AMD Reports 2015 Second Quarter Results

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
​AMD today announced revenue for the second quarter of 2015 of $942 million, operating loss of $137 million, and net loss of $181 million, or $0.23 per share. Non-GAAP1 operating loss was $87 million and non-GAAP1 net loss was $131 million, or $0.17 per share.

“Strong sequential revenue growth in our EESC segment and channel business was not enough to offset near-term challenges in our PC processor business due to lower than expected consumer demand that impacted sales to OEMs,” said Dr. Lisa Su, AMD president and CEO. “We continue to execute our long-term strategy while we navigate the current market environment. Our focus is on developing leadership computing and graphics products capable of driving profitable share growth across our target markets.”
 
If a company was looking to buy AMD, or a part of, I would think they would do so soon. This is a potentially good time to buy.
 
If a company was looking to buy AMD, or a part of, I would think they would do so soon. This is a potentially good time to buy.

Agreed. I was going to buy some of their stock back when it was around $2.25 (before the Fury announcement). Thankfully I didn't. When it hits below $1.50 I'll probably pick up a few hundred shares and sit on them for a while. I'm figuring that either AMD will get bought out or the upcoming Zen CPU will make them at least somewhat competitive and I'll make a little bit of a return. If not, at least it was worth a shot. :)
 
Agreed. I was going to buy some of their stock back when it was around $2.25 (before the Fury announcement). Thankfully I didn't. When it hits below $1.50 I'll probably pick up a few hundred shares and sit on them for a while. I'm figuring that either AMD will get bought out or the upcoming Zen CPU will make them at least somewhat competitive and I'll make a little bit of a return. If not, at least it was worth a shot. :)

$1.50 would be a nice place to jump in. However I don't think they will be over 2$ for quite some time.
 
AMD stocks broke all time low (since inception today)... once you reach that point... it will take tremendous effort to get up. Unless AMD does serious reorganizations... I don't think it will happen.

It will take at least 18 months until AMD stocks will be able to raise due to the upcoming Zen... that's IF AND ONLY IF, it's competitive with Intel product... until then I can only see AMD stock price struggling near 1$.
 
Buy a stock now that's only going to go down? knock yourself out

Lots of people will blow more than that in an evening at the casino. Your chances of not losing all your money on AMD stock is a bit better too.
 
Lots of people will blow more than that in an evening at the casino. Your chances of not losing all your money on AMD stock is a bit better too.

I'd also venture to say the odds of making a return on investment are better at a casino, unfortunately. Hope Zen delivers.
 

“Strong sequential revenue growth in our EESC segment and channel business was not enough to offset near-term challenges in our PC processor business due to lower than expected consumer demand that impacted sales to OEMs,” said Dr. Lisa Su,
I wonder how much of EESC's bump came from dumping remaining SeaMicro inventory.
 
I really hope that I'm wrong, but looks like this is it for AMD as we know it.

Meaning, unless someone buys or merge with them, they are just done.

But as I said, I really, really hope that I'm wrong.
 
Lots of people will blow more than that in an evening at the casino. Your chances of not losing all your money on AMD stock is a bit better too.

It will be if they can recapture market share, but there's no guarantee of that. The trend isn't looking good. Zen has to hit hard and restore it to the glory days of the K7. Maybe it will, it's the same chip designer.

Oh how I miss my Thunderbird with my ATI Radeon 9700 (soft hacked from the 9500, saved myself some good cash!!!)!
 
It will be if they can recapture market share, but there's no guarantee of that. The trend isn't looking good. Zen has to hit hard and restore it to the glory days of the K7. Maybe it will, it's the same chip designer.

Oh how I miss my Thunderbird with my ATI Radeon 9700 (soft hacked from the 9500, saved myself some good cash!!!)!

What happens to console makers (Microsoft/Sony) if AMD goes under? I presume there's some agreement in there that gives them royalty free rights to produce the chips as well as handing over tape outs?
 
I'm not sure when AMD's long term strategy is going to turn the tide for them. I would like to see what their long term strategy is all about, because so far it doesn't seem to be working for them. For the past couple of years it seems AMD has been losing hundreds of millions each quarter.
I don't think one single new product is going to save them. Even if it's a 'home run' so to say, it may lessen the sting a bit, but that's about it.
No matter what happens, I guess it will be interesting to watch.
 
I'd also venture to say the odds of making a return on investment are better at a casino, unfortunately. Hope Zen delivers.

I'm no expert in these stuff, but do shareholders usually make something when a company is acquired? If they do, then perhaps some people might bet on AMD being acquired by someone else in the near future.

I think AMD being bought out is a very likely scenario at this point. It's hard to see how they could carry on bleeding money, but at the same time they have stuff that other companies may want, such as their graphic tech.
 
