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
TheMailMan78
Big Member
by: cadaveca
Well since, then, JF-AMD is posting here now, so maybe he'll end that confusion. ;)
:laugh: I wouldn't hold my breath. TPU don't sign the mans paycheck.
Posted on Reply
#2
CDdude55
Crazy 4 TPU!!!
by: TheMailMan78
:laugh: I wouldn't hold my breath. TPU don't sign the mans paycheck.
I agree.

I think it's going to be just like the Pailt rep that used to post here, they'll come and correct some people and do some clean up work, but after a day or so they'll never log back in again.(unless something big with their company happens and they have to regulate)
Posted on Reply
#3
trt740
Thats not how it worked Palit guy posted for along time but during the USA financial melt down Palit closed up shop in the USA and he lost his job. So thats not really a good comparison but we get your point.
Posted on Reply
#4
cadaveca
My name is Dave
by: TheMailMan78
:laugh: I wouldn't hold my breath. TPU don't sign the mans paycheck.
NO, but AMD does, and his job is marketing...and according to him, we are included in that marketing. Likewise, it should be his job to correct any misconceptions based on data he has given. After all, something is better than nothing(well, not to me, but ya know.)
Posted on Reply
#5
CDdude55
Crazy 4 TPU!!!
by: cadaveca
NO, but AMD does, and his job is marketing...and according to him, we are included in that marketing. Likewise, it should be his job to correct any misconceptions based on data he has given. After all, something is better than nothing(well, not to me, but ya know.)
Even if that is the case, he as a ''marketer'' has a lot of other big fish to catch, like i said, he came here for a day or so to regulate on some rumors or all out false facts and then bam, you won't see him here possibly until everyone is speculating about another new architecture or another new line up or cards or CPU's.

But who knows, he may keep posting here and doing his best to answer some of our questions. I just don't think first on his list is '' Stay logged into TPU and regulate on some speculation and rumors for weeks on end''.


by: trt740
Thats not how it worked Palit guy posted for along time but during the USA financial melt down Palit closed up shop in the USA and he lost his job. So thats not really a good comparison but we get your point.
Ahhh, that's a good point.
Posted on Reply
#6
cadaveca
My name is Dave
OH, I agree...BullDozer seems to be his "baby".


Time will tell though...I hope for way more than I should, but am always prepared for the worst. ;) I know he's travelling right now, for example...
Posted on Reply
#7
bear jesus
I think we need to remember JF posts on other forums, has a blog he writes/replys to, he has a job (that i assume takes him away from home at times) and a life in general, i'm sure he will be around in the future but i'm also pretty sure he is busyer than most of us :laugh:

I just wish it was not so far away from the server chips launch, i hated waiting through istanbul's release for thuban to come out and show some overclocking numbers.
Posted on Reply
#8
wahdangun
yeah i hope it not as long as shanghai to deneb or istanbul to thuban, i hate to wait that long.

and btw if anyone know if amd will launch fusion chip with bulldozer core ?
Posted on Reply
#9
Neo4
by: trt740
Thats not how it worked Palit guy posted for along time but during the USA financial melt down Palit closed up shop in the USA and he lost his job. So thats not really a good comparison but we get your point.
Too bad about Palit USA and the rep who posted here. Glad you can still buy them as I have a Sonic Platinum 1 gig that does some crazy numbers Folding@Home and it's running default speed undervolted to .950v. I may even have it when I buy my Bulldozer mobo next year..
Posted on Reply
#10
TheMailMan78
Big Member
Green or Red I hope both sides bring it soon. These prices are getting out of hand.
Posted on Reply
#11
JF-AMD
AMD Rep (Server)
If you look at my forum coverage, I am on about 8 or 10 different ones and have thousands of responses.

While "marketing" is my job, forums are not in my job description and I don't get paid to be here. You'll notice that I specifically try to stay away from speculation on the competitor, I only deal with facts and you also should note that I start very few threads. I generally only respond to questions because I don't want to be a shill for my products.

I do most of my forum surfing after hours. For those of you that think this is the workday right now, I am in Tokyo right now and it is not even breakfast time. Been in forums for the past 2-3 hours.

Time will tell, reputations are not earned overnight.

What was the original question you were asking about, I can't track that back. If someone can repost I will get you an answer.
Posted on Reply
#12
TheMailMan78
Big Member
by: JF-AMD
If you look at my forum coverage, I am on about 8 or 10 different ones and have thousands of responses.

While "marketing" is my job, forums are not in my job description and I don't get paid to be here. You'll notice that I specifically try to stay away from speculation on the competitor, I only deal with facts and you also should note that I start very few threads. I generally only respond to questions because I don't want to be a shill for my products.

I do most of my forum surfing after hours. For those of you that think this is the workday right now, I am in Tokyo right now and it is not even breakfast time. Been in forums for the past 2-3 hours.

