AMD FX-Series "Bulldozer" will ship end of June

I think AMD will get some ground with what they have in store this year.

Llano has a potential of being a really good option for many non-enthusiasts and non-system builders, which is a lot of people out there.
Intel simply doesn't have anything to compete against Llano. Even Zacate's little GPU beats SB's on die GPU.

Bulldozer is a big unknown as far as single thread performance goes. But as far as server multi threaded performance goes it is hard not to expect for it to completely stomp Intel's offerings. Even with the aging architecture AMD is still doing pretty well with the Opteron/Magny-Cours line.

AMD is still on top of the top500 super computer list, with the fastest CPU based super computer in the world. Something they have been dominating for awhile. Recently Intel based solution took the first spot for the first time in a long time but it is a CPU+GPU solution so a lot of the credit should go to nVidia.

Zacate is simply better than Atom in every way possible.

This of course is the enthusiast forum where single thread performance is the king. It's only to be expected that many will just hate on AMD.

Whether you like Intel or AMD this will be a year of price wars between the two rivals, and things will get interesting.
 
r00t61 said:
Has there been any info released on the 900-series chipset associated with BD? Number of PCIe lanes, native USB3 support, number of SATA2/3 ports, etc.?
Click to expand...
BD uses HT 3.1 instead of 3.0, which is just a speed bump. I would expect from that to see higher stock CPU-NB and HT Link speeds although they may still be well below the actual specification like we have now (2.0 actual vs 2.6 GHz specification).

I saw a third-party slide in another forum with one of the interconnects (SB950?) doubling it's PCI-E lanes but since it wasn't an official slide and I haven't heard anything about that elsewhere it could have easily been a typo.

I'm also still not 100% on whether or not USB 3.0 is included in the chipset or still requires an additional chip. I've seen plenty of articles that go both ways on the subject.

As far as is public knowledge, EVERYTHING else is the same with 990FX & SB950 vs 890FX & SB850 unless there are some secret features AMD is keeping a tight lip on. One thing I am certain on is that until we have concrete knowledge of the changes I wouldn't suggest buying one of the early release AM3+ boards that use the 8xx chipset.
 
Anyone here think like I do?

Maybe the FX-6110 Bulldozer CPU will unlock to one of the 8 Core versions (FX-8110 possibly) :eek:

Never know, might be that the 6 Core FX is just a disabled 8 Core FX ready for us AMD enthusiasts / fanboys. :)
 
This teletran8 mofo is a trolling tool. Someone moderate this guy plz
 
Last edited by a moderator:
teletran8 said:
Anyone here think like I do?

Maybe the FX-6110 Bulldozer CPU will unlock to one of the 8 Core versions (FX-8110 possibly) :eek:

Never know, might be that the 6 Core FX is just a disabled 8 Core FX ready for us AMD enthusiasts / fanboys. :)
Click to expand...

nope, chances i would guess are low that two cores end up being "defective".....especially since they are doing a modular design, the ones you can "unlock" will be odd in core count i would bet....
 
CaptNumbNutz said:
Re-read what I typed. I said Socket AM3+ is backwards compatible with AM3 CPU's. I DID NOT say that socket AM3 is compatible with AM3+ CPU's.

Thank you.
Click to expand...

Theres no compabilty at all. It's a diffrent socket for god's sake.
 
I don't understand. If its a diffrent socket/pin layout, how the hell can you pop a AM3 cpu in an AM3+ board?

And if it's not a totally different, then Its just a case of AMD being greedy and setting artificial limits.
 
Alex. said:
I don't understand. If its a diffrent socket/pin layout, how the hell can you pop a AM3 cpu in an AM3+ board?

And if it's not a totally different, then Its just a case of AMD being greedy and setting artificial limits.
Click to expand...

same thing as AM2+ to AM3 ... different pin count....however they are having the pin placement similar so Am3's will be compatible (atleast from all the information we have soon)
 
W.Feather said:
same thing as AM2+ to AM3 ... different pin count....however they are having the pin placement similar so Am3's will be compatible (atleast from all the information we have soon)
Click to expand...

So why even change pin count?

(And don't feed me bullshit about there not being enough pins to establish a connection for new faster chips)
 
Alex. said:
I don't understand. If its a diffrent socket/pin layout, how the hell can you pop a AM3 cpu in an AM3+ board?

And if it's not a totally different, then Its just a case of AMD being greedy and setting artificial limits.
Click to expand...

They still have the original pin count (941 pin contacts) as the white AM3Socket with one more pin added.

The newer black AM3+ Socket has an added pin (942 total pins, you can see in those two pictures side by side where they added the 1 pin to the bottom left socket section. That there used to be two pins covered in plastic bubble area now there is just one covered with plastic. This socket pattern (AM3+) supports the newer Bulldozer CPU's, but it's also backwards compatible with the AM3 original pin pattern on the AMD CPU's...if that makes any sense.
 
Alex. said:
I don't understand. If its a diffrent socket/pin layout, how the hell can you pop a AM3 cpu in an AM3+ board?

And if it's not a totally different, then Its just a case of AMD being greedy and setting artificial limits.
Click to expand...
Notched differently. Look at the bottom of the CPU's, AMD has a pin configuration that will allow a AM3 CPU to fit in a AM3+ Motherboard (where every pin on a AM3 is accounted for on the AM3+ socket), but has a couple of pins in area's that are blocked off on an AM3 on AM3+ chips. So when we see a AM3+ Mobo the motherboard socket will have more pin holes then either the AM3 or AM3+ boards have.

My guess is that in the future they could have more notches left over if they have a planned upgrade path to a future AM3 version, or AM4. My guess is that the next version will be first in a new attempt of upgrade-ability and that the upgrade chain from AM2 will be completely broken.
 
Alex. said:
So why even change pin count?

(And don't feed me bullshit about there not being enough pins to establish a connection for new faster chips)
Click to expand...

It's the only way Bulldozer can support the newer features, and be a better product than the Phenom II's were...
 
teletran8 said:
It's the only way Bulldozer can support the newer features, and be a better product than the Phenom II's were...
Click to expand...

Not completely. Most sockets sizes changes are from Memory pinoutrs and channels due to on die memory controllers. That's why there has been little movement from 939-AM3+ as DDR2 and DDR3 had roughly the same amount of pins and still only 2 channels. All they need is one pin not available in another CPU to decide what feature set to use. AM3 CPU's for example can using the same pins for memory switch between DDR2 and DDR3 by being able to tell what board its in because of single pin marker.

That's also why there is such a little difference between the base i7 pin configuration and the i7 SB one. Yet the triple channel needed more and the 2011 (quad channel) probably to ecc support.

You see the same thing with AMD's server chips with the pin counts sky rocketing from 940 to 1944. Considering the similarity in chip counts its easy to see where the memory plays a part.
 
I think the pinout change mostly has to do with the different power plane configuration. Zambezi has at least 3 separate power planes (core, L3+uncore, mem controller) that AM3 doesn't have.
 
JMccovery said:
I think the pinout change mostly has to do with the different power plane configuration. Zambezi has at least 3 separate power planes (core, L3+uncore, mem controller) that AM3 doesn't have.
Click to expand...

That's why it won't work in AM3 boards, but the pins it self has no affect on this. The mobo just changes the signal and communication pathing based on which marker pin it detects on power up.

In the end AMD could still be using same 939 pinouts from almost what seems like a decade ago if they weren't worried about CPU's put into a system they shouldn't be in.

That's why Intel went from 1155 to 1156.
 
Michaelius said:
http://www.techpowerup.com/142251/A...-ASUS-AM3-Motherboards-Using-BIOS-Update.html

Asus says bios update wil be enough?
Click to expand...

Until it actually happens, I still wouldn't count on any existing non-890FX boards getting AM3+ support. Check out the image in that link. The current inexpensive Asus AMD lineup is all M4x motherboards, and that list is all M5x model numbers for everything but the 890FX boards.

Heck I need a new motherboard right now and I would gladly plunk down some cash for a AM3(+) motherboard *if* a company would openly state that the board would definitely be updated to support AM3+ CPUs. Instead we get little bits of information strewn about here and there or buried in web pages few people click on.
 
Last edited:
Alex. said:
I don't understand. If its a diffrent socket/pin layout, how the hell can you pop a AM3 cpu in an AM3+ board?
Click to expand...

That much is obvious. But it didn't stop you from acting like you did and in a completely arrogant, dickish fashion no less.

Jeebus.
 
Michaelius said:
http://www.techpowerup.com/142251/A...-ASUS-AM3-Motherboards-Using-BIOS-Update.html

Asus says bios update wil be enough?
Click to expand...
That article is spreading misinformation like wildfire. ASUS is releasing new revisions of existing boards with the AM3+ socket. This does not mean in any way, shape or form that existing products on the shelf with AM3 sockets will be able to run a BD chip with a BIOS flash. AM3 sockets still have a pin blocked off that will exist on BD chips... Who here wants to break a pin off their shiny new BD chip just to see if they can get something to work that AMD says won't?

http://www.xtremesystems.org/forums/showpost.php?p=4780923&postcount=35
 
Look at the pin layout it's posted all around.

The AM3 socket has a blocking section of 2 pins
The AMD3+ socket has only one blocked pin in that area.

So basically you can fit the round peg in the square hole since the peg is small enough... but you can't fit the square pen in the round hole get it because of the extra pin on the bulldozer (socket amd3+). Have a close look at the picture, do you see it???
 
