NVIDIA GeForce RTX 5000 Series “Blackwell” TDPs Leaked, Entire Lineup Unified with 12+4 Pin Power Connector

[Shou Miko]

as leaks have suggested the RTX 5070 might not even bet the RTX 4070 but will still cost properly more.
(I know several users will correct me on this, but I am still saying this, and only time in 2025 can tell what will happen fra NV's side if they don't just stick to their nickname NGreedia....)

I really hope that the "rumored" 8800 XT if AMD do not change the name is as good performance wise as power efficiency looks with rumours started with 220-275watt it would have been cool I believe to see AMD do something like this.

Energy Saving Mode: 220W
Normal Mode: 250W
OC Mode: 275W

It could be a physical switch on the board or a software toggle that just required a reboot than users could test which mode was the best for them power/performance wise.

 

[londiste]

Energy Saving Mode: 220W
Normal Mode: 250W
OC Mode: 275W

It could be a physical switch on the board or a software toggle that just required a reboot than users could test which mode was the best for them power/performance wise.

What? This has always been an option as is as easy or easier today. Open Adrenalin (or Nvidia app, or Afterburner) and set power limit to where you want it. In downwards direction it should work perfectly. Doesn't even require a reboot.

 

[Shou Miko]

What? This has always been an option as is as easy or easier today. Open Adrenalin (or Nvidia app, or Afterburner) and set power limit to where you want it. In downwards direction it should work perfectly. Doesn't even require a reboot.

My RTX 4070 refuses to budge from 195W under full load, even tried undervolting it, it doesn't give a damn it just keeps going up to 195W even adding about is it 2000Mhz on the ram still just runs at 195W.

Plus it was meant for normal people that doesn't really do much and it would be easy.

I never had an issue really with my AMD cards they did great but my current Nvidia card from Asus just refuses.

 

[Bomby569]

as leaks have suggested the RTX 5070 might not even bet the RTX 4070 but will still cost properly more.

leaks just want clicks, and the more they make you upset the better for rage views, that makes no sense. Sure it happened before but there was the insane amount of unsold cards from the covid/crypto days to get rid off. That will not happen again, that's why they are no longer making 4000 series cards and stocks are dropping.

 

[MxPhenom 216]

The "normal" TDP has increased a LOT in 2 decades. And gains from overclocking have gone down, also a LOT. From outside factors, cost of power has increased in most countries since then.

2 decades ago was 2004.
Cream of the crop GPUs were 6800 Ultra with its horrible 105W and X800 XT at 85-90W.
Best CPUs were Athlon 64s with 90W TDP. On Intel side the much-maligned Prescott P4s and older Gallatin P4EEs at 115W.
Potential overclocking gains, especially for 24/7 usage, were significant.
In most cases, TDP was not the limiting factor for performance and often enough parts did not consume up to TDP.

Compare this with today where high-end GPU consumes 300W and more. High-end CPU consumes over 200W.
Gains from overclocking are kind of there but for 24/7 usage the power cost for any performance increases is very bad.
Oh, and TDP is generally the limiting factor as well

We have gone from 250W or so for a high-end PC in 2004 to 600W or so in 2024. Or potentially much more if you have a 4090 and something like "TDP is just a suggestion" Intel CPU.

It would be a totally different story if performance hasnt improved much, but thats not the case. Performance these days is absolutely bonkers compared to what we used to have.

 

[Zach_01]

It would be a totally different story if performance hasnt improved much, but thats not the case. Performance these days is absolutely bonkers compared to what we used to have.

True
In 20years power increased 5-10x and performance... 300-400x if not more

 

[Sir Beregond]

I think you're mistaken about the jump in IPC from Ampere to Ada. The SMX is essentially the same; it was Ampere that saw nearly a 25% increase in clock normalized throughput per SMX compared to Turing. Turing also improved performance per SMX when compared to Pascal; just compare a 2070 Super to a 1080.

Ampere also improved the data pathing from Turing. Where Turing had a split of INT32 and FP32 per clock cycle, Ampere modified this pathing where the INT32 path could also do FP32, so you either had INT32/FP32 or FP32/FP32 per clock cycle per core. Ada is basically the same as Ampere (at least in how it manages that data per clock cycle). Biggest change from Ampere seems to be the massive increase in L2 cache and then of course new Tensor and RT cores and some other minor changes.

