NVIDIA GeForce RTX 5090 Founders Edition

[Fonduetime]

Those are there to have the continuity with older benchmarks, that is fine!

Also people still play 10 year old games.

 

[wNotyarD]

What I missing from the review is screenshot/comparison between the modes actual quality.

I guess that could make a separate review on its own.

 

[mkppo]

Almost as expected. I did expect them to improve efficiency a bit though, they've pretty much flatlined. Needs more testing at different power levels suppose.

So this is basically a 4090 Ti Super of sorts when it comes to performance. ~30% more power for 30% more performance at 25% higher price. Zero efficiency improvements.

 

[Cowboystrekk]

Nice performanceboost in 4k, nice cooler although 40dB is too much for me. Efficiency and pricing=

 

[BigMack70]

If they're basing that award on how efficient it is per frame then it fully deserves it. According to the review there are only three GPUs that are more efficient per frame.

It's the same perf/watt as the 4090. Why on earth are we praising ZERO improvement after two years?

I love TPU reviews, but they always find the most random nonsense to praise Nvidia for.

 

[csendesmark]

I guess that could make a separate review on its own.

I am not afraid, they will.

 

[Vya Domus]

Why on earth are we praising ZERO improvement after two years

That's really the problem here, I don't think there has ever been a new generation with virtually zero improvements in perf/watt or perf/price, even the rebrands usually at least had lower prices. It's quite an achievement.

 

[BoggledBeagle]

Why on earth are we praising ZERO improvement after two years?

Technical progress is supposed to bring better value for consumers. In this particular case they got 30% more performance, RAM and power draw for 30% more money.

Gain for a consumer does not exist after two and half years of development.

The only positive thing is the frame generation and upscaling improvements.

 

[phints]

Okay fastest GPU money can buy, tons of VRAM for workstation tasks, 2-slot card is great, but insanely expensive and power consumption is horrendous!

Even idle power consumption is way too high. Wtf has Nvidia done here? This needs to be made on TSMC 2N so badly.

 

[BigMack70]

Technical progress is supposed to bring better value for consumers. In this particular case they got 30% more performance, RAM and power draw for 30% more money.

Gain for a consumer does not exist after two and half years of development.

The only positive thing is the frame generation and upscaling improvements.

Yup. This generation is a flop as a "next generation" of graphics cards.

I really expected at least some improvements, even knowing it was all going to be on the same manufacturing node as Ada. Instead, it's just a 40 series refresh.

 

[redeye]

Okay fastest GPU money can buy, tons of VRAM for workstation tasks, 2-slot card is great, but insanely expensive and power consumption is horrendous!

Even idle power consumption is way too high. Wtf has Nvidia done here? This needs to be made on TSMC 2N so badly.

cliché number one… if you can afford the price of the card, you can afford the price of electricity.
do Ferrari Lamborghini Bentley Rolls-Royce etc buyers worry about the gas consumption?

 

[JustBenching]

Okay fastest GPU money can buy, tons of VRAM for workstation tasks, 2-slot card is great, but insanely expensive and power consumption is horrendous!

Even idle power consumption is way too high. Wtf has Nvidia done here? This needs to be made on TSMC 2N so badly.

The overall power draw in non gaming tasks is probably due to the excess ram (and gddr7). Another reason why spamming ram in lower tier cards doesn't make much sense.

 

[SOAREVERSOR]

Technical progress is supposed to bring better value for consumers. In this particular case they got 30% more performance, RAM and power draw for 30% more money.

Gain for a consumer does not exist after two and half years of development.

The only positive thing is the frame generation and upscaling improvements.

WRONG. Technical progress means more costs attached. That is the nature of it. Progress means costs. Costs in research, costs in complexity, costs in production. Progress means pay more and pay a lot more. That is what it means. What drives down cost is getting rid of progress when it comes to performance. Want better value, deliver the same performance with newer nodes. Never progress. Put a complete halt on progress and then focus on manufacturing. But PC gamers don't want this. They want their cake and they want it free and they are to blame for the situation they are in not nvidia. T

 

[Dragam1337]

Correct. I think people confuse absolute power draw with efficiency

How does the 5090 undervolt ?

