AMD Fast-tracks 7nm "Navi" GPU to Late-2018 Alongside "Zen 2" CPU

[oxidized]

Please AMD don't f*** this up please!

 

[kings]

7nm high-end gaming GPUs in 2018? Good luck with that...

I'm going to pull a chair so I don´t get tired.

 

[T1beriu]

AMD Fast-tracks 7nm "Navi" GPU to Late-2018 Alongside "Zen 2" CPU

Reading comprehension. Someone needs it.

1. AMD doesn't say Navi is coming late 2018. They're talking about Vega (20) on 7nm (MI Instinct) that we already know about for many months.

2. Zen 2 isn't coming in 2018, but 2019 as AMD clearly says in the press release.

our first 7nm server CPU that we plan to launch in 2019

Ruff weekend, eh?

 

[TheGuruStud]

The word consumer is nowhere to be found, how can you clickbait something without even mentioning it.

Irrelevant. Enterprise comes first for profits (since zen is now established and that's the way it works by default with vid cards). Nothing is coming in 2018 for consumer, so neither is Navi.

 

[Jism]

7nm polaris 30 or 40 with GDDR6 as a equivalent to the 1070ti please.

 

[sergionography]

The next major milestone is zen 2 and navi.
7nm vega is not part of this "major milestone" because 1- its already taped out and probably ready for mass production soon. And 2- its a niche product for a new market; AMDs way of hitting 2 birds in one stone where they start 7nm with a shrink to test the waters rather than a new architecture, and also tackle a new market while at it (low demand/low competition/low expectations = lower risk).

All in all, the title is wrong. Unless amd renamed 7nm vega to navi. Regardless, I dont expect new consumer/gaming AMD gpus anytime soon

 

[john_]

Please AMD don't f*** this up please!

Buy a Ryzen, or next time a Radeon. Help them not to "f*** this up", please!

 

[DeathtoGnomes]

Typical thread development here, as with any Press release, the fanbois from both sides come out of the woodwork and say absolutely nothing productive. Its really getting old.

IMO, the PR team at AMD is doing a fine job, but I want to know if the so called fast tracked GPU will actually be ready for launch this year. Short of a direct response from AMD, I seriously doubt anyone here can answer that. Try as you might.

 

[john_]

I dont think it was when i posted. Besides, my point still stands. The title is clickbait.

It's AMD's press release from about a week ago. It was there.
Expanding our High-Performance Leadership with Focused 7nm Development | Advanced Micro Devices

IMO, the PR team at AMD is doing a fine job, but I want to know if the so called fast tracked GPU will actually be ready for launch this year. Short of a direct response from AMD, I seriously doubt anyone here can answer that. Try as you might.

AMD is going for the professional market with it's next 7nm GPU, so whatever response from the Red team will be focusing on profesionals. For gamers we will probably see a bigger Polaris, just to counter 2050 and 2060 cards.

 

[medi01]

Clickbait eh?

For gamers we will probably see a bigger Polaris, just to counter 2050 and 2060 cards.

Include (at least) 2070, perhaps non-ti 2080 and it will kinda sorta makes very much sensa as why bother developing cards that people won't buy even if they are competitive.

 

[TheGuruStud]

Clickbait eh?



Include (at least) 2070, perhaps non-ti 2080 and it will kinda sorta makes very much sensa as why bother developing cards that people won't buy even if they are competitive.

Ah, see, someone gets it. Can't sway the stupid populace.

 

[notb]

If their 7nm CPU's make it to AM4 and my x370 will take it, I know already that I'll buy one, even if I don't need it. At that point, they'll have sold me one CPU from each Zen generation. And the only reason I would think to even do that is because of how affordable their CPU's generally are and just the simple fact that you don't have to change boards each gen. Much lower barrier to entry makes one less likely to stop and think... ...just spend the little chunk of cash, drop it in, and enjoy the gains. It's almost too easy to upgrade. It's brilliant. The reality is that this way, I actually spend more with them than I would have been able to justify with Intel.

You're literally the perfect AMD customer. Exactly as you said: you're attracted by AMD products being affordable and easy to replace, but you end up spending more than you would normally do.
The whole point of this strategy is to make people do yearly upgrades. We've seen many examples of this approach working (flagship smartphones, Sony cameras etc).
And it's a great strategy... for consumer electronics. I have no idea how AMD plans to approach corporate segment (both servers and business PCs). EPYC sales are fine (at least we're told they are), but the Ryzen PRO lineup is a complete flop.

No mention of ryzen 2 this year only navi this year apparently,and the first 7nm server cpu in 2019 which is when ryzen 2 is also slated for

Navi seems very unlikely this year. A new architecture on a new node? Is the arch even ready? We know the node isn't (not for high volume production, anyway).
Wasn't AMD going to adopt a safer tick-tock approach? What would mean 7nm Vega first.

Moreover, 7nm Navi seems like an expensive, high-performance product tuned to compete with RTX's finesse. This still leaves AMD with very old budget lineup.

