The future of AMD in graphics

[Topweasel]

I dunno what you saw.

It showed up waaaaaaay before, I think someone actually leaked their roadmaps into public.

Both Navi dies that will launch in ~August will be small (and fast).

It replaces everything, top to bottom.
Bigger gaming part is probably late'19-early'20 though.

I can look them up, but my memory is people guessing at a Vega20 and a Polaris30. But actual roadmaps didn't have it till about 2 years ago and well after what we know as Navi showed up. Just because people were right that there would be a Vega20 using AMDs naming convention doesn't mean that AMD had already made a decision to go forward with one.

Look I know you feel confident in your opinion. But it's just that. I feel otherwise. Can we just leave it at that.

 

[Yotsugi]

but my memory is people guessing at a Vega20 and a Polaris30

Vega20 showed up way, way, way earlier, on leaked ROCm roadmaps.

AMD had already made a decision to go forward with one.

They made that decision ages ago, since they needed a Hawaii replacement, which they got.

Look I know you feel confident in your opinion. But it's just that. I feel otherwise

You only have to wait some more.

Can we just leave it at that.

That's not how you have fun.

 

[Mopetar]

They may not need to even release a NAVI die that will be able to compete against Vega 20. They could keep Radeon Vii for 12-16 months and then in Q2-Q3 2020 they may release a next gen chip to compete against NVIDIAs 7nm gpus.

If that's the case they can't compete against NVidia on 7nm because the Radeon VII can't event compete with NVidia on 12nm. Navi doesn't have to be phenomenal for AMD to succeed with it, but it does need to be more than a stopgap, because they eventually need a compelling product. The 4000 and 5000 series showed that when they have a great product, the market will respond favorably.

I didnt know where else to put this,

Intel Discrete GPU Boss Raja Koduri Explains Why He Left AMD
https://hothardware.com/news/intel-...-explains-why-he-left-amd#AX6gCVXcB4tQBuYo.99

If you buy that, I've got a bridge to sell you.

 

[Yotsugi]

Radeon VII can't event compete with NVidia on 12nm

That's not exactly a consumer-focused product.
GK110 wasn't amazing as a consumer card either, and GK210 never was one.

Navi doesn't have to be phenomenal

It does, for they need to dig themselves into people's brains.

The 4000 and 5000 series showed that when they have a great product

4k was meh, but 5k was truly good, and actually gave AMD some profits and decent mss for once.

 

[Topweasel]

Vega20 showed up way, way, way earlier, on leaked ROCm roadmaps.

They made that decision ages ago, since they needed a Hawaii replacement, which they got.

As of here, I can make some mental room for Vega20 that was 14nm+ (12nm) and that getting scrapped and a new Vega20 as 7nm starting work, we heard reports shortly after that Vega20 is dead and maybe that was the 14nm+ version. But Vega20 as a 7nm chip is relatively new from what I gathered.

That's not how you have fun.

Having fun is playing back and forth, I think this, you think that, and each gives a reasoning for that. You have presented everything as cut and dry facts, interjecting your "fact" in every conversation regardless of topic, but with nothing to back it up. It's not a conversation and honestly I am kind of tired of it. You are free to do so, its a free country, but I am stepping out of conversations with you if that is the way you choose to continue.

 

[Yotsugi]

But Vega20 as a 7nm chip is relatively new from what I gathered.

Vega20 was a 7nm chip even on ancient leaked ROCm roadmaps.
With xGMI and 1/2 DPFP, the very exact product you've got shipping Q4'18.

but with nothing to back it up

You only have to wait.

 

[Topweasel]

Crap forgot to post the link

 

[Yotsugi]

Crap forgot to post the link

That's not the leaked ROCm roadmap.

 

[Topweasel]

That's not the leaked ROCm roadmap.

I am not searching for your mystery Roadmap. I am showing you an official roadmap that AMD presented to the press and their investors in the middle of 2017. Can you link anything more concrete then that?

 

[Yotsugi]

I am not searching for your mystery Roadmap

It isn't that hard.

I am showing you an official roadmap that AMD presented to the press and their investors in the middle of 2017

It's a puff piece with no actual chips listed.

Can you link anything more concrete then that?

"Spoonfeed me pls".



Yeah, we're not spoon feeding you.



esquared
Anandtech Forum Director

 

[Topweasel]

Says 2 Vega's both under 14nm. Then Navi as 7nm.

