Tuesday, August 24th 2010

AMD Details Bulldozer Processor Architecture

AMD is finally going to embrace a truly next generation x86 processor architecture that is built from ground up. AMD's current architecture, the K10(.5) "Stars" is an evolution of the more market-successful K8 architecture, but it didn't face the kind of market success as it was overshadowed by competing Intel architectures. AMD codenamed its latest design "Bulldozer", and it features an x86 core design that is radically different from anything we've seen from either processor giants. With this design, AMD thinks it can outdo both HyperThreading and Multi-Core approaches to parallelism, in one shot, as well as "bulldoze" through serial workloads with a broad 8 integer pipeline per core, (compared to 3 on K10, and 4 on Westmere). Two almost-individual blocks of integer processing units share a common floating point unit with two 128-bit FMACs.

AMD is also working on a multi-threading technology of its own to rival Intel's HyperThreading, that exploits Bulldozer's branched integer processing backed by shared floating point design, which AMD believes to be so efficient, that each SMT worker thread can be deemed a core in its own merit, and further be backed by competing threads per "core". AMD is working on another micro-architecture codenamed "Bobcat", which is a downscale implementation of Bulldozer, with which it will take on low-power and high performance per Watt segments that extend from all-in-One PCs all the way down to hand-held devices and 8-inch tablets. We will explore the Bulldozer architecture in some detail.
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.

Parallelism: A Radical Approach?
Back when analysts were pinning high hopes on the Barcelona architecture, their hopes were fueled by early reports suggesting that AMD was using wide 128-bit wide floating point units, leading analysts to believe that AMD may have conquered its biggest nemesis - floating point performance, in turn its pure math crunching abilities. However, that wasn't exactly to be. That's because the processor's overall number crunching abilities were pegged to its floating point performance, ignoring the integer units.

AMD split 8 integers per core into two blocks, each block having four integer pipelines, an integer scheduler for those, and an L1 Data cache. These constitute the lowest level of "dedicated components", dedicated to processor threads. There is a shared floating point unit between the two, with two 128-bit FMACs, arbitrated by a floating point scheduler. The Fetch/Decode, an L2 cache, and the FPU constitute "shared" components.

AMD is implementing a simultaneous multithreading (SMT) technology, it can split each of the "dedicated" components (in this case, the integer unit) to deal with a thread of its own, while sharing certain components with the other integer unit, and effectively make each set of dedicated components a "core" in its own merit of efficiency. This way, the actual core of the Bulldozer die is deemed a "module", a superlative of two cores, and the Bulldozer die (chip) features n-number of modules depending on the model.
So now you have a chip with eight cores with much lower die sizes and transistor counts compared to a hypothetical 32 nm K10 8-core processor. It is unclear whether AMD wants to further push down SMT to the "core" level and run two threads simultaneously over dedicated components, but one thing for sure is that AMD has embraced SMT in some form or another. In all this, the chip-level parallelism is transparent to the operating system, it will only see a fixed number of logical processors, without any special software or driver requirement.

So in one go, AMD shot up its integer performance. Either a thread makes use of one integer unit with its four pipelines, or deals with both the integer units arbitrated by the fetch/decode, and the shared FPU.

Outside the modules
At the chip-level, there's a large L3 cache, a northbridge that integrates the PCI-Express root complex, and an integrated memory controller. Since the northbridge is completely on the chip, the processor does not need to deal with the rest of the system with a HyperTransport link. It connects to the chipset (which is now relegated to a southbridge, much like Intel's Ibex Peak), using A-Link Express, which like DMI, is essentially a PCI-Express link. It is important to note that all modules and extra-modular components are present on the same piece of silicon die. Because of this design change, Bulldozer processors will come in totally new packages that are not backwards compatible with older AMD sockets such as AM3 or AM2(+).
Expectations
Not surprisingly, AMD isn't talking about Bulldozer as the next big thing since dual-core processors (something it did with Barcelona). AMD currently does have an 8-core and 12-core processors codenamed "Magny-Cours", which are multichip modules of Shanghai (4-core) and Istanbul (6-core) dies. AMD expects an 8-core Bulldozer implementation (built with four modules), to have 50% higher performance-per-watt compared to Magny-Cours.

Market Segments
As mentioned in the graphic before, AMD's modular design allows it to create different products by simply controlling the number of modules on the die (by whichever method). With this, AMD will have processors ready with most PC and server market segments, all the way from desktop PCs, enthusiast-grade PCs, notebooks, to servers. AMD expects to have a full-fledged lineup in 2011. The first Bulldozer CPUs will be sold to the server market.


