Tuesday, June 7th 2011

AMD FX 8 Core and 4 Core Processor Systems Seen Running at E3

At the Electronic Entertainment Expo (E3) 2011, AMD made its revival of the FX brand identifier official. The company steered clear of actually launching anything, but reran the audience through the AMD Bulldozer architecture, something AMD first did way back in August 2010 (yeah, it's been that long!). Knowing the audience needed a lot more than just that, AMD ran live demos of gaming PCs running the new FX series processors, again, without giving away any performance figures.

AMD first showed the final box art design. The box of the eight-core FX Black Edition processor is a classy metal canister, while the quad-core FX chip is housed in a more common-looking paperboard box, the design of which matches the one revealed in a box-art exposé back in March. The gaming rigs shown run the eight-core FX processor on an ASUS Crosshair V Formula motherboard, with Radeon HD 6900 series graphics, with an Eyefinity display setup.
An instance of next-generation AMD Overdrive software is running, displaying a surprisingly low 19°C temperature on all cores. This could be a glitch, probably because AOD doesn't support the sensor interface of the new FX chips properly, yet. The other thing AOD reveals is that each of the eight cores is running on its own BClk multiplier value, ranging from 1.00 GHz (5 x 200 MHz), to 3.20 GHz (16 x 200 MHz). The core voltage for all the cores is displayed as 1.4V, again we suspect a low-level interface glitch.

Source: 4Gamer.net
Add your own comment

178 Comments on AMD FX 8 Core and 4 Core Processor Systems Seen Running at E3

#1
[H]@RD5TUFF
X1REME said:
imho AMD is going to rape Intel (Multitasking, dunno about games or single) this round, the stepping B2 is to make sure it competes with intels incoming cpu`s (speed bumps mostly)... IB wont do much for intel next year, where amd BD has the advantage of major improvements for 2012....:nutkick:
I'll take 2 of whatever your smoking / snorting .. . . it's nice to dream . . .
Posted on Reply
#2
cdawall
where the hell are my stars
inferKNOX said:
You sir, need to study some physics (perhaps during that self prescribed break you speak of in your sig) before proposing such an erroneous statement.
That's the last I'll say on the matter.:)
without a heat exchanger of sorts explain to me how you plan on lowering something temperature. air does not just loose its temp. i kind of deal with air conditioning everyday. cool >800F bleed air down to <-30F. it doesn't happen by magic gotta use something in between to dissipate the heat.
Posted on Reply
#3
inferKNOX
@X1REME:
I've had a silent hope that AMD is holding back the numbers so that Intel won't be prompted to be over-aggressive on preparing their counter-attack and is hence trying to catch Intel... off-guard, so to speak.
I'd realistically estimate it, though, to be more-or-less on par with SB.
Sigh... the sheer suspense!!
Posted on Reply
#4
[H]@RD5TUFF
inferKNOX said:
@X1REME:
I've had a silent hope that AMD is holding back the numbers so that Intel won't be prompted to be over-aggressive on preparing their counter-attack and is hence trying to catch Intel... off-guard, so to speak.
I'd realistically estimate it, though, to be more-or-less on par with SB.
Sigh... the sheer suspense!!
I would like to see the 8 core model come in at around 2-3% faster than a 2600k Sandy Bridge, in my opinion anything less than this is just AMD falling short of the mark. I also have my concerns with how these chips will overclock, with Sandy Bridge chips hitting 4.5+ on air (some as high as 5 Ghz), that's a huge mountain to climb right out the starting gate.

But all the speculation in the world is worth less than this demo, time will tell. I do hope AMD can surprise us all and get a chip that is actually competitive, and cause intel to drop it's prices, but given AMD's history, I am sad to say that is just not likely to happen. :shadedshu
Posted on Reply
#5
largon
kid41212003 said:
No one seemed to notice that even though the frequency for each core did go down, but the voltage did not...
That's because it would not be feasible to make a motherboard that allowed each core to have a dedicated voltage source. No CPU/platform does that, anyways.
Posted on Reply
#6
pantherx12
cdawall said:
without a heat exchanger of sorts explain to me how you plan on lowering something temperature. air does not just loose its temp. i kind of deal with air conditioning everyday. cool >800F bleed air down to <-30F. it doesn't happen by magic gotta use something in between to dissipate the heat.
No need for a heat exchanger.

Thing is he's right the fans needed would have to be phenomenal :laugh: Basically if there is less hot objects in their ( atoms and such) then there will be less heat.

As say you had 20 hot things in something x size. You'd have a tdp of 20 ( sorry for easy numbers, can't be arsed)

Where as if you had 10 hot things in x size you'd have a tdp of 10.


