If I were the CEO of AMD... (rate)

[lamedude]

I’d shut it down and give the money back to the shareholders.

 

[Tsavo]

I’d shut it down and give the money back to the shareholders.

I'd make Pentium Pro clones and iPad cozies.

 

[ShawnD1]

I’d shut it down, pocket the cash, and flee to Belize.

I just went ahead and fixed that for ya

 

[Ferzerp]

*sigh*

Okay, let me explain this yet another way;

Say I'm drowning in debt - I'm under water on my mortgage, I've missed a car payment, and I'm only making the minimum on my credit card. My credit score (market cap/valuation), accordingly, is crap.

In order to take care of basic finances through the next month, I sell a few things on Craigslist; my stereo, my TV, some old clothing... whatever.

At no point during those Craigslist transactions is a potential buyer going to say to me "Hey, I'm interested in that 37" HD TV you have for sale, but I would like to ask how your personal finances are? Because if you're indebted, I'm only going to offer $20 for the TV you're asking $200 for because I think you're shady." And do you know why they're not going to say that? Because my FICO score has nothing to do with the value of my bedroom television.

The assertion to which I responded was that debt affects the value of assets. That is patently untrue. Moreover, the assertion was that a company couldn't possibly possess salable assets worth in excess of their market valuation - of course they could, because the values of those salable assets, like fabs and IP, are completely unaffected by the company's debt level. The value of an asset would only be affected by the owner's debt load were the buyer to inherit debts related to that asset; like taking over a car loan.

The only entity to whom debt level would factor in determining the value of a salable asset would be a creditor. But I'm not selling my TV to MasterCard, I'm selling it to some dude named Joe who wants a used TV.

Or; Even though the value of my personal possessions is $20,000, American Express, upon assessment, will only approve me for a $500 credit card because I also have $100,000 in debt. That does not mean, however, that I could not possibly raise more than $500 by selling stuff from around the house.

Why are you guys having such a hard time understanding this?

The picture is different when you're heavily in debt, bleeding cash, and start selling off assets to do something other than service that debt.

 

[krumme]

@2timer

Points colected as follows:

1 point for the plan
2 points for having the courage to post your plan
2 points for taking feedback and reflecting on it
1 point for having a good mood and a friendly attitude
2 points for starting a tread without any hidden agenda, half-trolling, sarcastic pitch
1 point for the results in the thread, its not about the OP, but the results discussed in the tread
-1 point for backing out, stay on it !

 

[mrmt]

This entire conversation is pretty moot though because AMD has already emptied out the closets. If they wanna sell anything else, it will literally be a limb.

Nope, the closet isn't empty. Quite the opposite, if they are going to sell, they will have to sell the meaty parts: IP.

And IDC is quite correct, the seller and the buyer themselves makes a difference in the perceived value of a item and the situation of then also dodoes impact too (hence the reason I pointed out that value isn't intrinsic). I would also add other objective points like legal terms, payments (when, how, terms), guarantees, etc.

When they reach the point of selling their IP, and they are not too far away from it, it will be quite a test for AMD management.

 

[mrmt]

Isn't that what happened to AMD when they sold off some of their IP to Qualcomm a few years back? Didn't they sell a business unit for something like $65m that was viewed as being worth a few multiples of that, but Dirk was desperate to raise cash to fund bulldozer development at the time? (motivated seller situation)

AMD sold the division because it was loosing money, and at the time the only thing that could lose money at the company was the CPU division in the eternal crusade against Intel. The division had potential, but would require a significant amount of capital and time for investment maturation, both things that AMD didn't want to afford.

 

[pablo87]

@2timer

Points colected as follows:

1 point for the plan
2 points for having the courage to post your plan
2 points for taking feedback and reflecting on it
1 point for having a good mood and a friendly attitude
2 points for starting a tread without any hidden agenda, half-trolling, sarcastic pitch
1 point for the results in the thread, its not about the OP, but the results discussed in the tread
-1 point for backing out, stay on it !

