ATI overtakes nVidia in discrete graphics marketshare

[Nemesis 1]

Not sure at 40nm . But at 28nm yes . Heres a link that will tell ya anything you want to know about the present shape of said company . We are still on topic I hope.

http://www.imgtec.com/

 

[Wreckage]

It's interesting that Microsoft just invested in ARM. A Windows 7 port would be a huge boost to Tegra.

 

[3DVagabond]

That would mean JHH was either dead or in charge of VIA. And I don't foresee WenChi (CEO of Via) stepping down for another 10 years.

Just curious about the legalities and if they could then just combine licenses. I would assume that if nVidia could buy VIA they could easily come to a severance agreement with it's current CEO to make room for the merger?

 

[cbn]

I A Windows 7 port would be a huge boost to Tegra.

Well at one time ASUS Eee Pad (the smaller 10" Tegra 2 one) was supposed to come with Windows 7 Embedded Compact. Apparently now it will be Google Android

http://androidcommunity.com/asus-ee...android-instead-of-windows-embedded-20100720/

 

[cbn]

It's interesting that Microsoft just invested in ARM.

I didn't find any info on MS investing in ARM, but recently they did sign up for an ARM license.

 

[Scali]

I didn't find any info on MS investing in ARM, but recently they did sign up for an ARM license.

Yes, they want to make their own phone hardware, I believe.

 

[cbn]

Yes, they want to make their own phone hardware, I believe.

I wouldn't doubt it and I'll bet they would like to follow Apple's footsteps by using integrating user aimed advertising into the package.

EDIT: But I am not sure why they would need their own hardware for that purpose?

Maybe they are planning on making a hand held gaming device as a companion to Xbox?

 

[cbn]

http://unplugged.rcrwireless.com/in...archaic-mobile-is-the-future-declares-nvidia/

PC is archaic, mobile is the future declares Nvidia

Google Android and Tegra 2 sounds interesting....

......But how about a CUDA enabled, 2Ghz ARM CPU netbook with proprietary Nvidia OS? Then Nvidia could make its own money via a CUDA/Physx marketplace? (I don't browse CUDA Zone very often, but it seems like they have a lot of developer interest over there.)

 

[Idontcare]

Just curious about the legalities and if they could then just combine licenses. I would assume that if nVidia could buy VIA they could easily come to a severance agreement with it's current CEO to make room for the merger?

To maintain x86 license it would have to be Via acquiring NV.

Same issue NV (JHH really) had when AMD wanted to buy NV...JHH insisted he replace Ruiz after the acquisition and remain CEO of the new business, Ruiz of course declined to cede his thrown so they went after buying ATI once spurned by NV.

 

[Acanthus]

To maintain x86 license it would have to be Via acquiring NV.

Same issue NV (JHH really) had when AMD wanted to buy NV...JHH insisted he replace Ruiz after the acquisition and remain CEO of the new business, Ruiz of course declined to cede his thrown so they went after buying ATI once spurned by NV.

wasn't nv wayyyyyyyyyyyyy out of AMD's price range at the time?

 

[jvroig]

wasn't nv wayyyyyyyyyyyyy out of AMD's price range at the time?

They didn't have the cash from their own bank account, so they had to seek financing elsewhere.

They already had this money (debt money of course, nobody just donated money to them), which is why when the nV deal fell through, they scrambled to ATi, because they already had a truckload of debt money and it better be put to good use. I'm not sure but I think it even had Hector Ruiz lie to investors (maybe even inflating some projections) to get the investors, then when they had the money, Ruiz told them the bad news (the real story) of AMD's projections.

That is why AMD's finances look bleaker than they are, due to the debt they incurred when they decided to purchase a graphics company. They aren't healthy by any means, but a significant portion of what made it look terribad was due to that.

 

[Borealis7]

so does this mean that AMD's cards are faster now than they were before? no? OK.
i don't see how this is exciting to anyone who isn't in AMD's sales department.

 

[jvroig]

so does this mean that AMD's cards are faster now than they were before? no? OK. i don't see how this is exciting to anyone who isn't in AMD's sales department.

1.) Shows market responds to execution of players - great execution gets rewarded, while poor execution gets punished. Keeps all players on their toes. [Good thing]
2.) AMD's cards aren't faster now, obviously, but this may mean AMD's cards in the future will be better. More marketshare = more money = more R&D budget = more engineering budget = more dev relations budget. This is something nVidia has enjoyed for a long time, and despite Fermi being a little late, it has served them very well (they are far more diverse than ATi, they have GPGPU and Tegra aside from gaming cards, and have excellent dev relations).

If the opposite happened (AMD's graphics marketshare lessened despite the great execution of the 5xxx series and the Fermi delays), that would have been terrible news. Since it didn't happen, this is good news.

 

[Idontcare]

wasn't nv wayyyyyyyyyyyyy out of AMD's price range at the time?

Evidentally not given that the deal fell thru because of JHH's power-play requirements and not because of pricing...although one could argue that JHH could have been paid off if enough money had been brought to the table.

I'm not sure but I think it even had Hector Ruiz lie to investors (maybe even inflating some projections) to get the investors, then when they had the money, Ruiz told them the bad news (the real story) of AMD's projections.

That is why AMD's finances look bleaker than they are, due to the debt they incurred when they decided to purchase a graphics company. They aren't healthy by any means, but a significant portion of what made it look terribad was due to that.

Yes that was an awful fourth quarter analyst meeting. They (AMD) gave all kinds of wildly creative revenues projections all the while knowing that Intel's C2D was just devastating their sales and completely eroding marketshare of their premium ASP SKU's.

They leveraged the analysts from that meeting to give AMD a good debt/credit rating so the banks could in good faith loan AMD the some $5B needed at the time, two weeks after AMD got what they wanted they then came out with a drastic downward revision to their Q4 projections.

It stank all the way around and the analysts turned and punished AMD stock severely as the institutional holdings dropped and made the float quite excessive. Too much supply and not enough demand, price fell some 50% over the course of the year with an average trade volume about 2x above the prior year.

Technically AMD never cooked their books, it was just blatantly obvious manipulation of the analysts by way of cooking their revenue projections. Since it wasn't done as a means to manipulate the market, it wasn't done with intent to manipulate AMD stock price, there was no action to be taken by the SEC. Ruiz knew what he was doing, or his legal counsel did, they made sure they never crossed the line of the law or regulation but the moral and ethics line was definitely not sacrosanct.

 

[ZimZum]

Isnt ATI the only portion of AMD thats actually making money?

 

[Scali]

Isnt ATI the only portion of AMD thats actually making money?

Yup.. For the past few years, the CPU division has mainly dependent on the Arab investment group and the Intel settlement. Deduct those sources of income, and the company as a whole would have operated at a loss.

 

[bryanW1995]

True, Via cannot sell or transfer their x86 license. If they get bought out their X86 license becomes invalid.

yes, but what if VIA buys Nvidia? how large of a company can Via acquire as a % before causing x86 problems, ie the acquired company can only be 5% of via's size, or 50%, etc? you can bet in an nvidia x86 scenario that intel would drag it out forever, too.

 

[Acanthus]

yes, but what if VIA buys Nvidia? how large of a company can Via acquire as a % before causing x86 problems, ie the acquired company can only be 5% of via's size, or 50%, etc? you can bet in an nvidia x86 scenario that intel would drag it out forever, too.

VIA is about 80% the size of Nvidia.

 

[Idontcare]

yes, but what if VIA buys Nvidia? how large of a company can Via acquire as a % before causing x86 problems, ie the acquired company can only be 5% of via's size, or 50%, etc? you can bet in an nvidia x86 scenario that intel would drag it out forever, too.