 

[john_]

Include (at least) 2070, perhaps non-ti 2080 and it will kinda sorta makes very much sensa as why bother developing cards that people won't buy even if they are competitive.

I don't know if they can counter Nvidia's new cards without going for a product that uses HBM and I don't know if that product will be bringing them any profits or even sales. It will be good for other reasons, than profits, it will be good for their partners and from a marketing perspective, but will they offer such a card to gamers? Especially when Nvidia will be heavily promoting ray tracing and everyone will be happy to pay for a new feature that will change the world(just trying to be in the mind of a new buyer)? On the other hand a cheap to make Polaris at 12nm, will not cost them much, will sell at OEMs and will also help covering whatever minimum waffer buys they have to make to GlobalFoundries this year.

 

[TheGuruStud]

I don't know if they can counter Nvidia's new cards without going for a product that uses HBM and I don't know if that product will be bringing them any profits or even sales. It will be good for other reasons, than profits, it will be good for their partners and from a marketing perspective, but will they offer such a card to gamers? Especially when Nvidia will be heavily promoting ray tracing and everyone will be happy to pay for a new feature that will change the world(just trying to be in the mind of a new buyer)? On the other hand a cheap to make Polaris at 12nm, will not cost them much, will sell at OEMs and will also help covering whatever minimum waffer buys they have to make to GlobalFoundries this year.

Vega will do RT and potentially at good perf (relatively speaking to turding). I don't think future cards will have any issue.

As useless as RT already is, it'll be unusable on 2070, so it's not like anything midrange needs it.

 

[Vayra86]

I don't know if they can counter Nvidia's new cards without going for a product that uses HBM and I don't know if that product will be bringing them any profits or even sales. It will be good for other reasons, than profits, it will be good for their partners and from a marketing perspective, but will they offer such a card to gamers? Especially when Nvidia will be heavily promoting ray tracing and everyone will be happy to pay for a new feature that will change the world(just trying to be in the mind of a new buyer)? On the other hand a cheap to make Polaris at 12nm, will not cost them much, will sell at OEMs and will also help covering whatever minimum waffer buys they have to make to GlobalFoundries this year.

That's just all in your head, not in anyone else's.

There is NO news of a Polaris refresh of any kind and neither would it make sense. People need to get it through their thick skulls: RTG is not worrying about your gaming Radeons anymore. Waste of time, waste of budget. Polaris was an attempt to attack the midrange but guess what, it can't even cover that anymore these days, not even with a shrink. In the same way I don't see Vega 20 change anything; it doesn't even target gaming and neither did its predecessor. All hope is on Navi to bring some sort of revival to AMD gaming GPUs. Or Intel.

Basically this press release confirms what we already knew when Vega first launched. Its over, RTG isn't playing ball anymore and anything they do now serves to support the APU/CPU push.

 

[oxidized]

Buy a Ryzen, or next time a Radeon. Help them not to "f*** this up", please!

I totally will if performance improve, not that they're bad but well...

 

[TheoneandonlyMrK]

That's just all in your head, not in anyone else's.

There is NO news of a Polaris refresh of any kind and neither would it make sense. People need to get it through their thick skulls: RTG is not worrying about your gaming Radeons anymore. Waste of time, waste of budget. Polaris was an attempt to attack the midrange but guess what, it can't even cover that anymore these days, not even with a shrink. In the same way I don't see Vega 20 change anything; it doesn't even target gaming and neither did its predecessor. All hope is on Navi to bring some sort of revival to AMD gaming GPUs. Or Intel.

Basically this press release confirms what we already knew when Vega first launched. Its over, RTG isn't playing ball anymore and anything they do now serves to support the APU/CPU push.

Yeh but no, guess what ,i had polaris since day one, many said Nvidia were better but Amd still sold a lot of Polaris, you seriously think they're happy to wave that income off.
No ,to me Rtx is Nvidias big play for customer retention, because they know what's coming.

 

[notb]

Vega will do RT and potentially at good perf (relatively speaking to turding). I don't think future cards will have any issue.

There's no way Vega will get close. At least not with similar results, because - obviously - you can lower precision and speed up calculations using GPGPU.

At this point the only thing AMD has to offer is their ProRender ray tracing implementation, which can also be used with Nvidia GPUs.
It's been shown that rendering on Vega 64 is 10-20% faster than 1080ti or P6000, which means it's nowhere near RTX and really not fast enough for real-time rendering at 1080p and above.
So whenever you see AMD being proud of their RTRT performance:
a) it's pre-RTX,
b) they mean tiny live previews - not high resolution games / animations.

RTRT for gaming was meant to be usable on next get Nvidia GPUs (so whatever comes after Turing around 2020).
That said, if what we've heard about RTX is true, it's more like 2 generations ahead of GPGPU.

As useless as RT already is, it'll be unusable on 2070, so it's not like anything midrange needs it.

