Rumour: AMDs 8core "Bullsharks" coming close to Gulftown.

[Arsynic]

And Fuad didn't say it, he heard it from a friend who heard from a industry source who heard it from a guy named Hank !
lol, sounds like more hit whoring, at its finest.

Hit whoring or stock bumping...one of the two.

 

[Markfw]

To me, I don't care if it takes 100 cores to match a 6 core Intel W/HT, and at what mhz, I just want a chip that performs at the same level as an Intel chip, that is cheaper.

 

[veri745]

To me, I don't care if it takes 100 cores to match a 6 core Intel W/HT, and at what mhz, I just want a chip that performs at the same level as an Intel chip, that is cheaper.

Agreed.

As a consumer, I care about performance per dollar

As a company, AMD cares about performance per mm^2

No one really cares about performance per core beyond a purely academic level.

If you want to compare the 980X to a BD, compare the die sizes; that is a much better indicator of whether BD will be competitive

 

[Rubycon]

My biggest concern is and has always been: chipset support. What good is a fantastic chip (and the FX60 was no doubt) if your motherboard options are nary and often end up with mediocre and even unusable results? I guess it depends on what you do and my applications are very latency sensitive in regards to system i/o and data manipulation.

 

[nonameo]

*sigh* okay... I'm not an expert on all this stuff or anything but I can tell there's some misunderstanding as far as cores, amd, intel, hyperthreading, multithreading, etc... goes...

For one, we have hyperthreading, which is what intel has had since the P4 days.

Suppose I'm in the kitchen and I'm doing my thing. The kitchen is the core and I'm the thread, okay? If I'm the only guy in the kitchen, I may be using the blender but not the sink, which isn't too efficient right? Well, intel processors put TWO people in the kitchen. However, we can't both use the sink at the same time, but one person can use the blender and the other can use the sink. It's not as fast as 2 kitchens, but the cost of adding an extra guy is pretty low and the efficiency is good.


AMD's kitchen looks like this.

AMD's kitchen is more like a cubicle space. So, you have one big room and 2 cubicle kitchens with their own sinks in them. There are 2 cubes in the cubicle space, so you can have 2 guys, each with their own little cubicle They don't share what's in the cubicle, it's just one guy, one cubicle. Now, the cubicle isn't exactly huge. It has just the basics in it. The big part of the kitchen is the countertop, and the countertop is pretty cool. the 2 guys can share the countertop, or one guy can use the whole countertop.

Now, AMD's cubicle space is a bit bigger than intel's kitchen, but still smaller than 2 kitchens. AMD's kitchen is more powerful though(well, in theory anyway) because each guy has its own sink. So yeah, is the cubicle the kitchen or the cubicle space? Hmm.... Hard to say.

In any case, HT and AMD's bulldozer approach are not equivalents. It's just 2 different ways to handle a similar problem: efficiency. So yes.... let's just take a chill pill and wait for benchmarks. In the end, what matters is getting the performance you want for less money.

 

[Red Storm]

To me, I don't care if it takes 100 cores to match a 6 core Intel W/HT, and at what mhz, I just want a chip that performs at the same level as an Intel chip, that is cheaper.

How exactly do you expect AMD to provide the same level of performance and price it cheaper? It cannot be overstated just how big of an advantage Intel has over AMD when it comes to income, R&D, manufacturing, etc. The best we can hope for is similar performance, and if AMD manages that, then you can bet it will be priced to match as well.

 

[Edrick]

The best we can hope for is similar performance, and if AMD manages that, then you can bet it will be priced to match as well.

+1

If AMD can produce a CPU that matches a 980X, then we can expect it to be priced similar to the 980X (perhaps a little cheaper). They already do this price dance with Nvidia on the Graphics side of the house (because they can). The people looking for a $200 980X killer are going to be very disappointed.

 

[Castiel]

+1

If AMD can produce a CPU that matches a 980X, then we can expect it to be priced similar to the 980X (perhaps a little cheaper). They already do this price dance with Nvidia on the Graphics side of the house (because they can). The people looking for a $200 980X killer are going to be very disappointed.

Exactly. If an 8 Core BD can beat Gulftown expect it around 700 dollars

 

[cbn]

I dont have any problem with that since the 8core 16MB Cache "Bullshark" from AMD comes with a die size a little larger than a quad core Sandy Bridge.

Its also a new uarch on a new process.

 

[aigomorla]

lulz.. i have to agree with the majority of the people on this thread.

IF AMD had a gulftown simular spec'd processor, whats to stop them from selling it near gulftown?

You guys are hopping for too much for a gulftown like processor priced @ 965BE levels.

 

[Makaveli]

Alot of truth in here.

I for one want to see AMD become a choice again at the highend just to give us more options.

However when it comes time for my next build which looks to be maybe 2012 Q2 if they have nothing to offer I will go intel again in a heartbeat.

If BD does compete with GulfTown you can expect AMD to price it accordingly. If for some miracle it is able to match 1155 SB prices will creep up. I do believe that SB-E will be too much for BD tho.

 

[JFAMD]

Cores are cores.

We are confident in our cores. Each core has its own dedicated integer pipelines. The hardware sees them as cores. The OS sees them as cores. The applications see them as cores.

Every core today has dedicated and shared resources. All we have done is determine how to share components that allow for lower die space and lower power consumption. But we leave all components discrete that could potentially bottleneck performance.

 

[BD231]

If HT can offer performance improvement on 4 imaginary cores think about what 4 added physical cores would be capable of guys, its really not that complicated.

1 FPU per two cores. One has to wonder if each FPU might get bogged down under high load but I'm sure AMD has considered this and would not have proceeded if it wasn't worth their time.

I get the feeling AMD's first run at Bulldozer won't be its best, I'll probably be waiting for the 2nd revision.

 

[Dark Shroud]

I get the feeling AMD's first run at Bulldozer won't be its best, I'll probably be waiting for the 2nd revision.

Which is why AMD's first gen APUs are not using Bulldozer technology. If Bulldozer at least shows a good improvement over AMD's current CPUs I'll be happy to use 2nd gen Bulldozer based APU's in PCs for my various family members & friends who ask me to build PCs for them.

I'll even get some AMD discreate video cards to make use of Fussion for my newphew & baby brother since they do moderate gaming.

 

[BD231]

Which is why AMD's first gen APUs are not using Bulldozer technology. If Bulldozer at least shows a good improvement over AMD's current CPUs I'll be happy to use 2nd gen Bulldozer based APU's in PCs for my various family members & friends who ask me to build PCs for them.

I'll even get some AMD discreate video cards to make use of Fussion for my newphew & baby brother since they do moderate gaming.

Better yet, build something for your unborn children!!

 

[RussianSensation]

LOL 8 cores to "approach" six core in performance does not sound like IPC match unless AMD is stuck in 2GHz land. A rumor is just that though. Does not sound like they have anything revolutionary if that's the case. Gulftown is over a year old now (comparing ES to ES - if you can get BD ES chips yet?!)...

If they release a $300 8 core chip that can come close to Gulftown without overclocking, that would be amazing imo. Right now the X6 1100T is only about as fast as a Core i7 870/950: http://www.computerbase.de/artikel/...-sandy-bridge/46/#abschnitt_performancerating

As has been said in the thread by other posters though, if their chip can come that close to Gulftown, then it will most likely beat the 2600k as well. In that case, the FX designation and much higher pricing will follow. We'll see. Either way more competition is good. Perhaps Intel will hurry up already and introduce a lower end $350 6-core processor on LGA2011 by the end of the year. 4-core processors should really be only $150-200 by now. Ah Intel and their 60% profit margins

 

[Rifter]

Ill start by saying ill believe the performance when anandtech benchs it.

That said im sure golftown is the next logical step for AMD to compete with. Right now they are competing with the Core2quads and low end i5/i7. So obviously the next step is to compete with gulftown, I Was hoping they would compete with SB but oh well.

 

[Powermoloch]

Just be glad AMD is still competing, otherwise we'll be spending high amount of dollar for a mediocre CPU

"crosses fingers"

 

[AC2]

If HT can offer performance improvement on 4 imaginary cores think about what 4 added physical cores would be capable of guys, its really not that complicated.

1 FPU per two cores. One has to wonder if each FPU might get bogged down under high load but I'm sure AMD has considered this and would not have proceeded if it wasn't worth their time.

I get the feeling AMD's first run at Bulldozer won't be its best, I'll probably be waiting for the 2nd revision.

+1 on all of this...

 

[Riek]

Ill start by saying ill believe the performance when anandtech benchs it.

That said im sure golftown is the next logical step for AMD to compete with. Right now they are competing with the Core2quads and low end i5/i7. So obviously the next step is to compete with gulftown, I Was hoping they would compete with SB but oh well.

competing with gulftown is competing with sandy bridge since gulftown is still faster than SB (50% more cores you know..)

 

[Dadofamunky]

well we have until end of Q2 to see how well BD performs. I'm hoping it competes nicely.

Boy, I sure hope so. We really need a strong AMD in the marketplace.

 

[TomSeek]

For those who said that AMD would price it around $700 if the performance is near that of Gulftown, think about this: Why would I buy a $700 AMD chip if I can get similar performance with a 2600K SB chip which probably would OC better?

Unless I am missing something, I would think that AMD has to price that BD chip in the $300's range.

Tom

 

[Ben90]

*snip* I have terrible reading comprehension

 

[JFAMD]

1 FPU per two cores. One has to wonder if each FPU might get bogged down under high load but I'm sure AMD has considered this and would not have proceeded if it wasn't worth their time.

Each CPU has its own dedicated FPU. They are only shared if you want to run AVX code. Intel shares registers for AVX as well, they just choose to use the SSE registers so that you can't do SSE and AVX together, you actually have to recompile your code to convert all SSE instructions to AVX-128 to effectively use it.

And floating point is ~10% of the processing load in most cases, integer is 90%, which is why we have a high number of integer cores.

Which is why AMD's first gen APUs are not using Bulldozer technology.

The first APUs were targeted at low power and small form factor, which is why they use the Bobcat core.

 

[PlasmaBomb]

20lb eh? Talk about bulk, alternator-sized L3

Beat me to it...

:thumbsup: