If they needed a scapegoat, they could just point at Keller. I'm with IDC, this is damage control- "we're still fine without Keller, steady as she goes".
Yep. Its quite clear someone else was put in front to take the blame. Their own confidence in Zen is very low to put it mildly.
delays and cancellations followed by lackbuster products.
What's the point of launching a product in 2016 if any meaningful revenues - if any revenue at all - doesn't come in 2016?
Translation:
I am a naive engineer. AMD was milked till the last drop by the management. We are now given command by the higher establishment to sit on the deck of the sinking ship.
Not sure I would recognize him--- is Keller in that picture? I thought Zen was supposed to be *his* baby.
Back left of the picture, pretty sure that's him walking out the door.
They could still sell a sizeable amount of the CPUs to OEMs in 2016 and make money in 2016, products would still start to appear in 2017,I am pretty sure he was talking about the final consumer seeing something in a store,not about when it's gonna be available for the industry.What's the point of launching a product in 2016 if any meaningful revenues - if any revenue at all - doesn't come in 2016?
Same reason Intel launched a tiny volume of Core M in 2014, when real volume wouldn't arrive until 2015. Marketing, mindshare, and convincing investors that they are "on track".
Except that it didn't, investors kept asking about the 14nm schedule and it wasn't a non-issue for Intel only because 22nm was healthy enough to churn out competitive products in the first place. In fact, the ashamed tone every time 14nm was mentioned in thel Q&As said everything.
That said, that's not the point I wanted to bring. If Zen launches in Q117 or Q416 that's largely irrelevant, what's relevant is whether they will reach their performance targets and whether the subpar foundry will be able to produce it with healthy yields rates. Those two variables are far more critical than any quarterly variation on the launch date.
Very true.
Also relevant is what it will be competing against at the time it is available for purchase in stores by end-users. Will that be Kaby-lake timing?
They could still sell a sizeable amount of the CPUs to OEMs in 2016 and make money in 2016, products would still start to appear in 2017,I am pretty sure he was talking about the final consumer seeing something in a store,not about when it's gonna be available for the industry.
Due to heterogeneous computing or because gamers want to do hybrid GPU rendering pairing it with Pascal?No OEM is going to touch Zen without an IGP.
Due to heterogeneous computing or because gamers want to do hybrid GPU rendering pairing it with Pascal?
Kabylake is an APU with up to 4 cores 8 threads,
First ZEN SKUs (with up to 16 threads ??) will be competing against Socket 2011-V3 and Broadwell-E .
Because even if they launch on time in late 2016 they will start showing up in products starting in 2017.
They could still sell a sizeable amount of the CPUs to OEMs in 2016 and make money in 2016, products would still start to appear in 2017,I am pretty sure he was talking about the final consumer seeing something in a store,not about when it's gonna be available for the industry.
No OEM is going to touch Zen without an IGP.
First Zen cpu is not for mobile/laptop.No OEM is going to touch Zen without an IGP.
Also, 8C/16T Zen CPU will only have 16 PCIe Gen. 3 lanes on the CPU
OK, but how do you think the mix of user demand will look like? AMD surely can't deliver something which fits the requirements above. But even for office users not the power efficiency during load is important but during idle. That's where they are today.More power and cost savings than anything else. Plus miniaturization. See what Intel is doing with 5x5, that's going to be the on the bigger end for OEM desktops by the time of Zen's release.
Maybe for, like, a quarter. 8C/16T Zen CPU will have to duke it out with Skylake-E.
Also, 8C/16T Zen CPU will only have 16 PCIe Gen. 3 lanes on the CPU while the Skylake-E parts will have up to 48 PCIe Gen. 3 lanes.
Are you sure that it is competing against Socket 2011-V3?Kabylake is an APU with up to 4 cores 8 threads,
First ZEN SKUs (with up to 16 threads ??) will be competing against Socket 2011-V3 and Broadwell-E .
Are you sure that it is competing against Socket 2011-V3?
I was under the impression that it will feature just dual channel and borrow/sit on the same socket as APU's. It looks more like a competitor to intel's midrange 115X which is not a necessarily a bad thing.
Where did you read that?
From my perspective, "dual channel" says everything about how high they are aiming the total platform cost wise. 2011 is expensive, quad channel high TDP mainboards are expensive.
But what possible gain could AMD get from competing against a niche enthusiast platform product?
If ZEN turn's out highly efficient and fast, Intel will just unleash the kraken: drop more cores on a die, sell them for almost 0 profit and obliterate AMD without braking a sweat.
The other way around, 115x teritory, probably where most pc sales come from anyways,
AMD should challenge this market as I do not think that it stands a chance competing against the enthusiast platform no mater how good ZEN may turn out. Right there is: "drop more cores on a die, sell them CPUs for almost 0 profit and obliterate AMD".
Integrate as much as you can on chips, bring total cost of the mainboards down by deploying an efficient uarch that could challenge a cpu like the 6700, ...
I guess he saw it here:
The rest may only have x8 btw.
This is an old FM3 slide, and it most probable referring to the 14nm APUs and not the High-End Server chips.