 

[AnotherReader]

Ampere also improved the data pathing from Turing. Where Turing had a split of INT32 and FP32 per clock cycle, Ampere modified this pathing where the INT32 path could also do FP32, so you either had INT32/FP32 or FP32/FP32 per clock cycle per core. Ada is basically the same as Ampere (at least in how it manages that data per clock cycle). Biggest change from Ampere seems to be the massive increase in L2 cache and then of course new Tensor and RT cores and some other minor changes.

For the most part, Ada is to Ampere what Pascal was to Maxwell: higher clocks with the same microarchitecture leading to much higher performance. For changes outside the shaders, your list covers it: the one standout feature is shader execution reordering.

 

[Sir Beregond]

For the most part, Ada is to Ampere what Pascal was to Maxwell: higher clocks with the same microarchitecture leading to much higher performance. For changes outside the shaders, your list covers it: the one standout feature is shader execution reordering.

Thanks. The shader execution reordering was what I was thinking of for "other minor changes" but couldn't remember what it was called.

Also going from Samsung 8nm (a 10nm node) to TSMC N4 (5nm node) made a huge difference too. Very much like Maxwell to Pascal, exactly like you say.

 

[tfdsaf]

Why would anyone want to buy a new PSU in order to run a 5060 or 5070? I don't get it. Why push a new PSU on cards that need up to 300w?

 

[AusWolf]

Even entry level cards requiring power through the new standard connector is a real bummer. The 5050 could have been a nice HTPC upgrade.

 

[:D:D]

tdp is nuts with gpu today.

Yeah, drives me nuts too when TDP is used in place of nvidia TGP.

 

[Bomby569]

Even entry level cards requiring power through the new standard connector is a real bummer. The 5050 could have been a nice HTPC upgrade.

what?! That most likely wont need a connector at all, and i don't care what any leaks say. Especially a PSU manufacturer that never uses the exact values

 

[AusWolf]

what?! That most likely wont need a connector at all, and i don't care what any leaks say. Especially a PSU manufacturer that never uses the exact values

I don't expect any card that doesn't need external power these days, especially from Nvidia, but fingers crossed.

 

[Bomby569]

I don't expect any card that doesn't need external power these days, especially from Nvidia, but fingers crossed.

the 3050 doesn't need one. And don't tell me is one generation old, because nvidia is actively selling them new on their own website. If you had a 4050 it most likely wouldn't require one either

 

[Zach_01]

Ay GPU card above 70-75W requires external power and even 3050 8GB FE is 130W.
Newer 4060 8GB FE is 115W and laptop oriented 4050 is 100W.

Very hard today for a GPU to be below 100W and can do even basic gaming.

NVIDIA GeForce RTX 3050 8 GB Specs

NVIDIA GA106, 1777 MHz, 2560 Cores, 80 TMUs, 32 ROPs, 8192 MB GDDR6, 1750 MHz, 128 bit

www.techpowerup.com

NVIDIA GeForce RTX 4060 Specs

NVIDIA AD107, 2460 MHz, 3072 Cores, 96 TMUs, 48 ROPs, 8192 MB GDDR6, 2125 MHz, 128 bit

www.techpowerup.com

Cant tell where 5050 will land or if it will be a dGPU for desktops.

 

[nguyen]

Ay GPU card above 70-75W requires external power and even 3050 8GB FE is 130W.
Newer 4060 8GB FE is 115W and laptop oriented 4050 is 100W.

Very hard today for a GPU to be below 100W and can do even basic gaming.

NVIDIA GeForce RTX 3050 8 GB Specs

NVIDIA GA106, 1777 MHz, 2560 Cores, 80 TMUs, 32 ROPs, 8192 MB GDDR6, 1750 MHz, 128 bit

www.techpowerup.com

NVIDIA GeForce RTX 4060 Specs

NVIDIA AD107, 2460 MHz, 3072 Cores, 96 TMUs, 48 ROPs, 8192 MB GDDR6, 2125 MHz, 128 bit

www.techpowerup.com

Cant tell where 5050 will land or if it will be a dGPU for desktops.

3050 6GB LP seems to be doing okay for not needing a power connector

There is also the RTX 2000 ADA Workstation GPU without power connector