Friday, February 10th 2012

AMD 2012 CPU Roadmap Unveils FX-X300 and A10 Series

AMD is pushing on with a desktop product lineup that's leveraging its Piledriver CPU and Graphics CoreNext GPU architectures in 2012. Apparently, the company will have a faster product development cycle to catch up with Intel's "Tick-Tock", as revealed in a roadmap slide scored by DonanimHaber. The current product lineup will remain unchanged in the first quarter of 2012. Then in the second quarter, AMD will launch a few more socket AM3+ FX-8000, FX-6000, and FX-4000 series eight, six, and four-core processors; along with the much talked about "Trinity" accelerated processing unit.

The fastest "Trinity" APUs will get a new brand identifier, the A10-5000 series. These APUs will pack next-generation "Piledriver" modular cores and Radeon HD 7600D series graphics. Around this time, AMD will also launch the Brazos 2.0 low-power APU for netbooks, nettops, and embedded computing devices. Brazos 2.0 will get the E2-1000 series branding. The big change is reserved for the third quarter of 2012, when AMD launches the successor of its less-than-lucky AMD FX "Bulldozer" processor family.
Codenamed "Vishera", AMD's new FX-x300 family (where x = 8 in case of eight-core, 6 in case of six-core, and 4, in case of quad-core), will likely be built on the same AM3+ platform, but based on the "Piledriver" core architecture, which brings in about 15% IPC increase over Bulldozer. The roadmap slide talks about FX-8350 being the top-end part, followed by FX-8320, FX-6300, and FX-4320. Around that time, AMD will replace its A6 and A4 "Llano" parts with new A6 and A4 "Trinity" ones. The A6-5400 APU features Radeon HD 7540D graphics, while the A4-5300 features Radeon HD 7480D.Source: DonanimHaber
Add your own comment

43 Comments on AMD 2012 CPU Roadmap Unveils FX-X300 and A10 Series

#1
nt300
mastrdrver said:
Actually your still wrong.

The BIOS does not disable the other core in the module since it is the whole module that is power gated.

The BIOS just does not expose the other core to the OS. AMD only power gates the module not the cores since the cores share parts and to shut down the only part that is not shared would save little to no power.

edit:The Tech Report's Bulldozer Review
They are talking about a modified BIOS where it lets you disable one core per module. It was tested in benchmarks and had better performance per clock with having 1 core per module disabled.

Example:
FX-8120 @ 1 Core per Module = 4 single cores = Quad-Core (4/4) performed MUCH faster than the FX-4100 (2/4) at same clock speed. AMD needs to setup Piledriver so it can disable and enable both ways, which ever way gives you better performance.
Posted on Reply
#3
Super XP
user21 said:
the way i see it? world's first 4.2ghz clocked processor. AMD FX-4170

http://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series%20FX-4170%20FD4170FRW4KGU.html
I think AMD cancelled that CPU in favour for newer Piledriver based CPU's. I may be wrong though. They were suppose to also release the FX 8170 which is also based on the newer B3 stepping based on Bulldozer. I believe Piledriver is C3 or C4 stepping which would have a different naming scheme IMO.
Posted on Reply
#4
mastrdrver
nt300 said:
They are talking about a modified BIOS where it lets you disable one core per module. It was tested in benchmarks and had better performance per clock with having 1 core per module disabled.

Example:
FX-8120 @ 1 Core per Module = 4 single cores = Quad-Core (4/4) performed MUCH faster than the FX-4100 (2/4) at same clock speed. AMD needs to setup Piledriver so it can disable and enable both ways, which ever way gives you better performance.
I understand what they are saying and I'm telling them that it is not going to happen and never will because of the way AMD designed the module.
Posted on Reply
#6
Nil Einne
mastrdrver said:
I understand what they are saying and I'm telling them that it is not going to happen and never will because of the way AMD designed the module.
I agree we're not going to see a tri core Bulldozer based CPU. What I'm a bit surprised is they didn't have 6 core (tri module) for the high end APU, 4 core for the mid and perhaps some of the high end and 2 core for the low and perhaps some mid end. That way in terms of performance the A10-5800K should beat the A8-3870K by a handy margin in pretty much most highly multithreaded, CPU bound apps whereas I'm not sure it will based on what I've seen in Bulldozer reviews (although I admit I haven't looked that much). But I guess in terms of yields, power consumption, core size, performance etc it didn't make sense. It will be interesting to see how the performance boundary (and price-performance) between A6 and A8 will be.
Posted on Reply
#7
mastrdrver
There's no need for a 6 core APU.

