AMD Bulldozer Thread

[i5L4ND]

Cant think of anything positive, to say about these chips at all.

[SMTB1963]

yes, people missed a lot. maybe this is what you should read next: "in the face of clear evidence to suggest to the contrary, why do some people remain so optimistic about the future?" http://medicalxpress.com/news/2011-1...c-reality.html

I've got news for you: rumor and speculation don't rise to the level of "clear evidence". Maybe this is what YOU should read next...

[Hornet331]

So now its offical... AMD produced a netburst like µarch and thats a failure on the consumer base...

Hell its P4 vs P3 all over again, if you look back then P3 also was better in many apps/games and offered better IPC and consumed less (despite beeing on a higher node)... Wow you really think someone won't reapeat the same mistakes your competitor makes...

The real bummer is game preformance and power consumption... its not faster then a X6 and consumes more...

[FischOderAal]

Goddamnit I really hoped BD would be at least close to Intel's offerings in the fields I care about (gaming / power draw).

I wanted my next rig to be all-AMD. Well, I think I still have some time to kill before I need to upgrade.

[h0bbes]

It's quite disappointing. I certainly expected more and was looking forward to greater competition. Oh well, best of luck to AMD

[SoF]

Funny to see everybody that surprised - what have you expected after an epic marketing stunt with 8.4+ GHZ but no other results?
10 days back was the first time I saw a bulldozer. We ran some 3DM11 and remarked the low CPU-scores instantly...running near 6.5 we all were like WTF?!
Then we did some AM3 - something around 270k LOL! It's probably the worst new CPU-generation I've ever seen

[Mechromancer]

As a PC gamer I am very disappointed. AMD spent the last half decade working on a CPU that is a good deal slower than their previous generation. Bad call AMD! Now the ousting of Dirk Meyer and all the execs jumping ship make sense. It's sad that we all felt this coming and here it is. I have to got to work today and tell my coworkers that I have to build a gawdamn Intel rig just get play my current lineup of games! The Witcher 2 does not show mercy to chump CPUs man. Neither does ArmA II Operation Arrowhead or Total War: Shogun 2 (which Bulldozer crashes on for some reason).

AMD needs to make a separate product line for the desktop like shrinking the 1100T, adding two more cores, and tweaking power efficiency. Hopefully the GPU side of AMD can keep things going there. The 7000 series is their only real hope.

[Kai Robinson]

Yep - AMD, you screwed the pooch on this. So much for going back to the AMD camp when i upgrade. I don't want more cores, i want more IPC.

[bamtan2]

Hopefully the GPU side of AMD can keep things going there. The 7000 series is their only real hope.

it cannot. they cannot produce enough llano, their new server chips suck, and their new enthusiast chips suck. they are not a rich company, and their sales will continue to falter all next year. this has happened before to tech companies. there is a limit to how many times a company can screw up their sales for a given amount of debt / cash on hand. they cannot run on fumes much longer. I will not be surprised if AMD does not exist in 2013.

[Hornet331]

Funny to see everybody that surprised - what have you expected after an epic marketing stunt with 8.4+ GHZ but no other results?
10 days back was the first time I saw a bulldozer. We ran some 3DM11 and remarked the low CPU-scores instantly...running near 6.5 we all were like WTF?!
Then we did some AM3 - something around 270k LOL! It's probably the worst new CPU-generation I've ever seen

THe problem is we didn't saw anything till now (well the leaked stuff, which where decleard fake by certain people )

For myself I am not really that surprised, but the clock for clock comparrisons are horrible up to 40% less performance then previouse generation? Holy moly, did AMD hope that they can gain 40% more clock compared to the X6?

Btw. where is movieman? I am quite intrigued what he has to say...

[kromosto]

Bulldozer is the confession of AMD being out of the competition.

[xdan]

First of all, some people here who offended Monstru and lab501 should give apologies.
Even OBR need aplogies, he told the truth.
JF AMD should be banned for all his lies
Second the Bulldozer is a an EPIC FAIL, crap whatsoever


Why, let' see....

Very less IPC than Thuban even with 8 cores.

For a single instruction thread, Bulldozer offers more front end bandwidth than its predecessor. The front end is wider and just as capable so this makes sense. But note what happens when we scale up core count.
Since fetch and decode hardware is shared per module, and AMD counts each module as two cores, given an equivalent number of cores the old Phenom II actually offers a higher peak instruction fetch/decode rate than the FX. The theory is obviously that the situations where you're fetch/decode bound are infrequent enough to justify the sharing of hardware. AMD is correct for the most part. Many instructions can take multiple cycles to decode, and by switching between threads each cycle the pipelined front end hardware can be more efficiently utilized. It's only in unusually bursty situations where the front end can become a limit.

Compared to Intel's Core architecture however, AMD is at a disadvantage here. In the high-end offerings where Intel enables Hyper Threading, AMD has zero advantage as Intel can weave in instructions from two threads every clock. It's compared to the non-HT enabled Core CPUs that the advantage isn't so clear. Intel maintains a higher instantaneous decode bandwidth per clock, however overall decoder utilization could go down as a result of only being able to fill each fetch queue from a single thread.