The theory is there it's just not possible with conventional designs.

(think about air high up in our atmosphere, it's cold as there's less air up there)
Posted on Reply
#7
Jizzler
Air temp, yes. But then wouldn't the CPU itself be hotter as there is umm... less air in the air? :)

We need to compress our cases so there's more medium to move the heat away!
Posted on Reply
#8
cdawall
where the hell are my stars
pantherx12 said:
No need for a heat exchanger.

Thing is he's right the fans needed would have to be phenomenal :laugh: Basically if there is less hot objects in their ( atoms and such) then there will be less heat.

As say you had 20 hot things in something x size. You'd have a tdp of 20 ( sorry for easy numbers, can't be arsed)

Where as if you had 10 hot things in x size you'd have a tdp of 10.


The theory is there it's just not possible with conventional designs.

(think about air high up in our atmosphere, it's cold as there's less air up there)
its less dense in the upper atmo hence cooler that wont happen on the ground with conventional fans

largon said:
That's because it would not be feasible to make a motherboard that allowed each core to have a dedicated voltage source. No CPU/platform does that, anyways.
each core is adjustable volts wise on AMD server chips
Posted on Reply
#9
wolf
Performance Enthusiast
X1REME said:
I cant remember AMD saying anything about the Radeon 6 series trashing thier opponents, although they clearly knew that would be the case until nvidia releases 580/590 etc. why?
what exactly did the 6 series trash? don't get me wrong it's a fantastic series. and the only card that really trashes anything is the 6990. the 6970 is on par with a GTX480, hardly a trashing. AMD are kicking ass with the top top card (dual GPU) and thats it as far as brute force GPU's go.

the perspective of trashing in AMD's recent years is all about what you get for your money, there they are making a killing.

[H]@RD5TUFF said:
I'll take 2 of whatever your smoking / snorting .. . . it's nice to dream . . .
some for me too! gotta share bro :pimp:

thing is at least for most of us here at TPU, were either enthusiasts (read: overclockers, or at elast call it optimisers if you will)or at least 'in the know', I just dont think BD will be for us. again I feel they will take the value segment of the market and have utterly staggering performance for your hard earned cash, but when it comes to the fastest, SB and IB will still be on top IMO. especially with overclocking added to the mix.

Like you say in a later post, most 2500/2600K's hitting 4.5ghz ++ on air is something AMD is going to have a ridiculously hard time matching, and trying to sell their unlocked chips as competition to that...
Posted on Reply
#10
seronx
wolf said:
what exactly did the 6 series trash? don't get me wrong it's a fantastic series. and the only card that really trashes anything is the 6990. the 6970 is on par with a GTX480, hardly a trashing. AMD are kicking ass with the top top card (dual GPU) and thats it as far as brute force GPU's go.

the perspective of trashing in AMD's recent years is all about what you get for your money, there they are making a killing.



some for me too! gotta share bro :pimp:

thing is at least for most of us here at TPU, were either enthusiasts (read: overclockers, or at elast call it optimisers if you will)or at least 'in the know', I just dont think BD will be for us. again I feel they will take the value segment of the market and have utterly staggering performance for your hard earned cash, but when it comes to the fastest, SB and IB will still be on top IMO. especially with overclocking added to the mix.

Like you say in a later post, most 2500/2600K's hitting 4.5ghz ++ on air is something AMD is going to have a ridiculously hard time matching, and trying to sell their unlocked chips as competition to that...

and that is why AMD 32nm SOI processors use

http://www.intel.com/technology/silicon/high-k.htm

http://www.maximumpc.com/article/features/fast_forward_hkmg_masses

HKMG
Posted on Reply
#12
seronx
wolf said:
yes I knew that, but that in itself does not guarantee massive overclockability like SB already has. AMD has a lot of work to do to appeal to the enthusiast-extreme overclocker.
They had 5 years to figure out what to do and how to do it

2007 Bulldozer the architecture was annouced

2009 was when Bulldozer was going to release with 45nm SOI HKMG but since GloFo decided not go for mass production of 45nm HKMG AMD had to wait for the next SOI version of HKMG (45nm could have been used for ES samples though)

2011 5 years since Bulldozer announced
Posted on Reply
#13
Velvet Wafer
wolf said:
yes I knew that, but that in itself does not guarantee massive overclockability like SB already has. AMD has a lot of work to do to appeal to the enthusiast-extreme overclocker.
Lets see... AMD already reached 4-4.2 ghz maximum clock, for a good percentage of their processors (since 965 c3) even if NONE of them sports HKMG.... so i guess, with it, the maximum clockability will be quite noticeable.... given, that it works correctly, and as intended (like in Intel processors)
5 Ghz doesnt sound SO unrealistic anymore, if you look at that fact;)

Only thing that we dont know, which is even more important, is the performance clock for clock... i suppose for BD beeing a little weaker there, than SB, especially in singlethreaded apps.

