Discussion: AMD's last hope for survival lies in the Zen CPU architecture

Page 7 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.


Why would they want to? Outside of the Desktop market, which EVERYONE agrees doesn't have any growth potential, no one uses X86 for anything. There's no money to be made.
 

I don't see anything in particular. What should I be looking for?

There are news reports that AMD recovered market share in the GPU department during Q3 2015. I expect this to continue during Q4. On the CPU side things are quiet however.
 

If ARM's campaign to whip up interest in server-side chips succeeds, the high-efficiency server market will be overrun with relatively inexpensive and mostly interchangeable chips too.
 


didn't you notice desktop FX to be launched in 2017 .. ??!!!!
 


ARM in servers isn't going anywhere. The performance isn't there, and the software isn't there. POWER and SPARC will continue to reign supreme in that space.

didn't you notice desktop FX to be launched in 2017 .. ??!!!!
Click to expand...

Desktop Zen APUs are scheduled for Q1 2017; the non-APU parts are still scheduled for Q4 2016.
 

Not every server needs record-breaking IOPS or computing power. Earlier this year, Intel launched the 14nm Xeon D-1500 based on the Atom C2000 for applications where low power is more important than raw performance.

For example, if the IoT becomes a thing, there will be high demand for low-power servers that do little more than aggregate, buffer and log sensor data.
 

Yes, but I was mostly referring to the CPU/APU market share, not their future products. Sorry if that was unclear.
I just bought myself a relatively cheap APU laptop for light gaming, and it's great. An A10-7400p APU with 8GB RAM and 1TB HDD for $430 is a great deal. It runs Mass Effect 3 at max settings easily (not the most recent game, but hey, that's what I play nowadays lol). Only now are we seeing APUs becoming more popular in cheap laptops. So we might see some improvement there also in the market share in Q4. But the gains won't be significant until Zen.

I'd also like to mention that no revamp of the FX CPUs took place due to problems with the die shrink that were not (entirely) in AMD's hands. They were forced to stay at 28nm all this time and this really hurt them. It's a miracle AMD is able to stay competitive while being stuck at 28nm, especially on the APU side. If they get the die shrink right this time, I really don't think Zen can fail in any category. If they can match Intel's IPC while maintaining the integrated GPU performance that they currently have, I don't think they can go wrong. They don't really have to surpass it.
 
Here's my opinion: In the "best case" that AMDs 40% IPC gains are accurate for all applications (which I doubt), that still only puts Zen at Haswell level performance, so Intel will still be 20-30% ahead on that front. And if those gains happen to include the benefit of the SMT cores, then the numbers are even worse for AMD.

I typically take "best case" numbers, and assume the "typical case" numbers are half that. So I'd expecting ~20% typical IPC gains over SR, which isn't that impressive.
 
I'll assuming the 40% IPC gain are for programs or benchmark that can take advantage of SMP efficiently.

I generally do not bother with trying to guess performance increase of future AMD / Intel processors. I remember seeing posts of people boasting that Skylake was going to give a 20% increased performance over Broadwell...

Too much hype or over enthusiasm can lead to devastating disappointment.
 

Those boasts were misquotes. The original sources were comparing Skylake to Haswell but Skylake fell short even from that.
 


I think AMD simply mentioned that 40% for the sake of saying something ..

The new architecture probably has no similarities to Bulldozer so how could they have guessed ??!!!


And besides at that time they didn't have any working silicon either so it's impossible to estimate performance over excavator ..

AMD themselves do not know how Zen is going to perform LOL !!
 
lol it's not like AMD is going out of business anytime soon, they did win the peasant wars... I mean console wars meaning they are selling a shit ton of shit CPU's.
AMD will NEVER be on my radar for a gaming PC for their CPU's or GPU's and I am not a fanboy I've just seen the difference. A few years back I had a AMD Phenom x4 840 something with HD 6870 2GB and my computer struggled to run games at med-high settings @1080p the whole build cost me around $700 I changed up and went with i5 2500k and GTX 660 FTW and started playing many of the same games at high-max and cost me around $800. The performance difference is clear and AMD simply can't beat it.

