The AMD Execution Thread [2007 - 2017]

Status
Not open for further replies.
I should say in support of UT that I agree AMD has had a long history of misleading investors, analysts, and being generally disingenuous in their "predictions". Nobody has been more critical of AMD than me in the past of this practice...particularly under Hector Ruiz, Dave Orton, and Henri Richard.

However, I would posit that the new AMD senior management team under Rory doesn't have that same intent or burden to feel compelled to do so. Rory blew the whole thing up...and restructured. He has met or beaten all of his restructuring metrics including: market share, profitability, cost control, and most product launches. While predicting the market adoption of a new product is dicey at best, I would argue that Rory and his team have forecasted quite well...especially in contrast to the former AMD senior management teams.

As a result, I think they will get a pass from this blatant attempt to extort a company on the rebound. The lawyers see the new cash...so they strike. Opportunism at it's best...err I mean worst.
 
overclocked_enthusiasm said:
I should say in support of UT that I agree AMD has had a long history of misleading investors, analysts, and being generally disingenuous in their "predictions". Nobody has been more critical of AMD than me in the past of this practice...particularly under Hector Ruiz, Dave Orton, and Henri Richard.

However, I would posit that the new AMD senior management team under Rory doesn't have that same intent or burden to feel compelled to do so. Rory blew the whole thing up...and restructured. He has met or beaten all of his restructuring metrics including: market share, profitability, cost control, and most product launches. While predicting the market adoption of a new product is dicey at best, I would argue that Rory and his team have forecasted quite well...especially in contrast to the former AMD senior management teams.

As a result, I think they will get a pass from this blatant attempt to extort a company on the rebound. The lawyers see the new cash...so they strike. Opportunism at it's best...err I mean worst.
Click to expand...

I have been following this thread for a while and I am eagerly awaitig your overview O_E
 
Well, my optimism was misplaced. I am either early or wrong on my analysis for AMD. I am expecting to open up in the $3.65 range tomorrow and we could retest lower still...not sure what I am going to do (hold or sell) as my cost basis is around $3.50. Anyways, here is the good/bad highlights that I heard:

BAD

1. AMD's exposure to the PC slowdown was dramatically worse than Intel. Their exposure to low end notebook is too big and they got CRUSHED by tablets....CS down 9% sequentially...loss of $7 million
2. They missed buying the full $1.15 billion due to GF for the 2013 WSA as they only bought $960 million. 2014 WSA isn't done yet.
3. Guidance for Q1 revenue down 16% (+/- 3%). CS down "in line with seasonality" and GVS down "coming off a strong Q4 for our semi-custom SOCs". I guess this means significantly fewer consoles. No discussion of Litecoin or other mining.
4. They are lowering optimal OpEx target from $450 million to $420 million and lowering "optimal cash" to $1 billion from $1.1 billion. This implies either cash burn coming OR window dressing for lower Q1 cash total.
5. OpEx rose from $450 million in Q3 to $462 million...miss

GOOD

1. The guidance was conservative. Rory said "we see some stabilization in the PC space but have not modeled that into our guidance as we are still calling down 10% for 2014. We will take advantage of any incremental revenue opportunities that exist in this regard".
2. The 2013 WSA is FULLY completed and there will be no penalties...or preconditions going forward concerning the 2013 WSA...DONE. They paid the remaining $200 million fine in Q1 2014 for the "take or pay" penalty from the 2012 WSA.
3. Kaveri being well accepted and moving up the stack...I'll believe it when I see it.
4. Server strategy moving forward to ARM-64 and look forward to pushing advantage as the only x86/ARM/ARM-64 vendor available.
5. Margins on consoles sounded like they were north of 20%...or at least close.
6. GVS was up 29% sequentially due to consoles and R7/R9. $121 million profit
7. Inventory down 4% to $884 million
8. Guided 2014 up in revenue...profitable...inventory flat to down...OpEx down to $420-$450 million per quarter.
9. Q1 "Break even or better"

My takeaway was VERY mixed. The PC market is still a HUGE drag on AMD and if Kaveri, Beema, and Mullins cannot gain significant OEM traction I see no reason to own the stock...PERIOD. The ARM-64 Seattle opportunities are largely a year or more off. The semi-custom business is nice but can't carry the water by itself. The OEM APU business MUST increase...and must increase NOW...it is as simple as that. They really didn't miss on anything huge but the guidance was VERY SOFT.
 
