AMD Bulldozer Finally!

[drunkenmaster]

Indeed!
And the server parts don't have to have as high clock speeds so it was easer to produce them.

This is true in a sense, but using a fact to prove a false point. Server parts aren't easier to produce due to lower clock speeds, they have lower clockspeeds largely to hit the TDP's they need, IE stick 2 octo cores together with 4.2Ghz turbo and you'll blow any tdp budget you want, also they are targeting cloud servers and all kinds of servers which generally prefer threads over clockspeed.

They are the same chips, they need to hit a certain TDP points to enable them to produce a lower tdp at the lower clock speed.

Theres also the little fact that, if you can produce however much you want, and pretty much match demand easily, then you aren't loosing money, when you've got limited supply, server chips sell for a crapload more, and OEM server builders, in terms of revenue, are the guys to get your product to first and with as many chips as they can use.

IE one chip for £2000, or 20 chips for £100, when you have limited chips, £2000 a chip is the very very easy choice. Cray in particular are already advertising and planning to upgrade lots of computers with Bulldozers, with a few supercomputers basically just waiting on the chips.

Long term server market share gain will revenue wise, be worth billions more than desktop market share gain so again thats the people to sell too first.

One of the best things AMD have coming, is Glofo having other fabs up and running, if right now today they had Malta up and running, they could have quite easily adding some further capacity at Malta, which would help AMD out on capacity no end. The reason I don't know about 32nm at Malta is, if they were up and running a year ago, it would be a no brainer, being done halfway through next year, 32nm isn't necessarily the right move, and getting to 28/22nm soi is the smarter move.

 

[pmc25]

Distilling what Drunkenmaster said into a few words:

Interlagos is low volume, high margin. So AMD can launch with a few thousand / tens-of-thousand chips in hand.

Zambezi is vastly higher volume and much lower margin, therefore they have to produce many more of them before they launch.

Realistically, final retail version Zambezi chips have probably been sitting in secure AMD warehouses for circa 5-6 weeks now, if not more.

 

[Cooper]

Oh dear......

http://webcache.googleusercontent.c...zer&cd=2&hl=de&ct=clnk&gl=de&client=firefox-a

Sandybridge it is then, if this is accurate...BD is looking a bit of a turd...30% slower clock for clock.

 

[pmc25]

Oh dear......

http://webcache.googleusercontent.c...zer&cd=2&hl=de&ct=clnk&gl=de&client=firefox-a

Sandybridge it is then, if this is accurate...

Another engineering sample. Who cares?

Even if this was representative of retail silicon (which it won't be), the test is total ********. They're running 1333Mhz on a platform that has 1866Mhz as standard, and by all accounts needs the extra memory bandwidth.

So, this is a test of a crippled chip which they've decided to cripple further with slow RAM.

 

[Subliminal Aura]

Another engineering sample. Who cares?

Even if this was representative of retail silicon (which it won't be), the test is total ********. They're running 1333Mhz on a platform that has 1866Mhz as standard, and by all accounts needs the extra memory bandwidth.

So, this is a test of a crippled chip which they've decided to cripple further with slow RAM.

Huh ? What difference does it make ?

 

[FrenchTart]

Oh dear......

http://webcache.googleusercontent.c...zer&cd=2&hl=de&ct=clnk&gl=de&client=firefox-a

Sandybridge it is then, if this is accurate...BD is looking a bit of a turd...30% slower clock for clock.

Assuming that is completely accurate then... who cares? If the BD chips are much cheaper than an equivalent SB then they're worth it. People seem to forget that AMD chips for a long time have been about bang for buck, not having the absolute best.

Personally I've gone SB and I'm quite happy with it. If I hadn't done I would've been very tempted by BD as I'm sure it would've been quite a cheap upgrade.

 

[tommybhoy]

November 15 Launch Date for Sandy Bridge-E
'Even as AMD struggles to get its FX series processors out, and into to the market, Intel seems to be doing the opposite: trying to get Sandy Bridge-E (enthusiast) out as quickly as possible. The new high-end desktop/workstation platform was slated for December, and then there was news that chipset delays could push it into 2012. Fresh news pin-points the launch date to Week 46 of 2011, more specifically, November 15.'

http://www.techpowerup.com/151759/November-15-Launch-Date-for-Sandy-Bridge-E.html

Took a gamble last month and bought a 990fx Sabertooth, should have stuck with my Q6600 a bit longer.

It may be out this month, next month, but at the same time it may still not be out also.

I have considered getting a hex core to keep me going, but it may be time to cut my loses!

Looks like the Sabertooth may well be going in the members market I think, as there is still no official word yet on BD.

Before anyone tells me it's not delayed as there has been no official word, don't bother, we have been reading for how long now on BD, and quite literaly getting fed up about the whole sorry story.

Been using top end AMD gpu's for the last 3 years as they have been best bang for buck, but I'm starting to get a sour taste in my mouth with AMD now.

JF AMD, if you read this, then between this forum and the others you frequent, you must know that with the release of Sandy Bridge-E imminent (with no firm release date for BD), AMD is going to start hemorrhaging sales to the opposition due to lack of information alone.

 

[Martini1991]

I'm calling crap on that benchmark set.

The 8 cored BD should beat the 2600k in applications that 8 cores are utilised.
It'd be devastating if it's true.

