AMD Could Launch New Navi GPUs Soon

[Mephis]

...

I dont think they are releasing Ray Tracing card with NAVI. Maybe a refresh later on with RT based on NAVI but this NAVI launch will be without RT for sure. Console first with RT to see how things go and then focus on PC. Besides they still need to catch NV with the performance and adding RT feature, is not going to accomplish this.

You might be right about this. My only point is that Navi with RTRT exists. The only downside that I see to this strategy is that consoles won't be out until holidays of 2020. That would mean, no AMD RTRT for pc until 2021 at the earliest. That gives Nvidia basically almost 2+ years to establish a dominant standard that AMD would have to play catch up to. We are seeing how hard that is with Cuda.

 

[ratirt]

You might be right about this. My only point is that Navi with RTRT exists. The only downside that I see to this strategy is that consoles won't be out until holidays of 2020. That would mean, no AMD RTRT for pc until 2021 at the earliest. That gives Nvidia basically almost 2+ years to establish a dominant standard that AMD would have to play catch up to. We are seeing how hard that is with Cuda.

Honestly I wouldn't care more about RT now. There's no rush into the RT cause it is new and AMD's got other things to worry about. Besides AMD has been talking about RT for a long time and maybe they've got something and now they want to try it with the consoles. For me NV released RT case they wanted to be 1st doing it. Remember, AMD released first 7nm gpu node and somehow I see NV being "allergic" to whatever AMD does first in the GPU department.
If the RT from AMD hits the market, I wonder, if this is going to be AMD's development or something similar to what NV released in terms of the RT cores of some sort.

 

[bug]

I don't think so. Many, including me thought it was, but it seems Navi is just an answer to a new generation of GPU's. AMD is still working on their RTRT offerings. My guess is that they want to do a better launch than the RTX launch. While I think the RTX launch was successful, no one can deny that it could have been better. AMD is not done, as Ryzen has proven, they just getting started.

But with Navi out the door, when do you figure the real answer to Turing will come? Before or after Ampere launches?

 

[ratirt]

But with Navi out the door, when do you figure the real answer to Turing will come? Before or after Ampere launches?

Does it really matter if it is before or after? You won't buy it anyway cause like some other people here you are all attached to NV's nervous system. Why mock AMD launch for the big Navi?
You better focus on NV's ampere

 

[bug]

Does it really matter if it is before or after? You won't buy it anyway cause like some other people here you are all attached to NV's nervous system. Why mock AMD launch for the big Navi?
You better focus on NV's ampere

Well, I've been in the green camp for years because of Linux support. Now that AMD has (mostly) cleaned that up, I am actually paying attention to their launches as well. I don't game anymore, so there's no urgency for me to upgrade, but I still don't want to fall behind much.

And I'm not mocking the "launch for big Navi" because officially, there is nothing to mock so far.

 

[ratirt]

Well, I've been in the green camp for years because of Linux support. Now that AMD has (mostly) cleaned that up, I am actually paying attention to their launches as well. I don't game anymore, so there's no urgency for me to upgrade, but I still don't want to fall behind much.

And I'm not mocking the "launch for big Navi" because officially, there is nothing to mock so far.

Are you sure you weren't mocking? Giving the impression that no matter what AMD will do it is still going to be laughable cause of Ampere? Cause for me it does sound like it. It has been confirmed it is coming so not sure what you are after here. The dates haven't been set yet but they will come later. Officially, it is coming. Official release, you need to wait.

It is always good to keep track of what's in the market and what will be in the near future

 

[bug]

Are you sure you weren't mocking? Giving the impression that no matter what AMD will do it is still going to be laughable cause of Ampere? Cause for me it does sound like it. It has been confirmed it is coming so not sure what you are after here. The dates haven't been set yet but they will come later. Officially, it is coming. Official release, you need to wait.

It is always good to keep track of what's in the market and what will be in the near future

Can you post a link that says big Navi is coming?
Because Lisa Su only said they have something in the pipeline beyond what is already announced for the next quarters. Two quarters from now is already 2020, which is why I said if they launch then, they will probably be going against Ampere.
Nothing is known about Ampere either, but for the past few launches Nvidia led with the big silicon around April-June.

 

[lexluthermiester]

At this point until AMD pulls of something huge, it will be a vicious cycle.

It always is regardless of what comes out of the companies involved.

