News Nvidia Ada Lovelace and GeForce RTX 40-Series: Everything We Know

One nitpick with this way of phrasing: "That means the big Infinity Cache gave AMD a 50% boost to effective bandwidth".

The Cache on the GPUs doesn't make it so the card has a higher bandwidth, much like AMD's 3D VCache is not making DDR4 magically have more bandwidth. I know what the implied point is, but I think it shouldn't be explained that way at all. Preventing using the GDDR/DDR BUS to fetch data is not the same as increasing the effective bandwidth of it. You saturate that cache and you're back to using the slow lane. On initial load, you still use the slow lane. Etc...

Other than that, thanks for the information. I do not look forward to 600W GPUs. Ugh.

Regards.
 
Reactions: thisisaname

escksu

Respectable
Aug 8, 2019
604
224
2,260
0
I reckon 1000w gpu isnt that far away...

Not a good thing for power consumption to keep going up when pple are all talking about climate change and going green
 

spongiemaster

Estimable
Dec 12, 2019
1,803
907
3,560
0
Other than that, thanks for the information. I do not look forward to 600W GPUs. Ugh.
Unless you're shopping for a $2000+ GPU, you're not going to have to worry about 600W any time soon. These new flagships are going to be the equivalent of SLI setups from years ago minus the headaches of needing SLI profiles for proper performance. You'll only need one physical slot, but the cooler is going to take up 4 like old school dual slot card SLI.
 
Reactions: Sluggotg

Tom Sunday

Prominent
Jul 24, 2020
69
17
535
0
The one thing we know for sure is it is not going to be cheap!
I will be happy to snagging a basic RTX 3090 (dreaming it costing me around $700 for a GPU generation almost 2-years old) perhaps next year in January or so? Then a 4K TV as more money becomes available. The RTX 40-series is totally crazy in my view as how much power can one ever need. Besides you are right, it will also not be cheap.
 

warezme

Distinguished
Dec 18, 2006
2,416
29
19,840
20
I have been using a 1000w PS for many years but it's getting really long on the tooth so I purchased a new 1200w, waiting to get installed one of these days. I don't really like the idea of needing so much power but I remember the days of 2 and 4 card SLI , I used to run and that was excessive. Now a single card can run circles around all that without the driver and game compatibility issues so it is better.
 
Reactions: Tom Sunday
Sep 16, 2021
17
6
15
0
What are GPUs going to do in the next (50) generation? If power increased this much again, we'd be bumping up against the maximum wattage for a north american wall outlet.
 
What are GPUs going to do in the next (50) generation? If power increased this much again, we'd be bumping up against the maximum wattage for a north american wall outlet.
Until the current generation there had been slight decreases or staying about the same for several generations. It seems after all other consumer electronics have upped their game for decreasing their products power requirements the gpu industry has gone the other way.
 
Reactions: RodroX

JarredWaltonGPU

Senior GPU Editor
Editor
Feb 21, 2020
1,143
1,025
4,070
1
One nitpick with this way of phrasing: "That means the big Infinity Cache gave AMD a 50% boost to effective bandwidth".

The Cache on the GPUs doesn't make it so the card has a higher bandwidth, much like AMD's 3D VCache is not making DDR4 magically have more bandwidth. I know what the implied point is, but I think it shouldn't be explained that way at all. Preventing using the GDDR/DDR BUS to fetch data is not the same as increasing the effective bandwidth of it. You saturate that cache and you're back to using the slow lane. On initial load, you still use the slow lane. Etc...

Other than that, thanks for the information. I do not look forward to 600W GPUs. Ugh.

Regards.
Cache hits reduce going to GDDR6/GDDR6X memory, which means you get the equivalent of more bandwidth. That's what "effective bandwidth" means. Or put another way, a large L3 cache had a hit rate of something like 50% higher than no L3 cache for AMD, which means 50% of memory accesses that formerly went to the GDDR6 didn't need to go there. AMD even said "effective bandwidth" with some of its RDNA 2 presentations, so if it's good enough for AMD I figure it's good enough for us.

It's the same as saying 16Gbps for GDDR6 memory speeds, with an "effective clock" of 8GHz. Technically, the base GDDR6 clock is actually 2GHz. It then sends eight bits per clock (quad-pumped and DDR), which gives us 16Gbps. That's the data rate, but the clock speed is nowhere near 8GHz. And yet, that's what most GPU utilities will report as the memory speed.
 
Cache hits reduce going to GDDR6/GDDR6X memory, which means you get the equivalent of more bandwidth. That's what "effective bandwidth" means. Or put another way, a large L3 cache had a hit rate of something like 50% higher than no L3 cache for AMD, which means 50% of memory accesses that formerly went to the GDDR6 didn't need to go there. AMD even said "effective bandwidth" with some of its RDNA 2 presentations, so if it's good enough for AMD I figure it's good enough for us.

It's the same as saying 16Gbps for GDDR6 memory speeds, with an "effective clock" of 8GHz. Technically, the base GDDR6 clock is actually 2GHz. It then sends eight bits per clock (quad-pumped and DDR), which gives us 16Gbps. That's the data rate, but the clock speed is nowhere near 8GHz. And yet, that's what most GPU utilities will report as the memory speed.
I know how it works, but it's just not factually correct to portray it like that. Just like MT/s vs the Mhz when reporting the RAM speeds. It is misleading, that's is all. In that light, then if you have a big enough cache, then you have infinite VRAM bandwidth "equivalent" since you'll never use it? That sounds iffy, even if you can draw that parallel.

