AMD Zen Performance Preview

[muxr]

Intel got the best cost structure of any semiconductor company. Because they can afford designs that got higher utilization rates. Intel is the only company in the world that got lower transistor cost at 14nm. You can only lose against them in a price war.

When you have 90% of the market share entering the price war affects the market leader the most.

How much do you think Ryzen would sell if it had the same price as the competing 6900K? Not very much, there is literally no reason to switch to AMD if Intel offers same perf/watt. This means AMD has to price their products lower.

Chasing AMD to the bottom would hurt Intel more than letting AMD take its marketshare back.

Same reason why Nvidia doesn't lower 1060 6gb price to under $200. They totally could (it's cheaper to manufacture than the rx480), but it makes them less money if they do that, because 1060gb already outsells rx480 2:1.

Similarly Intel has the mindshare. They have to be more careful not to canibalize their own margin by entering the price war with AMD. AMD has no marketshare, they got nothing to lose. Only way for them with Ryzen is up.

 

[drescherjm]

When you have 90% of the market share entering the price war affects the market leader the most.

In a price war Intel could easily put AMD out of business if it wanted buy selling its CPUs at cost or even a loss.

 

[muxr]

In a price war Intel could easily put AMD out of business if it wanted buy selling its CPUs at cost or even a loss.

You say this as if throwing away 10s of billions dollars of profits like that would not have ramifications for Intel. For one the board would axe the CEO, right off the bat. If he let AMD cost them that much money. The investors would be pretty pissed as well.

 

[drescherjm]

You say this as if throwing away 10s of billions dollars of profits like that would not have ramifications for Intel.

It certainly would have serious ramifications. Far less than the ramifications for AMD however.

 

[muxr]

It certainly would have serious ramifications. Far less than the ramifications for AMD however.

And that's why it will never happen. Neither Intel nor AMD can afford the price war. But Intel stands to lose more money (potential revenues) than AMD from a price war, because they have the established leadership position and marketshare.

 

[razor1]

When you have 90% of the market share entering the price war affects the market leader the most.

How much do you think Ryzen would sell if it had the same price as the competing 6900K? Not very much, there is literally no reason to switch to AMD if Intel offers same perf/watt. This means AMD has to price their products lower.

Chasing AMD to the bottom would hurt Intel more than letting AMD take its marketshare back.

Same reason why Nvidia doesn't lower 1060 6gb price to under $200. They totally could (it's cheaper to manufacture than the rx480), but it makes them less money if they do that, because 1060gb already outsells rx480 2:1.

Similarly Intel has the mindshare. They have to be more careful not to canibalize their own margin by entering the price war with AMD. AMD has no marketshare, they got nothing to lose. Only way for them with Ryzen is up.

Quite the opposite, the market leader can sustain a war or attrition, the smaller company can't.

AMD doesn't have pricing flexibility, Intel does keep that in mind AMD is not in any position to go to a price war with Intel, they have what 1 billion in cash which they need to keep around because of their debt 500 million has to be in their account when the bonds mature? They don't want to go into a price war nor cut prices because they are still at the breaking point.

edit:

nV doesn't need to price lower because the rx480 was selling higher before and the 1060 is selling just as well ore actually most likely better as marketshare shifts towards nV this quarter, as the rx480 now. Even though the 1060 is priced a bit higher, this is partly due to mind share but also because of over all performance and power consumption with the fact that nV has halo products too which goes to mind share.

Do you agree this is a monopoly right now?

If you agree to this, during monopolistic times in a market there is no price elasticity.

This is what competition will bring back. But price elasticity there is a balance from supply and demand, and if performance is the same as whats out there, the change is very little. Cause its already out there . So AMD's ability to change the market to their favor by using price, just doesn't work. This is basic economic theory within a monopoly and new comers into the market (AMD is virtually a new comer). This theory has been seen in many many markets and with same results.

 

[razor1]

And that's why it will never happen. Neither Intel nor AMD can afford the price war. But Intel stands to lose more money (potential revenues) than AMD from a price war, because they have the established leadership position and marketshare.

If Intel and AMD go into a price war, they both lose money but AMD has very little money to lose. This has nothing to do with market perception who is the leader, that is well know and if a price war ensues it won't change it cause well Intel will still sell more as long as price and performance stack up, just like in the past AMD with Athlon were market leaders and they priced their chips accordingly, Intel had to cut prices to try to keep marketshare. Do you really think AMD will sacrifice going into the black by going into a price war? As stated anything less than 20% that Intel's prices are at right now, will give them losses. So that is the hard limit AMD can push 20% less than Intel (if performance and features are all equal). Intel can, and will match that.