:thumbsup::thumbsup::thumbsup:

 

[pablo87]

AMD sold the division because it was loosing money, and at the time the only thing that could lose money at the company was the CPU division in the eternal crusade against Intel. The division had potential, but would require a significant amount of capital and time for investment maturation, both things that AMD didn't want to afford.

AMD sold because they're......(not fit to print), not that I mind, I have a stake in Qualcomm doing well.

 

[zephyrprime]

Selling ATI off would be a death sentence for AMD. Nobody but enthusiasts want to buy a CPU only and enthusiasts are a tiny market. To make money, you need to sell to 98% of the market and those customers want cpu+gpu combos.

 

[sm625]

AMD needs radeon to beat intel in gaming. They could have done this 5 years ago but failed. They needed to completely rework their x86 cores with new instructions specifically tailored for game devs. ie they put their decade of graphics IP to good use by building a cpu core that actually is a graphics core. So instead of processing a frame and sending it to the gpu over a bus and loading and filling and storing a bunch of memory blocks, they just seamlessly process and render the frame all in one step, using a driver written to take advantage of the new uA. If done right it revolutionizes what you can do with per watt of computing power. They have not done this and have not even come close to doing this, despite almost 10 years of preparation. (I say 10 years because the plans were surely floating around the bowels of AMD and/or ATI 10 years ago). Nothing. Just an epic fail. HSA is a joke. Sure, they had to get around microsuck and directx, but hey if they couldnt do that then it made no sense to go forward with the deal.

 

[BenchPress]

With the infusion of cash from restructuring and Wall Street, pour money into TSX R&D.

TSX doesn't cost billions to implement. And besides, AMD already experimented with transactional memory technology before.

That said, AVX2 is currently a much better way to achieve higher performance than more cores. In Haswell, AVX2 doubles the throughput per core. Haswell will also achieve significantly higher Hyper-Threading performance, thanks to having two pairs of symmetrical execution ports for each type of scalar instruction, so the threads don't have to share things. AMD might catch up with say Sandy Bridge for multi-threaded software by implementing TSX, but not with Haswell, which is the real target for software that uses TSX. Improving Hyper-Threading goes hand-in-hand with TSX, and there's nothing AMD can easily do to match that.

Last but not least, AMD should not sell all of its graphics departments. The CPU and GPU are converging, and after the APUs come the UPUs (unified processing units). AVX2 is a leap ahead but it's part of Intel's plan to pave the way to a UPU. AMD would have to keep up with that, or better yet be the first to implement AVX-512 or AVX-1024. But note that Intel already has 512-bit instructions in Xeon Phi...

 

[2timer]

OP is back. (lol)

Great thread guys. I really appreciate the humor and insight of the comments. It kind of spun out of my control, though, because I waded into a topic that I don't have real world experience with. I read benchmarks and I enjoy tech blogs. But I guess I am uninformed about the industry. That's why I decided to back off - I don't know enough about the CPU industry as a whole to really debate.

Thanks for the feedback everyone!

 

[Wall Street]

If I were running AMD I would fire some underperformers, go into crisis mode and offer big bonuses to the Radeon driver team.

1. The frame latency articles are bad press and hurting sales.
2. Radeon 7000 series is your one saleable product currently out there and will be in the market for at least the next year.
3. They are rumored to have an official dual-GPU card that hasn't been released yet, probably because if released as-is the reviews will focus on the multi-GPU frame latency driver problems

The best think that AMD could to is spend a half-million or so to incentivize the driver team to work 6 days a week 12 hours a day in "crunch mode" for the next two months. The drivers are knee capping this product and a small investment would yield tens of millions in additional sales over the next 12 months. The memory manager rewrite needs to be done like yesterday, they have been talking about it for months. I know that the drivers are complicated and take a while, but if you have the hardware and don't have the drivers, you might as well not even bother with the hardware.

