AMD's next-gen RDNA 2 rumor: 40-50% faster than GeForce RTX 2080 Ti

[EarthDog]

Personally I don't mind having a 350W card

That's you and likely not some of these users who double talk through power (complain when NV uses more, but OK when AMD does, lol). Surely that is from both sides of the track, but AMD users here are surely a vocal group... lol.

EDIT: Cute comment

 

[BoboOOZ]

That's you and likely not some of these users who double talk through power (complain when NV uses more, but OK when AMD does, lol). Surely that is from both sides of the track, but AMD users here are surely a vocal group... lol.

Well, for a general consumer product, or a pro-oriented product, I understand the problem, but high-end gaming cards are, well, for gamers, so why complain about a few extra watts?

And there's somebody driving the boat? I haven't noticed then, I've seen so much fanboy flame around here, I was certain it's only the wind

 

[EarthDog]

Well, for a general consumer product, or a pro-oriented product, I understand the problem, but high-end gaming cards are, well, for gamers, so why complain about a few extra watts?

And there's somebody driving the boat? I haven't noticed then, I've seen so much fanboy flame around here, I was certain it's only the wind

It's like wallpaper... changes with the times and who's doing it... but I digress.

A few extra watts? My man... Nvidia's flagship was 225W and you're good up to 350W? People just don't want to deal with cooling such loads... most gamers rock 1080p and good with mid-range and below. So yeah, if the flagship start at 350W, lol on that.

That said, I know you used 350W as an example... and I agree that if it is within 'a few watts' people won't care. But, it wasn't within a few watts (5700XT to 2070) it was 20W/10%. If that gap increases, anyone who doesn't have bias will complain.

 

[BoboOOZ]

A few extra watts? My man... Nvidia's flagship was 225W and you're good up to 350W? People just don't want to deal with cooling such loads... most gamers rock 1080p and good with mid-range and below. So yeah, if the flagship start at 350W, lol on that.

That said, I know you used 350W as an example... and I agree that if it is within 'a few watts' people won't care. But, it wasn't within a few watts (5700XT to 2070) it was 20W/10%. If that gap increases, anyone who doesn't have bias will complain.

Well, I agree that for 1080p/144FPS 150W is enough. But we don't need flagships for that.

But for high refresh ultra-wide or for 4k/120 FPS I am prepared to accept 350W or even more, as long as the card is reasonably silent and it doesn't cost an arm.
An arm is 700 euro for me, btw .

 

[bug]

Well, I agree that for 1080p/144FPS 150W is enough. But we don't need flagships for that.

But for high refresh ultra-wide or for 4k/120 FPS I am prepared to accept 350W or even more, as long as the card is reasonably silent and it doesn't cost an arm.
An arm is 700 euro for me, btw .

You have to account for what a 350W high-end card does to the mid-range, too
Fortunately, 375W is the most you can get out of a PCIe card (and that's with two 8 pin connectors), no one will go so close to that limit.

 

[TheLostSwede]

When it comes to radeon performance - rumours have been proven to be fake. At least my memory doesnt go back enough to recall a point in time when they were true.
They boil down to "radeon amazing otherworldly performance incoming" only to be proven false.

I'll believe it when i see it.
(it'd be great if it does to be clear)

Rumours ≠ fact.
Fake news is something else entirely.

Rumours should always be taken with a healthy dose of NaCl.

 

[BoboOOZ]

You have to account for what a 350W high-end card does to the mid-range, too
Fortunately, 375W is the most you can get out of a PCIe card (and that's with two 8 pin connectors), no one will go so close to that limit.

Hahah, you say that because you don't listen to rumors.

But I was hearing yesterday night some crazy rumors about Nvidia using a new power connector on some high-end card just because of that (going above 350W).

 

[HD64G]

Wow, you're so good you actually know the power draw. Kudos, dude.

I, for one, hope the rumor is true (yeah, yeah, I know better), because we really need more HP in the mid-range for 4k gaming to actually become mainstream.

We have enough elements to calculate on both performance and power draw after official slides from AMD showed the +50% efficiency vs Navi gen1 and 80CUs of Navi21 is known for some time now. Minimum clocks of RDNA2 is given from the next-gen consoles also.

