It's easy to confuse higher MT ratings as giving the FPS (from bandwidth), but since you can measure how much bandwidth a game or application is using in various tools, it's pretty clear that it's something other than raw bandwidth.

More bandwidth helps resolve the latency issue in a non direct way, because you can make the entire memory transaction in one cycle, rather than taking several.

This is why it's not as simple as comparing first word latency of DDR4 vs DDR5, because the way data is read/written to the banks is different, so 1:1 latency/bandwidth comparisons don't work. There's also the issue of CPU memory controller gears.

Something most people probably aren't aware of for example, is due to the motherboard setting all subtimings, some of which can't be changed by the user in the BIOS, the same memory kit will be running at different settings on different boards/platforms. E.g. CMD rate 1T on AMD and 2T on Intel.

This is one of the reasons it's particularly prudent to manually set timings and voltages, auto settings for memory are as trash as auto settings for CPU overclocking, it's just most people don't realise XMP/EXPO only covers the main timings and main voltage.

For DDR5 especially, the subtimings are more influential on performance.
 

Toothless

Tech, Games, and TPU!
Supporter
Joined
Mar 26, 2014
Messages
9,292 (2.52/day)
Location
Washington, USA
System Name Veral
Processor 5950x
Motherboard MSI MEG x570 Ace
Cooling Corsair H150i RGB Elite
Memory 4x16GB G.Skill TridentZ
Video Card(s) Powercolor 7900XTX Red Devil
Storage Crucial P5 Plus 1TB, Samsung 980 1TB, Teamgroup MP34 4TB
Display(s) Acer Nitro XZ342CK Pbmiiphx + 2x AOC 2425W
Case Fractal Design Meshify Lite 2
Audio Device(s) Blue Yeti + SteelSeries Arctis 5 / Samsung HW-T550
Power Supply Corsair HX850
Mouse Corsair Nightsword
Keyboard Corsair K55
VR HMD HP Reverb G2
Software Windows 11 Professional
Benchmark Scores PEBCAK
Most games only use a few GB/s of bandwidth.

It's latency that causes performance improvements for games. Latency improves with higher MT at the same timings, so faster kits are generally lower latency, unless it's a particularly egregriously bad XMP/EXPO profile. It's easy to confuse higher MT ratings as giving the FPS (from bandwidth), but since you can measure how much bandwidth a game or application is using in various tools, it's pretty clear that it's something other than raw bandwidth.

More bandwidth helps resolve the latency issue in a non direct way, because you can make the entire memory transaction in one cycle, rather than taking several.

This is why it's not as simple as comparing first word latency of DDR4 vs DDR5, because the way data is read/written to the banks is different, so 1:1 latency/bandwidth comparisons don't work. There's also the issue of CPU memory controller gears.

Something most people probably aren't aware of for example, is due to the motherboard setting all subtimings, some of which can't be changed by the user in the BIOS, the same memory kit will be running at different settings on different boards/platforms. E.g. CMD rate 1T on AMD and 2T on Intel.

This is one of the reasons it's particularly prudent to manually set timings and voltages, auto settings for memory are as trash as auto settings for CPU overclocking, it's just most people don't realise XMP/EXPO only covers the main timings and main voltage.

For DDR5 especially, the subtimings are more influential on performance.
Yeah.. no.. current ram kit limits tweaking quite a bit. If I was running my 4000 2x32 kit then yeah I could get latency down quite a bit while keeping 3600-3800 for memory clocks. However my wife's 14900k needs it more for her stuff. I get about 70ns~ with the four dimms and at some point the chiplets of the 5950x are the stopping point of dropping latency more.

But yes, latency and all that helps.
 
Top