So that's the 180% performance scale using CMT compared to Intel HT.
In fact 2cores(1 module) has less performance than Intel 1core+ HT. Performance may be 80%

The numbers in this table speak louder than words. The performance of Bulldozer microarchitecture has become way lower than that of the previous-generation processors. The simplification of Bulldozer microarchitecture by combining a pair of cores into a single module with shared resources led to a significant (25-40%) drop in specific performance compared with the previous-generation AMD microarchitecture. As a result, Bulldozer cores do not just work at half the speed of Sandy Bridge cores. In addition to that the performance of the Bulldozer processor module with two cores is even lower than that of a single Sandy Bridge core with enabled Hyper-Threading technology. Should we expect any performance records from a CPU with such microarchitecture? This is more of a rhetorical question…

Last drop which filled the cup...
Overclocking and power consumption....Performance/watt...

The results are shockingly bad to say the least. On overclocked full load, the power usage of the slower FX-8150 was nearly TWICE that of the faster i7-2600K. We haven't seen such monstrous power usages except in some dual socket systems or extreme overclocking (which ours isn't). We redid the tests multiple times using different motherboards and power measuring equipment but the results were the same everytime. Using a multimeter, the 12v rail on the extra 8+4pin CPU connectors also measured a whopping ~23A. With overclocked power loads like these, its no wonder AMD has to ship a water cooling solution together with the chip.

[Leeghoofd]

I will not try 7Ghz 8 cores under load with the press kit, maybe someone else will still need it for a review... Brian will probably be not afraid.

I had high hopes for a performance boost over the early silicon, even being aware months back that the CPU was not ideal for 3Dbenching, it was kind of a bummer. Even at 6.5Ghz I had like 9850CPU score in 06...sigh... Wprime was awfull too sigh x 2.. i won,'t talk about superPi as I will get flamed at :p so I stopped wasting LN2 to go or higher clocks...

BUT : For gamers this is not a bad CPU at all. OC it lightly (4.5-4.8) and it will serve you well ( it's not all about raw performance as the GPU is the limiting factor once you cranck up the detail and Res ) If the soft is adapted, it can be a good encoding CPU...

But the prices have to be far lower then what they are at... 2500K has dropped to 185 euros, so the FX8150 has to be at 175 or similar in a few weeks ( unless they included the all in one cooling setup )... also 990FX boards are not cheap either...

I hope Piledriver is at least 20% better then Zambezi...

[Micutzu]

Attachment 121183

So by 2013 when Intel will launch Haswell, AMD will be back at Phenom II level IPC.

That looks very dissapointing if you take Moore's Law into consideration.

[gosh]

it cannot. they cannot produce enough llano, their new server chips suck

For servers bulldozer looks really good!

[SoF]

Who the F*** cares about SuperPi???

There is enough old software out there based on classic x86 performance - why we should not care?

How bout 6.5Ghz vantage or '06 8C/8T???
Just hope it can compete with 5.7Ghz 2600k 4C/8T

Haha not even close - not sure about vantage but 06 is total fail

JF AMD should be banned for all his lies
Second the Bulldozer is a an EPIC FAIL, crap whatsoever
[IMG]2vnhzra.gif[/IMG]

Haha one GIF tells everything about Faildozer

[Tao~]

Point #1 : Monstru, you were right

Point #2 : Bulldozer WILL gain in performance with newer OS / kernels . I have a feeling this is what is happening: the OS is unaware of the two integer cores in a module, so one of them is being underutilized - the shared L1 between the two might be causing cache invalidation problems (as looncraz reported). Overall, the two integer cores end up performing worse than a single core, which in turn might be smaller than the single integer unit in K10.5
If these two problems get fixed Bulldozer should gain atleast +10-20% performance in Integer operations. Now why the FPU is in such a bad shape I have no idea. I would love to see an indepth analysis of the architecture and where they might have gone wrong (yeah, maybe AT will do it soon ) Probably they are waiting for software updates. Hopefully Piledriver will see a better time when it releases six months later. Bulldozer, as it is, is DEAD.

[Hornet331]

For servers bulldozer looks really good!

Yeah because we have seen so many benches.. oh wait Linpack which is a HPC benchmark right now shows abysmal performance... are you really gona try to produce a hype again... Wait and see till the actualy product is here before spouting nonesens.

[Monstru]

Guys, if you think about it, in the end the biggest fault is the marketing's team fault. Ok, the engineers got into some problems along the way, and things haven't turned out like they should. Nevertheless, for sure this must be the base of something better, otherwise there is no point in launching it. So somewhere along the way, we will see something better based on this arch. Dunno when, dunno how much better, but this has to be a starting point for something.

But...in this situation, at least NOT name this FX, and at least put a price tag 10-20 Euros lower then 2500K if you want to be in any way competitive and if you do not want to deceive the buyers...If FX-8150 would have been 20 Euros cheaper then i2500K from the start, then the only problems would be the power consumption and the comparison with Phenom II. Still much better then how this turned out in the end...

[gallag]

For servers bulldozer looks really good!

And here we go again, the same hype from the same people.

[Hornet331]

I said for ages dessau needs to go.. but nope he is still there...

srsly someone needs to slap amds marketing devision:
http://www.youtube.com/watch?v=8rDwXuAINJk

[Kristoferr]

I said for ages dessau needs to go.. but nope he is still there...

srsly someone needs to slap amds marketing devision:
http://www.youtube.com/watch?v=8rDwXuAINJk

Trying to make pile of look like a bread. Thats what it is.

[kik]

Regarding that video: http://i.imgur.com/H8e92.jpg

[STaRGaZeR]

I was going to write a somewhat long post about BD, but I think it's not even necessary, so I'll do it in 3 words:

AMD Bulldozer sucks.

If you see Hans de Vries, JF-AMD, or any of the "big" guys posting anything about this, tell us!

[AliG]

Well.... I guess all we can say at this point was that silence and delays just about always = a bad thing