And, I wish people would stop saying "AMD are price/performance"

I'm sorry, we need performance from AMD, as without it Intel are free to charge what they like for the performance products.

 

[FrenchTart]

I'm calling crap on that benchmark set.

The 8 cored BD should beat the 2600k in applications that 8 cores are utilised.
It'd be devastating if it's true.

And, I wish people would stop saying "AMD are price/performance"

I'm sorry, we need performance from AMD, as without it Intel are free to charge what they like for the performance products.

No we don't - we've been fine for years with Intel having the performance lead. AMD are competitive in the mainstream segment.

It'd be *nice* if AMD could take the performance crown again but it really doesn't matter. Look at graphics cards.

 

[easyrider]

Assuming that is completely accurate then... who cares? If the BD chips are much cheaper than an equivalent SB then they're worth it. People seem to forget that AMD chips for a long time have been about bang for buck, not having the absolute best.

Personally I've gone SB and I'm quite happy with it. If I hadn't done I would've been very tempted by BD as I'm sure it would've been quite a cheap upgrade.

Any ideas on pricing?

 

[FrenchTart]

Any ideas on pricing?

No idea. Hence use of the word "if"

The second part of my post was just based on previous AMD pricing. Mid range mobos tend to be cheaper for an okay spec, same with processors. Obviously this isn't true across the board.

 

[easyrider]

And, I wish people would stop saying "AMD are price/performance"

I'm sorry, we need performance from AMD, as without it Intel are free to charge what they like for the performance products.

Me too.

I'm sick of the whole "well you can keep your old AM3 mobo etc..

What AMD need to do is get a new socket and have a part that rivals intel.

I reckon BD will be a massive let down tbh.

You will be able to get a BD for less money than SB sure, but if its slower than who cares tbh

 

[Martini1991]

No we don't - we've been fine for years with Intel having the performance lead. AMD are competitive in the mainstream segment.

It'd be *nice* if AMD could take the performance crown again but it really doesn't matter. Look at graphics cards.

What has that to do with anything?
There's not a gulf between AMD and Nvidia like there is AMD and Intel.

While we're fine now, what happens in a few years when we need the performance, and AMD can't offer it?

 

[FrenchTart]

What has that to do with anything?
There's not a gulf between AMD and Nvidia like there is AMD and Intel.

While we're fine now, what happens in a few years when we need the performance, and AMD can't offer it?

It shows that you don't need to have the fastest products to be competitive and/or successful. I'd have thought that would be obvious.

I've no idea where you're imagining up the idea that AMD won't be able to offer the needed performance in a few years. They'll advance well enough to keep pace with technology. It really doesn't matter if Intel has chips that are 10 (or more) percent faster, as long as the prices reflect this.

 

[FrenchTart]

Me too.

I'm sick of the whole "well you can keep your old AM3 mobo etc..

What AMD need to do is get a new socket and have a part that rivals intel.

I reckon BD will be a massive let down tbh.

You will be able to get a BD for less money than SB sure, but if its slower than who cares tbh

The majority of AMD's customers.

Of course, from an enthusiast point of view we want something exciting and super fast. We're a tiny tiny percentage of the customer base though.

 

[Martini1991]

It shows that you don't need to have the fastest products to be competitive and/or successful. I'd have thought that would be obvious.

I've no idea where you're imagining up the idea that AMD won't be able to offer the needed performance in a few years. They'll advance well enough to keep pace with technology. It really doesn't matter if Intel has chips that are 10 (or more) percent faster, as long as the prices reflect this.

But it's not 10% difference.. Clock for clock it's more like 40%? AMD are only at Q6600 level clock for clock at the moment, although, that should change with BD. But then Intel have Ivy in a few months...

What gives me the idea that AMD can't keep up? The fact that the difference between the two keeps widening.

 

[FrenchTart]

But it's not 10% difference.. Clock for clock it's more like 40%?

What gives me the idea that AMD can't keep up? The fact that the difference between the two keeps widening.

I strongly doubt we will end up with a 40% performance difference for the equivalent priced processors. Perhaps in a few edge situations that particularly favour one architecture but you'd be pretty dumb to base your purchases on a situation that you'll use your processor for 0.000000000000000000001% of the time.

 

[Martini1991]

I strongly doubt we will end up with a 40% performance difference for the equivalent priced processors. Perhaps in a few edge situations that particularly favour one architecture you'd be pretty dumb to base your purchases on a situation that you'll use your processor for 0.000000000000000000001% of the time.

Where did I say that?
I'm saying their clock for clock difference at the moment is like 40%.

 

[FrenchTart]

Where did I say that?
I'm saying their clock for clock difference at the moment is like 40%.

Their clock for clock difference means absolutely zero (outside of enthusiast circles) if the performance is the same or cheaper for the same price, along with TDP/etc in server spaces.

I'm not saying that BD will be awesome and for all I know it could fail massively. If it does, though, it won't be for the reasons that seem to be coming up in this thread.

 

[Subliminal Aura]

But it's not 10% difference.. Clock for clock it's more like 40%?

No can't see it being as bad as 40% or else AMD can just write themselves off and go defunct.

Hopefully the gap will be 20-25% and SMP will be better than intel - If so I'll take a punt with AMD and build a machine for virtual hosts.

Can't remember the last time I bought an AMD chip (2003 / 2004 ?)