The AMD Execution Thread [2007 - 2017]
- Thread starter overclocked_enthusiasm
- Start date
-
- Tags
- amd
- Status
I never really understood the way people believe that a single man can make a huge difference at a company like AMD.
Not that I don't respect the abilities of somebody like Jim Keller. I'm sure he's an exceptional engineer. But I'm also sure that this same Jim Keller could always create something even better at a company that is not running on fumes.
Because the reality of today's chip design is that great overall architecture is necessary but far from sufficient for a competitive product.
Take power consumption: you will always need an army of grunt engineers who comb through each flipflop and interface to ensure that none of them do a toggle too much. If you're going against Intel who can throw 3 times the number hours at that same optimization problem, it's going to be very hard to compensate for that with a better architect, especially since Intel is not incompetent at architecture as well.
There was a time when a few brilliant men in a corner could make a difference, but that time has been passed years ago.
Not that I don't respect the abilities of somebody like Jim Keller. I'm sure he's an exceptional engineer. But I'm also sure that this same Jim Keller could always create something even better at a company that is not running on fumes.
Because the reality of today's chip design is that great overall architecture is necessary but far from sufficient for a competitive product.
Take power consumption: you will always need an army of grunt engineers who comb through each flipflop and interface to ensure that none of them do a toggle too much. If you're going against Intel who can throw 3 times the number hours at that same optimization problem, it's going to be very hard to compensate for that with a better architect, especially since Intel is not incompetent at architecture as well.
There was a time when a few brilliant men in a corner could make a difference, but that time has been passed years ago.
D
Deleted member 13524
Guest
He's probably not cheap and AMD doesn't have a lot of cash to begin with, so maybe this was part of his contract from day one?
Get in, stay for 3 years, help designing the new core and build a team that will take over the project, get out.
Got any source for that?
That sounds shady as hell. And even if they did state that, I don't believe that's everything there is about the subject.
They couldn't find a better way to spend $2.5B on a company that generates $130M annually (and on a dated title that's bound to fade nonetheless)?
Even if they kept the money overseas, they couldn't get a better deal off the cash alone than ~2% interest? That's bollocks.
Just don't expect Minecraft 2.0 to be out on PS4 though.
Get in, stay for 3 years, help designing the new core and build a team that will take over the project, get out.
Got any source for that?
That sounds shady as hell. And even if they did state that, I don't believe that's everything there is about the subject.
They couldn't find a better way to spend $2.5B on a company that generates $130M annually (and on a dated title that's bound to fade nonetheless)?
Even if they kept the money overseas, they couldn't get a better deal off the cash alone than ~2% interest? That's bollocks.
They had to do a $7.6B write-off over Nokia's acquisition, so it can't have been all roses like you describe.
Were you expecting Microsoft to come out and say "we're boycotting all other platforms from now on"? Most ports were already under way when the purchase happened. Of course the Vita and PS4 versions were still coming out.
Just don't expect Minecraft 2.0 to be out on PS4 though.
To understand it you only need to realize that most people have no idea what they're talking about and have no comprehension of what a major engineering project entails. They instead look at it like a sports hero thing. It's like how Carmack was the only engineer that mattered at id Software. On and on.
Last edited:
I don't think anyone here is under the illusion that Jim Keller can work wonders on his own. However, people in decision-making positions matter. When they make the right decision, you get K8 or Swift; when they don't, you get Netburst or Bulldozer.
Beyond this, Jim Keller seemed pretty enthusiastic about being at AMD, and doing bold things. Sure, some of that was probably boilerplate corporate bullshit, but it seemed at least partially sincere. His departure suggests that the circumstances that drew him back to AMD in 2012 may have changed. I suppose it might have something to do with the rumors of financial changes at AMD, but this is pretty speculative.
Silent_Buddha
Legend
It was dated the minute it was released to the public. That didn't matter. And the brand and sales continue to grow. It is consistently in the NPD top 10 for software sales. It is now the 3rd best selling software gaming title in the history of video games. And it's continuing to sell well. If sales continue to grow it won't be long until it's the number 1 best selling video game of all time. At this point it is also likely the most well known gaming IP on the planet.