 

[muxr]

Quite the opposite, the market leader can sustain a war or attrition, the smaller company can't.

AMD doesn't have pricing flexibility, Intel does keep that in mind AMD is not in any position to go to a price war with Intel, they have what 1 billion in cash which they need to keep around because of their debt 500 million has to be in their account when the bonds mature? They don't want to go into a price war nor cut prices because they are still at the breaking point.

I explained why the Ryzen part costs less than $40 to manufacture (reality is more like $20). AMD has plenty of wiggle room.

If intel decided to sell Core i7 and other CPUs at $40 a piece, to squeeze AMD out of the market, they would lose in excess of $30B+. If they did nothing they might lose $4-5B, due to AMD taking some marketshare back. There is no way Intel wants to lose $30B. Simple as that. Heads would roll if that happened.

This is how duopolies co-exist. Coke and Pepsi. AMD and Nvidia.

 

[razor1]

I explained why the Ryzen part costs less than $40 to manufacture (reality is more like $20). AMD has plenty of wiggle room.

If intel decided to sell Core i7 and other CPUs at $40 a piece, to squeeze AMD out of the market, they would lose in excess of $30B+. If they did nothing they might lose $4-5B, due to AMD taking some marketshare back. There is no way Intel wants to lose $30B. Simple as that. Heads would roll if that happened.

This is how duopolies co-exist. Coke and Pepsi. AMD and Nvidia.

Intel has the same flexibility and more than Zen!

They CAN"T go into a price war period!

The person starting the price war has never gained marketshare in semiconductors PERIOD!

You can see it with Intel in the past, AMD, ATi, nV, anytime any of these companies started a price war, they got the short end of the stick with marketshare, even if they were the marketshare leaders.

Lets add Qualcomm, Apple, Samsung, all of them, do you see their lower end product introductions in smart phones gaining them any marketshare? Apple it hurt them outright within the first quarter, the other two took them time but yeah nothing there too.

Duopoly's only exists if neither of the two companies get ahead of each other. And in this case AMD has fallen behind enough to be pretty much ignored for the past 10 years as a competitor in either CPU or GPU's so that is something they have to get around first before they start being a serious competitor again.

 

[Pieter3dnow]

When you have a vertical company, like Intel, and since their 14nm process is much more mature not to mention they are on their 3rd version of 14nm, its 100% likely they have a cost benefit that AMD could only wish for, remember why AMD started their own foundries? Because they knew they needed them to compete with Intel. It wasn't because it was a whimsical thought, spending billions of dollars per year on fab tech is not a frivolous thought.

You give a new dimension to the phrase comparing apples with oranges ... More money does not equal better processing power in the x86 field another sign is how Zen (Ryzen) was came to light , AMD was more cash strapped then ever in that period.

The time period for AMD when they used their own fabs was when the only problem they had was their management Back then the flow of money was a lot different for that sector when people still were stuck on buying a desktop or laptop if they wanted a computer that ship has sailed now.

Incorrectly assuming that the maturity of the 14nm process is important meant that ARM would not stand a change against Intel and yet the mobile market proven that your assumption is just dead wrong.

 

[muxr]

AMD didn't gain more market share due to Polaris. And they priced it where its performance metrics was.

I disagree on both points.

 

[razor1]

You give a new dimension to the phrase comparing apples with oranges ... More money does not equal better processing power in the x86 field another sign is how Zen (Ryzen) was came to light , AMD was more cash strapped then ever in that period.

The time period for AMD when they used their own fabs was when the only problem they had was their management Back then the flow of money was a lot different for that sector when people still were stuck on buying a desktop or laptop if they wanted a computer that ship has sailed now.

Incorrectly assuming that the maturity of the 14nm process is important meant that ARM would not stand a change against Intel and yet the mobile market proven that your assumption is just dead wrong.

Vertical companies have nothing to do with money, money is a side effect of their capabilities. And when there is a singular vertical company in a singular market the penetration of others companies can only happen when they have superior products (and this only happens if they are able to keep up with market demand, manufacturing capabilities, otherwise they can't gain as much as they can). I suggest you read up on some economics.

I haven't assumed shit man, I suggest you read up on EE and nodes and why they are harder then ever to produce and when and why the sustaining of semiconductor markets is so volatile when mistakes are made.