There's very good reason to believe that the 17w ULV Trinity parts will equal 25/35w Llano parts (talking notebooks here). 25/35w Trinity parts will blow Llano out of the water.

Realize that Trinity looses the slow L3 cache that definitely does not help BD on the desktop.
Posted on Reply
#8
eidairaman1
The Exiled Airman
mastrdrver said:
there's no need for a 6 core apu.

There's very good reason to believe that the 17w ulv trinity parts will equal 25/35w llano parts (talking notebooks here). 25/35w trinity parts will blow llano out of the water.

Realize that trinity looses the slow l3 cache that definitely does not help bd on the desktop.
l3 cache is designed as a compensator for ddr3. It worked for phenom 2...
Posted on Reply
#9
nt300
Hustler said:
"based on the "Piledriver" core architecture, which brings in about 15% IPC increase over Bulldozer."


Great, so that makes them roughly as fast as a Phenom II, but not until nearly 2013....:rolleyes:
Wrong, 25% faster over Phenom II and 15% faster over Bulldozer. In most benchmarks Bulldozer outperforms PII by about 70% to 80% of the time.
mastrdrver said:
I understand what they are saying and I'm telling them that it is not going to happen and never will because of the way AMD designed the module.
http://www.techpowerup.com/forums/showthread.php?t=161031
Posted on Reply
#10
Mulderer
the first brazos amazed me,cheap and quite powerful..
so i'm looking forward to brazos 2.0
Posted on Reply
#11
trickson
OH, I have such a headache
Yo_Wattup said:
*yawn*
I am with you. *yawn*
Posted on Reply
#12
Super XP
trickson said:
I am with you. *yawn*
Would you prefer Intel kill off AMD and AMD kill off NVIDIA :eek:
Then we can kiss Competition GoodBye :eek:
That's how you sound in some of your posts :D
Posted on Reply
#13
trickson
OH, I have such a headache
Super XP said:
Would you prefer Intel kill off AMD and AMD kill off NVIDIA :eek:
Then we can kiss Competition GoodBye :eek:
That's how you sound in some of your posts :D
LOL. There is NO way this will ever happen. You sound like it is the end of the WORLD!
AMD is doing great, Just not a GREAT as they once were is all. Get over this BS of Intel Killing AMD and the end of competition will you?! :mad:
Posted on Reply
#14
Goodman
nt300 said:
Wrong, 25% faster over Phenom II and 15% faster over Bulldozer
Wrong piledriver 15% faster then the FX would put it 5-10% faster then PIIx6... :laugh::rolleyes:

nt300 said:
In most benchmarks Bulldozer outperforms PII by about 70% to 80% of the time.
Maybe but not by that much & as for the rest of the 20-30% PII kicks out Bulldozer butt like there were no tomorrow :shadedshu

Anyhow it's hard to tell if Piledriver will be better than the current FX line , we'll have to wait & see
I don't care if they don't get it out before Sept 2012 as long as they "fixed" the god damn thing , i want higher performance at same clock speed first , after that they can start raising clock speed all they want
Posted on Reply
#15
trickson
OH, I have such a headache
Till the Chips hit the market there will be all the fuzzy math and speculation as to how great it will be. just like the Bulldozer! More paper launching and more hype from AMD is all I see here.
Posted on Reply
#16
xenocide
Goodman said:
Wrong piledriver 15% faster then the FX would put it 5-10% faster then PIIx6...

Maybe but not by that much & as for the rest of the 20-30% PII kicks out Bulldozer butt like there were no tomorrow
Depends on the workload and the model numbers. A Phenom II X6 will crush an FX-6xxx across the board, but still lose out to the FX-8xxx line in heavily threaded applications. If we're talking about per thread performance, Phenom II is about 5-10% faster. PD is basically intended to push per thread performance just past Phenom II, whether or not that actually happens is another story.
Posted on Reply
#17
Super XP
Here is logic. A quote from friend off Hardware Analysis.

A_Pickle said:
I think Bulldozer is pretty good, actually. It's just not a conventional multi-core implementation, which was a big gamble on AMD's part (one which may not pan out) -- but if the software begins to make use of it, which I think it will, we'll see some interesting stuff.

I don't think AMD is really gunning for the top performance crown anymore, and frankly, I think it's ridiculous for people to expect them to try. Intel pulls down billions of dollars more than they do -- they just don't have the resources to compete with that. They need to focus, like they did with the Athlon 64, on what Intel is f**king up -- which is to say, graphics.
Posted on Reply
#18
tacosRcool
I love all the AMD hate here. Everybody seems to forget that they do not compete with what everybody says they do, they compete at lower price points.
Posted on Reply
Add your own comment