JF-AMD said:
AMD is supporting BD on AM3+ sockets only. Period.
Click to expand...

That certainly isn't what ASUS is saying... They've listed some boards that are currently available that they say with a BIOS update will support AM3+ CPUs. They listed the Crosshair IV Formula and Extreme boards, for instance. I don't think ASUS would do this unless it was possible somehow.

I know, I know. You work for AMD and that makes you credible (at least to me and most people on the forums) but a company like ASUS doesn't just blow smoke like that unless they've tested things. I mean, we're talking about ASUS, not some start-up or oddball (here's looking at you, Asrock).
 
Not to be mean, but half of ASUS boards don't support Thuban C-states properly. Their BIOS updates on the AMD side at least have been notoriously few and far between. I havent seen 1100T or 975 BIOS updates for any of their high-end boards.

I don't exactly have confidence in what they say at this point.
 
My understanding is that some manufacturers are releasing new revisions of current models but with the new socket. For example on Gigabyte's website the Revision 3.1 boards have a dark grey/black socket with an extra pin.
 
DedEmbryonicCe11 said:
That article is spreading misinformation like wildfire. ASUS is releasing new revisions of existing boards with the AM3+ socket. This does not mean in any way, shape or form that existing products on the shelf with AM3 sockets will be able to run a BD chip with a BIOS flash. AM3 sockets still have a pin blocked off that will exist on BD chips... Who here wants to break a pin off their shiny new BD chip just to see if they can get something to work that AMD says won't?

http://www.xtremesystems.org/forums/showpost.php?p=4780923&postcount=35
Click to expand...

Maybe someone will come along with a socket adapter. Remember Powerleap and Lin-Lin?

After all, it's just one pin. The sockets share quite a lot in common, and still have to support AM3- processors. It's entirely possible that the mobo makers were informed early on of AMD's direction so they could begin development, and they planned some flexibility into their later mobo designs to ease the transition.

Honestly, this keying sounds quite artificial. Why not make it compatible with AM3, but with fewer power states/power planes, and no DDR2 support? What's the big freakin' deal, AMD?
 
I wonder if you could drill out the hole rather than yank off the pin. You'd need a very fine drill, like jewelers use.
 
Last edited:
CaptNumbNutz said:
No. No possibility of quad or triple channel ram on the desktop side. It also has nothing to do with the chipset or the motherboard, and most likely won't improve performance desktop side anyways. Quad channel is socket C32 and G34 in the server and high end workstation marketplace only.

The confirmation that you won't see an increase in RAM channels has to do with the socket AM3+ itself. Socket AM3+ is backwards compatible with socket AM3 CPU's (it will accept older AM3 cpu's), therefore AM3+ has a similar amount of pins in the socket (approx. 940). The memory controllers are on the cpu's themselves. Therefore any new channels means you would have to add pins to the CPU and the socket.
Click to expand...

It might be possible to add memory and bandwidth via an HT-attached controller or a 2nd cpu. The AMD platform has supported non-uniform memory arrangements since K8, and that's exactly what the new quad-channel G34 socket is doing - 2 channels to each cpu die.

So say they have CPUs where all the cores are defective, but the hypertransport and memory controller (and cache?) still work. That can be used as an HT memory switch. Of course, the consumer chips don't support cache coherency this way (afaik).
 
Last edited:
i dont know why people care so much about upgrading a CPU with their old motherboard.. do you want to have a new chipset to take advantage of the new features of the system??
 
mjz_5 said:
i dont know why people care so much about upgrading a CPU with their old motherboard.. do you want to have a new chipset to take advantage of the new features of the system??
Click to expand...

Exactly! I want this new chip to shine! No reason to compromise performance. Besides, your looking at add'l $150-200. This ain't [H] for nothin!:D
 
mjz_5 said:
i dont know why people care so much about upgrading a CPU with their old motherboard.. do you want to have a new chipset to take advantage of the new features of the system??
Click to expand...

Name me one feature 9 series chipset has that makes huge diffrence to 8 series.
 
Oh my, for years AMD has allowed backwards compatibility with their CPU's and for once your crying over a board refresh with a new CPU? Some people are just too spoil!:eek:
 
Awww screw it! Just get the needle nose pliers out and take that extra pin off. There, it will fit in all AM3 sockets then. :D
 
jamesrb said:
Not to be mean, but half of ASUS boards don't support Thuban C-states properly. Their BIOS updates on the AMD side at least have been notoriously few and far between. I havent seen 1100T or 975 BIOS updates for any of their high-end boards.

I don't exactly have confidence in what they say at this point.
Click to expand...

I have a friend using an 1100t on a system I built for him with a m488td-m/usb3 right now with no issues.
Posted via Mobile Device
 
You must log in or register to reply here.
Back
Top