HardOCP Readers Ask AMD Bulldozer Questions

This part made me laugh out loud..."The new CPU core delivers higher frequency while maintaining IPC,"

They think it's a good thing by increasing MHz but not the IPC...

Well.. Does it actually increase MHz? Going from 4.2Ghz (which seems to be what a lot of the PIIX6's could hit) to 4.5GHz is not really THAT much of an improvement, given the lower IPC, increased cost in power and purchase price, and increased cooling required, which means a noisier PC
 
Having thought about this for a few hours I'm still struggling to understand why AMD answered these questions. How could they possibly think it would be beneficial to come to a tech enthusiasts website with marketing spin, hype, and question evasion? I don't get it :confused:
 
I make a prediction that Amd will go the way of the Gaming industry. Large volume sale. Aka
Consoles, handheld devices and entry netbook , tablets and laptops. Why ? Because thats where the money is.

Goodbye enthusiast videocards! 7 to 10 years at most.

Nvidia's already got Tegra.

Not predicting the end of computers.... just how it'll evolve and change.
It's not personal, it's business.

Problem is, BD is sure as heck not well suited for that, even scaled down... Llano/Zacate are ok but still based on older designs and they don't scale well beyond budget systems either. Intel's in a better place to move in on ARM's turf, and they're already doing it. Intel's gonna keep letting AMD play catchup in the desktop (and budget laptop) space without completely obliterating them in a price war because it suits them, it allows them to maintain a higher profit margin and it keeps the government at bay.

I just don't know if AMD can keep treading water forever... They gotta find a better way to tread, heh, maybe a long term future looking server play is the ticket, who knows. Enthusiasts are definitely not the driving force we once were tho, at least not as far as desktop sales (because let's face it, desktop sales decline every year), so while the Q&A might've seemed like a nice gesture it's pretty irrelevant to AMD's broad strategy.
 
Last edited:
I'm kinda torn here.... i know northern island will be good.... graphic parts haven't failed to disapoint.
but when LLano ipc will beat pile driver's.

I'll have a tear running down my cheek on the side of a dusty road.
 
I just can't understand why a chip that is supposedly optimized for server workloads can be so power-hungry.

Also, if multithreading is the focus here, why isn't Bulldozer much better than older 4-core or 6-core parts with its 8 fancy new cores. It just doesn't add up. If multithreading was the goal, it should be better at it than it currently is.

Also, looking at this...

7. Why would I buy a $275 Bulldozer cpu when the $170 1090t seems to equal its performance or actually do better at every benchmark and game we've seen?

Adam Kozak, Product Marketing Manager, AMD - We understand our customers make purchase decisions based on how they use their PCs, and in many cases our AMD Phenom™ II processors are a great (purchase).

For those ready for a more modern architecture, who want a desktop for high resolution gaming and to tackle time intensive tasks with newer multi-threaded applications, the AMD FX processor is a great upgrade.

... makes me want to say one thing: New is not better, better is better!

Just give us a 32nm die shrink of the Phenom II please!
 
Last edited:
I would have seriously bought a BD if they would have cut the crap and been straight with us and said, this chip is way too ahead of it's time, but we'r trying our best to get developers to optimize software for our chip in the meantime. We over estimated the market, please give us another chance. The last thing any of us wants to see, no matter who's camp your in is a complete cpu monopoly, which will ultimately only lead to higher prices and a stagnation in innovation. Sorry if we dissapointed anyone, just realize we gave it our all and in no way are we throwing in the towel. Rest assured that AMD is committed and suited for the long haul and last but not least i want to wish all of our die Hard Forum fans a happy holiday.
 
Last edited:
Wow! Maybe AMD needs to follow up with those answers with some meat and potatoes. Showing is more convincing then telling, kinda a let down to OCP community with those shabby non-clear answers. Future software? Show us some, at least redone benchmarks working great with BullDozer, I am sure there are some at least, even if internal to AMD.

Win7 improvement for scheduling is needed ASAP, I know from my own testing configuring programs with Bulldozer can really make things improve. At least show some Linux results when configured for Bulldozer which supposenly is good. Good to here there maybe an update shortcoming: Good news from AMD on this.

