AMD VEGA confirmed for 2017 H1

[TNA]

I still remember "it;s an overclockers dream"

Yea. That was funny. But lets be honest, that seemed more like a silly employee slip up than an AMD official statement. Still was very funny at the time though

...Yeah and I still remember GTX970 4GB VRAM Touche

lol. True. What was funny about that one was the sheer amount of people ready to defend it and even funnier was how they felt shafted and went and gave Nvidia even more money for a 980

 

[silversurfer]

And there in lies the issue. Looks fairly likely Nvidia are going to drop the 1080ti three months before you can even buy Vega and will be 20%-40% faster than a 1080 depending on resolution and game and then finally AMD limp to the market with a card only as fast as last years 1080.

Im thinking it may be like the Zen dynamic, more about giving power to the people. It appears Zen will deliver a lot of performance to the lower end of the market, whether its actually faster top end we're not sure but its fairly certain its a game changer overall. I mean in terms of changing a market, the vast majority of people are not buying the very top end.
Also Nvidia has game makers who give Nvidia a bias where as AMD cards mature with driver support, the two companies vary in their approach. They said they'll bring a higher and lower Vega right so I'm interested across the whole range how they'll alter things

 

[Greebo]

Wouldnt that put it about 10% or so faster than the Titan X Pascal? I cant see Nvidia doing that can you.....unless they know something we don't and are more than a bit worried

...Yeah and I still remember GTX970 4GB VRAM Touche

a Titan Pascal overclocked to the same as a 1080 is 20% to 40% faster than a 1080 depending on res and game. Okay you have to put a waterblock on the Titan to get the same high clocks as the 1080 but with the 1080ti you wont need to as they will have the same high end AIB air coolers as the 1080. If the remove the voltage lock then perhaps the 1080ti will be faster than the Titan Pascal as 2.3 should be doable.

Any reviews you see with the Titan with stock cooler vs 1080s boosting to 2k arent valid. The Titan will quickly throttle down to a lowish boost.

 

[Greebo]

Yea. That was funny. But lets be honest, that seemed more like a silly employee slip up than an AMD official statement. Still was very funny at the time though

To be fair that "claim" was made by Corporate Vice President and Product CTO of AMD Joe Macri so your would expect him to know. Its not like it was some lowly employee claiming that.

 

[maonayze]

To be fair that "claim" was made by Corporate Vice President and Product CTO of AMD Joe Macri so your would expect him to know. Its not like it was some lowly employee claiming that.

Actually isnt it normally the other way around. Most People on the top rungs of companies usually know diddly squit about the technical or engineering details of their products....they are just Managers (NO offence to any Managers out there...but lets be honest the tech and engineering of a product isnt your job)..Now if Raja had said it then.....fair dinkum...fire away.

Lowly employees within the heirarchy of a company can be Engineers and know far more about the tech aspects than any CTO.

 

[Panos]

...Yeah and I still remember GTX970 4GB VRAM Touche

shhhh don't start another GTX970 why it was advertised as a 4GB VRAM with 64 ROPS and 2048KB level 2 cache.
While it actually had 3.5GB at full speed, 56 ROPs and 1792KB L2 Cache

 

[Steampunk]

What's interesting to me is that (I think I'm right in saying) RTG and AMD are now being run by people with engineering backgrounds, rather than the business backgrounds of previous CEOs who didn't perform. I think the current CEOs may well appreciate the benefits of aiming high and bringing out competitive products that will give them a halo product/effects. I think there has been a change of attitude at the top of AMD, and Polaris was a transition product, almost a filler product while they worked on a "true" next gen 14nm product. RTG didn't waste time on a high end 490, because they could see the prize of a proper next gen part and worked towards it

I think it may be justifiable to have high hopes for Vega, as I think it's aiming for more than we've previously seen. It represents a big change in how AMD approach the market.

 

[TNA]

What's interesting to me is that (I think I'm right in saying) RTG and AMD are now being run by people with engineering backgrounds, rather than the business backgrounds of previous CEOs who didn't perform. I think the current CEOs may well appreciate the benefits of aiming high and bringing out competitive products that will give them a halo product/effects. I think there has been a change of attitude at the top of AMD, and Polaris was a transition product, almost a filler product while they worked on a "true" next gen 14nm product. RTG didn't waste time on a high end 490, because they could see the prize of a proper next gen part and worked towards it

I think it may be justifiable to have high hopes for Vega, as I think it's aiming for more than we've previously seen. It represents a big change in how AMD approach the market.

+1

 

[Kaapstad]

What's interesting to me is that (I think I'm right in saying) RTG and AMD are now being run by people with engineering backgrounds, rather than the business backgrounds of previous CEOs who didn't perform. I think the current CEOs may well appreciate the benefits of aiming high and bringing out competitive products that will give them a halo product/effects. I think there has been a change of attitude at the top of AMD, and Polaris was a transition product, almost a filler product while they worked on a "true" next gen 14nm product. RTG didn't waste time on a high end 490, because they could see the prize of a proper next gen part and worked towards it

I think it may be justifiable to have high hopes for Vega, as I think it's aiming for more than we've previously seen. It represents a big change in how AMD approach the market.