After searching on the web and finding nothing valuable, does anyone know anything about GDDR6? there are plenty of articles from 2012 that say "GDDR6 coming in 2014" but nothing recent that actually discusses the improvements the new memory will provide.

The lack of information currently available is disappointing, though that seems to be the trend with AMD.
 
Last edited by a moderator:
UniversalTruth said:
You mean they are supposedly using other methods for providing higher memory bandwidth...

There will be significant changes with future memory configurations and layout, no?
Click to expand...

AMD has been pretty clear about wanting to use High-Bandwidth Memory (cf. presentations with SK Hynix) while NVIDIA has announced that Volta would use stacked memory.

Both IHVs may use standard DDR4 in lower-end offerings, but I think it's pretty clear that some kind of stacked memory is what they both want, which means there's little point in developing GDDR6.
 
Alexko said:
AMD has been pretty clear about wanting to use High-Bandwidth Memory (cf. presentations with SK Hynix) while NVIDIA has announced that Volta would use stacked memory.

Both IHVs may use standard DDR4 in lower-end offerings, but I think it's pretty clear that some kind of stacked memory is what they both want, which means there's little point in developing GDDR6.
Click to expand...
Why do you assume GDDR6 will not be stackable? Do you have information to support that claim?
 
Last edited by a moderator:
Wynix said:
Why do you assume GDDR6 will not be stackable? Do you have information to support that claim?
Click to expand...

I don't assume that GDDR6, if it existed, couldn't be stacked.

But GDDR has historically been about reaching high clock speeds. Stacked memory, as far as I can tell from what's been announced at this point, is about reaching high bandwidths with very wide buses at low clock speeds (to optimize power-efficiency). So I don't see how GDDR could fit into that picture.
 
]'GDDR' is merely a label used for high bandwidth memory, you keep making assumptions about it when we have little/no information about it.
The history of it improving via frequency increases does not matter as that was before stacked memory became a viable option.
 
Last edited by a moderator:
GDDR is a label for memory with a typical use as graphics memory.
It's an open question whether that niche is big enough to justify another special standard, and the demand for high bandwidth with low power kind of aligns GPUs with other systems.

Whether HBM will find widespread acceptance is an uncertainty that runs counter to the other trend, but I believe the intent is that it won't be a graphics-only standard.
The near and medium term show an interesting split in the desired memory standards, with HMC and things like on-package eDRAM getting in the way. HMC as described is definitely more feature-rich and it seems like the list of players that like it is longer than HBM.

AMD's history of memory partnerships with marginal partners is another, see Qimonda (GDDR5) and Elpida (GDDR5M secondary source) for various attempts.
It may not be a causative factor, but there's a correlation where AMD's initiatives are not born into a situation marked by health.
 
After going back and looking at those AMD/HYNIX slides again it does seem obvious that HBM will be the replacement for GDDR5.
I just wish AMD were clearer about information.

Sorry for wasting all our time.
 