But with Navi out the door, when do you figure the real answer to Turing will come? Before or after Ampere launches?

While that's a fair point, who knows... We know they're working on RTRT so it's only a matter of time and sooner rather than later.

Are you sure you weren't mocking?

Bug was being a smartalec. He does that often, it's only playful. No worries mate!

 

[ratirt]

Can you post a link that says big Navi is coming?
Because Lisa Su only said they have something in the pipeline beyond what is already announced for the next quarters. Two quarters from now is already 2020, which is why I said if they launch then, they will probably be going against Ampere.
Nothing is known about Ampere either, but for the past few launches Nvidia led with the big silicon around April-June.

Link? Lisa Su said they have something in the pipeline. I know AMD will have a big Navi. BTW. can you show me a link that AMD states there will be no Big Navi? exactly I can ask you the same thing. You dont seem to get it. I don't care if they are going against ampere or not (same goes for AMD). Why do you people always have to bring what NV is launching as a marker. These two are totally different companies. NV launched RT amd didn't because there is no point for that. There is other stuff in the pipeline and there will be a "Big Navi". Who cares if it will go against Turing or ampere. I don't. Do you know why AMD is bringing "big Navi"? AMD is a GPU and CPU producer. CPU's came first now it is GPUs' turn. That's why they will get it out. 2019 of course not. Did I say AMD will launch it 2019? 2020 for sure.

Bug was being a smartalec. He does that often, it's only playful. No worries mate!

Why would I be worried? Because a whippersnapper is arguing with me? I don't think so

 

[bug]

Link? Lisa Su said they have something in the pipeline.

She said they have something high-end in the pipeline beyond what's already announced for the next quarters. Which amounts to nothing, because it's not like anyone was expecting AMD would just stop working after that.

I know AMD will have a big Navi.

I don't.

BTW. can you show me a link that AMD states there will be no Big Navi?

Obviously not. When I say I don't expect a big Navi, I'm going by two things:
1. Pushing 5700 to 5700XT decreases perf/W by 12%, which means 5700XT is already past the architecture's sweet spot. Thus, it's not going to be easy to scale this up.
2. AMD is already working on RTRT which is probably not going to be retrofitted to Navi.

That is why I guess (as opposed to you, who seem to just know), AMD's next moves will be to put Navi into $200 or less cards and come up with something else (Arcturus?) next year to compete at the top.

 

[ratirt]

She said they have something high-end in the pipeline beyond what's already announced for the next quarters. Which amounts to nothing, because it's not like anyone was expecting AMD would just stop working after that.

I don't.

Obviously not. When I say I don't expect a big Navi, I'm going by two things:
1. Pushing 5700 to 5700XT decreases perf/W by 12%, which means 5700XT is already past the architecture's sweet spot. Thus, it's not going to be easy to scale this up.
2. AMD is already working on RTRT which is probably not going to be retrofitted to Navi.

That is why I guess (as opposed to you, who seem to just know), AMD's next moves will be to put Navi into $200 or less cards and come up with something else (Arcturus?) next year to compete at the top.

It is a brand new arch. Which part you dont understand with this? AMD is a GPU company. They said there is more coming out and as you said high-end. They will increase the number of cors or make them more efficient which is obvious but they need to work on this. Why do you think they are releasing it after few months later than the 5700 non-xt and xt release? They've got some issues or it is not ready yet. 7nm, new arch and you already wanna have top notch product? Come on man. You are smarter than this.
They work on RT for consoles not PC Navi. There's absolutely no point for RT now in the PC gaming. The performance is not there. Isn't this obvious?

 

[Assimilator]

Many people are forgetting that Intel's current Ice Lake-U/-Y are still on the first generation 10nm node.

Incorrect, ICL is on 10nm+. There was only a single SKU released on the original 10nm node, the CNL Core i3-8121U, and the yields on that relatively simple chip were so poor that the iGPU didn't even work - it's such a POS that Intel has purged all evidence of its existence from their website.

Intel's 10nm DUV node isn't dead or broken, and the yield issues are solvable, but the "bigger" problem is achieving high enough production volume.