I really don't think Bulldozer is going to be around for long, less then a year, would be nice to know if a future upgrade will be possible with 990fx chipsets.

Still most of the software out there is not going to be rewritten for Bulldozer, so the thought that it will perform better with future software to me tells everyone it doesn't perform well with today's software :eek:.
 
so people had to wait a ridiculous amount of time to get Bulldozer. now they have to wait for the os and apps to be better suited for it? basically its always a waiting game as for as their cpus are concerned. I just scratch my a head when I see ignorant people actually "upgrade" to this cpu from a Phenom X6 or X4 especially for gaming.
 
I know a few Intel groupies that are highly disappointed with the 2700k's modest bump in performance, but in all honesty can you blame Intel. It's not like they felt any heat to increase the gap, they'r only competition for miles id their own chip and at least they slightly improved their last one and the price bump, while hardly a bargain, is not completely unreasonable for a new platform. Bulldozers price needs to do exactly what the name implies, because the only thing I currently own that has no use for me in the present but will come in handy in the future is a casket.
 
Wow the interview bit where the tough questions are asked are answered so stupidly, I wonder if Adam Kozak knows that what AMD had going for them (price/performance) is not happening.

Anyone at AMD can see this anyone not from AMD can see this yet we get answers that aren't answers but a "deep" traumatic response. As if someone at AMD will tell it how it is that the desktop part is overpriced and not due to popular demand rather then manufacturing problems.

It is like the old west where beads and mirrors were traded for far exceeding valuables ....
 
The level of non-answer and marketing BS in these responses is hilarious. Expected better after a letdown like Bullshitdozer. Still, it's good that it was posted since it shows us AMD's unwillingness to just come out and say "Ok, we messed up here and here, and this is how we're fixing it going forward." Instead they want to pretend like the CPU is just awesome and that its everyone elses fault for not being hipster enough to take advantage of it properly.
 
Just read this review over in another forum and it was so spot on that I just had to repost the quote:

"It never works that way. I've been overclocking since the IBM XT was out (got it from 4.7Mhz to 8.x Mhz), and I have yet to see a time when the supporting software caught up with the new instruction sets while I still had the "new" CPU. I was the first to get a 32-bit processor, waited years before NT technology took advantage of it, then my chip was old. I had the first AMD with 3DNOW, waited for games to support it, then my chip was old. I had the first 64bit processor, XP 64bit was a flop and had to wait for Vista 64, then my chip was old.

The only time this ever worked for me was with the i7 920. My longest running chip between upgrades, still have it three years later, blazing at 4.2Ghz with apps that can use it's cores.

Buy what's fast today. Don't worry about tomorrow. AMD dropped the ball on BD
."
 
I think we may be missing the biggest point yet. From my understanding AMD cherry picked the questions they would answer from all submitted! Think of how much more spin there would have been if they tried to answer only difficult questions.
 
I think we may be missing the biggest point yet. From my understanding AMD cherry picked the questions they would answer from all submitted! Think of how much more spin there would have been if they tried to answer only difficult questions.

:D maybe in this case they should have answered less questions then they did.
 
hell yeah, kick the guy while he's on the floor

and why so much whining over the product you're not gonna buy?


Gabe Gravning, Senior Product Marketing Manager, AMD -

The question is not "will this transition to multithreaded applications happen" but rather "how soon?" AMD drove this same type of inflection point in the industry with 64-bit computing and APU platforms.

You could have at least suggested developers working on AMD Evolved titles to give us 64-bit executable every now and then.
Instead of what... Far Cry was it?
 
Last edited:
hell yeah, kick the guy while he's on the floor

and why so much whining over the product you're not gonna buy?




You could have at least suggested developers working on AMD Evolved titles to give us 64-bit executable every now and then.
Instead of what... Far Cry was it?

LOL...I think your missing the point. Were whining because we WANTED them to succeed and buy them, but not after reading the reviews.

