Barcelona to have 40% floating Point & 10+% Int Gains over X5355

[motiv]

Just been reading EE times and it has an AMD exec stating that there are going to be double digit gains over the 2.66 Clovertown. It's a little difficult to work out which Chip will, so you can only hope it's the 2.3ghz Barcelona and not a future 2.7 

Anyway it's an interesting read as they demonstrated the Quad Core with a R600 combo (probably a AMD/ATI motherboard too).

Separately, AMD gave one of the first public demos of the R600, its next-generation graphics controller that uses 320 multiply-accumulate units. The company showed a Barcelona-based system using two 200W R600 graphics cards to hit a terabit/second benchmark.

http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=197700269

Also, anandtech have a very interesting article about the architecture of the K10, check it out here

http://anandtech.com/cpuchipsets/showdoc.aspx?i=2939

[fudo]

thanks a good one thanks for keeing your eyes on it

should write something about it

[motiv]

Some more info

According to Mario Rivas, executive vice president of AMD's computing products group, the company's Barcelona will yield 42 percent better "floating-point" performance than Intel's Xeon X5355 Clovertown in SPEC_fp benchmark tests.

For SPEC_int, which measures integer-processing tasks, Barcelona will have a leg up on Clovertown again with a performance advantage of more than 10 percent, Rivas said.]According to Mario Rivas, executive vice president of AMD's computing products group, the company's Barcelona will yield 42 percent better "floating-point" performance than Intel's Xeon X5355 Clovertown in SPEC_fp benchmark tests.

For SPEC_int, which measures integer-processing tasks, Barcelona will have a leg up on Clovertown again with a performance advantage of more than 10 percent, Rivas said.

http://www.extremetech.com/article2/0,1697,2099613,00.asp

[fudo]

I meet Mario at CES he is a great chap

[roosterbaby!]

Just curious but, I thought the K10 nomenclator was put aside for the K8L or some such mess? Amd should have stuck with the K9 thing as this probably would have gotten some free press i.e. -AMD's K9 is no dog or, K9 the Pitbull of chips, or AMD may have been sitting with the dogs ,but now their out to chomp the competition.

[flippin_waffles]

Yeah, K9 would have been a killer name for these chips! Imagine Greyhound, Pitbull, Bulldog, Rotweiller.  That would have been awesome for marketing, provided they are as revolutionary as I think they are. But from what i've read, K9 is actually the name given to dual core K8's. Of course I could be completely wrong, but it does make sense. And K8L for mobile, which I do believe was mentioned by AMD themselves also.

[ThePate]

Sorry if this sounds likea noobish question, but I do not know! So it ask.

So, K8 are A64? K8L are low-power? K9 are Dual cores, So called X2? and now K10 is new architehture?

[roosterbaby!]

Here ya go Pate-http://en.wikipedia.org/wiki/List_of_AMD_microprocessors

[roosterbaby!]

By the way; could someone find a viable and up to date AMD proc raodmap. I've looke and can't find one. Pllleeeaase ....and I'm not talking about this one-http://www.amdcompare.com/prodoutlook/, I want this one-http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_608,00.html, but up to date.

[gOVEDO]

There is no single evidence that can confirm or debunk AMD's bold claims. But I really, really don't trust to the marketing managers. If we look at the statistics in the past 10 years in 100% cases they are overestimating the performance and the capabilities of their unseen upcoming products. For example, Intel were claiming that clock for clock, C2D will be at least 40% faster than K8 in the worst scenario, but clock for clock C2D is only 20% faster in average; AMD were claiming K8 to be twice fast as same clocked Northwood, but it is only 40% faster in average.
If we take a look at the K10 architecture, it has no theoretical chances to be faster than or as fast as C2D for integer, at same frequency.
IMO, Barcelona will be up to 40% faster clock for clock than C2D for floating point calculations, will perform roughly same as C2D for SSE, but it will be slower for integer code. It will be a beast and a nice chip, but not even close to what AMD marketing machine is BS-ing about. If it is clocked less than 2.8GHz, than it will be #2 player in the 2P and 4P server computing.
Anyway I would appreciate if AMD publishes some real-world benchmarks instead of their bold claims and unargumented estimations.

[King of Pilots]

There is only one bench i trust to say so: my pc, a good mobo, and 3dmark06. Before it will be launched you can say that it can fly, make you roast-beef, make you pay less taxes and hires top-models to do sex with you. We'll see when it will be really tested by someone who is not the one who will sell it.

[viperiii]

Transition to a 45nm is still possesses certain technical issues. according to TSMC. (Taiwan Semiconductor Manufacturer Corporation)

[fudo]

in at least one tests SPEC 2000 AMD is 47 percent faster but this doesnt means that AMD is always that much faster. at least it has a fighting chance

[Xajel]

Well.. it's very promising indeed

AFAIK, Penryn @ 3.3GHz has an 40% advatage over 2.88GHz Clevertown, while AMD says that barcelona will have up to 50% advantage over clock2clock Clevertown, thats mean that even Penryn can't compete with Barcelona, I just hope AMD realease it's CPU's in faster clock starting at 2.5 not ending with 2.5 as Intel is already preparing 3.0GHz+ Chips, I know it's hard for Intel to go for 3+ with 65nm unless pumping up the TDP alittle higher, and this will kick Intelsass very hard specially in server/workstation market

[Thund3rhans]

The new news to the K10 parts sound great (66% more fp power, w00t), i hope they´re real
I hope that will help amd to sail through the storm