BS they were using their own fabs since the 70's prior to when they were in the CPU industry! Don't even know the history of AMD for all the love you show them? Don't remember when they took on Nexgen they Nexgen guys saying it was a bit hard to use AMD's fab's because their previous chips were designed for IBM fabs? Yeah, AMD was making their own CPU's in their own Fabs since the beginning of Intel coming to AMD to make socket compatible CPU's. After the buyout of Cryrix, Cryrix had issues going to AMD fabs too!

I can show you many many links on this, and have talked about this in this very forum in other threads, so time for you to some leg work and search google, you will find them in the hundreds. EEtimes had some great articles about it. This has nothing to do with Intel vs AMD, it is just a natural occurring process of any technology market.

 

[razor1]

I disagree on both points.

They didn't gain any marketshare, if you look at when AMD gained marketshare and then lost again, they gained when Polaris was out for a like a week, and then they lost when they were out for a quarter, that just shows us Polaris had little to do with AMD's marketshare gain in the board view of things it was nV's drop due to season effects. Polaris volume sales were higher yes, because the drop of marketshare was less then the gain that was seen in the prior quarter (couple % points), but if they truly were out selling nV's cards 1060 and lower we would have seen very little drop, and this is not factoring in the 1050 which came a quarter later which hasn't shown up in numbers yet.

 

[muxr]

They didn't gain any marketshare, if you look at when AMD gained marketshare and then lost again, they gained when Polaris was out for a like a week, and then they lost when they were out for a quarter, that just shows us Polaris had little to do with AMD's marketshare gain in the board view of things it was nV's drop due to season effects. Polaris volume sales were higher yes, because the drop of marketshare was less then the gain that was seen in the prior quarter (couple % points), but if they truly were out selling nV's cards 1060 and lower we would have seen very little drop, and this is not factoring in the 1050 which came a quarter later which hasn't shown up in numbers yet.

I am looking at an overall yearly sales. Not the week by week which will fluctuate, especially when you have new products being introduced. I also include Polaris 11 Mac Books for instance. Workstation chips and AMD Instinct. In the end Polaris will contribute to AMD's overall regain of market share from 20% last year.

For instance I don't remember the last time I saw an AMD GPU as a top seller, and rx480 has topped the list a few times at least on Amazon.

https://i.imgur.com/EF08G4b.jpg

Nvidia still outsells AMD 2:1 pretty much, but 66% to 33% is 2:1..

 

[razor1]

and that is the only place you are looking at? What about the JPR numbers that came out last quarter and the quarter before? Lets wait and see what happens with 4q marketshare numbers?

Do you see a difference in your amazon to mine? Is it really 3 to 1, looks to me much higher than that.... So as I stated, the trend you can read, but you aren't going to get anything outside of a trend.

This is just the first page, if you go through all the pages, its more like 4 to 1 in nV's favor, but we know that isn't right.

 

[muxr]

I took my screenshot a few weeks ago as anecdotal evidence.

You mean this JPR research?

Where AMD is up to 30% from last years 18%?

As long as Q4 results are above 18% I am right. And I am confident they will be.

 

[razor1]

those marketshare numbers (not the graph) include notebook shares, where AMD did gain.

You need to the breakdown of the numbers......

https://www.jonpeddie.com/press-rel...-q3-2016.-total-gpu-shipments-up-a-whopping-2

 

[muxr]

those marketshare numbers (not the graph) include notebook shares, where AMD did gain.

You need to the breakdown of the numbers......

So? Polaris is in notebooks too. I mentioned Apple design wins. So yes I am right, Polaris contributed to AMD's marketshare gain this year. From 18.8% to 30% is no small feat.

 

[razor1]

So? Polaris is in notebooks too. I mentioned Apple design wins.

That wasn't from Polaris, geez man Apple makes AMD only like 100 million a year! Look at Hexus.net report on that one. And AMD got nothing extra from Apple they just renewed the contract. Do you see polaris anywhere else in notebooks? They got rid of old chips at bargain bin prices.

You guys really need to look into things a bit more deeply then what you feel.

 

[muxr]

That wasn't from Polaris, geez man Apple makes AMD only like 100 million a year! Look at Hexus.net report on that one. And AMD got no extra from Apple they just renewed the contract. Do you see polaris anywhere else in notebooks? They got rid of old chips at bargain bin prices.

You guys really need to look into things a bit more deeply then what you feel.

lol, 30% market share is greater than 18.8%. It's math not feelings.

Perhaps it's your feelings you should check if that's clouding your judgment here.

If AMD didn't release Polaris how much would their share be today? Take a guess.

 

[razor1]

lol, 30% market share is greater than 18.8%. It's math not feelings.