I agree on your second point as well (cough), if those asshole software developers would just code specifically for AMD/BD 64 bit processors then we wouldn't be having all these issues! :rolleyes:
 
I think the age old dilemma "which came first, the chicken or the egg?" applies here. Which came first, the software or the hardware? If you wrote the software first, you wouldn't have the hardware to support it. If you have the hardware first, there isn't software to support it.

I think the latter statement applies to AMD, in order to make multithread application faster, the architecture needed to change. Current software hasn't been written to take advantage of the architecture changes. This is where AMD "dropped" the ball. AMD didn't focus on supporting software developers to get the software out at about the same time as their hardware. Although, I think they had intended to cater first iteration BD chip to the super-computing crowd and hoping that the second iteration BD chip will be available when the software is out there written for it.

How AMD supports software development from now on will greatly determine if AMD will succeed.
 
Athlon 64 which implemented x86-64 instructions was released in 2003,

And all we got is:

Far Cry
Crysis Warhead
Chronicles of Riddick: Escape from Butcher Bay
UT2004

and some forgotten builds of HL2 and STALKER:SoC
(feel free to add if I missed some garage game :))
 
Too many whiners on this site, must be Democrats....

If you don't like Bulldozer, buy Intel.... But by all means please stop whining... it's just a CPU.
 
Athlon 64 which implemented x86-64 instructions was released in 2003,

And all we got is:

Far Cry
Crysis Warhead
Chronicles of Riddick: Escape from Butcher Bay
UT2004

and some forgotten builds of HL2 and STALKER:SoC
(feel free to add if I missed some garage game :))

ahh, there are a lot more than that, just type in 64bit games into google.
Also games are not the only things that are 64bit, i'd say 99% of the apps I use are 64bit.
 
I'm glad that everyone here was just as offended at these responses as I was.
 
Gabe Gravning, Senior Product Marketing Manager, AMD
Well there's your problem. I have a marketing degree and worked briefly in a marketing capacity for a few years. I learned to put a positive spin on unpopular products. It's the nature of the beast that a marketing person isn't going to give a straight answer. Their job is to make the product look attractive.
 
I think we need to put this thread to rest.
Amd didn't up to the expectation and hype it generated.

And this confirms that Amd's has given up the being competitive vs intel.

http://www.tomshardware.com/news/AMD-APU-Z-Series-ARM-Tegra-3,14114.html

QFT
"Our strategy is to accelerate our growth by taking advantage of our design capabilities to deliver a breadth of products that best align with broader industry shifts toward low power, emerging markets and the cloud."

I hate being right!

On the other end.... HD 7000 is right around the corner... good old ATI!! i mean Amd....
 
Thanks AMD, i am going to skip the whole 7000 line now like i did with the 6000. When the 8000 line comes out and the CPU's isn't fixed by then, i am going to be rocking Intel and Nvidia.
 
fuck you amd and your marketing answers, we're not dumb.
 
Too many whiners on this site, must be Democrats....

If you don't like Bulldozer, buy Intel.... But by all means please stop whining... it's just a CPU.

It's a forum for discussing exactly this kind of thing... Should there just be a big silent vacuum when a company releases a turd? Heck no. If you don't wanna read it you can just not click on the thread/link y'know.

Frankly, if AMD's really trying to claim that BD is waiting for software developers to catch up, then that's even worse than the alternative (like say, they screwed up and miscalculated). Their track record in working with developers on the GPU side is abysmal, so that's not a good sign for BD (ignoring all the other logistics against it for a moment).

The 6950 sold in spite of their laggard driver development, I bought two, the base silicon was still a good value. BD isn't even remotely good enough to overcome any software bias against it... And unfortunately CPU architectures are usually in place for much longer than GPU architectures.
 
I'm amazed that someone didn't seem to have a word with the AMD execs to say "err look guys this is a tough crowd and any hint of marketing BS wont fly here!"

Didn't they read through their answers or do a little research into how this whole thing came about?

Of course not.

What this really shows is that AMD/Intel etc. really don't have a clue how to handle the enthusiast market or just don't care about it. If that's not clear to folks here then it never will be.
 