AnandTech just did a review of ICL retail silicon and the takeaway is that 10nm+ still takes a massive frequency hit compared to 14nm+++++++. ICL the microarchitecture has about 20% IPC uplift, but 10nm+ imposes a 20% frequency penalty. That means Intel can either launch ICL on 10nm for almost no net performance gain (great tactic considering Zen 2), or they can launch it on 14nm+++++++ and actually have new CPUs with a decent performance uplift that allow them to actually make some sales over Christmas. Which option do you think they're gonna pick?

 

[bug]

AnandTech just did a review of ICL retail silicon and the takeaway is that 10nm+ still takes a massive frequency hit compared to 14nm+++++++. ICL the microarchitecture has about 20% IPC uplift, but 10nm+ imposes a 20% frequency penalty. That means Intel can either launch ICL on 10nm for almost no net performance gain (great tactic considering Zen 2), or they can launch it on 14nm+++++++ and actually have new CPUs with a decent performance uplift that allow them to actually make some sales over Christmas. Which option do you think they're gonna pick?

Hint: The CPU on Ice Lake is about the same as Whiskey Lake, but the IGP is vastly better. So, which option do you think they're gonna pick?

But keep in mind these are 15W parts, they tell little about how this scales when TDP get more legroom. Then again, there's no danger of a desktop Ice Lake CPU sneaking up on us anytime soon...

 

[medi01]

meaning that AMD's answer to Turing is almost ready

Because faster, cheaper, more efficient 5700 / 5700XT cards are not an "answer".
Makes sense, right.

 

[bug]

Because faster, cheaper, more efficient 5700 / 5700XT cards are not an "answer".
Makes sense, right.

That would be just the 5700. The 5700XT is like 5% faster than 2060 Super, but 13% less efficient. And they're both missing many hardware capabilities compared to Turing.
But you are correct, that is AMD's answer to Turing. If only they would follow up with some nice $200-ish parts...

 

[medi01]

I wouldn't care more about RT now.

NV dude literally stated that the he expects first AAA game to _require_ hardware RT to come in 2023.

 

[bug]

NV dude literally stated that the he expects first AAA game to _require_ hardware RT to come in 2023.

But if you're a hardware enthusiast and want to check it out sooner or simply show support, you don't have to wait that long

 

[efikkan]

AMD isn't done, but neither is Nvidia. Navi is their response to Turing. It doesn't have RTRT yet, but that doesn't mean it isn't their response.

I generally try to avoid this type of wording, especially since it challenges the definition of the word "response". Navi was meant to compete with Pascal, but was delayed 1.5 years so it ends up competing with Turing instead, and even that is a stretch, because Pascal was after all a backup plan due to delays to Volta.
Let's take another one to make it even worse; Ice Lake(Sunny Cove) was intended to launch a few months after Zen 1, and was designed long before Intel knew any details of Zen at all. Now, the first Ice Lakes actually launched after Zen 2, and the rest will arrive around Zen 3, so can we then call Ice Lake/Sunny Cove a response to Zen(1/2/3)?

You might be right about this. My only point is that Navi with RTRT exists. The only downside that I see to this strategy is that consoles won't be out until holidays of 2020. That would mean, no AMD RTRT for pc until 2021 at the earliest. That gives Nvidia basically almost 2+ years to establish a dominant standard that AMD would have to play catch up to. We are seeing how hard that is with Cuda.

If any Navi based GPUs will have some kind of accelerated RTRT, it will be something AMD threw in there "last minute", and will be nowhere close to the power of Nvidia's solution.

But remember that while RTRT will eventually become widespread, the process will be slow, and it will primarily be a nice to have feature until then. In the second half of 2020, we expect Nvidia to launch their next generation, and if AMD can't match this in efficiency, it wouldn't matter if they have some primitive RTRT or not. Then AMD will continue to be stuck only really competing in the low-end, having <10% of the mid-range market share and no presence in the high-end.

 

[medi01]

if AMD can't match this in efficiency, it wouldn't matter if they have some primitive RTRT or not.

(it might be some other RT "efficiency" or I somehow don't see it in current gen RTX)

AMD clearly plans to do it differently than nVidia.
Both of its major semi-custom clients want RT and will be getting AMD's version of it.
Consoles sell at around 20-40 million units a year, so, combined, 40+ million + AMD's GPUs, APUs.

If AMD's solution is not compatible, too bad for NV.
If AMD's solution is much faster than NVs, too bad for NV.
And, wait a sec, if it is somehow much slower, again, too bad for NV< as it would be consoles setting the target for game devs.