I'm no expert in these stuff, but do shareholders usually make something when a company is acquired? If they do, then perhaps some people might bet on AMD being acquired by someone else in the near future.

I think AMD being bought out is a very likely scenario at this point. It's hard to see how they could carry on bleeding money, but at the same time they have stuff that other companies may want, such as their graphic tech.

Only if you sell your stock and they aren't bankrupt in the process.
 
IT IS NOT DEMAND IT IS A PRODUCT WORTH BUYING THAT IS STAGNATING THE MARKET

We will not sit by and clamour for the 5-15% gains on the previous generation of your products anymore we want something that will set paradine slash the power use and increasing ipc is not revolutionary it is expected. If amd wants to quit sliding down the walls of urinal they are going to need a foot hold and that will be something that smashes these upgrade tracks and sets a new point of contention for how a pc should run...

Now would be a very good time to make a cpu that could run at 4-6 ghz and have a Flash sector that could be upgraded to newer instructions. Or instead of dual threaded cores which is coming with zen make triple or quadruple threaded cores and call it quantum threading or super threading. Or maybe come up with some method of making several logical cores able to work together on a single thread. to eliminate a single threaded difference completely by making even single threads finished by multiple cores...
 
Now would be a very good time to make a cpu that could run at 4-6 ghz and have a Flash sector that could be upgraded to newer instructions.
You're describing an RFPGA, and they exist at much lower clock speeds. Since they are general devices, they are not optimized to the specific task. The way blocks are organized limits clock speed and circuit efficiency. Over the years many universities have claimed performance breakthroughs using massively parallel architectures on FPGAs as a replacement for general CPUs, but every single one never lived up to the hype. ASICs and CPUs rule not because there's a dark conspiracy holding down alternatives, but because they're best at meeting manufacturing scale, price and performance. Want to rapidly develop a new processor without the expenses of custom silicon?

An FPGA is the way to go, then you'd export and validate it using standard cell libraries and tools and wind up with a final product that's faster, more energy efficient and 1 or 2 orders of magnitude cheaper to make at scale.

For the rest you're simply hand-waving away limits of computational theory, formal verifiability of radical execution model architectures and probably making Amdahl's corpse silently weep too.

tl;dr If it were that easy, we'd already be there.
 
You're describing an RFPGA, and they exist at much lower clock speeds. Since they are general devices, they are not optimized to the specific task. The way blocks are organized limits clock speed and circuit efficiency. Over the years many universities have claimed performance breakthroughs using massively parallel architectures on FPGAs as a replacement for general CPUs, but every single one never lived up to the hype. ASICs and CPUs rule not because there's a dark conspiracy holding down alternatives, but because they're best at meeting manufacturing scale, price and performance. Want to rapidly develop a new processor without the expenses of custom silicon?

An FPGA is the way to go, then you'd export and validate it using standard cell libraries and tools and wind up with a final product that's faster, more energy efficient and 1 or 2 orders of magnitude cheaper to make at scale.

For the rest you're simply hand-waving away limits of computational theory, formal verifiability of radical execution model architectures and probably making Amdahl's corpse silently weep too.

tl;dr If it were that easy, we'd already be there.
The fact that both amd and intel have done nothing to advance computing outside of silicon optimization over the past 5 or so years is still true.

And the hand waving over current methodology was attempts without a background or knowledge of low level design at thinking of a new direction that could be taken to set a paradigm and change the way computers work for the future...

I still think if amd could design a chip that could take a single threaded process and somehow split the workload of that single process to multiple cores without the process being a true multi threaded process. Amd could then fix the lackluster single threaded performance they have shown for the past couple years... But i know that something like that is more software side of things and that if they are dealing with software side there is no reason to not just make it multithreaded to begin with.

My idea of triple or quad threaded hyperthreading over the standard 2 thread hyperthreading.
 
The fact that both amd and intel have done nothing to advance computing outside of silicon optimization over the past 5 or so years is still true.

That's just not true. Take a computer organization course and you'll realize your arguments come from the arrogance of ignorance.
 
facepalm

Maybe samsung will buy them for a fat discount

Samsung could have bought them last year with the change they find in their couch.

AMD is worth less than Minecraft. If someone truly wanted to buy them they would have by now.
 
Whatever they're paying Lisa Su it's too much. She needs to do like Jobs did and take a $1 salary until she turns things around. She's coming up on being there a year, so the "Well she hasn't been there that long" excuse is going to become exceedingly stale.
 
That's just not true. Take a computer organization course and you'll realize your arguments come from the arrogance of ignorance.

Yea but people can only see superfluous benchmarks. I don't blame them, they think if they get 5 billion more points in a replay their Facebook experience goes way up.
 
Whatever they're paying Lisa Su it's too much. She needs to do like Jobs did and take a $1 salary until she turns things around. She's coming up on being there a year, so the "Well she hasn't been there that long" excuse is going to become exceedingly stale.