But if this is what it has come to I am done. No use in replying to me I won't be able to read it.

 

[IllogicalGlory]

"Spoonfeed me pls".

If they haven't seen it before, how do you expect someone to find the slides you're referring to? If you want to make a point with them, you better post them yourself.

Anyway, you can find them here: https://videocardz.com/65521/amd-vega-10-and-vega-20-slides-revealed

 

[Yotsugi]

If they haven't seen it before, how do you expect someone to find the slides you're referring to?

Like the very first picture in Google.

 

[IllogicalGlory]

Like the very first picture in Google.

One wonders how one who's never seen it before recognizes the one you're referring to, or has any idea whether it's anything meaningful. Also, google image search isn't the first place you look. A normal search turns up any number of things.

How about you instead take the 5 seconds it takes to back up your point? Assertions about roadmaps and articles someone has seen are useless without the source.

 

[Yotsugi]

One wonders how one who's never seen it before recognizes the one you're referring to

Typing words into Google isn't *that* hard.

Also, google image search isn't the first place you look

Where else would you look for roadmaps? Not clickbait articles that's for sure.

How about you instead take the 5 seconds it takes to back up your point?

Spoonfeeding is always bad, and searching stuff is ez.



Yeah, we're still not spoon feeding you.
This is your second troll post so far. It's now stopped.


esquared
Anandtech Forum Director

 

[ozzy702]

One wonders how one who's never seen it before recognizes the one you're referring to, or has any idea whether it's anything meaningful. Also, google image search isn't the first place you look. A normal search turns up any number of things.

How about you instead take the 5 seconds it takes to back up your point? Assertions about roadmaps and articles someone has seen are useless without the source.

I wouldn't waste my time with him.

 

[guskline]

I have been following this thread for awhile and thought I would jump in with my thoughts.

Before that let me say I own a GTX1080TI, GTX 1080, GTX1070TI, GTX1070, GTX780TI classified, AMD Radeon VII and have owned AMD RX480s, R9-290s, 5850s and many more both on the Nvidia and AMD side.

Nvidia is in the graphics arena to stay but so is AMD's RTG division.

There is no doubt that Nvidia is ahead in gpu performance. I think the Rad VII was rushed to market to give AMD a higher end card to stay relevant with the GTX1080TI, GTX2080 and perhaps even the GTX2080TI. I said relevant - in the same ballpark.

I skipped the Vega 56/64 releases, in part because the GTX1080TI, when I purchased it early on was such a good deal and I wanted to upgrade from the RX480s I had in CF. I already had a GTX1080 so it was easy. That's not a slap at Vega56/64 so much as it is a statement of the reality at the time.

The Rad VII is a strange card that I'm having loads of fun with. Though 7nm, it is still a refined GN card but with 16g of HBM2.

I switched my 4k 42" Acer monitor from my GTX1080TI/5960x rig to my Rad VII/2700x rig just because the Rad VII does so well in 4k. Almost as well as the GTX1080TI. My Dell 34" 3415W is servicing the GTX1080TI.

I suspect Navi should appear mid year to "battle it out" with Nvidia in the mid tier level-especially the hot GTX 1660TI. The Rad VII will have to hold down the fort at the top end until AMD refines a faster 7nm Navi based card to supplant it.

The predictions of AMD fading from the graphics scene were probably fueled by the leaving of Raji Koduri. I suspect he was disappointed that Dr. Su was so focused on Ryzen, and his perception that she undercut his development team. Plus a lucrative offer from Intel is hard to turn down, especially when Jim Keller is already there.

I find Dr. Su to be a brilliant strategist who is righting the AMD cpu ship (the mother ship as AMD was a cpu maker first) before fully funding the RTG division. Rory Read deserves credit for cost cutting and Dr, Su deserves credit for staying focused on the prize. I also notice that she has replaced persons with marketing "flambouyance" with quieter engineering-oriented leaders.

Thus, RTG will grow stronger. Console sales help the bottom line.

Will AMD go "toe to toe" with Nvidia in every gpu line? I doubt it. But it doesn't have to since it also includes a growing and robust cpu division.

Nvidia has to have the gpu sales it does to keep the ship moving, as gpu sales appears to be it's primary source of revenue.

If RTG stays competitive it helps Nvidia also.

 

[DrMrLordX]

Typing words into Google isn't *that* hard.