Hotchips 22 Presentation by AMD on the Bobcat Architecture
Below are as-is slides from AMD's Hotchips presentation on the Bobcat architecture.
Add your own comment

283 Comments on AMD Details Bulldozer Processor Architecture

#1
pantherx12
Bloodcrazz said:

AMD by law have given false and misleading information regarding the Bulldozer.
.... No they have not, a road map is not a legally binding statement.

But anyway, chaps who have issues with the AM3 thing wanna take your discussion to a new thread instead?


All it takes is clicking the start new topic button : ]
Posted on Reply
#2
inferKNOX
Bloodcrazz said:
If u read the earlier posts he was talking about AMD saying there releasing the AM3 socket on the bulldozer and AMD changed there minds.
After Millions of ppl bought AM3 socketed cpus believing they would be compatible.
And now they have changed the socket on the Bulldozer.
AMD by law have given false and misleading information regarding the Bulldozer.
I read the whole thread before posting.
AMD did not officially announce anything, so people riding on rumors made a mistake just as pantherx12 said.
Anyway, it's not like they can't sell to upgrade, it's not a consumable.:wtf:
Posted on Reply
#3
naram-sin
Bloodcrazz said:
If u read the earlier posts he was talking about AMD saying there releasing the AM3 socket on the bulldozer and AMD changed there minds.
After Millions of ppl bought AM3 socketed cpus believing they would be compatible.
And now they have changed the socket on the Bulldozer.
AMD by law have given false and misleading information regarding the Bulldozer.
In essence, it said they "plan", not "we take a pledge to insure backward comp...", you get the idea. And I can't believe that anyone bought the board only to be able to upgrade it 2 years after... that, my good sir is foolish. I get a new MOBO's more often than my car get's new tires. And it's never on a premium. Anyone that upgraded wanted 8xx because it was a larger no. than 7xx, because it had from scratch support for DDR3 and supported new interfaces...

You AMD boyz are just spoiled, what with AM2/AM2+/AM3 backward compatibility... :D

inferKNOX said:
I read the whole thread before posting.
AMD did not officially announce anything, so people riding on rumors made a mistake just as pantherx12 said.
Anyway, it's not like they can't sell to upgrade, it's not a consumable.:wtf:
Good frelling point.
Posted on Reply
#4
Valdez
Bloodcrazz said:
If u read the earlier posts he was talking about AMD saying there releasing the AM3 socket on the bulldozer and AMD changed there minds.
After Millions of ppl bought AM3 socketed cpus believing they would be compatible.
And now they have changed the socket on the Bulldozer.
AMD by law have given false and misleading information regarding the Bulldozer.
I think only a very low percentage of buyers chosed an am3 board because of bulldozer. Anyway backward compatibility is not a guarantee, never was. Do you remember the am2/+/3 compability? There were a lot of am2 boards which can accept the phenom II series, but not only the socket matters. There are other things like bios size, power design, manufacturer support quality.

If bulldozer would be backward compatible to am3 socket, and you have a popular am3 board from a decent manufacturer, you'll likely have bios support for the new processor.
Posted on Reply
#5
newtekie1
Semi-Retired Folder
I'm interested to see how this performs. I really hope it has something for Intel, because their high end prices have been insane thanks to no competition.

But people need to not be fooled. If this really does outperform Intel, don't think AMD is going to keep prices low, they will jack prices up just as fast an Intel when they can. The only reason prices are cheap now is because they have to be cheap.

PaNiC said:
X48 realsed for the die shink core 2 65nm to 45nm and got bunch of new chips
i own a X48 and have nothing bad to say about it.
X48 was board i had befor upgrading to my 890fx, i could have gone 1366 drop a i7 930 and later i7 970.
PaNiC said:
trolling would mean i'm an intel fan boy or something, im just making my point that there was no need for the 890fx and to mislead people. like AMD did.
fact: people call me an emo first and i own a amd. im not a fanboy
I hate to be the one to burst your bubble, but the move to the 800 series was done to add USB 3.0 and SATA 6Gbps natively to the chipsets.