We were talking about Polaris. Polaris did shit for notebook sales, cause they aren't in new notebooks (just replaced old notebooks), take notebook out of the equation when talking about Polaris.

Don't sit here and try to fool me with lackadaisical insight of notebook + desktop marketshare, when I just linked you to desktop marketshare figures, just search this forum and B3D, I have everything split up based on Q calls, JPR, Mercury numbers, and trends from retailers for the past few quarters.

 

[muxr]

We were talking about Polaris. Polaris did shit for notebook sales, cause they aren't in new notebooks (just replaced old notebooks), take notebook out of the equation when talking about Polaris.

Don't sit here and try to fool me with lackadaisical insight, just search this forum and B3D, I have everything split up based on Q calls, JPR, Mercury numbers, and trends from retailers for the past few quarters.

I don't need to fool you. Polaris as an architecture contributed to AMD's GPU share gain this year. It's reflected in the independent numbers you posted. Also I am not even counting Polaris in PS4 Pro and Xbox Slim.

 

[razor1]

I don't need to fool you. Polaris as an architecture contributed to AMD's GPU share gain this year. It's reflected in the independent numbers you posted. Also I am not even counting Polaris in PS4 Pro and Xbox Slim.

Well I'm not counting those because they aren't in PC gaming period. If you want to add those in, be my guest cause I'm not going to talk about those, we are talking about PC desktops, and that is it.

 

[muxr]

Well I'm not counting those because they aren't in PC gaming period. If you want to add those in, be my guest cause I'm not going to talk about those, we are talking about PC desktops, and that is it.

I said I am not. I think you just like to argue.

 

[razor1]

AMD’s overall unit shipments increased 15.38% quarter-to-quarter, Intel’s total shipments increased 17.70% from last quarter, and Nvidia’s increased 39.31%.

This is from the JPR Q3 report.

Now how does that make it AMD's Polaris gained marketshare when nV's shipments gains were higher then AMD's shipments gains more than doubled?

how can you even correlate that with Polaris making up ground? That is Polaris losing ground!

And this is what I stated, AMD's volume shipments went up, but nV's did too at a higher rate though!

 

[muxr]

This is from the JPR Q3 report.

Now how does that make it AMD's Polaris gained marketshare when nV's shipments gains were higher then AMD's shipments gains more than doubled?

As I said I am comparing this year's market share to last years market share. Feel free if you want to argue quarter to quarter numbers, but that wasn't my point.

 

[razor1]

Polaris was only out last quarter!

How the hell are you sitting here and trying to look at last year?

I disagree on both points.

This was your response to Shintai talking about Polaris marketshare gains/no gains.....

This was his qoute

AMD didn't gain more market share due to Polaris. And they priced it where its performance metrics was.

So you want me to take you seriously or are you just flipping to something else?

Any case Q4 numbers are going to give much better understanding because Pascal's full line up is out (still Vega needs to come out, but that we won't see till q3 next year in the figures).

 

[muxr]

Polaris was only out last quarter!

How the hell are you sitting here and trying to look at last year?



This was your response to Shintai talking about Polaris marketshare gains/no gains.....

This was his qoute



So you want me to take you seriously or are you just flipping to something else?

Any case Q4 numbers are going to give much better understanding because Pascal's full line up is out (still Vega needs to come out, but that we won't see till q3 next year in the figures).

As long as AMD has higher numbers than 18% my statement is correct. Polaris did gain market share as opposed to Tonga. Seeing how Polaris is their only product in the last 2 quarters. So yes, let's wait for Q4 results, I am sure they are higher than 18%.

 

[razor1]

As long as AMD has higher numbers than 18% my statement is correct. Polaris did gain market share as opposed to Tonga. Seeing how Polaris is their only product in the last 2 quarters. So yes, let's wait for Q4 results, I am sure they are higher than 18%.

Dude Polaris came out just before last Q one week before, they will not be included a year earlier lol, so yeah if you want to use 18% from last year's Q3 as your baseline, by all means go ahead, but its just ridiculous to say Polaris netted them the increase lol.

The only Q numbers that are valid to look at when talking about Polaris is last Q's to this Q's to next Q's. And last Q's to this Q's, they lost over all marketshare (desktop discrete), so, so far Polaris hasn't done anything for AMD from a marketshare perspective.

Just to put your "looking at last years Q3" as the starting point in perspective. Lets look at 2004 AMD's Financial report and compare that to 2014 markeshare numbers for BD, is that ok?

 

[muxr]

Dude Polaris came out just before last Q one week before, they will not be included a year earlier lol, so yeah if you want to use 18% from last year's Q3 as your baseline, by all means go ahead, but its just ridiculous to say Polaris netted them the increase lol.