I run my 4090 at 2700 mhz @ 925mv, which gives 95% of the stock performance at roughly 300 watts.

5090 doesn't strike me as a great undervolter.

 

[wNotyarD]

My worry with the power draw isn't the power draw per se, but the fact it is actually hitting the ceiling of the 12V-2x6 specs. God forbid any manufacturing fault on cables, and fatigue/mishandling of the connectors.

 

[W1zzard]

@W1zzard Which software did you use? Looks like it's limited to +2000 (32Gbps) in most other reviews.

My own in-house

 

[BigMack70]

WRONG. Technical progress means more costs attached. That is the nature of it. Progress means costs. Costs in research, costs in complexity, costs in production. Progress means pay more and pay a lot more. That is what it means. What drives down cost is getting rid of progress when it comes to performance. Want better value, deliver the same performance with newer nodes. Never progress. Put a complete halt on progress and then focus on manufacturing. But PC gamers don't want this. They want their cake and they want it free and they are to blame for the situation they are in not nvidia. T

This makes very little sense.

It's completely reasonable to expect performance and value improvements even when the manufacturing node is the same. Maxwell was manufactured on the same node as Kepler, yet brought meaningful performance and efficiency improvements.

Nvidia has just jumped the shark this time because there is zero competition from AMD and all their profits are tied up in AI, so they deliver a product which is just "more of the same" to gamers.

 

[W1zzard]

@Solaris17 @Steevo

Card has been reassembled, only 3 screws left over. Runs perfectly fine, temps are only 2°C higher than LM. The fixme has been updated.

 

[phanbuey]

Massive meh, not worth it really at all until the new features are leveraged and even then... wow. 4090 MSRP buyers pretty frikken happy rn.

Yeah exactly - 4090 was really well priced @$1600, so if you bought that in 2022 or 2023, you now have a 5080+ performance... vs if you bought a $1100 4080 and now a $1100-1200 5080 - you would have had less performance for $2400.

 

[mtosev]

Any stores in the EU have the card listed with the price?

 

[THU31]

Happy to see that the minimum voltage is 0.8 V, down from 0.9 V on Ada. That should allow for more aggressive undervolting. That's pretty much the main thing that will determine if I'm going to buy the 5070 Ti. I'd love to be able to keep it at 200 W in the summer.

As for the 5090, RT performance looks rather disappointing. And it seems that the power limit is the culprit. The GPU clock is over 200 MHz lower in Cyberpunk RT vs. raster.
The 4090 would max out the power limit pretty much only in RT games, while the 5090 does it basically across the board (when not CPU limited).
It also seems the all these architectural improvements aren't really doing much. This is the 4th generation of RT cores, and the performance hit is still massive.

 

[Chrispy_]

What I missing from the review is screenshot/comparison between the modes actual quality.

I suspect that is coming later, as it requires quite a bit of work.

 

[dir_d]

Can't really judge this thing on price because its a Halo product but the power and temps.... I agree with many that said this should have been a 3 or maybe 4 slot card like the 4090.

 

[Steevo]

Can't really judge this thing on price because its a Halo product but the power and temps.... I agree with many that said this should have been a 3 or maybe 4 slot card like the 4090.

Makes you wonder what the core could do with liquid cooling. I bet with AM cards we see well into the 700W range and better cooling.

@Solaris17 @Steevo

Card has been reassembled, only 3 screws left over. Runs perfectly fine, temps are only 2°C higher than LM.

Cost savings in BOM and assembly, Nvidia will contact you shortly.

 

[W1zzard]

Cost savings in BOM and assembly, Nvidia will contact you shortly.

Hope these are not the hardware units required for DLSS