Samsung and Global Foundries will produce AMD's Next Gen Greenland GPU and Zen CPU

[ShintaiDK]

There was also a PDF link BTW. I could see 2 of 3 pics. Interestingly, why does the most DX11 load run on the same core as the most loaded game threads? Possibly due to necessary communication and DX calls done by these threads? In such a scenario these threads (or better: jobs) pull the DX11 API and driver load to a core which already has some load (in the example the most game related load).

Its PR material. It will always show the worst/best case. You should know better.

The beating is simple, as I explained before: with weak performance of a single core, the game thread + 80% DX11 or 13% DX12 related load (8c) still creates a bottleneck.

That's the reason, why asking for a proof of FX beating the i3 due to DX12 is stupid and shows a missing understanding.

Why? Because everything so far in the real world point to that ST performance is still the most important factor with DX12?

 

[AtenRa]

If(/when) Zen flops or goes awol, AMD will be in real trouble.

When ZEN comes out in late Q4 2016, AMD will create an extra 1.5B per year of revenue from the semi-custom segment from its new deals on top of what they already have now (close to 2B). So even if ZEN completely flops, they will be far better than what they are today.

But, IF they will manage to gain a puny 5% increase of Server market share in 2017, when Server market will be more than 50B, with a 50% margins they will gain a huge amount of profit.

Edit: Realistically im not expecting to gain 5%, but close to 2-3% increase of server market share will be enough for some nice profits.

 

[ShintaiDK]

What extra 1.5B are those? Its obviously not the NX deal since it doesn't hold that value.

Also AMDs current semicustom revenue is shrinking.

 

[AtenRa]

What extra 1.5B are those?

From the two new semi-custom deals starting H2 2016.

 

[ShintaiDK]

From the two new semi-custom deals starting H2 2016.

Links please. I have a feeling you compressed several years into 1 year.

 

[AtenRa]

Links please. I have a feeling you compressed several years into 1 year.

I dont have the link now ill find it later, its was estimated to be 250M+ per Quarter but i think Devinder Kumar said 1.5B in his last interview.

 

[coercitiv]

Links please. I have a feeling you compressed several years into 1 year.

If the object of discussion is the SA Transcripts then the relevant part should be this:

$1.5 billion starting in the second half of 2016 and from a modelling standpoint you can look at it as a three-year cycle for those products to go ahead and generate revenue over that time frame.

 

[AtenRa]

Ahh yes thats the one thanks

 

[Mercennarius]

You pictures dont work. And DX12 wont change game logic.

Hence why an i3 beats a FX8350 in AOTS. And how was Fable? Oh right, same thing.

So the challenge is up to you, show me a DX12 game where your dream comes true.

DX12 does indeed become more efficient in regards to multi-threading. AOTS benchmark DX11 vs DX12 multithreaded performance:
DX11:

DX12:

In my own testing performance was worse when I limited the benchmark using Windows Affinity to 8 threads instead of allowing all threads to be accessible.

 

[ShintaiDK]

If the object of discussion is the SA Transcripts then the relevant part should be this:

Ahh yes thats the one thanks

That's 500M$ per year. Or 125M$ per quarter. Not 1.5B$ per year.

1.5B$ per year would more or less mean the same as PS4+Xbox one combined to put it in perspective.

 

[AtenRa]

Well, even is this is 500M per year they will still be better than what they are today.

 

[ShintaiDK]

I wouldn't be so sure on that. Seems the end score for 2015 will be a 28% shrinkage give or take.

The current console deals is shrinking too. +500M$/year may only reduce the next drop rather than remove it.

 

[AtenRa]

I wouldn't be so sure on that. Seems the end score for 2015 will be a 28% shrinkage give or take.

The current console deals is shrinking too. +500M$/year may only reduce the next drop rather than remove it.

In 2016 they will have a complete new product portfolio from APUs to GPUs and new Semicustom. Im expecting they will increase market share in every segment vs 2015 except from Servers. Revenue will only increase in 2016 due to the new products and im expecting they will be in 2014 numbers again.
2017 onward will see another revenue increase from Server as ZEN will be able to gain a small increase for that segment.

Current console deals are not shrinking.

 

[ultimatebob]

If I need to explain that to you you've got a problem.

How long have you been predicting AMD's "inevitable" Chapter 11 bankruptcy?

Never read so much crap in all my life. You guys have no idea about how business works.

You really need to try reading the AMD balance sheet instead of AMD's PR slides. When a company is losing almost $200 million a quarter and has less than $800 million in cash on hand, you quickly realize that they can only run the business as they have been for about a year before needing to sell part of the business or taking on additional debt. I'm not sure who would be crazy enough to give them additional money to burn at this point.

 

[Fjodor2001]

The AMD stock price has increased by nearly 100% in just the last 3 months:

http://www.google.com/finance?chdnp...q=NASDAQ:AMD&ntsp=0&ei=ho2BVuHBM8ycsAHqt4XQCQ

Starting from low level or course, but it's the relative increase that is important if you're investing. So the market's faith in AMD has increased considerably lately.

No way you're going to see a 100% increase in the Intel stock price over 3 months.

 

[AtenRa]

The AMD stock price has increased by nearly 100% in just the last 3 months:

http://www.google.com/finance?chdnp...q=NASDAQ:AMD&ntsp=0&ei=ho2BVuHBM8ycsAHqt4XQCQ

Starting from low level or course, but it's the relative increase that is important if you're investing. So the market's faith in AMD has increased considerably lately.

No way you're going to see a 100% increase in the Intel stock price over 3 months.