The only Q numbers that are valid to look at when talking about Polaris is last Q's to this Q's to next Q's. And last Q's to this Q's, they lost over all marketshare (desktop discrete), so, so far Polaris hasn't done anything for AMD from a marketshare perspective.

Just to put your "looking at last years Q3" as the starting point in perspective. Lets look at 2004 AMD's Financial report and compare that to 2014 markeshare numbers for BD, is that ok?

Dude the whole argument is silly beyond belief. AMD grew market share this year. There was no new GPUs in the last 18 months other than Polaris. Of course Polaris is part of it.

 

[lolfail9001]

Dude the whole argument is silly beyond belief. AMD grew market share this year. There was no new GPUs in the last 18 months other than Polaris. Of course Polaris is part of it.

Actually there was a whole bunch of rebranded low end GPUs AMD dumped on the market mid-year. Their entire Q2 market share was created by them.

 

[muxr]

Actually there was a whole bunch of rebranded low end GPUs AMD dumped on the market mid-year. Their entire Q2 market share was created by them.

Sure the r9 380 and Polaris in Q3 and Q4.

 

[razor1]

Sure the r9 380 and Polaris in Q3 and Q4.

Exactly Q3 216 to Q4 2016, Polaris So why would you look at something before Q2 2016 to figure out Polaris's gains?

Q2 2016 AMD was already at 29 some percent overall prior to Polaris's release. Polaris released last week of Q2 so its numbers in Q3 2016 for a full Quarter, and what was the change? Negative......

So now lets talk about last year to this years Q2 2016, why did AMD go from 18.8 % to 29 and odd % in Q2 2016 ALL PRIOR TO POLARIS RELEASE? Hmm ok so when talking about Polaris we should talk about from last year seems logical lol?

 

[muxr]

Exactly Q3 to Q4, Polaris So why would you look at something before Q2 to figure out Polaris's gains?

Q2 AMD was already at 29 some percent overall prior to Polaris's release. Polaris released last week of Q2 so its numbers in Q3 for a full Quarter, and what was the change? Negative......

So now lets talk about last year to this years Q2, why did AMD go from 18.8 % to 29 and odd % in Q2 ALL PRIOR TO POLARIS RELEASE? Hmm ok so when talking about Polaris we should talk about from last year seems logical lol?

There is sequential growth and year to year growth. I am arguing that Polaris played a role in the year to year growth. Sequential growth can be seasonal and is only one way to look at growth.

 

[razor1]

There is sequential growth and year to year growth. I am arguing that Polaris played a role in the year to year growth. Sequential growth can be seasonal and is only one way to look at growth.

How the hell do you have a train of logic for year to year growth when 3 out of the 4 quarters we are talking about, Polaris isn't even there! You can't attribute it to Polaris, it was something OTHER THAN Polaris. And the Quarter Polaris was there they lost marketshare (although they gained in volume sales)

 

[razor1]

Actually there was a whole bunch of rebranded low end GPUs AMD dumped on the market mid-year. Their entire Q2 market share was created by them.

That is exactly what it was!

 

[muxr]

How the hell do you have a train of logic for sequential growth when 3 out of the 4 quarters we are talking about, Polaris isn't even there! You can't attribute it to Polaris, it was something OTHER THAN Polaris. And the Quarter Polaris was there they lost marketshare (although they gained in volume sales)

Your train of thought is sequential growth, not mine, lol.

 

[lolfail9001]

Your train of thought is sequential growth, not mine, lol.

Sorry, but his point stands: timeline moves from quarter to quarter, not quarter over 3 quarters. You can't explain year-to-year growth with a GPU that existed for the last quarter of 4, especially if it is combined with quarter over quarter drop in market share.

 

[razor1]

Your train of thought is sequential growth, not mine, lol.

sorry meant year to year growth. edited it to reflect that.

 

[muxr]

Sorry, but his point stands: timeline moves from quarter to quarter, not quarter over 3 quarters. You can't explain year-to-year growth with a GPU that existed for the last quarter of 4, especially if it is combined with quarter over quarter drop in market share.

Yes you can lol. https://www.thebalance.com/year-over-year-yoy-growth-calculation-definition-3305970

Year to year growth compares the same time period of one year to the next. You can compare Q3 and Q4 of one year to the next, that's called year to year growth.

I think Polaris played a part in increased year to year market share growth by AMD. I highly doubt they would be above 20% if Polaris wasn't released.

razor1 is arguing the opposite which is just silly IMO.