Also of note is the idea that AMD should stop investing in strong core and only invest in low power cores. This is a bad idea. Charlie D at Semiaccurate highlights the weakness of this argument correctly: the problem with this is that today's strong cores are tomorrows low power cores. Right now ARM is putting in the investment money in Out-of-Order execution and other aspects that AMD and Intel have had to invest in. If AMD falls behind the curve they may end up like VIA, who years ago decided to target only the low power / low speed space but now only have anemic chips and no path to being competitive. In five years Intel will be selling a Celeron with many cores, on die graphics, TSX and AVX 2, so if AMD doesn't keep investing in competitive cores they will be a non-player at that time.

 

[mrmt]

Selling ATI off would be a death sentence for AMD. Nobody but enthusiasts want to buy a CPU only and enthusiasts are a tiny market. To make money, you need to sell to 98% of the market and those customers want cpu+gpu combos.

And why can't AMD license GPU technology from others like a lot of ARM players do?

 

[mrmt]

If I were running AMD I would fire some underperformers, go into crisis mode and offer big bonuses to the Radeon driver team.

So you would focus in the division that answers for one 1/3 of your sales, and inside this division in products that gives you, let's say, 10% of those revenues? AMD has a CPU problem that cannot be solved by the consumer GPU division. It's nothing but a waste of time and money to focus there.

If you really want to make money with GPU, then the strategy of giving the extra bonus to the GPU driver guys is ok, as long as they deliver good *Firepro/Firestream* drivers. It is there that Nvidia is making money, and if AMD wants to eat Nvidia's bacon they must have a compelling line up here.

 

[AnandThenMan]

And why can't AMD license GPU technology from others like a lot of ARM players do?

Because you instantly lose all the benefits of having your dev teams inhouse, which your main competitor will still have. ARM licenses only, so all the ARM players are at the same competitive starting point. Plus it's not really a valid comparison, a GPU is much more complex than a CPU on the software end. If you license the tech, you have to hope that what you are provided with is going to stay at least on the same pace as your competitor (Nvidia) or you're done.

That is a very scary way to do business.

Let's backtrack. Do you honestly believe if AMD had been licensing GPU IP for the last 5 years, they would have been first with GDDR5 for example? You seem to be suggesting that AMD can have their cake and eat it too. They can sell off ATI's IP to raise cash, but still have all the potential competitive advantages they have now. But the spin off ATI company is also going to license the GPU tech to others, so now AMD has to also compete with them.

This is a dilution of your business at best, the only way AMD should sell off ATI IP is if they don't want to make any GPU based products anymore. Now if AMD wants to license out their GPU tech that is a different story. And actually they are doing just that with the various console contracts.

 

[mrmt]

Let's backtrack. Do you honestly believe if AMD had been licensing GPU IP for the last 5 years, they would have been first with GDDR5 for example? You seem to be suggesting that AMD can have their cake and eat it too. They can sell off ATI's IP to raise cash, but still have all the potential competitive advantages they have now. But the spin off ATI company is also going to license the GPU tech to others, so now AMD has to also compete with them.

No, I'm suggesting that AMD is picking too many fights for the size of the stick they have. You cannot reasonably expect that AMD will develop a chip to compete with Intel, another to compete with Nvidia and jump in the middle of the ARM crowd with the amount of resources they currently have.

Regardless of the technical achievements you mentioned, right now, their iGPU is just dead weight for them. It's just cost, they cannot monetize their iGPU prowess, and they can't make any significant amount of money with their GPU offers. AMD needs to develop the professional market, but they cannot afford it right now because they are in full crisis mode, and they need money, a lot of money to fix their CPU business.

What I'm saying isn't that AMD can have the cake and eat too, it is obvious that AMD will lose some things if they sell ATI, but the money and the reduced management overhead might be worth it.

 

[AnandThenMan]

I still can't see why you want AMD to sell the only part of themselves that is making money. Without GPU tech (licensing is simply not going to work with a highly complex GPU) AMD is only left to compete with Intel head on! I agree AMD needs to pour more resources into the professional market, but they need to stabilize their revenue stream first.