AMD stocks can make you big money or make you broke

 

[Fjodor2001]

AMD stocks can make you big money or make you broke

Yes, I know. The AMD stock is much more volatile than big "safe" companies like Intel. AMD is high risk, but also potential for high reward.

Anyway, AMD's stock trend is sharply up in the last few months, which is interesting.

 

[AtenRa]

Yes, I know. The AMD stock is much more volatile than big "safe" companies like Intel. AMD is high risk, but also potential for high reward.

Anyway, AMD's stock trend is sharply up in the last few months, which is interesting.

yea true,

Im expecting a good Q4 result, increase of revenue and stabilization in some segments. That could be a reason for the stock increase.

 

[ShintaiDK]

In 2016 they will have a complete new product portfolio from APUs to GPUs and new Semicustom. Im expecting they will increase market share in every segment vs 2015 except from Servers. Revenue will only increase in 2016 due to the new products and im expecting they will be in 2014 numbers again.
2017 onward will see another revenue increase from Server as ZEN will be able to gain a small increase for that segment.

Current console deals are not shrinking.

We heard this for how many years now? What happens every single time?

Look on yearly semicustom revenue, its shrinking.

 

[Arachnotronic]

Yes, I know. The AMD stock is much more volatile than big "safe" companies like Intel. AMD is high risk, but also potential for high reward.

Anyway, AMD's stock trend is sharply up in the last few months, which is interesting.

You can make very big gains with large "stable" stocks just as well as with smaller ones via options.

Let's use Intel as an example. Here are some of the call options at various strikes on INTC stock that will expire on Jan. 20, 2017:

If I buy the Intel $35 strike calls, I have to pay $3.05 per share (stock is at $34.88 right now) for those calls, meaning that to break-even at expiration date, the stock needs to see $38.05.

However, if by expiration, Intel stock goes to $41.10 -- a mere 17.8% in the stock price -- I double my money because those $35 strike calls will be worth $6.10, twice what I paid for them. On the flip side, though, if Intel is below $38.05 by expiry, then you will lose money. At or below $35 and those options are worth zero.

Inexperienced investors think that small caps/penny stocks are the only ways to make big gains, but if you think a "high quality/stable" company will go up by a solid, though not crazy, amount then options are a good way to get small-cap like gains.

Now...

HUGE DISCLAIMER: The above analysis was meant completely and entirely for informational purposes only to illustrate the mechanics of options. I am NOT making any recommendation to buy/sell any sort of financial instrument.

 

[Fjodor2001]

Sure, I know. But I was not talking about stock options, just normal stocks. I'm aware that you can leverage the outcome of stocks using options, and that goes both ways (profit/loss).

Now having said this, if you're a real gambler, you should have bought AMD stock options 3 months ago...

 

[Arachnotronic]

Sure, I know. But I was not talking about stock options, just normal stocks. I'm aware that you can leverage the outcome of stocks using options, and that goes both ways (profit/loss).

Now having said this, if you're a real gambler, you should have bought AMD stock options 3 months ago...

Hah!

Yeah, this is a very big run in AMD stock. I can only make educated guesses as to why the stock has run up like this, but congratulations to anybody who made money on this run!

 

[mrmt]

Links please. I have a feeling you compressed several years into 1 year.

What Lisa Su said in conference calls is that both undisclosed semi-custom designs would yield 1 billion for the entire life of the projects.

It seems that 1 billion for the entire life of both projects suddenly became 1.5 billion per year when translated to some idioms.

 

[VR Enthusiast]

You really need to try reading the AMD balance sheet instead of AMD's PR slides. When a company is losing almost $200 million a quarter and has less than $800 million in cash on hand, you quickly realize that they can only run the business as they have been for about a year before needing to sell part of the business or taking on additional debt. I'm not sure who would be crazy enough to give them additional money to burn at this point.

You really need to try understanding the balance sheet instead of looking at the bottom line. AMD is not losing $200 million a quarter, 1/3rd of it was an inventory write-off and a lot went on restructuring charges and other one-offs.

They have already sold off their packaging plant for $320 million and that will be seen in 2nd quarter next year.

They have plenty of time and money to get to Zen.

 

[Dresdenboy]

^upto 40%...

"upto" makes no sense. They neither stated "up to" nor "at least". Like it happened in the past, "on average" is most likely. This time even more so, as this was based on internal simulations, but is frequency-agnostic (more real silicon related). And everybody is invited to criticize the chosen app mix. It will never be right.

Its PR material. It will always show the worst/best case. You should know better.

It clearly demonstrated that even with DX12 ST performance (or load balancing) is causing a bottleneck. Accusing companies of steady lying even while using material about APIs and game engines from other companies, looks like a genetic fallacy. And CPU load logs support the things shown in the diagrams (even while being simplified).

Why? Because everything so far in the real world point to that ST performance is still the most important factor with DX12?

Non dichotomic logical thinking should lead to that conclusion. DX11 with a non-MT driver (AMD) made things worse by a factor of nearly 2x. DX12 kind of clears the sight to find the next bottleneck: the heaviest game engine load on a single core. If a game runs at 35 fps w/ DX11 and at 60 fps w/ DX12, it's both an improvement with DX12 and limited by ST performance at the same time.

Now comes the run for the last percents to perfection: if the FX delivers 60 fps (=playable) and the i3 delivers 70 fps (=playable), one group of people will be satisfied with the FX, while another group will tout the 17% more fps at lower power as the most important thing. I'll leave this as it is.

Also AMDs current semicustom revenue is shrinking.

Common console sales seasonality.