As I said, it's just a nitpick.

Regards.
 

blppt

Distinguished
Jun 6, 2008
500
29
19,010
0
I will be happy to snagging a basic RTX 3090 (dreaming it costing me around $700 for a GPU generation almost 2-years old) perhaps next year in January or so? Then a 4K TV as more money becomes available. The RTX 40-series is totally crazy in my view as how much power can one ever need. Besides you are right, it will also not be cheap.
The problem with the 3090 is that even now its not great performing with its "killer feature" (that being raytracing). Better than the AMD RX6900XT, but AAA games that implement raytracing like CP2077 and DL2 pretty much require you to use DLSS to get good framerates at 4k with RT on.

Honestly, if you're rocking a 2xxx series nvidia right now, I wouldn't even bother with the 3 series. Its just not that much better.
 
Reactions: Tom Sunday

JarredWaltonGPU

Senior GPU Editor
Editor
Feb 21, 2020
1,143
1,025
4,070
1
The problem with the 3090 is that even now its not great performing with its "killer feature" (that being raytracing). Better than the AMD RX6900XT, but AAA games that implement raytracing like CP2077 and DL2 pretty much require you to use DLSS to get good framerates at 4k with RT on.

Honestly, if you're rocking a 2xxx series nvidia right now, I wouldn't even bother with the 3 series. Its just not that much better.
If he means getting "RTX 3090-like performance" rather than an actual RTX 3090, I think waiting for RTX 4070 or 4080 or whatever would be a good plan. The 3070 basically matches or exceeds the performance of the 2080 Ti, at theoretically half the cost. I hope we'll get a card for around $700 that will be faster in most games than an RTX 3090 with Ada. The card might only have 10–12GB of memory, but that should be sufficient.
 
Reactions: Sluggotg and blppt
Wow, 600 watts consumer grade GPU on the horizon. This is nuts

Go to wonder, if the cooler on some RTX 3090 can take up to 3.5 slots, this 600watts monster will probably require an hybrid design, or a 5 slot cooler?, And how much hot will the vram run? ..... crazy times.
 
Reactions: Sluggotg

JarredWaltonGPU

Senior GPU Editor
Editor
Feb 21, 2020
1,143
1,025
4,070
1
Wow, 600 watts consumer grade GPU on the horizon. This is nuts

Go to wonder, if the cooler on some RTX 3090 can take up to 3.5 slots, this 600watts monster will probably require an hybrid design, or a 5 slot cooler?, And how much hot will the vram run? ..... crazy times.
The "fix" for the hot running GDDR6X is simply to get them covered by the main vapor chamber that cools the GPU. The memory isn't consuming anywhere near as much power (and thus generating as much heat) as the GPU itself. But most cards only have thermal pads and basically heatspreaders on the memory. The RTX 3090 Ti cards seem to have been tweaked to cool the memory better. Asus has a larger heatsink bracket for just the memory, that has relatively large fins. It's still not going to be as effective as the heatpipes and vapor chambers used on the GPU itself, but it's better than nothing. The RTX 3090 for example had 12 chips on the back of the cards that basically just pressed up against the backplate of the card. There was no active cooling, and while the memory definitely got hot, a wraparound heatsink would have done wonders.

Basically, 3-4 slots with a large heatsink and multiple fans, plus a vapor chamber, should be capable of handling 600W of heat. But skip out on the vapor chamber and it will be problematic.
 
The "fix" for the hot running GDDR6X is simply to get them covered by the main vapor chamber that cools the GPU. The memory isn't consuming anywhere near as much power (and thus generating as much heat) as the GPU itself. But most cards only have thermal pads and basically heatspreaders on the memory. The RTX 3090 Ti cards seem to have been tweaked to cool the memory better. Asus has a larger heatsink bracket for just the memory, that has relatively large fins. It's still not going to be as effective as the heatpipes and vapor chambers used on the GPU itself, but it's better than nothing. The RTX 3090 for example had 12 chips on the back of the cards that basically just pressed up against the backplate of the card. There was no active cooling, and while the memory definitely got hot, a wraparound heatsink would have done wonders.

Basically, 3-4 slots with a large heatsink and multiple fans, plus a vapor chamber, should be capable of handling 600W of heat. But skip out on the vapor chamber and it will be problematic.
Yeah I know, to bad its only now, many months after the launch, that some makers are putting a decent cooling for the vram.
 

Tom Sunday

Prominent
Jul 24, 2020
69
17
535
0
I don't really like the idea of needing so much power. Now a single card can run circles around all that!
Yes indeed I myself never thought that I would require more than a 1000 Watt PSU under any circumstances. At the latest local computer show here people around the tables were talking about 1600 Watt units to future proofing ones new system. Then with GPU prices still not being in reach for the man on the street...my new system build budget is now clearly out of the window. The only way out for me is now working harder on my 'third-shift' and getting more overtime opportunities. 2023 in this respect cannot come soon enough and perhaps affording me in buying former and or older generation hardware at much reduced pricing.
 

ASK THE COMMUNITY