I hope AMD manages to increase their market share on the top tier of the GPU market as that will push prices down both for new and used GPUs. We as customers need that as much as we needed Zen to arrive and create a healthy CPU market as the one we have now with $400 for 12C/24T for 3900X.

 

[cucker tarlson]

Hahah, you say that because you don't listen to rumors.

But I was hearing yesterday night some crazy rumors about Nvidia using a new power connector on some high-end card just because of that (going above 350W).

I heard about the new connector too,it'll charge your phone at 225w striaght off the single 12-pin

We have enough elements to calculate on both performance and power draw after official slides from AMD showed the +50% efficiency vs Navi gen1 and 80CUs of Navi21 is known for some time now. Minimum clocks of RDNA2 is given from the next-gen consoles also.

I hope AMD manages to increase their market share on the top tier of the GPU market as that will push prices down both for new and used GPUs. We as customers need that as much as we needed Zen to arrive and create a healthy CPU market as the one we have now with $400 for 12C/24T for 3900X.

if 50% perf/wat is the only known fact it'd be wise to see what amd achieved with +50% perf/wat last time they had it.you may be disappointed that it wasn't 1.5x performance of a card that is 1.5x faster (2.25x).not even 1.5x performance actually.
and no,80cu is not "known for some time".it's "rumored for some time" with no basis except clickbait tech sources on TT/YT

 

[BoboOOZ]

I heard about the new connector too,it'll charge your phone at 225w striaght off the single 12-pin

As long as the RT performance of your phone is 10X that of a 2080Ti, who cares . we're trolling too much at this point.

if 50% perf/wat is the only known fact it'd be wise to see what amd achieved with +50% perf/wat last time they had it.
and no,80cu is not "known for some time".it's "rumored for some time" with no basis except clickbait tech sources on TT/YT

The thing is, RDNA2 doesn't need 80CU to have 2X the performance of the Navi10, there's also IPC and clock speeds in the equations. 64 CU could be enough.

 

[kapone32]

Wow, you're so good you actually know the power draw. Kudos, dude.

I, for one, hope the rumor is true (yeah, yeah, I know better), because we really need more HP in the mid-range for 4k gaming to actually become mainstream.

Exactly we need a card to serve the masses. I have a 1.2KW PSU so the power draw is a non factor for me. The price has to be right though because if it is too high in the stratosphere it will fail, regardless of the 2080TI being faster or not. I actually do not like when people reference the 2080TI vs RDNA2 that is not AMD's official guidance.

 

[cucker tarlson]

As long as the RT performance of your phone is 10X that of a 2080Ti, who cares . we're trolling too much at this point.

The thing is, RDNA2 doesn't need 80CU to have 2X the performance of the Navi10, there's also IPC and clock speeds in the equations. 64 CU could be enough.

doesn't perf/wat include ipc ?

 

[bug]

Hahah, you say that because you don't listen to rumors.

But I was hearing yesterday night some crazy rumors about Nvidia using a new power connector on some high-end card just because of that (going above 350W).

Using two 8 pin connectors is already outside PCI-SIG specs, I doubt anything on top of that will make its way into consumer space.

doesn't perf/wat include ipc ?

Nope. IPC is perf/clock. Higher clocks mean higher power draw, but one does not include the other.

 

[cucker tarlson]

Using two 8 pin connectors is already outside PCI-SIG specs, I doubt anything on top of that will make its way into consumer space.


Nope. IPC is perf/clock. Higher clocks mean higher power draw, but one does not include the other.

but ipc is a way to achieve higher perf/wat

 

[BoboOOZ]

doesn't perf/wat include ipc ?

Perf includes already everything: IPC gains, efficiency from shrinking, better efficiency from the maturing process, improving the architecture, improving clock speeds, etc.
But you don't necessarily need to have any IPC gains to improve perf/watt. The simplest way is just to downclock the chip.

Anyways, what I pointed out to you is that 80CU is not needed for RDNA2 to achieve 2X Navi10 performance.

 

[EarthDog]

Well, I agree that for 1080p/144FPS 150W is enough. But we don't need flagships for that.

But for high refresh ultra-wide or for 4k/120 FPS I am prepared to accept 350W or even more, as long as the card is reasonably silent and it doesn't cost an arm.
An arm is 700 euro for me, btw .

That's you bud... you. 350W+ single GPU... lol.. I hope not.

The simplest way is just to downclock the chip.