There is always risk associated with buying IP. And as mentioned the price was a bit on the high side. But in terms of risk? It wasn't the riskiest (not even close) purchase a company has ever made. Buying a company or IP that is trending down is far riskier than paying a bit too much for a company or IP that is trending up.
There was also pressure from influential stockholders for Microsoft to do something with its stockpile of cash. There aren't many low risk buyouts available that could have provided them with something that potentially synergizes with where they want to go in the future. Or that give them as much mindshare as Minecraft.
That doesn't guarantee it won't eventually fail, of course. Nothing in business is guaranteed. Anyone that tells you otherwise is smoking some good stuff.
BTW - it isn't making 130 million annually. That was for the year prior to the buyout. The brand and sales have grown since then. It'll likely be significantly more than that for this FY. Although with it being part of MS now, we're unlikely to see sales/revenue numbers for just Minecraft.
Regards,
SB
Last edited:
Actually the write off might not have anything to do with profitability or lack of, the cash used as DB pointed out was over seas cash, the write off now helps MS to save cash in the US market, so its more like consolidation of money here in the US.
I'm still not sure Bulldozer itself is a problem and not the process. Bulldozer has stayed on the same process since its inception while intel has continued to pull ahead in process jumps. I would have loved to have seen an updated bulldozer on a 22nm or 16nm process.
Hopefully it turns out to be worthwhile to have spent those R&D resources on Zen and APUs. We did get to see plenty of Bulldozer warm overs though in those APUs.
Yea but didn't they only go from 32nm with bulldozer to 28nm on the newer APUs ? So its not like the jumps we've seen with intel.
I would suspect Zen will be worth while simply from going from a 32nm to a 16nm chip before even accounting for any improvments in the cores. Next year will be a huge upgrade year for me. If zen comes close to what intel is doing then I will buy it. If its far behind then I will skip.
If MS bought AMD it would be interesting to me , I use a lot of MS's products. I have a surface pro 1 that needs replacing (perhaps the 4 ) and I have the Microsoft band. I'm also interested in the rumored surface phone. I would not mind seeing AMD in all 3 of those.
If he was poached, his leaving could signal a significant change if others follow him.
AMD insists that there is a strong organization in place right now, but a notable portion of the organization did not see sufficient reason to stay, and if he's not an island he might be a bellwether for what AMD has to offer for those that worked with him.
AMD's Jaguar team seems to have been lost this way.
I'm curious if that explains in part why parts of the now consolidated graphics group were pulled from under Papermaster. Stepping in to manage the CPU side more directly might not leave as much head space to manage whatever portion he managed that was moved out.
I was puzzling over what parts that could have been, and what it would mean for graphics if the CTO--ostensibly in charge of managing the company's technological future--no longer looked over it.
The original 32nm Bulldozer didn't compare very favorably to Intel's 32nm Sandy Bridge, or even to AMD's own 45nm Phenom II X6, for that matter.
I wouldn't go that far. It did better than the phenom II x6 in the majority of the tests and beat sandy bridge on a few tests.
It looks like Keller last worked at AMD from '98 to '99: https://en.wikipedia.org/wiki/Jim_Keller_(engineer) If he had a hand in everything from the x86-64 instruction set, K8, and Hyper Transport in a span of less than two years, it'll be interesting to see what he's influenced this time from '12 to '15.
Not core for core. Sure, it had more cores than the PII X6, but only because the finer process allowed it.
D
Deleted member 2197
Guest
I don't think cores is the right way to compare considering the diffrences between the two.
Neither do I think that comparing those architectures from an absolute standpoint is quite right. Some effort should be put into factoring in the difference in process tech. On paper, 32 nm space-wise represents a 40% disadvantage for Phenom II. Energy-wise it might be in a similar order of magnitude but i am not sure about that.
- Status