On the other hand, the move from x38 to x48 was for nothing other than a name change. In fact some x38 boards could be flashed to x48 without a problem because the manufacturers simply put a sticker on the x38 board and called it an x48. Yes, 1600FSB was offically supported with the x48 chipset, and not the x38, but anyone that ran an x38 knows it easily did 1600FSB.
Posted on Reply
#7
link2009
Bloodcrazz said:
I dont think that "PaNiC" insulted anyone at all.
Im pretty sure that he was having a go at AMD not at any person.
Its called freedom of speech hence this being a FORUM.
inferKNOX said:

Can you just quit it, we're not interested in your rant!:shadedshu
I can't believe people like you that complain about AMD's new CPU not being backward compatible, when they've been making all their others backward compatible for so long! Get over it, the world can't be on hold forever, the only way to make significant changes is to make things radically different and not backwards compatible.:nutkick:
wolf said:
this thread has been going downhill since page 1.
I think everyone knows he is trolling.
Posted on Reply
#8
Valdez
btarunr said:
Old roadmap is old. AMD told us it's a different socket just last week.
Well, this is the newest roadmap so far (for 2011)...

Which socket did amd told you? It is an existing or a new socket?
Posted on Reply
#9
inferKNOX
btarunr said:
Old roadmap is old. AMD told us it's a different socket just last week.
Ooohh... what is the socket called?!... tell us, tell us, tell us, tell us!! *Giant Goo-Goo Eyes*
Posted on Reply
#10
CDdude55
Crazy 4 TPU!!!
Because of this design change, Bulldozer processors will come in totally new packages that are not backwards compatible with older AMD sockets such as AM3 or AM2(+).
Finally a new socket!

Hopefully we can actually see some performance from these things pretty soon, getting tired of AMD fanboys cuming there pants over speculation.
Posted on Reply
#11
JATownes
Valdez said:
Well, this is the newest roadmap so far (for 2011)...

Which socket did amd told you? It is an existing or a new socket?
inferKNOX said:
Ooohh... what is the socket called?!... tell us, tell us, tell us, tell us!! *Giant Goo-Goo Eyes*
If I am not mistaken, they call it "Socket G"...anyone know if I am remembering that correctly? :o
Posted on Reply
#12
inferKNOX
JATownes said:
If I am not mistaken, they call it "Socket G"...anyone know if I am remembering that correctly? :o
Do you have details? Does it take pins or have them and contact the CPU like with Intel. Are physical dimensions the same? Number of contact-points/pins/whatever-you-call-them?

Really concerned about a possible different HS retention layout...:o
Posted on Reply
#13
JATownes
I cannot remember where I read that at, but I think it was a derivative of the G34 Socket used for the Opteron Magny-Cours.

Socket G34 also supports 4 channel DDR3, so it seems like a natural extension to modify the socket for consumer desktop usage.
Posted on Reply
#14
btarunr
Editor & Senior Moderator
This is pure FUD, but:

I personally expect the consumer Bulldozer package to have nearly 1000~1400 pins, just not arranged like AM3/2 or compatible with it.

The processor may continue to have a dual-channel DDR3 memory controller (maybe higher memory clock speeds of 1833 MHz support to give higher bandwidth).

Processor will need pins to give out 40 PCI-Express 2.1 lanes (incl. the A-Link III which is x4).

No HyperTransport pins on the consumer packages. The 2P/4P Opteron package might be bigger, as it needs pins for 1 or 2 16-bit HyperTransport links (to neighbouring sockets).
Posted on Reply
#15
PaNiC
newtekie1 said:
I'm interested to see how this performs. I really hope it has something for Intel, because their high end prices have been insane thanks to no competition.

But people need to not be fooled. If this really does outperform Intel, don't think AMD is going to keep prices low, they will jack prices up just as fast an Intel when they can. The only reason prices are cheap now is because they have to be cheap.





I hate to be the one to burst your bubble, but the move to the 800 series was done to add USB 3.0 and SATA 6Gbps natively to the chipsets.

On the other hand, the move from x38 to x48 was for nothing other than a name change. In fact some x38 boards could be flashed to x48 without a problem because the manufacturers simply put a sticker on the x38 board and called it an x48. Yes, 1600FSB was offically supported with the x48 chipset, and not the x38, but anyone that ran an x38 knows it easily did 1600FSB.
most of this was coverd in older posts
and as for the new AMD chip it will be good and AMD know that hence the socket forcing people to upgrade, seem intel dosn't have much they put all there eggs in the Larrabee basket and anything they bring out will only be a plan b
Posted on Reply
#16
bear jesus
JATownes said:
I cannot remember where I read that at, but I think it was a derivative of the G34 Socket used for the Opteron Magny-Cours.