overclocked_enthusiasm said:
My takeaway was VERY mixed. The PC market is still a HUGE drag on AMD and if Kaveri, Beema, and Mullins cannot gain significant OEM traction I see no reason to own the stock...PERIOD. The ARM-64 Seattle opportunities are largely a year or more off. The semi-custom business is nice but can't carry the water by itself. The OEM APU business MUST increase...and must increase NOW...it is as simple as that. They really didn't miss on anything huge but the guidance was VERY SOFT.
Click to expand...
Your analysis are great. What is bothering with AMD is that they are doing not' for their products.
They would not have to look far away to find inspiration, Nvidia is a good example on the matter (trying to make a niche for its product with the shield, tegra note, etc. whether it is a success or not it is better than doing not').
Now in fact it is worse than not making any effrot to showcase their products actually baring their gpus none or their product really aim for a particular usage, form factor or niche.
It is not a technological issue but it is their business plan. First they have to stop to kid them-selves with their TDP figures, they do it right in the GPU realm they have to do it right in the APU/CPU one.
It may trump reviewers, make benchmarks better but it is not trumping OEM.
THey also have to stop creating chips for which there is no market, whenever I search products embarking Kabini for example I find mostly cheap, salvaged parts.
The same applies to their higher performance APUs, most reviews agrees that Kaveri most interesting sku are salvaged one to (actually once available I think the most interesting ones are going to be the ones salvaged to the point where the whole gpu is disabled...).
It is a waste of money.

I can't see how their APU line at large is going to do better if the chips they designed aim at not existing markets (be it because of performances, TDP or price).
 
Last edited by a moderator:
Does AMD actually see anything of that above their normal distributing price? Newegg automatically adjust prices based on demand but don't they keep all the profits? Are the brand IHV's and distributors the only ones that see increased profits from high demand?
 
Malo said:
Does AMD actually see anything of that above their normal distributing price? Newegg automatically adjust prices based on demand but don't they keep all the profits? Are the brand IHV's and distributors the only ones that see increased profits from high demand?
Click to expand...

Hmm, if they don't see anything above the suggested prices, then it is quite a weird problem for them and they should find a way to control it somehow...

Have no clue what is happening but the article itself suggests that AMD will be in a not so pleasant situation once the bubble explodes

Litecoins, like Bitcoins, are produced by 'mining', which involves computers using processing power to solve difficult equations. Once a person, or a group of people, have solved the equation, they successfully 'mined' Litecoins.

http://www.cbronline.com/news/what-is-litecoin-4178333

What equations? What exactly are they using so much power for?
 
Last edited by a moderator:
I know for Bitcoin that they're hunting for inputs to SHA256 whose results begin with a certain number of 0's by brute force. In the grand scheme of things, it's an utter waste of the watts pumped in and those cards (unless someone found a good use for such solutions of SHA256). It would be nice if the were a crypto-currency whose work-units were actually useful to people, like protein folding or some type of costly but verifiable simulation for research.
 
I will try to understand it.
If the US government, for instance, uses the solutions (the SHA-2 functions), why don't they use supercomputers for that purpose but very badly influence the normal consumer market?

And if these unreal prices are so high only in the US, why don't people go to other countries and buy the cards from there for much cheaper?
 
Malo said:
Does AMD actually see anything of that above their normal distributing price? Newegg automatically adjust prices based on demand but don't they keep all the profits? Are the brand IHV's and distributors the only ones that see increased profits from high demand?
Click to expand...

I doubt AMD gets anything extra per GPU, because selling them for more money without updating MSRPs would be rather disingenuous. Besides, it would increase prices everywhere, unless partners and retailers were willing to eat the cost—but why would they be?—and that hasn't happened. You can still find 290(X)s at MSRP or even (sometimes well) below in some countries.

Still, AMD is selling more GPUs, so they're making more money than they normally would.
 
Status
Not open for further replies.

Similar threads

F
Dedicated gaming GPUs bottom of the priority list?
2
Replies
34
Views
3K
Seanspeed
S
F
RDNA5 Speculation
Replies
2
Views
1K
Erinyes
E
T
Re-litigating Radeon market performance *spawn
2 3
Replies
40
Views
3K
Dictator
D
D
Intel, AMD, Microsoft & Others Team Up For Develop The “UALink”, Direct Competitor To NVIDIA NVLink
Replies
3
Views
868
pcchen
P
xpea
Datacenter GPU market supply and demand (2024)
2
Replies
28
Views
3K
Erinyes
E
Back
Top