Well, to be fair to her, she was made captain of a sinking ship. Maybe the ship is beyond repair at this point, and her job is to keep it afloat until someone comes along and rescue it.
 
Whatever they're paying Lisa Su it's too much. She needs to do like Jobs did and take a $1 salary until she turns things around. She's coming up on being there a year, so the "Well she hasn't been there that long" excuse is going to become exceedingly stale.

She is probably golden parachuting she says she is there to bring it to profit but if she sinks it samsung buys it she gets paid elevendy billion dollars as she walks out the door. Intel then will need to go to whoever ends up with certian patents and negotiate a new cross licensing deal if they can't just buy out amds patent portfolio to begin with...
 
The fact that both amd and intel have done nothing to advance computing outside of silicon optimization over the past 5 or so years is still true.

That's just not true. Take a computer organization course and you'll realize your arguments come from the arrogance of ignorance.
I was typing on a tablet last night, potato keyboard style, so I didn't get to all I wanted to type.

A few web sites have great analysis of micro architecture changes in each generation. Johan's articles at AnandTech are great examples. x86 still has the bleeding edge advances first in production processors, but the details can be technical. A few sites offer analyses of these changes in accessible language.

The problem really is that it becomes "fact" that nothing changes, when that view is driven by people who are woefully ignorant of understanding or appreciating how each improvement, x86 extension and optimization are advances. The information is out there for anyone to learn.

And to get back to the earlier argument, AMD's CPU series suffers because it basically ditched hand tuned circuits in favor of automated design tools. It's a trade-off. An advantage is seen in how AMD can squeeze more transistors into a given area, helping to reduce the manufacturing cast. The disadvantage is that automated tools rob efficiency, leading to poor IPC and specifications (like cache bandwidth) that severely lag much older Intel chips. Fixing this is a engineer recruitment/retention/culture/money problem, so it's not likely to change as AMD's R&D/OpEx budgets shrink.
 
And to get back to the earlier argument, AMD's CPU series suffers because it basically ditched hand tuned circuits in favor of automated design tools. It's a trade-off. An advantage is seen in how AMD can squeeze more transistors into a given area, helping to reduce the manufacturing cast. The disadvantage is that automated tools rob efficiency, leading to poor IPC and specifications (like cache bandwidth) that severely lag much older Intel chips. Fixing this is a engineer recruitment/retention/culture/money problem, so it's not likely to change as AMD's R&D/OpEx budgets shrink.

I see this argument floating around occasionally, but I think it's pretty unfounded. Both Intel and AMD are implementing their processors in a mix of RTL and schematics. Critical timing blocks are going to be full-custom for both Intel and AMD; other circuits will be automated. Intel may have more (and better) engineers dedicated to physical design, but it's not a fundamental methodology difference.

For the mainstream CMOS processes, it's simply not possible to design anything but trivial circuits in the 3+ GHz range without going custom. Hypothetically, if AMD were somehow able to implement their processors using only push-button synthesis and PnR tools they should get out of the microprocessor industry and just sell EDA tools.
 
AMD is worth less than Minecraft.

ZZNuQEc.gif
 
I see this argument floating around occasionally, but I think it's pretty unfounded. Both Intel and AMD are implementing their processors in a mix of RTL and schematics. Critical timing blocks are going to be full-custom for both Intel and AMD; other circuits will be automated. Intel may have more (and better) engineers dedicated to physical design, but it's not a fundamental methodology difference.
It makes no sense to hand design all units. Sorry if that was unclear because I didn't mean to suggest it.

But it is a very fundamental difference between AMD and Intel now. AMD switched to fully automated layout and RTL and Intel still hand tunes critical sections, in addition to automation.
 
AMD switched to fully automated layout and RTL and Intel still hand tunes critical sections, in addition to automation.
As of late 2011 when I was still at AMD as an IC designer, critical blocks were still full-custom (typically anything much above 1 GHz). As I mentioned, it's not possible to make modern 3 GHz microprocessor without going custom for the critical blocks that are running at-speed.

Conversely, I've worked with Intel more recently on a few low-frequency designs and they are fully invested in automation just as the rest of the industry is.

Also, see this post and several other in the same thread by an ex-AMD designer who explains how AMD got into this mess by going automated: http://forums.macrumors.com/threads...for-upcoming-macs.898409/page-25#post-9746191
I assume Cliff's comments are somewhat uncontextualized, or maybe a by-product of the era he worked in. They're the EE equivalent of an assembly programmer criticizing the use of a compiler. If Cliff was working at Intel during the same era, he likely would have seen the same transition.

Yes, sometimes automation won't be sufficient - part of the job of integrated circuits designers is to identify how circuits need to be built.
 
Back
Top