http://www.semiaccurate.com/forums/showthread.php?t=759
:o
How does that to compare to say an Xeon X5570? I'd love to see Bulldozer benchmarked with 8 cores disabled, just so thread for thread and clock for clock comparison is possible.
sexy if true.
Funny how since quadcore everyone's saying "more cores doesn't matter" and AMD and Intel are just increasing the number of cores :D
Wow, very impressive.
wonder what these are gonna cost :rolleyes:
OT:
how does one get an ES chip exactly? i thinks they are planted by semi companies....i smell a conspiracy :hitself:
Is that voltage real?? :eek:
I don't believe this for a second.
Seems to be a fake for me. Too soon for an Interlagos/Orichi ES chip...
Fake or not, there are some things that dont seem right.
According to the cpu-z screen there are two processors. That means 32 cores and 32 threads for that wprime score? And what about CMT or whatever its called?
i have some doubts too...
the 2 and the 8 of 28MB L3 look a bit far away from each other... photoshop?
and bulldozer was supposed to be an 8core chip, not native 16core...
seing a 16core mcm before the 8core shows up is weird...
double l1 cache sizes?
double l3 cache sizes?
28mb for 16 cores... on magny cours 12core mcm there is 12mb and cpuz only shows 10 as 2 are reserved for the snoop filter.
12 mb for 12 cores, here its 28mb for 16cores, assuming 4mb is reserved for the snoop filter that would mean 32mb, so 2mb per core L3 cache...
hmmm the numbers certainly make sense...
no, its mcm, 2 cpus on one package... same thing as on magny cours, itll show two processors but in the main window of cpuz itll show all cores and all cache...
if its a fake, its def a good one and not an amateur... the numbers seem about right from what i know...
but this is weird... why would they even bother with magny cours 12core if they have an improved 8core and 16core mcm sampling already... strange...
And yes, Bulldozer architecture suposed to have some SMT technology implementation, that CPUz shot doesn't shows any kind of multithreading cores.
It just seems impossible to me for too many reasons to list. This screams photoshop to me.
edit - reading that thread:
JF-AMD is an AMD rep that posts over at AMDzone.com.Quote:
Look at this
http://img41.imageshack.us/img41/7678/wprime155b.jpg
now look at this
http://img89.imageshack.us/img89/9111/obero.png
The 10Mb also seems to be far apart and we know thats not a photo-shop. As you said the numbers add up very well huh :confused:
I think magny cours 12core is about ready while the bulldozer is a early sample aka maybe even a prototype "Stepping 0 :shrug:" The magny cours is coming 2010 and the bulldozer is coming in 2011. The silicon has to go through debugging, etc etc "AMD does not need another TLB bug in their brand new architecture :yepp:"
EDIT:- Just noticed there is something written after AVX, its not a comma maybe its AVX-A lol "AMD's gets a special AVX version just like SSE4a"
maybe
2 opterons 8 cores with 14MB of L3 each. But 14 seem strange too.
Then i think it's a single ship with 3 die with 2 die of 8 cores, and one die of L3 of 16MB with 2MB disable for redondancy.
Hello guys,
cpuz (even version 1.52.3) does not know any 32 nm AMD processor, so it is impossible that it displays "32 nm".
Besides, the cache associativities are missing ... and finally, the program would not display "AVX" in the instructions sets (it reads avx feature but does not display it).
:-(
it is a fake! if you look at all numbers in CPU-Z you can see that there is no space between them, only if there is a "1" because every number use 6 pixel and only the "1" use 4 pixel, so there is a space of 1 pixel to each side..
The 28MB L3 is fake, so the screenshot is fake.
http://www.abload.de/img/fake111ls2a.png
and you donīt need to be a photoshop pro... this is done in 2 min with paint..
http://www.abload.de/img/fakespco.png
@Chri$ch when did u get the i12 lol, that must really be able to run crysis...
Ya this does make the pic more questionable!!!! Hope s7e9h3n can check it up with his sources!!
after a few minutes i realised something strange in the line of the cache L2.
I agree the fake.