Well there's your problem. I have a marketing degree and worked briefly in a marketing capacity for a few years. I learned to put a positive spin on unpopular products. It's the nature of the beast that a marketing person isn't going to give a straight answer. Their job is to make the product look attractive.

The point is that we need a cpu with good price/performance not marketing people explain (or better put not explain anything) to us what they want us to think.

I never seen such stumbling , not even in "The Keystone Kops".
 
Neither really needs the enthusiast market much anymore tho.... We don't drive sales like we used to, it's all about mobile now, portability, form factor, battery life, etc. The performance battle on the consumer space is not even in the top three as far as priorities go. Difference is, Intel has the luxury of keeping enthusiasts semi-happy by throwing us a bone with the K series and whatnot... Whereas every mistake for AMD just puts it deeper into the sinkhole.
 
The problem with the answers is that AMD chose to ask Marketing/PR guys to answer some of the questions when most of the information folks here wanted was technical.

Even the answers from the AMD tech guys look 'smoothed over' by the PR team. As others have mentioned had they just been straight they would have still held onto some respect.

There was no need for Adam Kozak etc. to poke their noses in.
 
ahh, there are a lot more than that, just type in 64bit games into google.
Also games are not the only things that are 64bit, i'd say 99% of the apps I use are 64bit.

Nope. There aren't alot more 64 bit games.
Out of hundreds of games released each year, that's pretty much it.



Athlon 64 which implemented x86-64 instructions was released in 2003,

And all we got is:

Far Cry
Crysis Warhead
Chronicles of Riddick: Escape from Butcher Bay
UT2004

and some forgotten builds of HL2 and STALKER:SoC
(feel free to add if I missed some garage game :))
 
Clearly, the PR folks filtered which questions were suitable for engineers to answer. Or, the engineers realized they couldn't answer truthfully without blatantly trashing the product, so they redirected some questions to the PR guys.

I found the designer's answers fair. The man shouldn't put down his company's product because his words can easily find their way back to a major customer during a bid or negotiation (and these major customers matter far more to AMD than the entire HardOCP readership). The marketing dudes probably feel that the Opteron brand still has sufficient inertia to plow through some of BD's bad press and therefore don't want their engineers adding fuel to the fire. It sounds like he did the best he could, and when I read between the lines it was pretty obvious to me that he understands our disappointment.

As for the PR answers, frankly, I could've done better myself. It's like they don't even know about all the bad reviews.
 
Well we dont have any multi-threaded games, so getting and 8 Core CPU, when thier is zero games that utilize them is kinda what point ??

This release is their business model not for gamer's. AMD made claim on it for gamers by boosting the speed beyond intels, But that claim is a mute point as speed on a single core threaded games ( which is all our industry currently has ) is kinda piss poor.

IPC boost for single threaded items, intels hyper-threading is better then AMD's current instruction set. AMD is not able to compete vs intel on a single core basis, they are trying to move to get developers into multi core, and stop all the single core old ways of programming. We have had dual core and up CPU;s for a while now, many many years,

Yet we still dont have a single game that can run native multi-threaded yet, so dont blame AMD for all the issues people discussed here.

When the rest of the development world catches up and every thing works as multi core - multi threaded pure hardware/software, will we see new things happen, But with x86 that industry standard framework everyone uses, is piss poor for such actions to occur.
 
I think answering the questions that were directed towards some of the issues with performance we've seen should have been handled differently. Clearly, many of us aren't new here (excuse the join date, I've been reading [H] for damn near a decade) or to CPUs and architecture. I think they could have provided answers without admitting mistakes. It's one thing to say "this is where we screwed up and miscalculated" and another to say "this is where we're expecting performance increases and this is how." Other than the win7 scheduler, I saw very little of the latter.

They could have answered some of these questions that were obviously part of our original questions, like...

is the branch prediction in bulldozer modules up to par considering the depth of the pipeline?

What about the cache latency issues? Is a clock speed bump in order to increase some of the issues we've seen with cache latency or is there another way around that? Is it understandable considering the new architecture and a "give and take" sort of deal? What about the slip in L1 cache size? Can we expect an increase there or are you sticking to the L1D shindig you've introduced with the bulldozer?

