GPU upgrade for 9900k

[Aj5002]

Hello,

 

I am currently trying to decide between upgrading my gpu from a 2080ti to a 7900xtx or Rtx 4090 and could really use some outside opinions. The gpu would be used with a i9 9900k for 4k Gaming. Any opinions would be greatly appreciated. 

 

Thanks!

[Dedayog]

How much  money do you have?  7900XTX is $1000, 4090 is $1600+.

 

Really it's more about how much you want to spend.  The 9900K may not provide enough at 4K though, may want to look at a total upgrade.

 

 

[Tetras]

With the RDNA 2 / Ampere generation, HUB found that AMD cards performed better with slower CPUs, so I'd be inclined to go AMD if you can confirm this is still the case for RDNA 3 / Ada.

[Aj5002]

17 minutes ago, Dedayog said:

How much  money do you have?  7900XTX is $1000, 4090 is $1600+.

 

Really it's more about how much you want to spend.  The 9900K may not provide enough at 4K though, may want to look at a total upgrade.

 

 

I have a little over  2k to spend on some upgrades. I guess the most most pertinent question would be, if the 9 9900k would bottle neck the 4090 to the point I wouldn't really see a difference between the 7900xtx and 4090. I will also most likely be upgrading my mobo/cpu some time next year.

[Agall]

6 minutes ago, Aj5002 said:

I have a little over  2k to spend on some upgrades. I guess the most most pertinent question would be, if the 9 9900k would bottle neck the 4090 to the point I wouldn't really see a difference between the 7900xtx and 4090. I will also most likely be upgrading my mobo/cpu some time next year.

I'd rebuild a new system than dropping in an RTX 4090 into a 9900k system. A 9900k will limit an RTX 4090 regardless of resolution. Even in the most multithreaded optimized AAA single player games you'll still see a limitation from just raw single threaded/IPC performance.

 

A 7900 XTX and new platform, like a 7800x3D, 13600k or 13700k is a better choice with a $2k budget.

[Aj5002]

44 minutes ago, Agall said:

I'd rebuild a new system than dropping in an RTX 4090 into a 9900k system. A 9900k will limit an RTX 4090 regardless of resolution. Even in the most multithreaded optimized AAA single player games you'll still see a limitation from just raw single threaded/IPC performance.

 

A 7900 XTX and new platform, like a 7800x3D, 13600k or 13700k is a better choice with a $2k budget.

Thanks for the reply!  this is another option I will have to consider. looks like I could do a full rebuild with a 7800x3d and 7900xtx for about 3200 CAD, And this would allow me to sell my current build for around 1k. 

* Edit CAD price*

[Agall]

12 minutes ago, Aj5002 said:

Thanks for the reply!  this is another option I will have to consider. looks like I could do a full rebuild with a 7800x3d and 7900xtx for about 2500 CAD, And this would allow me to sell my current build for around 1k. 

Not a bad option, if you're already running a decent NVMe drive, you can simply retain it and put a new one in your old system with a fresh install of Windows. Windows 10/11 will gladly accept a drive transplant, actually impressive how well it works.

 

If you're curious about some of the data regarding CPU choice, I made a thread specifically for Warframe, which has a hybrid pseudo-MMO type game. Being there's large population open areas, 2-4 player locally hosted coop, and pure single player experiences. The framerate on a high end system being variable between 30-1200 fps.

 

 

There was only one scenario where a 4790k vs 7950x3D had the same framerate, and it was single player, enclosed area with no enemies. Keeping in mind that a 4790k isn't far off from a 9900k in single threaded performance. Intel didn't get off their butt when it came to improvements until 11th generation to keep up with AMD's 3rd generation Ryzen.

 

Traditional CPU benchmarking is almost purely academic, since the games they use to demonstrate the differences are supposed to be relatively consistent to show a proper delta. Unless you're looking at a direct comparison for a specific game, those benchmarks are only academic.

 

What they don't do is test games most people actually play that handshake to servers and have CPU draw calls for player positioning. Mostly because properly testing this involves substantially more work and you can never truly replicate the scenario without having every person on the server participate (and having a solid direct connection to the server which isn't always the case).

 

The somewhat academic benchmark I did in Warframe is a 'close enough' representation of the various scenarios that I learned from playing the game from over a decade while understanding how the backend works, and I also did it at 4K ultra to get the absolute best case scenario for the 4790k. It also helps in debunking the false premise that CPU choice doesn't matter as much for 4K ultra, which isn't even true for academic benchmarks unless the game is horribly GPU bound even with an overclocked RTX 4090.