No?

You are also lowering performance this way...

 

[BoboOOZ]

No?

You are also lowering performance this way...

Second part is true, but first part is not.
You are indeed lowering performance, but you are lowering electrical consumption more. The 5700XT is most efficient in its Apple variant at slightly above 1GHz, it's still decent at 1.6GHz and well, we know how "glorious" it is at 1.9GHz and above...

 

[EarthDog]

Second part is true, but first part is not.
You are indeed lowering performance, but you are lowering electrical consumption more. The 5700XT is most efficient in its Apple variant at slightly above 1GHz, it's still decent at 1.6GHz and well, we know how "glorious" it is at 1.9GHz and above...

More? How do you know? You list the 5700XT.... I can buy that because of its power use compared to performance... but what about every other card?

So again I say, NO as you lower performance and power this way... no way to tell which is "more" or how much per MHz/GHz it is actually lowering power. No. lol.

 

[cucker tarlson]

That's you bud... you. 350W+ single GPU... lol.. I hope not.

No?

You are also lowering performance this way...

yup.
even if you keep it quiet,which is actually easy on 300w cards with how good coolers have become,you still have to deal with heat output.
I'd say I'd draw a line at 250w,but ideally I wanna be closer to 200w.Can't afford it with 2070S,gotta be overclocked for 1440p.Maybe next time it'll be a better idea to get a bigger card but run it stock.

 

[BoboOOZ]

More? How do you know? You list the 5700XT.... I can buy that because of its power use compared to performance... but what about every other card?

So again I say, NO as you lower performance and power this way... no way to tell which is "more" or how much per MHz/GHz it is actually lowering power. No. lol.

Well, yes, just look at the mobile graphic cards. Look at Nvidias 2060, for instance, which variant is the mùost efficient? The 80W MaxQ version, of course. The numbers are out there, this isn't a difficult question to answer. I only took the example of the 5700XT because it's the most telling, you have the results for 2X the clock speed increase available.

 

[moproblems99]

I'm still sticking with my initial guess that big navi/rdna2 will fall somewhere between a 2080ti and Ampre's flagship (non titan). My biggest question is power use to get there on the same process...

Whereas on the NV side, its a new arch plus a die shrink.

Weren't the "rumors" of "Big" Ampere already over 300w, hence the 12 pin power connector?

I hope AMD isn't worse...on the flip side I'll be able to cook eggs and bacon while gaming. I can see it now:. RGB grease catch. I should patent that.

 

[EarthDog]

Well, yes, just look at the mobile graphic cards. Look at Nvidias 2060, for instance, which variant is the mùost efficient? The 80W MaxQ version, of course. The numbers are out there, this isn't a difficult question to answer. I only took the example of the 5700XT because it's the most telling, you have the results for 2X the clock speed increase available.

Sure, but each card is different man.. you just can lower the clocks and accomplish that goal. Sorry.

Weren't the "rumors" of "Big" Ampere already over 300w, hence the 12 pin power connector?

I hope AMD isn't worse...on the flip side I'll be able to cook eggs and bacon while gaming. I can see it now:. RGB grease catch. I should patent that.

I haven't come across that rumor... however, I've seen you (someone) say it twice now in this thread.

3 hours ago... https://www.pcgamer.com/nvidia-ampere-12pin-power-connector/

they aren't buying it.......

And think about it... what kind of a BEAST would Ampre have to be at over 350W? If we consider what we have now at 225W... plus a node shrink plus new arch, plus 50% more power to use.... I don't think any competitor stands a chance on the performance front if that is true.

So many rumors... so many people taking these as the gospel and standing by them... god bless the lemmings...... every one!

 

[BoboOOZ]

I hope AMD isn't worse...on the flip side I'll be able to cook eggs and bacon while gaming. I can see it now:. RGB grease catch. I should patent that.

High cholesterol AND gaming? That's a really unhealthy life you plan on leading...

 

[Vayra86]

but ipc is a way to achieve higher perf/wat

No, higher perf/watt is. IPC is instructions per clock, but if you can do more within a single tick that tick is likely also drawing more power.

 

[BoboOOZ]

Sure, but each card is different man.. you just can lower the clocks and accomplish that goal. Sorry.

No need to be sorry, just show me that card. the card which is more efficient at a higher clock speed.