What about the sub-par performance in SSE/SSE2? They perform well with AVX/FMA and SSE4, but today you won't run into many of those normally. Is the poor performance in older instruction sets in part due to the (at the time) unattainable 30% clock speed increase from the previous generation in order to maintain that IPC? I think the kicker here is that they stumbled over themselves and partially answered this, albeit they flat-out lied. They claimed to have maintained IPC but as soon as you see most benchmarks that don't utilize bulldozer's new architecture you quickly realize that it's slipped. IPC would have been equal at the originally planned +4ghz stock clock speed and really shouldn't have to rely on a perfect utilization of their new turbo tech, which is incredibly impressive but falls short of achieving the gains you'd see from a constant 4.5ghz stock clock.

Can we expect significantly higher clocks in order to address some of these issues? Or will they be cured with more aggressive turbo scaling? (which, by the way, is actually more impressive than Intel's So kudos on that one).

What about TDP and clock-speed scaling issues? If the clock speed goes up, the power consumption gets ugly very quickly with Bulldozer. Is mid 4ghz stock clocks achievable on 32nm process and still keeping the TDP under 140W? What about turbo? What about the rather piss-poor scaling we've seen with higher clocked bulldozers and performance increases? Can that be improved and how? If increase in mhz : increase in performance ratio doesn't increase, then you're just producing excess heat, using too much power and fighting an exponentially uphill battle (less room, more power thus tougher at higher clocks. throw the yield issue in with that and it suddenly doesn't sound like a great idea).

Regarding IPC and pipeline, can we expect even an even lower gate count? How?

What about a price drop considering how it performs to cheaper alternatives if volume increases? Frankly, unless the 8150 is priced under a 2500k, it's just not worth it. Is it even possible to drop it in price considering the transistor count and how demanding 32nm has proved to be for GloFo?

One of the biggest issues I've had with AM3+ and Bulldozer has been the 9xx-series chipset, which is essentially just an 8xx-series chipset rebranded but with SLi capability. As a long-time AMD fan I can tell you that was easily the worst decision you guys made and wreaks of Intel behavior. FM2 will have the improvements and bonus features and actually requires a new platform whereas AM3 > AM3+ seems just a money-grabbing attempt to sell the same motherboards but with a different number of pins. For those with AM3+ platforms now, just be aware that Piledriver will be the last new chip to fit into the am3+ platform and after that you'll require a new motherboard/CPU if you want to upgrade the AMD path again. So unless they've lit a fire under their ass, you're in for more disappointment

Finally, is the pursuit of more cores and maintaining IPC worth it considering the longer pipeline, higher clocks and higher power consumption figures? The idle power consumption numbers are very impressive considering the processors size (originally you guys claimed ~2B transistors, but apparently corrected yourselves and now claim it's ~1.2B, which is still quite large considering SB has a lower count including the on-die graphics). It's hard to justify this architecture at this moment considering how slowly software has advanced. But it's also not exactly the fault of developers as there hasn't been any significant improvements in passively, as opposed to how multithreading is currently only actively, inserted into code. The thubans did quite well with multithreaded processing and have higher single-threaded performance than the bulldozers (often equal multithreaded as well which is pretty fucking pathetic), had the new turbo core and power gating tech been added to a die shrink of the thubans on 32nm along with some minor tweaks I think we'd probably have seen a better chip overall.

EDIT: Future windows releases may help more than any scheduler will. As ARM scales up we'll inevitably see more multithreaded code and a merging of RISC/CISC on x86 processors (though I guess they're all some sort of hybrids nowadays anyway, but you get my point)
 
Last edited:
I have not ruled out Bulldozer yet. I don't understand this fully and I am looking into this but.. it seems that compiler optimization is what is really holding things back. When things are compiled using Open64, things improve around 30%. Look here:

http://www.phoronix.com/scan.php?page=article&item=amd_bulldozer_open64&num=4


Personally? I think Bulldozer is forward thinking not realized yet. When will things catch up? soon enough and that would not stop me from buying one.
 
Back
Top