But especially in multicore apps, the Technology implemented by AMD is supposed to bring much greater Benefits, than the Hyperthreading Technology by Intel.
It even was suggested by Anand, that Bulldozer maybe has AVX instructions, like SB, due to the pair of 128bit fmacs on die.

Then, there is Magny Cours....Just look at how far they already came, with 12 cores... dont you think that the technological sucessor, will sport even greater performance, with even less power consumption? Remember, even Magny Cours, was not HKMG.
Its still ordinary 45nm SOI;)
Posted on Reply
#14
pantherx12
cdawall said:
its less dense in the upper atmo hence cooler that wont happen on the ground with conventional fans
I know man.

I even said it in that post.
Posted on Reply
#15
seronx
Velvet Wafer said:
Lets see... AMD already reached 4-4.2 ghz maximum clock, for a good percentage of their processors (since 965 c3) even if NONE of them sports HKMG.... so i guess, with it, the maximum clockability will be quite noticeable.... given, that it works correctly, and as intended (like in Intel processors)
5 Ghz doesnt sound SO unrealistic anymore, if you look at that fact;)

Only thing that we dont know, which is even more important, is the performance clock for clock... i suppose for BD beeing a little weaker there, than SB, especially in singlethreaded apps.

But especially in multicore apps, the Technology implemented by AMD is supposed to bring much greater Benefits, than the Hyperthreading Technology by Intel.
It even was suggested by Anand, that Bulldozer maybe has AVX instructions, like SB, due to the pair of 128bit fmacs on die.

Then, there is Magny Cours....Just look at how far they already came, with 12 cores... dont you think that the technological sucessor, will sport even greater performance, with even less power consumption? Remember, even Magny Cours, was not HKMG.
Its still ordinary 45nm SOI;)
BD should have the same as SB to 4x SB

IPC = Integer Performance(ALU) (Right?)

A BD Module has 2 ALUs per core

I read somewhere not sure where that the SB only has 1 ALU per thread or was it just 1 ALU all together?
Posted on Reply
#16
Velvet Wafer
seronx said:
BD should have the same as SB to 4x SB

IPC = Integer Performance(ALU) (Right?)

A BD Module has 2 ALUs per core

I read somewhere not sure where that the SB only has 1 ALU per thread or was it just 1 ALU all together?
I dont think we can be really sure about that, as said, until there are benches to actually throw in some facts into the specualtion... but the signs are not bad, even tho, they arent too good either:D

IPC is Instructions per Clock if im not wrong :)

I guess what you mean is, that BD has only one FPU per 2 core module, so 4 total for 8 cores shared?
Posted on Reply
#17
seronx
Velvet Wafer said:
I dont think we can be really sure about that, as said, until there are benches to actually throw in some facts into the specualtion... but the signs are not bad, even tho, they arent too good either:D

IPC is Instructions per Clock if im not wrong :)

I guess what you mean is, that BD has only one FPU per 2 core module, so 4 total for 8 cores shared?
SB only has 1 FPU per 2 threads as well

FLOPC is the same as SB at 256bit while in 128bit FX is 2x that of SB


(Problem #1 with this slide is it referring to 1 Flex FPU or all 4 Flex FPU in Zambezi)

http://en.wikipedia.org/wiki/Instructions_per_cycle
Calculation of IPC

The number of instructions per second for a processor can be derived by multiplying the instructions per cycle and the clock speed (measured in cycles per second or Hertz [Hz]) of the processor in question. The number of instructions per second is an approximate indicator of the likely performance of the processor.

The number of instructions executed per clock is not a constant for a given processor; it depends on how the particular software being run interacts with the processor, and indeed the entire machine, particularly the memory hierarchy. However, certain processor features tend to lead to designs that have higher-than-average IPC values; the presence of multiple arithmetic logic units (an ALU is a processor subsystem that can perform elementary arithmetic and logical operations), and short pipelines. When comparing different instruction sets, a simpler instruction set may lead to a higher IPC figure than an implementation of a more complex instruction set using the same chip technology; however, the more complex instruction set may be able to achieve more useful work with fewer instructions.
Posted on Reply
#18
[H]@RD5TUFF
seronx said:
They had 5 years to figure out what to do and how to do it

2007 Bulldozer the architecture was annouced

2009 was when Bulldozer was going to release with 45nm SOI HKMG but since GloFo decided not go for mass production of 45nm HKMG AMD had to wait for the next SOI version of HKMG (45nm could have been used for ES samples though)

2011 5 years since Bulldozer announced
They spent a similar amount of time on Phenom 1 and that was a huge success .. . . oh no wait it wasn't.