How can you say that RT is useless? Do you even understand what it is?

 

[EntropyZ]

With Pascal cards going on deals and second-hand cards being steals. Wow, this is going to hurt to watch. I'll sit right beside people watching.

 

[Vayra86]

How can you say that RT is useless? Do you even understand what it is?

Ray Tracing isn't new. And with the current performance and implementation we've seen (very limited), it effectively still is as useless as it was 5 or 10 years ago. Nothing's changed, but Nvidia would love you to believe otherwise. Its comparable to VR right now: you can get some idea of the experience it should have been, but it really isn't there yet and comes at the cost of performance and quality of life. VR is another such tech that has been launched and re-launched a couple times but just won't stick because of its gross inefficiency.

Honestly, RT is fár, far away still. Implementation within DX12 is not going to take off either, contrary to what some may believe or say. This needs 10+ years to mature and it needs much faster and better hardware still, plus widespread adoption.

 

[RejZoR]

I frankly prefer good quality faked effects at super speed than high quality real ones with horrendous performance. I've seen enough amazing reflections and lighting systems in games from recent history to know we don't need ray tracing until it's really ready for general use. The problem is, most just don't care, they just smack a cubemap on it and call it a day. I'm pretty sure perspective corrected cubemap could be used for reflections, causing things to be correctly aligned depending on angle at which you're looking at reflections. But no one seems to be doing that for some dumb reason at least I can't think of any at this moment).

Take CS:GO as an example on that village map where you drive with vans in the beginning of round. When inside houses, at certain angles, window reflections on flooring looks perfectly aligned with the actual windows. But as you move around, reflections go totally off. I'm sure this could be addressed with minimal penalty, surely lower than ray tracing. And that's basically all there is to reflections and lighting. We don't need absurd details when hardware clearly isn't up to the task. Baby steps and doing those right, not huge leaps and then nothing works for 5 or 10 years... Unreal Engine is incredibly good with reflections and so is CryEngine. And they don't use any raytracing.

 

[megamanxtreme]

I can't wait for their 75W variant, so I don't have to connect cables and it's low wattage. But seriously, Late 2018? Man, still waiting for $100 brand new card to compete with the RX 570 and 8 Gigs, I'm good for a good while. My Amazon Gift Card money is ready.

 

[nemesis.ie]

For sure, I'm looking to put a card in my HTPC for better UHD playback than the awful thing in the Haswell provides, but the prices of the lowest-end cards are too high just for that.

 

[efikkan]

There is nothing in the press briefing supporting the claim of this article. You better come up with the source or retract this article immediately, or this will fall into the category of fake news. This is the same trash Wccftech published yesterday, and there is no need to compete with them in their race to the bottom.

The only interesting part of this memo is AMD's emphasis of their continued use of 12/14nm. The transition to 7nm will be gradual and very slow.

 

[notb]

Ray Tracing isn't new. And with the current performance and implementation we've seen (very limited), it effectively still is as useless as it was 5 or 10 years ago. Nothing's changed, but Nvidia would love you to believe otherwise. Its comparable to VR right now: you can get some idea of the experience it should have been, but it really isn't there yet and comes at the cost of performance and quality of life. VR is another such tech that has been launched and re-launched a couple times but just won't stick because of its gross inefficiency.

Aren't we mixing 2 things here? Or maybe 3?

1) Ray tracing (RT) is a basic rendering technique. It's been around for decades and is fundamental - not useless or unusable as one of AMD fanboys claims.
2) Real-time ray tracing (RTRT) is... RT in real time ;-), i.e. fast enough (whatever that means).
It's been around for a while, but used for previews - not final renders. Previews are greatly simplified - they ignore some materials and some effects. Also the resulting live render is usually low-res and under 30fps.
3) RTRT in games means it has to be efficient enough for processing all effects, at high-resolution (1080p+) and high frequency ( has to be acceptable for gaming, i.e at maybe 1080@60fps, maybe 4K@30fps...

Honestly, RT is fár, far away still. Implementation within DX12 is not going to take off either, contrary to what some may believe or say. This needs 10+ years to mature and it needs much faster and better hardware still, plus widespread adoption.

You're talking about general processing implementation, i.e. what standard GPU cores do.
Nvidia used an ASIC and it's just way faster - just like tensor cores are way faster for neural networks.

Everything else you've said is more or less correct.
If one wants to combine RTRT with 4K@60fps, then doing that on GPGPU is 10 years away from now. But on ASIC it should be possible withing 1-2 generations, i.e. 4 years tops.
But thanks to RTX cards, you don't have to wait 10 years. For mere $1200 you can already make your games look as if it's 2028 (just at 1440p tops).
And when you buy your next RTX card in 2021 for another $1200, it should be OK for 4K@60fps.

There's just no way around it. AMD will have to respond with a similar tech, ignore RTRT ("Who needs realism? We're so romantic!") or magically make Navi 4x faster than Vega.