Time will tell, reputations are not earned overnight.

What was the original question you were asking about, I can't track that back. If someone can repost I will get you an answer.
Did cadaveca create the TWKR chip?
Posted on Reply
#13
cadaveca
My name is Dave
by: TheMailMan78
Did cadaveca create the TWKR chip?
:roll:





I didn't say I created it, I said it was my idea. ;)


Normally I'd call you a troll, but that's seriously funny.

:D


Now wouldn't you be upset if he said yes.:laugh:
Posted on Reply
#14
wahdangun
by: TheMailMan78
Green or Red I hope both sides bring it soon. These prices are getting out of hand.
yupz, and it really retarded, without NV the price is just gonna be worst, i hope AMD still have some sense in their brain, and not price it like no tomorrow
Posted on Reply
#15
mastrdrver
While I don't visit XS too often I'm usually browsing S/A and it seems like JF posts quite a bit (at least to me) over on there. Though that forum can get highly technical at times.

by: JF-AMD
If you look at my forum coverage, I am on about 8 or 10 different ones and have thousands of responses.
Best one I remember recently is when you got called out over on OC.net for being a noob since your post count is so low. :roll:

Btw can you comment on whether all AM3 chips will be able to work in the Bulldozer socket for non server types like us?

I've just never seen anyone clarify if all AM3 chips will work or just certain ones (why some would and some wouldn't outside of bios support wouldn't make sense to me but maybe there is something more than what is currently known). Trying to be as general with that question in regards to future stuff as I can but I'm sure you know what I'm trying to get at.

by: TheMailMan78
Did cadaveca create the TWKR chip?
bwahahahaha :laugh:
Posted on Reply
#16
JF-AMD
AMD Rep (Server)
Actually I don't know the answer to this.
Posted on Reply
#17
bear jesus
by: mastrdrver
can you comment on whether all AM3 chips will be able to work in the Bulldozer socket for non server types like us?
If the desktop socket for bulldozer is AM3+ would it be logical to assume that it should be simmilar to socket AM2 and AM2+ and having likly the simmilar compatibility issues and kind of trade offs?

If that is correct then should not all AM3 chips be able to work in an AM3+ board and all AM3+ chips be able to work in an AM3 board with the only incompatibilities brought in by board makers depending on their bios updates?

I would hope so as over recent years and between 2 computers i have used an AM2 chip on an AM2+ board, an AM2+ chip on an AM2 board and am currently using an AM3 chip on an AM2+ board (of corse also had matching chips to sockets), it has been one of the big reasons it has been so easy for me to stay with AMD.
Posted on Reply
#18
crazyeyesreaper
Chief Broken Rig
bulldozer wont work period in AM3 end of story there but the pin layout may be such that an AM3 cpu can be dropped in and work we have to wait and see for now tho
Posted on Reply
#19
bear jesus
by: crazyeyesreaper
bulldozer wont work period in AM3 end of story there but the pin layout may be such that an AM3 cpu can be dropped in and work we have to wait and see for now tho
I would say there goes my processor/board swapping but to be honest i don't intend to get an AM3 board anymore (mainly due to asus being a pain in the butt and taking forever with the crosshair IV extreme), i just intended to wait for bulldozer and if its good enough to tempt me away from intels offerings then i will be buying a current board to go with a current chip. either way though i'm really looking forward to an all out upgrade to my gaming rig :D
Posted on Reply
#20
mastrdrver
From what I've understood is that AM3 cpus will work in the Bulldozer socket. Whether it has AM3+ or if that is for Llano idk. Though Bulldozer is not backward compatible with AM3 boards this has been made clear after the initial releases from Hot Chips.
Posted on Reply
#21
scaminatrix
by: JF-AMD
People seem to be really caught up in how many channels of memory there are, and not necessarily how efficient those channels perform.

What if you had 2 channels that could perform the same as 3? Would you still demand 3 or would you be ok with 2?
New chipset with dual channel DDR4 support next year? 990FX?
Posted on Reply
#22
JF-AMD
AMD Rep (Server)
DDR4 in 2011? You need to do a little more research on that one.
Posted on Reply
#23
scaminatrix
by: JF-AMD
DDR4 in 2011? You need to do a little more research on that one.
My optimism will be the death of me!
Posted on Reply
#24
bear jesus
by: scaminatrix
New chipset with dual channel DDR4 support next year? 990FX?
:laugh: unfortunatly it should be around 2015 we see DDR4, would be nice to have some DDR4 next year but i would be happy with some DDR3 if i can push it over 2000mhz on an amd board.
Posted on Reply
#25
nt300
by: scaminatrix
New chipset with dual channel DDR4 support next year? 990FX?
More like very late 2012 to 2013 that was already confirmed by many ram manufacturers. I still think DDR3 has a lot more legs to run on and now that price is comig down it very good.
Posted on Reply
Add your own comment