GPU Test System Update for 2025

[ribizly]

See @rainzor‘s reply. You are at the extreme end of the range, i tried to pick something more common, and intentionally mismatched settings. With 3 identical monitors running using the same timings you should see some power improvements. Cru might help


Very similar to 3080, and very little interest generally. After GeForce 50 releases I’ll drop most GeForce 30 cards anyway. Current+last gen+fastest from 2 gens ago is what i usually include

I don't agree. If there are little differences (within 1-2% for a resolution), then those could be in one line.
E.g. in this case for 1440p relative performance:
100%: 3090Ti/4070Ti
80%: 3080/7800XT/4070
67%: 7700XT/3070Ti/6800
61%: 3070/2080Ti/4060Ti
It would be much easier to read the chart as well, and could fit all the cards into the same performace level without problem.

But I understand how the things work, just don't like it.

 

[John974]

Nvidia has always been a bit harder on the cpu because of driver overhead. Is the performance difference vs AMD cards because the 9800x3d is fast enough to compensate for that and let Nvidia cards stretch their legs more?

Or is the latest AMD driver really slower?
On average the XTX was never slower vs the 4080 and now it is?

 

[EarthDog]

Silly question @W1zzard ... is it possible to have a dynamic chart for any that use the cost online as a variable? In that, every week (2 weeks? Month?), they can be updated? This work and these charts are all kinds of awesome, but those with money become almost instantly obsolete as prices change so frequently...especially when the new cards come out, last gen tends to drop. Or, perhaps two charts, one based on MSRP, the other 'real time'.

Just thinking out loud. Not sure if it's worth it.

As always, thanks for your dedication and efforts.

 

[W1zzard]

dynamic chart for any that use the cost online as a variable?

That's a good idea .. but realistically you'd have to be able to edit not only the tested card's price but all others, too?

Or, perhaps two charts, one based on MSRP, the other 'real time'.

an MSRP comparison has been included in the review for like year now I think

 

[10tothemin9volts]

I wanted to add Wukong Path Tracing, failed at the configuration and like half my results are broken. Will probably retest soon and add, because it is important for GeForce 50 and the market it addresses

Thanks, testing Cyberpunk 2077 pathtracing alone might be enough if u dont want to test too many titles bc any GPU that performs well there, is going to perform as well in other games' pathtracing? (of course, until games come out which use heavier pathtracing (Cyberpunk 2077 could do it too))

 

[EarthDog]

but realistically you'd have to be able to edit not only the tested card's price but all others, too?

That's the idea. Maybe it pulls pricing data from newegg or Amazon if the cards are on there. Dumps that data in the spreadsheet, forces a recalc and updates the chart. Pulls monthly, maybe. Sounds like a lot of effort for little return... bjt would be awesome if it exists and was accurate.

an MSRP comparison has been included in the review for like year now I think

I meant all cards in the chart compared by msrp, not solely the review item. Sorry if I missed it... repeatedly, lol.

 

[W1zzard]

missed it... repeatedly, lol.

XFX Radeon RX 7900 XTX Magnetic Air Review

The XFX Radeon RX 7900 XTX Magnetic Air comes with user-replaceable fans that pop out without any tools. It's a fantastic solution to clean your card and to simplify RMAs. The card achieved excellent performance results in our review, and the overclocking potential is highly impressive.

www.techpowerup.com

end of the page, this is what you are looking for, no?

 

[Mr_Engineer]

Nvidia has always been a bit harder on the cpu because of driver overhead. Is the performance difference vs AMD cards because the 9800x3d is fast enough to compensate for that and let Nvidia cards stretch their legs more?

Or is the latest AMD driver really slower?
On average the XTX was never slower vs the 4080 and now it is?

Historically AMD cards used to age better and gain more performance with newer drivers compared to Nvidia cards, but it seems like the roles are reversed now. Maybe AMD just gave up at this point, as they are focusing more on servers and AI.

Also, I've always felt that RDNA3 seems broken somehow, it did not have the performance that was intended/expected. Could be a bottleneck/inefficiency within the hardware. Let's hope AMD fixes the issues for RDNA4.

Great system and game selection @W1zzard . If it does get an RTX 5090 in the future, then I suggest the RAM is upgraded to 64GB.

 

[Rabit]

Will be nice to add second platform like Ryzen 5600 to compare GPU divers overhead in 2-3 games. * At-least for GPUs bellow $500

 

[ayurai]

which ssd is using in this setup?

 

[EarthDog]

XFX Radeon RX 7900 XTX Magnetic Air Review

The XFX Radeon RX 7900 XTX Magnetic Air comes with user-replaceable fans that pop out without any tools. It's a fantastic solution to clean your card and to simplify RMAs. The card achieved excellent performance results in our review, and the overclocking potential is highly impressive.

www.techpowerup.com

end of the page, this is what you are looking for, no?

Not really. But perhaps I'm still being dense...

Your first line says...."We looked up each card's current USD price on Newegg and used that"

These charts only use msrp..."If the tested card is at risk of being sold out and resold at higher prices, we estimate a price based on current market conditions and use it. In this case, the vendor MSRP is also indicated as an additional chart bar."

So, the only msrp listed is for the review card while the others are from newegg or estimations. An ALL msrp chart is what I'm asking about.

@Mr_Engineer

Great system and game selection @W1zzard . If it does get an RTX 5090 in the future, then I suggest the RAM is upgraded to 64GB.

Why? A full 87% of people have 32gb or less... 3% with 64 or more (steam stats).

 

[W1zzard]

Your first line says...."We looked up each card's current USD price on Newegg and used that"

Wrong section. Look for "Performance per MSRP" heading

 

[BlaezaLite]

This test rig needs firing up, surely?

 

[tajoh111]

Nvidia has always been a bit harder on the cpu because of driver overhead. Is the performance difference vs AMD cards because the 9800x3d is fast enough to compensate for that and let Nvidia cards stretch their legs more?

Or is the latest AMD driver really slower?
On average the XTX was never slower vs the 4080 and now it is?

Now AMD cards have a bit more overhead.

Look at the 1080p results with a 9800x3d vs a r5 5600 with a RTX 4060 vs a rx 7600. The RX 7600 and 7600xt lose a tad more performance with the weaker 5600.

 

[Everset]

when i was buying 6900xt, it was a bit behind of 3090 ti at 1080p, on new charts its roughly equal to non ti 3090. What happened ?

 

[TheinsanegamerN]

when i was buying 6900xt, it was a bit behind of 3090 ti at 1080p, on new charts its roughly equal to non ti 3090. What happened ?

AMD started prioritizing rDNA3 when it came out, in the months leading to launch. We're likely seeing the same thing now, AMD is gearing up for rDNA4, so rDNA3 is on the backburner and 2 is basically on life support.

Look at the last 10 games TPU tested, AMD dropped a whole tier in some titles.

 

[ben dover]

Would love to have the exact manufacturer tag on the DDR5 memory, plus at what voltage you are able to get those timings of 28-36-36-76 ???

Been wondering if Thermaltake RG33D516GX2-6200C32B with 32-38-38-80 6200 MT/s 1.3 V could possibly do it, at 1.45 or 1.5V maybe ?
Then you will need active cooling for prolonged sessions, or else you will have errors when getting above 55 C.

edit: because i just bought a F5_6000J2836G16GX2-FX5
32gb G.Skill kit which has 6000 MT/s and CAS Latency: 28-36-36-96, and i wondered why this Thermaltake stuff looks much better.

There are better kits than the one i bought, but they cost more:

CMT32GX5M2B6000Z30K or​

F5-6400J3039G16GX2-TR5G​