 

[bug]

(it might be some other RT "efficiency" or I somehow don't see it in current gen RTX)

AMD clearly plans to do it differently than nVidia.
Both of its major semi-custom clients want RT and will be getting AMD's version of it.
Consoles sell at around 20-40 million units a year, so, combined, 40+ million + AMD's GPUs, APUs.

If AMD's solution is not compatible, too bad for NV.
If AMD's solution is much faster than NVs, too bad for NV.
And, wait a sec, if it is somehow much slower, again, too bad for NV< as it would be consoles setting the target for game devs.

I'm pretty sure Xbox uses DXR, so AMD has their work pretty much cut out for them

 

[efikkan]

(it might be some other RT "efficiency" or I somehow don't see it in current gen RTX)

I wasn't referring to RT "efficiency", but efficiency in general.
Turing already scales better than Navi, despite Navi having the advantage of a more advanced node, an advantage which will evaporate when Nvidia launches their next generation. And if AMD can't compete with Nvidia in non-RT, it wouldn't matter if they have some primitive RTRT or not.

 

[Ran]

Upcoming quarters = read what she says without adding own wishes. It could be earliest in Q4 2019, perhaps Q1 2020 or in worst case scenario Q2.
Anyway this is good news for everyone, even if nVidia would tweak a super duper version of their GPUs.... consumers win.

 

[Darmok N Jalad]

Incorrect, ICL is on 10nm+. There was only a single SKU released on the original 10nm node, the CNL Core i3-8121U, and the yields on that relatively simple chip were so poor that the iGPU didn't even work - it's such a POS that Intel has purged all evidence of its existence from their website.



AnandTech just did a review of ICL retail silicon and the takeaway is that 10nm+ still takes a massive frequency hit compared to 14nm+++++++. ICL the microarchitecture has about 20% IPC uplift, but 10nm+ imposes a 20% frequency penalty. That means Intel can either launch ICL on 10nm for almost no net performance gain (great tactic considering Zen 2), or they can launch it on 14nm+++++++ and actually have new CPUs with a decent performance uplift that allow them to actually make some sales over Christmas. Which option do you think they're gonna pick?

Hint: The CPU on Ice Lake is about the same as Whiskey Lake, but the IGP is vastly better. So, which option do you think they're gonna pick?

But keep in mind these are 15W parts, they tell little about how this scales when TDP get more legroom. Then again, there's no danger of a desktop Ice Lake CPU sneaking up on us anytime soon...

I really wonder if we will see desktop SKUs of Ice Lake at all. The IPC gain is good, but the mobile SKUs seem to suggest there are issues getting the frequency up to 14+++ levels. Even the IGP isn’t seeing higher clocks, and it depends on faster RAM. Put mobile Ryzen on the same RAM as Ice Lake, and I don’t think it would be very close.

 

[Mephis]

(it might be some other RT "efficiency" or I somehow don't see it in current gen RTX)

AMD clearly plans to do it differently than nVidia.
Both of its major semi-custom clients want RT and will be getting AMD's version of it.
Consoles sell at around 20-40 million units a year, so, combined, 40+ million + AMD's GPUs, APUs.

If AMD's solution is not compatible, too bad for NV.
If AMD's solution is much faster than NVs, too bad for NV.
And, wait a sec, if it is somehow much slower, again, too bad for NV< as it would be consoles setting the target for game devs.

Except that right now the major consoles run AMD APUs and pc games still run better on Nvidia hardware. And as someone else mentioned, the next Xbox will definetly use DXR, which RTX is compatable with. AMD hasn't been able to use their console dominance to push devs to their camp yet, I seriously doubt they will be able to with RTRT.

 

[Darmok N Jalad]

Except that right now the major consoles run AMD APUs and pc games still run better on Nvidia hardware. And as someone else mentioned, the next Xbox will definetly use DXR, which RTX is compatable with. AMD hasn't been able to use their console dominance to push devs to their camp yet, I seriously doubt they will be able to with RTRT.

Current consoles are powered by Jaguar cores. What developers have been able to squeeze out of that hardware is impressive, IMO. Any modern desktop CPU would run circles around the x86 cores found in Xbox and PS4. And we all know how nvidia has gotten developers in their camp—many PC games launch with an nvidia logo on the splash screen. I’m sure that isn’t because they are nvidia fanboys.