Just because you give a monkey 5 years to do something doesn't mean he is going to be able to paint the Mona Lisa.

Again you can speculate all you want and talk about how AMD has some small architectural changes that will make little real world difference, but no benchies no proof!
Posted on Reply
#19
seronx
[H]@RD5TUFF said:
They spent a similar amount of time on Phenom 1 and that was a huge success .. . . oh no wait it wasn't.

Just because you give a monkey 5 years to do something doesn't mean he is going to be able to paint the Mona Lisa.

Again you can speculate all you want and talk about how AMD has some small architectural changes that will make little real world difference, but no benchies no proof!
Phenom I had cache issues(latency(L3->L2) and size(L3)) and silicon leakage issues(highest clock they could get is 2.6GHz (Stock)) and they also had to release this CPU to even get a competing cpu out

Phenom II fixed them while still having the same architecture

K10 also was in the bleak era of AMD (Scandals, Debts, and the usual)

Bulldozer is going to be in a profitable era(Intels profitable era Core 2(Wolfdale/Yorkfields) -> i7(Nehalem) -> i7(SB))

K8 Refresh announced 2006
K10(K8 Refresh) happened in 2007 <-- Production(But not optimal production rate)
1 year later Phenom I came out
2 years later Phenom II came out
4 years later 3.7GHz 980 BE II

New Architecture happened in 2007 <-- Announcement of the BD Architecture
Bulldozer happened in 2009 <-- Production(but not optimal production rate)(No HKMG)
3 years later first gen Bulldozer Enthusiast only FX
4 years later Bulldozer goes mainstream Trinity APU

It is only common sense a monkey learns to fix something once it gets punished

:pimp:
Posted on Reply
#20
trickson
OH, I have such a headache
I call all speculation BS I want real bench marks ! NOW AMD ! YOU HEAR ME ? NOW !
Posted on Reply
#21
xBruce88x
erocker said:
There's no way it's cooler than 66F in that convention center. It's most likely 72-80F. Looks like they're still using the old broken temp sensors.
If it were cooler than 66F, it would explain why the guy is wearing a long sleeve shirt this time of year...
Posted on Reply
#22
trickson
OH, I have such a headache
19c on all cores ? I mean come on man give me a break . I water cool my shit and can't get that ! Even if they are cooler running there is just no way in hell you are going to get an 8 core CPU to run at 19c with that POS HSF on it ! WTF do they take us for any way ? I have to say this is just pure BS ! Also why are the frequencies of the cores all over the place ? I see them as low as 983MHz and the temp is still 19c ? man I can not believe this at all ! good god even the multipliers are all over the place !
Posted on Reply
#23
mastrdrver
The clocks and multipliers all over the place is how it is suppose to work.
Posted on Reply
#24
seronx
trickson said:
19c on all cores ? I mean come on man give me a break . I water cool my shit and can't get that ! Even if they are cooler running there is just no way in hell you are going to get an 8 core CPU to run at 19c with that POS HSF on it ! WTF do they take us for any way ? I have to say this is just pure BS ! Also why are the frequencies of the cores all over the place ? I see them as low as 983MHz and the temp is still 19c ? man I can not believe this at all ! good god even the multipliers are all over the place !
Set your house to 64F and turn off your cpus overclock and turn on the intel's equivalent of Cool'n'Quiet, I think its called E1C or Speedstep

I haven't used intel since Pentium 4 so I wouldn't know the names, You should get the same results
Posted on Reply
#25
trickson
OH, I have such a headache
mastrdrver said:
The clocks and multipliers all over the place is how it is suppose to work.
Not how want mine to work . If one core is 3.2GHz then 16c would not be right at least it don't seem right . Looks like an up and down roller coaster ride to me . Wouldn't the temps be just as fluctuating as well ? I do know but to me this seems like a load of BS is all . I hope they come out soon so we can all see just what they are doing in real time real life apps . All this jumping the gun and looking at some half bald china man running the chip with windows system monitor ONLY seems a bit like giving a porn video with every one waring clothes ! NO MONEY shots ? ! WTH ? :cry:
Posted on Reply
Add your own comment