Friday, September 24th 2010

AMD Orochi ''Bulldozer'' Die Holds 16 MB Cache

Documents related to the "Orochi" 8-core processor by AMD based on its next-generation Bulldozer architecture reveal its cache hierarchy that comes as a bit of a surprise. Earlier this month, at a GlobalFoundries hosted conference, AMD displayed the first die-shot of the Orochi die, which legibly showed key features including the four Bulldozer modules which hold two cores each, and large L2 caches. In coarse visual inspection, the L2 cache of each module seems to cover 35% of its area. L3 cache is located along the center of the die. The documents seen by X-bit Labs reveal that each Bulldozer module has its own 2 MB L2 cache shared between two cores, and an L3 cache shared between all four modules (8 cores) of 8 MB.

This takes the total cache count of Orochi all the way up to 16 MB. This hierarchy suggests that AMD wants to give individual cores access to a large amount of faster cache (that's a whopping 2048 KB compared to 512 KB per core on Phenom, and 256 KB per core on Core i7), which facilitates faster inter-core, intra-module communication. Inter-module communication is enhanced by the 8 MB L3 cache. Compared to the current "Istanbul" six-core K10-based die, that's a 77% increase in cache amount for a 33% core count increase, 300% increase in L2 cache per core. Orochi is built on a 32 nm GlobalFoundries process, it is sure to have a very high transistor count.Source: Xbit Labs
Add your own comment

152 Comments on AMD Orochi ''Bulldozer'' Die Holds 16 MB Cache

#1
xrealm20
Very interesting - if these cards are much more efficient per clock cycle, a decrease in clockspeed wouldn't be a bad thing. Only time will tell at how well these new processors will fare, but here's hoping to them being an overclocker friendly processor unlike Intel's SB...
Posted on Reply
#2
bear jesus
xrealm20 said:
Very interesting - if these cards are much more efficient per clock cycle, a decrease in clockspeed wouldn't be a bad thing. Only time will tell at how well these new processors will fare, but here's hoping to them being an overclocker friendly processor unlike Intel's SB...
I hope for some nice overclocking as well as it will be time for my first water cooling setup if buldozer overclocks well and if i go for sandy bridge i will just get one of the K (overclocking) versions.

I really think the next year is going to be very exciting for pc hardware, sandy bridge and bulldozer and of corse amd's 6xxx and 7xxx cards and nvidia's kepler :rockout:
Posted on Reply
#3
JF-AMD
AMD Rep (Server)
cadaveca said:
Very hot...apparantly we'll see a clockspeed decrease(which I assume is due to the high levels of cache), but IPC will increase. I'm kinda expecting 2.4ghz or so...maybe lower...for launch chips.
Actually, power and thermals are exactly the same with our current products. Not sure where you are getting the clock speed info though.
Posted on Reply
#4
cadaveca
My name is Dave
afw said:
Well I read that Buldozer will do more instruction per clock ... so it will be interesting to see what its capable of
Yeah, i mentioned that too. Bulldozer, with the info we have now, sounds pretty good. BUt in reality, it has some competition, and we don't know too much about that either. Seems like a mixing of words from either side, trying to keep people interested. I truly hope that they can deliver...I have far more faith in GF's process than TSMC's, so I don't think I need to hope too much, but the TLB bug of Phenom 1 still rings fresh in my head.
Posted on Reply
#5
bear jesus
cadaveca said:
but the TLB bug of Phenom 1 still rings fresh in my head.
I must admit that made me hold off untill the 9850be (already had an am2 motherboard thus why i did not switch to intel) even though it was a very specfic error, i would hope amd will have learnt from that mistake as if i have to wait for bulldozer if it is not a great processor from the start then they will be pushing me over to intel for the first time in years.
Posted on Reply
#6
cadaveca
My name is Dave
JF-AMD said:
Actually, power and thermals are exactly the same with our current products. Not sure where you are getting the clock speed info though.
That's what alot of sites are saying, as you can see by the link "afw" posted. Anyway, that's good news, John, that you seem to hint at.
Bulldozer: The Turbo Diesel Engine
In many respects, the Bulldozer architecture is comparable to a diesel engine. Lower RPM (clock-speeds), high torque (instructions per second). When implemented, Bulldozer-based processors could outperform competing processor architectures at much lower clock speeds, due to one critical area AMD seems to have finally addressed: instructions per clock (IPC), unlike with the 65 nm "Barcelona" or 45 nm "Shanghai" architectures that upped IPC synthetically by using other means (such as backing the cores up with a level-3 cache, upping the uncore/northbridge clock speeds), the 32 nm Bulldozer actually features a broad integer unit with eight integer pipelines split into two portions, each portion having its own scheduler and L1 Data cache.
I did say I know nothing, and it's not like I work for you guys, so of course all I have is my opinion.

