The AMD Execution Thread [2007 - 2017]
- Thread starter overclocked_enthusiasm
- Start date
-
- Tags
- amd
- Status
Maybe we need to rename the thread. How about 'The AMD Execution Success Thread'?
All in all good results, I thought that the revenue at $1.6B was a little low for a C/GPU company. It is quite ironic that royalties from GPUs in two consoles are enough to move the needle for them. I guess it shows how important a console design win would be for the company as a whole, especially if they can sell them both the CPU and GPU design.
All in all good results, I thought that the revenue at $1.6B was a little low for a C/GPU company. It is quite ironic that royalties from GPUs in two consoles are enough to move the needle for them. I guess it shows how important a console design win would be for the company as a whole, especially if they can sell them both the CPU and GPU design.
Why is it ironic that 20 million units in a quarter would make a difference?
Because thats just a sideline business, they aren't like Rambus who make that their bread and butter of operations.
Silent_Buddha
Legend
That's really quite good when you consider that 2009 benefitted from the Intel settlement.
Q4 2010 GPU operating revenue rose to 68 million. AMD gross margin drop to 45% from 46% month to month was due primarily to lower ASP of CPUs so margins on GPUs remains high. Still, it's a solid 4% YoY rise in gross margins.
It would be interesting to see what margins would be if the GPU division wasn't shackled by the CPU division. It could be in the 50's or higher for all we know considering AMD's CPU division has very little pricing flexibility due to competition with a much stronger Intel.
They are also now finally seeing positive cash flow from Global Foundries related items. Looks like a positive 27 million from that. I wonder what the 283 million from legal settlements is related to? I can't recall any legal battles AMD was in the past couple quarters.
Regards,
SB
Q4 2010 GPU operating revenue rose to 68 million. AMD gross margin drop to 45% from 46% month to month was due primarily to lower ASP of CPUs so margins on GPUs remains high. Still, it's a solid 4% YoY rise in gross margins.
It would be interesting to see what margins would be if the GPU division wasn't shackled by the CPU division. It could be in the 50's or higher for all we know considering AMD's CPU division has very little pricing flexibility due to competition with a much stronger Intel.
They are also now finally seeing positive cash flow from Global Foundries related items. Looks like a positive 27 million from that. I wonder what the 283 million from legal settlements is related to? I can't recall any legal battles AMD was in the past couple quarters.
Regards,
SB
codedivine
Regular
In the earnings call they said they settled with Samsung but didnt say about what.
They got that much money or had to pay it out?
Silent_Buddha
Legend
It was positive cash flow so it was Samsung paying AMD.
Regards,
SB
Regards,
SB
Silent_Buddha
Legend
Even without that, it was a solid profit of, 471 - 236 (after taxes) = 235 million USD compared to losses in 2009 without the Intel settlement (304 - 1,267 = -963 million USD), and massive ~3 billion USD losses in 2008 and 2007.
Regards,
SB
Regards,
SB
So that makes it around a $500M dollar turnaround in profitability, right?
A lot of 2007 and 2008 loss was writedown, so no cash outflow.
Some profits are better than none, but low profits will limit AMD's ability to compete.
Also, wrt zacate/ontario, their target market is netbooks, which will come under pressure from tablets, for which amd still has no competitive chip.
Considering Atoms are going tablet too, Zacate/Ontario can fit there too.
Moorestown had much lower power than a bobcat system. It's 9W for the cpu/gpu alone plus more for southbridge.
single core bobcat set up hits 9 watts. Dual core is 18 watts. Later this year we will see them move from 40nm to 32/28nm which should allow them to either pack more cores in or reduce power usage.
So I think its safe to say the single and dual core bobcats will make it into tablets at some point this year.
Also lano seems like it will be a good low end desktop chip and i think that is coming out this quarter. It may not light the world on fire , but I gotta believe there will be alot of oems that would love to show off a 6 core phenom 2 with a decent on die gpu .
Hopefully bulldozer lets them sell a chip for more than $200 bucks though
There are dual core versions (lower clocked ofc) which take 9W.
ATM, they are an order of magnitude away from their power targets. CPU+NB+SB should be < 1W to go into tablet market, since you can't have even a passive heatsink and you need to last > 6 hrs on one charge. Even the best fusion dies take 9W excluding SB.
Bobcat is supposed to be <1W capable. Sure, Ontario is a 9W product, but it's also basically the only SKU at 9W (well, there's the 1.2GHz single-core too) which means there is probably very little binning going on. It seems to me that AMD has decided to emphasize lineup simplicity, at the expense of performance/W for "top" products. In fact, there are no "top" products to speak of, while Intel does have such "top" products, it the ULV family.
I bet that "ULV" versions of Ontario could achieve the same level of performance (or pretty close) at significantly lower voltages, hence power levels.
Anyway, Ontario probably isn't very well suited to tablets (the presence of two cores, for one thing, needlessly increases leakage) but AMD could probably design a tablet-capable chip based on Bobcat. Perhaps they ought to wait for 28nm before doing it, but I think it's possible. It might be necessary, or at least highly beneficial, to integrate the southbridge in such a chip, both for cost/integration reasons and to limit I/Os, which can be quite costly power-wise.
You have to pull the entire SoC <1W. T2 is <1W @peak, iirc.
Actually it's not about core counts, it's about clocks.
C-series has 9W, and there's both single & dualcore C-series models. E-series is 18W and again there's both single & dualcore models.
- Status