I don't know how prohibitive Intel could really make the x86 licensing process without attracting interest from the DOJ.

I personally believe the reason we don't see more players in the x86 space is simply that the barrier to entry resides in the upfront development costs involved combined with the need for access to absolute bleeding edge process technology or else you aren't going to see higher ASPs than what Via sees now.

Texas Instruments had/has an x86 license and we used to make our own 486 CPUs as well as fabbing Cyrix's 486 chips. But you never heard of a pentium-class TI x86 CPU because the design costs combined with our small market share just didn't justify trying to stay in that market. (same reason we got out of the dram business as well)

If Intel gave Nvidia a free x86 license today it would 4-6 yrs before their very first product could come to market and they'd have to invest even more money in its development than AMD will spend in the same time to develop their own Intel-competive chip (because AMD already has the staff, infrastructure, and patents plus an existing development pipeline that isn't starting from zero)...does NV even have the cash for that?

 

[cbn]

I don't know how prohibitive Intel could really make the x86 licensing process without attracting interest from the DOJ.

I personally believe the reason we don't see more players in the x86 space is simply that the barrier to entry resides in the upfront development costs involved combined with the need for access to absolute bleeding edge process technology or else you aren't going to see higher ASPs than what Via sees now.

Texas Instruments had/has an x86 license and we used to make our own 486 CPUs as well as fabbing Cyrix's 486 chips. But you never heard of a pentium-class TI x86 CPU because the design costs combined with our small market share just didn't justify trying to stay in that market. (same reason we got out of the dram business as well)

If Intel gave Nvidia a free x86 license today it would 4-6 yrs before their very first product could come to market and they'd have to invest even more money in its development than AMD will spend in the same time to develop their own Intel-competive chip (because AMD already has the staff, infrastructure, and patents plus an existing development pipeline that isn't starting from zero)...does NV even have the cash for that?

How about the ARM market? There is potentially more growth, but there are a lot more companies competing in that market right?

 

[Idontcare]

How about the ARM market? There is potentially more growth, but there are a lot more companies competing in that market right?

ARM is where everyone are going. It'll be huge, much much bigger TAM than the x86 space.

 

[cbn]

ARM is where everyone are going. It'll be huge, much much bigger TAM than the x86 space.

I have read about 2Ghz ARM CPUs being just as fast as atom while consuming much less power (although I am not sure how true that is). I'd imagine in the netbook space where people are critical about battery life this could be a huge boon to whomever decides to pick up the technology.

But what about operating systems? Are there any failing phone Operating systems that could find new life reconfigured as a pure smartbook/netbook Operating system when coupled to either AMD or Nvidia GPGPU?

 

[Phynaz]

A Windows 7 port would be a huge boost to Tegra.

Yes it would be.

Not ever going to happen though.

 

[cbn]

But I'm not really sure what mobile devices would want with GPGPU at this point. You're not going to run PhotoShop or anything on your phone or iPad, I think.

Well maybe we will eventually see external Video cards (with their own enclosures and Power supplies) as a way of supplementing ARM powered devices?

How much work can a cpu offload to the Video card? Could Photoshop be run almost entirely on a video card?

 

[jvroig]

Well maybe we will eventually see external Video cards (with their own enclosures and Power supplies) as a way of supplementing ARM powered devices?

External video card enclosures and power supplies on an iphone, ipod touch, or ipad? Not very likely.

Could Photoshop be run almost entirely on a video card?

Not at all. Last I heard, the OpenGL benefits were just zooming in and out faster while having the image much sharper and clearer than before, rotating images, scrolling etc. Basically, GPGPU added features and speed. That was in CS4. I am not sure if CS5 actually did more offloading, but through OpenGL that won't be much.

EDIT: Here, found the page from nVidia about the acceleration benefits: link

It added features, not really took over the CPU as far as I understood that page and the discussion about it several months ago here in AT.