I was kinda actually hoping for a reduction in both power and thermals as the process matures, as currently, my 965BE is overheating on stock cooling(65c+ load). It's a horrible sample though.

Any info you can give that, of course, is more than welcome. ;)
Posted on Reply
#7
bear jesus
cadaveca said:


I was kinda actually hoping for a reduction in both power and thermals as the process matures, as currently, my 965BE is overheating on stock cooling(65c+ load). It's a horrible sample though.
That is exactly why i dont use stock cooling with my processor, i stuck the 965's hsf on an old athlon x2 thats in my htpc.
I really hope amd brings a new hsf with the bulldozer

*hint, hint* JF-AMD :p (although yes i admit i don't have a clue on temps or cooling needs on the new core so the current hsf may be plenty for bulldozer)
Posted on Reply
#8
Jstn7477
I hope AMD gets this generation right and can fight Intel. I'm liking the looks of Zacate vs. Intel Atom, and hopefully these desktop parts will be powerful and get AMD back into the game. :toast:
Posted on Reply
#9
roberto888
SSE4.1, SSE4.2, the 16MB of L3 Cache, and the very high amount of transistors will make this CPU Kick Ass!!;)And also will make it very expensive! :)
Posted on Reply
#10
mechtech
I have a question. With Sandy Bridge and Bulldozer completely new architecture, will Windows 7 OS be able to take full advantage of it performance wise, or does the OS matter not??
Posted on Reply
#11
cadaveca
My name is Dave
mechtech said:
I have a question. With Sandy Bridge and Bulldozer completely new architecture, will Windows 7 OS be able to take full advantage of it performance wise, or does the OS matter not??
I do not think OS matters too much. The "lighter" the OS, the more power left over for apps, etc, but other than that...
Posted on Reply
#12
Dave63
I think this could be a game changer for ultra high end (blue-ray editing, composing sound tracks, servers, ect....). But software needs to get more in to the game to use more cores, most still do not use 3 cores at this point some do but very few. Thay are years away from fully using 6 cores this has been showin in bencemarks. I will be fun to see what happens. This is just my own view very new at this but learning alot as i go.
Posted on Reply
#13
JF-AMD
AMD Rep (Server)
roberto888 said:
SSE4.1, SSE4.2, the 16MB of L3 Cache, and the very high amount of transistors will make this CPU Kick Ass!!;)And also will make it very expensive! :)
I would not necessarily say that.
Posted on Reply
#14
WhiteLotus
JF-AMD said:
I would not necessarily say that.
Who is this JF-AMD, an actual AMD spokesperson?


And what would you not say, the expensive bit? I hope they are not...
Posted on Reply
#15
JF-AMD
AMD Rep (Server)
Director of product marketing for servers at AMD
Posted on Reply
#16
theonedub
habe fidem
JF-AMD said:
Director of product marketing for servers at AMD
You should get in contact with a Moderator or W1z to verify that and get you a title.
Posted on Reply
#18
JF-AMD
AMD Rep (Server)
theonedub said:
You should get in contact with a Moderator or W1z to verify that and get you a title.
Yeah, can't even edit my sig at this point :(
Posted on Reply
#19
bear jesus
JF-AMD said:
Yeah, can't even edit my sig at this point :(
pm w1zzard or another mod so people can see you are an offical amd rep... just because lol :D
Posted on Reply
#20
ERazer
great news no doubt but im more interested if its gonna be am3 compatible or new socket
Posted on Reply
#21
JF-AMD
AMD Rep (Server)
New socket for client, same sockets for server.
Posted on Reply
#22
ERazer
JF-AMD said:
New socket for client, same sockets for server.
ty sir, guess gonna wait till i switch one of my rigs to amd for wcg, been wanting amd cruncher
Posted on Reply
#23
cadaveca
My name is Dave
Can you confirm that this socket change was necessary to implement the expected performance improvements?

What I'm actually really interested in, of course, is 3D performance. What key areas are targeted to improve this?
Posted on Reply
#24
bear jesus
To be honest i kinda of wish the bulldozer would come in a g34 socket (1974 pin) so that it came with quad channel ram... although i doubt it would make much different apart from benchmarks and maybe virtual machines.
Posted on Reply
#25
JF-AMD
AMD Rep (Server)
yes. using AM3 would have meant performance compromises.
Posted on Reply
Add your own comment