Then do it yourself, and post a link.

Spoonfeeding is always bad, and searching stuff is ez.

Spoonfeeding? You make the claim, you provide the evidence. Don't expect other people to make your point for you. That's bad form, in the extreme.

I suspect Navi should appear mid year to "battle it out" with Nvidia in the mid tier level-especially the hot GTX 1660TI.

That's one thing I'm wondering about . . . if Navi manages to hit between Vega56 and Vega64, I think it can win. If it's equal to Vega56 at a lower price point, I think it'll turn a few heads, but nothing major. Regardless, I do not expect it to threaten anything RTX2060 or better, which is what some people seem to think AMD will do (or will need to do) all at a Polaris price point/power draw.

 

[guskline]

IllogicalGlory, the photo you posted appears to be thE MI-50?

Wasn't it was released in late 2018?
Admittedly the the Rad VII is a variation, more suited for consumers (3 fans vs 1, etc).

 

[guskline]

@guskline you are welcome to join us in the Radeon VII owner's thread in the AMD subforum if you'd like to share your experiences with the card! More data is always welcome.

Also, this is what happens when you start to overclock radeon VII:

https://www.3dmark.com/3dm/33836147

GPU settings: 1940 MHz GPU, 1.09vGPU, 1200 MHz RAM. Highest TJunct is 95C for that run.

Then do it yourself, and post a link.



Spoonfeeding? You make the claim, you provide the evidence. Don't expect other people to make your point for you. That's bad form, in the extreme.



That's one thing I'm wondering about . . . if Navi manages to hit between Vega56 and Vega64, I think it can win. If it's equal to Vega56 at a lower price point, I think it'll turn a few heads, but nothing major. Regardless, I do not expect it to threaten anything RTX2060 or better, which is what some people seem to think AMD will do (or will need to do) all at a Polaris price point/power draw.

Agreed DrMrLordX. navi need to go toe to toe with the GTX1660TI. I think from the reviews I have seen it will be a big volumn seller for Nvidia.

Rad VIIs are limited in Volumn. AMD's RTG needs a winner to replace the 570/580 and perhaps the Vega 56/64 class and fight it out with Nvidia.

 

[guskline]

Also missing, in large part in this discussion, is the revenue AMD has been able to secure for the RTG division by corraling the Xbox, Playstation market.

Google's new Stadia announcement will surely raise some eyebrows.

 

[NTMBK]

Between Microsoft's XCloud running on modified XBox hardware and Stadia running on Vega, AMD are in line to stitch up a big part of the streaming market.

 

[Mopetar]

Between Microsoft's XCloud running on modified XBox hardware and Stadia running on Vega, AMD are in line to stitch up a big part of the streaming market.

I'm not sure if that's a valuable market segment. There's no consumer awareness to be gained and I don't think that any of these streaming platforms are viable in the long term and will fizzle out, but even assuming that I'm wrong, AMD doesn't have any kind of lock-in since the service providers can easily switch to NVidia or even Intel GPUs in the future and the end-users won't even notice.

 

[Yotsugi]

Then do it yourself, and post a link.

Yikes.

Spoonfeeding? You make the claim, you provide the evidence. Don't expect other people to make your point for you. That's bad form, in the extreme.

Yikes^2.

Regardless, I do not expect it to threaten anything RTX2060 or better, which is what some people seem to think AMD will do (or will need to do) all at a Polaris price point/power draw.

You've just asked the next-gen consoles to have a dGPU upgrade so minor calling them next-gen would be a stretch.
Yikes^3.
That's a combo for a mere single post.

Between Microsoft's XCloud running on modified XBox hardware and Stadia running on Vega, AMD are in line to stitch up a big part of the streaming market.

Streaming isn't big, and won't ever be big with current infrastructure and average internet speeds.
Not even touching things like latency and whatnot.

AMD doesn't have any kind of lock-in since the service providers can easily switch to NVidia or even Intel GPUs in the future and the end-users won't even notice

Gamedevs will.
And they will tear a lot of hair after that.





This is your third troll post so far. It's now stopped.


esquared
Anandtech Forum Director

 

[NTMBK]

Streaming isn't big, and won't ever be big with current infrastructure and average internet speeds.
Not even touching things like latency and whatnot.

I think it's going to be a flop too, but even rolling out the services in the first place will get AMD several data centers' worth of GPU/APU sales. Nothing to sniff at.