Engineers are more likely to bring out low end products first as they like things that will work and sell in high volumes.

A classic example of this was the Japanese takeover of the motorcycle market where they started at the low end and worked up.

 

[drunkenmaster]

What's interesting to me is that (I think I'm right in saying) RTG and AMD are now being run by people with engineering backgrounds, rather than the business backgrounds of previous CEOs who didn't perform. I think the current CEOs may well appreciate the benefits of aiming high and bringing out competitive products that will give them a halo product/effects. I think there has been a change of attitude at the top of AMD, and Polaris was a transition product, almost a filler product while they worked on a "true" next gen 14nm product. RTG didn't waste time on a high end 490, because they could see the prize of a proper next gen part and worked towards it

I think it may be justifiable to have high hopes for Vega, as I think it's aiming for more than we've previously seen. It represents a big change in how AMD approach the market.

The problem with that is, it's incorrect. Meyer was an engineer, Bulldozer was done under his watch. Rory Read wasn't an engineer as such, more of a business guy with an IT background, he got Keller in, got the Zen project underway, got the company into semi customs which led directly to the console wins which is the main reason the company survived through the Zen R&D cycle. Like 70% of stuff currently going on at AMD and almost all the positive stuff began under Read.

Su hasn't done anything bad, nor necessarily good at this point. Hard to say really, with companies like AMD, plans can have a 5+ year turn around, meaning the current CEO(if they haven't been around for that long) wasn't involved in most of the plans for things just coming out now.

She had more involvement in making RTG but even then the graphics guys would have been working on a new architecture anyway.

HBM type plans of small scalable chips, interposers and a completely different style of designing architectures is something that has been in the works since before Read even joined, their first HBM prototype being from 2011 or so with work starting a fair amount of time before that.

Basically it's far far too simplistic to point at the current CEO in a business with 4+ year CPU design cycles and appoint all blame or success on whoever is currently in the job.

 

[drunkenmaster]

Engineers are more likely to bring out low end products first as they like things that will work and sell in high volumes.

A classic example of this was the Japanese takeover of the motorcycle market where they started at the low end and worked up.

Engineers don't care at all about volume and sales numbers, they care about tech. Engineers always care about the fastest thing, or the most interesting power saving tech. They care about pushing the limits and breaking performance numbers regardless of what that performance is. Sales people care about volume. If a company takes over goes from the low end up, that is almost certainly a business decision, not an engineering one.

 

[Tute]

Engineers are more likely to bring out low end products first as they like things that will work and sell in high volumes.

A classic example of this was the Japanese takeover of the motorcycle market where they started at the low end and worked up.

It almost reminds me of those generations where you get a "tester" part on the new process in order to get practice in and start improving yields. Sort of like what Intel do with their tick-tock strategy. I'm trying to remember the last time it happened in GPUs before Polaris, but i'm fairly sure AMD have done it before.

 

[Kaapstad]

Engineers don't care at all about volume and sales numbers, they care about tech. Engineers always care about the fastest thing, or the most interesting power saving tech. They care about pushing the limits and breaking performance numbers regardless of what that performance is. Sales people care about volume. If a company takes over goes from the low end up, that is almost certainly a business decision, not an engineering one.

You have not worked much with engineers have you !!!!

Engineers like to get things running properly, they don't like people trying to break records as ultimately they have to clean up the mess and fix it.

I have worked in factories all my life with loads of engineers, I also like breaking records the engineers don't.

 

[spoffle]

It almost reminds me of those generations where you get a "tester" part on the new process in order to get practice in and start improving yields. Sort of like what Intel do with their tick-tock strategy. I'm trying to remember the last time it happened in GPUs before Polaris, but i'm fairly sure AMD have done it before.

The 4770 was one such instance.

 

[LoadsaMoney]

The Furys were also tester parts.

 

[JediFragger]

The Furys were also tester parts.

Yup, basically just large Tonga's with HBM (to give them experience with interposer's, and also kept them somewhat 'in the game' til the next gen stuff arrives).

Tbf it's not a bad chip, I was fairly surprised by how strong it is at my resolution

 

[humbug]

wrong room

 

[JediFragger]

wrong room

Would you prefer us to talk about the Ti again?

 

[humbug]

Would you prefer us to talk about the Ti again?

No.... i posted something in the wrong room, it was there for a while before i realised and edited.... should have been clear, my bad

 

[akarypid]

No.... i posted something in the wrong room, it was there for a while before i realised and edited.... should have been clear, my bad

I think nobody would've noticed anyway... This thread has recently been 10% about Vega and 90% about... just about anything else!