What AMD actually needs to do is breakthrough the death grip Intel has on OEMs. This is the elephant in the room that people don't like to talk about, or are just in denial. Let's put it this way, AMD products are not 80% worse than Intel, yet that is about where the marketshare numbers sit at best. Put another way, if Intel had AMD's product stack, they would still command the lions share of the market. This has been proven more than once when AMD had the best products but it made little to no difference.

So the real goal for AMD should be to create market segments that Intel is not a part of, and work with suppliers that want to embrace these products. No small feat, but that is pretty much the only way ahead for AMD, they are simply never going to change the status quo at this point in the x86 arena.

 

[Wall Street]

So you would focus in the division that answers for one 1/3 of your sales, and inside this division in products that gives you, let's say, 10% of those revenues? AMD has a CPU problem that cannot be solved by the consumer GPU division. It's nothing but a waste of time and money to focus there.

If you really want to make money with GPU, then the strategy of giving the extra bonus to the GPU driver guys is ok, as long as they deliver good *Firepro/Firestream* drivers. It is there that Nvidia is making money, and if AMD wants to eat Nvidia's bacon they must have a compelling line up here.

I think that the console contracting business is all wrapped up. I also think they should continue to progress with CPU R&D however this won't pay off as they don't have much new in the pipeline in the near term. Of course better GPU drivers also impact the compute market and there is a halo effect where performance improvements in the top cards sell lower end cards (how many people comment on Titan and how many pageviews did the Titan review get vs. how many people actually will buy Titan).

The graphics drivers strike me as something where they are 95% of the way to having a great graphics product and a small investment now could pay multiples in the next 12 months. The original post suggested going to investors and promising them a quick payback, and the GPU drivers are one of the few things that would pay off in this time frame. Frankly, I don't think that any of the Bulldozer/Trinity refreshes will be Core i7 killers AMD doesn't have the time or capital to go all in on a new architecture.

 

[mrmt]

I still can't see why you want AMD to sell the only part of themselves that is making money. Without GPU tech (licensing is simply not going to work with a highly complex GPU) AMD is only left to compete with Intel head on! I agree AMD needs to pour more resources into the professional market, but they need to stabilize their revenue stream first.

I want to sell that part because AMD needs to tackle a CPU problem, not GPU, and whatever money they can make with GPU is insignificant for the CPU needs. If AMD cannot make bleeding edge CPUs, and it is clear that they don't have the resources to stay in the bleeding edge in the future, what's the point in bleeding edge graphics?

What AMD actually needs to do is breakthrough the death grip Intel has on OEMs. This is the elephant in the room that people don't like to talk about, or are just in denial.

Nobody does. But to break Intel grip with OEMs you must first have the line up and capacity to fill in case of Intel pull the plug at the OEM. This is not something feasible, it's not even worth discussing.

 

[Idontcare]

I still can't see why you want AMD to sell the only part of themselves that is making money. Without GPU tech (licensing is simply not going to work with a highly complex GPU) AMD is only left to compete with Intel head on! I agree AMD needs to pour more resources into the professional market, but they need to stabilize their revenue stream first.

What AMD actually needs to do is breakthrough the death grip Intel has on OEMs. This is the elephant in the room that people don't like to talk about, or are just in denial. Let's put it this way, AMD products are not 80% worse than Intel, yet that is about where the marketshare numbers sit at best. Put another way, if Intel had AMD's product stack, they would still command the lions share of the market. This has been proven more than once when AMD had the best products but it made little to no difference.

So the real goal for AMD should be to create market segments that Intel is not a part of, and work with suppliers that want to embrace these products. No small feat, but that is pretty much the only way ahead for AMD, they are simply never going to change the status quo at this point in the x86 arena.

Given that AMD has strong IP in the GPU arena, bested only by Nvidia and even then it is an even heat at times, it seems to me that AMD should leverage that IP in the way ARM does.