My point is, you can get better performance for roughly the same cost, and if you just can't spend an extra $100 wait and save a couple months, the performance gain is worth it.

That being said, AMD can't just keep adding to existing CPU/GPU's, the 300 series was an absolutely disappointing rebranded 200 series and at a much higher cost. My current GTX 970 (for $300) is approx. 4% slower with an average of 3 FPS less than R9 390x ($400) and that's a GTX with stock speeds no FTW or Ti or OC vs a OC'd 390x out of the box FOR $100 MORE!!! what makes sense!? For $400 I could get a GTX 970 that smokes an R9 390x and then OC it to make sure it simply can't keep up. Bottom line, AMD is already dead to me, they need to get out of the PC market and stick to the peasant market, but that just can't happen because of Monopoly laws or some stupid crap I probably should have read up on before commenting on. AMD can't keep up with the competition as far as performance goes, and I'm ok with that, I have never been happier than I am with my intel/Nvidia build and I haven't seen even close to similar performance from an AMD/AMD build in the same price range.
 


From what I have read in an unconfirmed analysis, AMD only makes about $10 profit per console sold. Meaning the amount of money left after subtracting manufacturing and overhead costs per unit.

In 2012 AMD's Net Income was a loss of $1.18 billion. In 2013 they lost $83 million. In 2014 they had a loss of $403 million. Assuming the $10 per console is correct, Sony & Microsoft would need to sell over 166 million consoles combined in order for AMD to wipe out their total loss of $1.666 billion from 2012 to 2014. That excludes inflation.
 

Selling "shit tons" of chips is almost pointless when the net margin is rather small and volume is not that high.

AMD has already lost over 300M$ for FY2015 and their semi-custom division which handles console SoCs, AMD's only profitable business unit, only turned in 27M$ in profit over the last quarter, down from 45M$ in Q1 and 97M$ in 2Q2014. At this rate, even this division might turn up a loss in the near future.

To top that off, AMD is losing about 10% in revenue per quarter across the board.

AMD will need a major miracle to get out of this.
 


When I Google searched when Zen will be released, it said October 2016. Maybe I misread it?
 


I like that they are claiming the "performance poer watt" for their market. I assume that is because no other x86 chip is in where they are because last I checked Intel has much faster CPUs at the same or lower TDPs.

I don't see VIA getting into it. They don't do much beyond audio/nics it seems.
 


As I've said many times, the total take for the console wins isn't really that much. You're maybe about $80 Million per year, which is a nice chunk of change, but not really going to move the financial needle in either direction.
 


The point of my post was basically to point out the fact that AMD cannot rely on the sale of consoles with their APUs to keep the company afloat.

Hypothetically speaking, over 166 million PS4 and Xbox One consoles would need to be sold to offset AMD's loss from 2012 - 2014 assuming the $10 profit margin is accurate. Based on the following links it took 8 years (2005 - 2013) for Sony to sell 80 million PS3 consoles. The same can be said of the Xbox 360.

http://venturebeat.com/2013/11/05/playstation-3-sells-80m-units-not-bad-but-far-short-of-150m-ps-2-sales/
http://vgsales.wikia.com/wiki/Xbox_360

Based on the rate at which those previous generation have sold, it is absurd to think that sales of consoles with AMD APUs in current generation consoles will keep AMD afloat since they are more or less hemorrhaging losses and debt.
 
I will be sad to see AMD go. I still find their chips to be a good value personally. I build many AMD vishera gaming PCs and they sell well because they can still handle most games with a decent GPU (specifically GTX 960s). Since the price is far lower than an intel build, and it still handles World of warcraft, Skyrim and Witcher 3, I think it's worth it. But looking at the financial position of the company, it makes me nervous as hell.
 


That's just the thing. If Intel WAS the same price, then we would all pick Intel. Price is currently what AMD has to offer over Intel.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom