Originally posted by: waffleironhead
GPU? Graphic Processing unit?
CPU? Central Processing unit?
Current sub forum: CPUs and Overclocking.
One could assume that was the processor he was talking about. I mean, to go out on a limb.
Current sub forum: CPUs and Overclocking.
One could assume that was the processor he was talking about. I mean, to go out on a limb.
Those bastards.
Blind-folded, both hands tied behind their backs, working with pennies on the dollar budgets, watching every other one of their colleagues being laid off around them...and yet despite such rich sweet rewards that such a work environment offered them they still went about there 3-day work-weeks taking coffee breaks at 9 and 3 and lunch from 10 to 2 and didn't get a dam thing done for the company's loyal, benevolent and over-adoring shareholders.
Yep...I see nothing in the money-chain to suggest it was anything but the engineers who wasted resources, mis-prioritized next gen architecture programs, mis-predicted what the competition was bringing to the table, etc.
All I see is a wasteland of fat incompetent boobs unfit of the title engineer.
Hector ? AMD had and still has great engineers. I'm pretty sure any blame in this case should be laid right where it belongs: at the friggin top.
I can't see AMD regaining the top spot in performance for a while (don't wanna say never, because honestly who knows what happens 5 years from now). HOWEVER I can see AMD being a solid choice for an average build if deneb pans out. Right now they are a very low end, or htpc only, answer really. And I used AMD chips exclusively since like 1999. But I'm no fanboy, I bought them because since the original Athlon they were just plain better chips. Now I'm running a core2duo. They are simply the better chips at the moment.
I can't see AMD regaining the top spot in performance for a while (don't wanna say never, because honestly who knows what happens 5 years from now). HOWEVER I can see AMD being a solid choice for an average build if deneb pans out. Right now they are a very low end, or htpc only, answer really. And I used AMD chips exclusively since like 1999. But I'm no fanboy, I bought them because since the original Athlon they were just plain better chips. Now I'm running a core2duo. They are simply the better chips at the moment.
frostedflakes
Diamond Member
- Mar 1, 2005
- 7,925
- 1
- 0
As others have pointed out, depends on what you want to do with your computer. Intel probably has better priced CPUs at the low-end now, but back when I put my system together in Jan 2007, the cheapest C2D cost twice that of the Athlon64 X2 3600+. I didn't need the extra performance the C2D offered, I just wanted a cheap dual-core for multitasking, so I ended up going with the 3600+.
Their high-end is pretty meh, though. Especially once you start overclocking the Core chips, it's just no contest.
Also, how is Ruiz to blame? Or the engineers? Other than the TLB erratum, I wouldn't say AMD has really screwed up. They have some pretty good products, but Intel has just been at the top of their game lately, Core is a fantastic architecture. It's tough to compete against that.
Their high-end is pretty meh, though. Especially once you start overclocking the Core chips, it's just no contest.
Also, how is Ruiz to blame? Or the engineers? Other than the TLB erratum, I wouldn't say AMD has really screwed up. They have some pretty good products, but Intel has just been at the top of their game lately, Core is a fantastic architecture. It's tough to compete against that.
taltamir
Lifer
- Mar 21, 2004
- 13,576
- 6
- 76
Originally posted by: frostedflakes
As others have pointed out, depends on what you want to do with your computer. Intel probably has better priced CPUs at the low-end now, but back when I put my system together in Jan 2007, the cheapest C2D cost twice that of the Athlon64 X2 3600+. I didn't need the extra performance the C2D offered, I just wanted a cheap dual-core for multitasking, so I ended up going with the 3600+.
Their high-end is pretty meh, though. Especially once you start overclocking the Core chips, it's just no contest.
Also, how is Ruiz to blame? Or the engineers? Other than the TLB erratum, I wouldn't say AMD has really screwed up. They have some pretty good products, but Intel has just been at the top of their game lately, Core is a fantastic architecture. It's tough to compete against that.
there have been erratum in the past, trying to hide this one as long as they can bit them in the ass though...
But no, Hecktor Ruin is to blame here. After nearly bringing motorolla to the ground he went with AMD, he torpedoed a merger with nvidia over squabbling with their CEO, he gave himself fat bonuses while cutting everyone's elses off. He fought with the major talent causing them to quit because of him... AMD has been hemmoraging their most talented staff for a while now. I mean, he give himself the biggest raise yet during AMDs biggest recorded loss (but of course, many were laid off due to that loss)...
With such lovely management morale couldn't get any lower.
BTW, when nvidia was on hard times, the management took a pay cut and not a single employee was laid off
If you don't overclock, AMD isn't too bad a choice at the low-to-mid range price points. Once you overclock, then Intel has a big advantage.
Lol true. I wish he had been sacked earlier. I hate to see AMD in trouble...we need them around. No doubt that Core2 was amazing...I got an E6400 when it launched back in '06 and kept it until now but AMD could have done much better than they did since Core2 launched. Is this the ultimate case of complacency while on top?? Or very poor management?? Probably both I suppose.
Well it probably feels like ages since Core2 was launched but I guarantee to you that internally at AMD (and Intel) 2 yrs is a mad rush to get anything changed. Process technology and architecture development is like a CPU with a massively huge/long multi-stage pipeline.
And even though the sucker is operating at 4GHz and new stuff is coming out every 1-2 yrs, when R&D gets a branch mispredict there is an amazing (to us outsiders) long pipeline flush and refill process that takes place and it can appear to us that time is standing still at the company and nothing is being done about it.
From a cost standpoint Intel can afford to have a more complex multi-threaded/multi-core R&D team (i.e. many projects being worked on in parallel with lots of timeline overlap) so an R&D branch mis-predict can be more readily masked by productivity of another already-in-progress processing thread. (C2D wasn't started when Prescott came to market, it was started well before Prescott even taped out)
But with the same analogy AMD doesn't have the resources to operate a multi-core multi-threaded R&D operation, not unless they make it a multi-cored 486 to compete with Intel's multi-core C2D R&D teams. So they must operate more in the single-core mode...put nearly all your eggs in one basket and hope you predicted correctly where the market (competition) is going to be 4 yrs from now. Get it wrong and its gonna take you nearly another 4yrs to attempt to correct it.
There are many reasons why such an analogy should never be made between R&D organization and CPU's, but from what I've seen working on the inside of this industry this is about the closest working analogy I can conceptualize when talking about.
I don't think there was any complacency or mismanagement at the top of AMD. They are competing with a company that has 400% more budget then they do. In the early 1980's no one expected Toyota to produce an automobile that would rival Porsche or Jaguar, they simply didn't have the resources to go there even if it was their business plan to do so.
If you find a processor, any processor, to be crushing bones then you are decidely using it incorrectly.
Dunnington seems like it would be destined to put out some stellar performance/watt numbers to be sure. What I haven't figured out is just how much it matters if a boutique processor scores well in any regard.
Surely Dunnington is not intended to fuel new system sales given Nehalem is right around the corner. Its for upgrading existing systems, yes? Not exactly stealing socket sales away from AMD.
I'm more interested in how gainstown performs on those server apps that AMD has soundly owned to date. Gainstown sales will definitely come at AMD's loss so they'd be more relevant (IMO) to focus on gainstown performance.
Phynaz
Lifer
- Mar 13, 2006
- 10,140
- 819
- 126
It's a drop in replacement, therefore enterprise customers don't have qualify a new platform. I can buy the exact same DL580 today as I bought yesterday, except I'll have 12 cores instead of eight. In the world of virtualization, this is hardly boutique. The DL580 is about as mainstream as you can get.
I don't have to qualify a new platform, as I will have to with Nehalem.
I don't have to qualify a new platform, as I will have to with Nehalem.
Even though it has 6 cores, Dunnington is limited to 2 sockets...
The performance IBM is talking about is for a single socket Dunnington if I'm not mistaken.
Perhaps I am incorrectly using the term boutique in this context.
By boutique I was meaning niche market applications, by which I mean to speak to total $-volume of the segment.
Owning 100% of a $25M/yr segment isn't exactly breathtaking. Owning 100% of a $25B/yr segment is breathtaking.
So what kind of $-volume does Dunnington represent for Intel? (and thus potential loss for AMD if its a zero-growth market segment)
I am sure dunnington means the world to the folks who don't want to re-qualify out of their DL580's, but are we talking about 6 people who feel that way or something of real gravity like 80% of the fortune500 companies?
Yep
8 socket!?
Nevermind...that's a not-so-small server segment. Dunnington is going to generate some revenue for Intel. Lots'o revenue.
According to this quote from the press release you linked to, they will be going up to 16 sockets!
See this Youtube video for presentation of benchmarks from these platforms....
http://www.youtube.com/watch?v=bV933rpD2Wc
Thanks for the correction...
However, I am very dubious about how well a >2 socket FSB-based system will do vs the Opterons. Intel has had many systems that were along these lines in the past (though they peaked at 4 cores per socket), and they weren't nearly as effective at 8 cores or higher.
Edit: This is exactly the type of system that Nehalem was designed for...
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 14K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 9K
-
Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)- Started by Vattila
- Replies: 13K
-
Facebook
Twitter