Become to licensible GPU tech what ARM is to licensible cpu tech.

Instead of trying to compete with the qualcomm's and the Intel's, license your tech out to them the same as ARM does and pull in revenue that way. It works for ARM, and unless Nvidia decided to compete in the same model there would be no one that could compete against AMD's IP.

 

[mrmt]

I think that the console contracting business is all wrapped up. I also think they should continue to progress with CPU R&D however this won't pay off as they don't have much new in the pipeline in the near term. Of course better GPU drivers also impact the compute market and there is a halo effect where performance improvements in the top cards sell lower end cards

Professional drivers aren't the same thing of consumer drivers. You are putting improvements in the professional drivers like a marginal benefit of improving consumer drivers, it doesn't work like that. Professional optimizations occur per app, must be thoroughly validated and there is no room for bugs. It is exactly this attitude of "develop and they will come" from AMD that allowed Nvidia to dominate the entire GPGPU and Professional market with a more proactive approach.

All the measures that you are suggesting would improve AMD's position in a market that is shrinking and both players are making almost no money.

 

[AnandThenMan]

I want to sell that part because AMD needs to tackle a CPU problem, not GPU, and whatever money they can make with GPU is insignificant for the CPU needs. If AMD cannot make bleeding edge CPUs, and it is clear that they don't have the resources to stay in the bleeding edge in the future, what's the point in bleeding edge graphics?

AMD needs to have the best blend of CPU/GPU, that is the future not just the CPU. You'll notice Intel has no GPU tech worth talking about, so one could apply your argument and say if Intel can't make bleeding edge GPUs, what is the point of having the best CPU? It is an antiquated mindset to only think of general purpose tasks only running on the CPU. Try Bitcoin mining on a CPU and see how far that gets you. Look at what Nvidia is doing with their professional line, said GPUs are used in the visual effects industry to great effect, cutting down render times exponentially.

In a nutshell what you are advocating is AMD compete directly with Intel in only CPUs. Why? AMD tried this for most of their history and look at the results. Why would they want to travel down the very same path that has nearly put them out of business, it makes no sense.

Nobody does. But to break Intel grip with OEMs you must first have the line up and capacity to fill in case of Intel pull the plug at the OEM. This is not something feasible, it's not even worth discussing.

This is a different subject that I think should be debated separately. The chicken and the egg scenario is something AMD has struggled with, and in fact was specifically targeted by Intel and the #1 motivation for them bribing OEMs. The critical mass scenario applies.

Given that AMD has strong IP in the GPU arena, bested only by Nvidia and even then it is an even heat at times, it seems to me that AMD should leverage that IP in the way ARM does.

Become to licensible GPU tech what ARM is to licensible cpu tech.

Suppose AMD did this starting tomorrow, who would step up and use the GPU tech? Be specific. Now this, "bested only by Nvidia and even then it is an even heat at times" are you serious? AMD has many times bested Nvidia, going back to the ATI days same thing. They have both traded performance crowns many times, lately AMD has generally been first with node shrinks, launches, and hardware tech such as GDDR5 which AMD essentially created.

 

[mrmt]

AMD needs to have the best blend of CPU/GPU, that is the future not just the CPU. You'll notice Intel has no GPU tech worth talking about, so one could apply your argument and say if Intel can't make bleeding edge GPUs, what is the point of having the best CPU?

You are seeing CPU and GPU as two equals, they aren't. You have just to look at the TAM for CPU and GPU market to see that CPU is *orders* of magnitude bigger than GPU. So yes, Intel can afford to not have bleeding edge GPU capabilities but AMD cannot afford to not have bleeding edge CPU capabilities to compete on the market.

You have to look no further than Nvidia to see that GPU isn't a strong differential as you think. Despite being ahead in graphics than any of the other ARM manufacturers, they are struggling in the mobile arena because they can't field a chip with bleeding edge CPU capabilities.