Socket G34 also supports 4 channel DDR3, so it seems like a natural extension to modify the socket for consumer desktop usage.
1974 pins :eek: that must be one fat package if bulldozer gives some good performance then quad memory channels might be quite useful for some virtual machines, i just hope we find out for sure what socket it will be and the sepc

I admit i am pretty glad i held off on going for an am3 board, im still running on an am2 790fx/sb600 asus m3a32-mvp that's on its third processor (currently an am3 pehnom ii) to be honest i would be happy with an all new socket as i have been very happy with my am2 upgrades and i saw little point going am3 yet unless buldozer was compatible.

either way i want more info on bulldozer and sandy bridge as im getting the full system upgrade itch (not that i realy need more power)
Posted on Reply
#17
DigitalUK
btarunr said:
This is pure FUD, but:

I personally expect the consumer Bulldozer package to have nearly 1000~1400 pins, just not arranged like AM3/2 or compatible with it.

The processor may continue to have a dual-channel DDR3 memory controller (maybe higher memory clock speeds of 1833 MHz support to give higher bandwidth).

Processor will need pins to give out 40 PCI-Express 2.1 lanes (incl. the A-Link III which is x4).

No HyperTransport pins on the consumer packages. The 2P/4P Opteron package might be bigger, as it needs pins for 1 or 2 16-bit HyperTransport links (to neighbouring sockets).
i think i remember seeing bulldozer was going to have quad-channel memory, i could be wrong. but i cant wait for bulldozer been waiting for years, wish i could pre order now..:)
Posted on Reply
#18
mdsx1950
Wonder if these chips will kick Intel buttocks.
Posted on Reply
#19
CDdude55
Crazy 4 TPU!!!
Man, 8 cores and 16 threads just seems so pointless right now. The software isn't keeping up.:(
Posted on Reply
#20
bear jesus
CDdude55 said:
Man, 8 cores and 16 threads just seems so pointless right now. The software isn't keeping up.:(
For the most part right now its is almost useless but there are more and more things that are geting more heavily threaded, one thing i would be interested in is how a folding@home smp client would do on a 16 threaded bulldozer although i dont know if it would scale over all 16 threads or do it well...... i geuss time should tell
Posted on Reply
#21
pantherx12
CDdude55 said:
Man, 8 cores and 16 threads just seems so pointless right now. The software isn't keeping up.:(
8 or 16 cores XD

I think it is, there's been massive improvements the past 5 months or so.

I've plenty of apps that max out my cores.
Posted on Reply
#22
CDdude55
Crazy 4 TPU!!!
bear jesus said:
For the most part right now its is almost useless but there are more and more things that are geting more heavily threaded, one thing i would be interested in is how a folding@home smp client would do on a 16 threaded bulldozer although i dont know if it would scale over all 16 threads or do it well...... i geuss time should tell
pantherx12 said:
8 or 16 cores XD

I think it is, there's been massive improvements the past 5 months or so.

I've plenty of apps that max out my cores.
No doubt software is moving forward and allowing more cores and threads to be utilized, but it's not as widespread still. Then again it depends on what you do, i mean for an average consumer or gamer, more then 2 cores still isn't needed in most programs. But on the flip side, the people doing massively CPU intensive things might actually need an 8 or 16 core CPU. there has definitely been improvements, but not enough to warrant a CPU with more then 4 cores. But even if the software isn't fully caught up, it's inevitable that hardware will just keep getting better regardless.(and that's what can be disappointing sometimes)
Posted on Reply
#23
bear jesus
CDdude55 said:
No doubt software is moving forward and allowing more cores and threads to be utilized, but it's not as widespread still. Then again it depends on what you do, i mean for an average consumer or gamer, more then 2 cores still isn't needed in most programs. But on the flip side, the people doing massively CPU intensive things might actually need an 8 or 16 core CPU. there has definitely been improvements, but not enough to warrant a CPU with more then 4 cores. But even if the software isn't fully caught up, it's inevitable that hardware will just keep getting better regardless.(and that's what can be disappointing sometimes)
Very true but the bulk of my current machine i have had for over 2 years, i would hope a full system upgrade next year would give me at least a couple more years if not maybe 3 years and i hope at some point in that time software will start to catch up to what will be an out of date machine in that time lol
Posted on Reply
#24
Batou1986
Kudos to AMD for finally trying something completely new i hope these can at least compete with intel for a change.

And congrats to PaNiC for making my ignore list on your first day way to go :shadedshu
Posted on Reply
#25
Mindweaver
Moderato®™
WHat! you mean i can't use my NEW 1055T in the new chipset MB's built for these processors?.. j/k don't ban me mussels! hehehe

Wow this has been the best news of the day! I can't wait to own the newness! (Maybe Intel will be pushed to bring out Core24veinticuatro!
Posted on Reply
Add your own comment