AMD Bulldozer / FX-8150 Desktop Performance Review @ [H]ardOCP

Wildace said:
anyone know where i can get the beta bios for the CHV so i can disable indevidual cores? the 0813 bios update ont he asus site only lets you disable cores 2 at a time.
Click to expand...
xorbe said:
I was wondering the same thing.
Click to expand...

You can't. The damn FX 8150 and FX 8120 don't have 8 real cores, but only 4. AMD decided to call cores modules because they implemented Chip Level Multi-Threading, or CMT for short. The modules are actually the cores, so when it says that you can only disable two cores at a time, you're actually disabling a module, which in fact is a core. Does this make sense?
 
wolfman3k5 said:
You can't. The damn FX 8150 and FX 8120 don't have 8 real cores, but only 4. AMD decided to call cores modules because they implemented Chip Level Multi-Threading, or CMT for short. The modules are actually the cores, so when it says that you can only disable two cores at a time, you're actually disabling a module, which in fact is a core. Does this make sense?
Click to expand...

actually you CAN disable just a single core from one of those "modules" the bios just needs to have the option for it. people have already done it and its been proved that disabling one of the cores so the one thats left active wont be sharing resources improves that cores performance ect.
 
wolfman3k5 said:
You can't. The damn FX 8150 and FX 8120 don't have 8 real cores, but only 4. AMD decided to call cores modules because they implemented Chip Level Multi-Threading, or CMT for short. The modules are actually the cores, so when it says that you can only disable two cores at a time, you're actually disabling a module, which in fact is a core. Does this make sense?
Click to expand...

Yet someone has a magic bios that does exactly that:

http://www.xtremesystems.org/forums...readed-Perf.&p=4970768&viewfull=1#post4970768
 
i think the bios version is 9911 which allows you to turn off cores 2 4 6 8 i still havnt found a link to it yet tho.
 
wolfman, they can't repair bulldozer and fix its 'issues' due to its modular design. This isn't a case of "we didn't do enough" but rather "we did too much and over-complicated things." Long pipeline. Why? CMT. Again, Why? 2 billion transistors and massive leakage. Why? A step backwards in IPC. Why? You don't take this thing back to its engineers and expect them to improve the IPC, leakage and lower the transistor count with a "fix" because it supposedly has modular design. In order to salvage any sort of hope for gaining market share Bulldozer needs to be scrapped or entirely reworked in order to be competitive with Sandy Bridge, and with Ivy Bridge around the corner I think expecting it to improve and become relevant is just wishful thinking. AMD has stated that there's a 10-15% increase in performance annually, which means after a couple of years it may catch up to the PhenomII in terms of IPC. Considering how they swindled us, I'd expect slightly better software support and a bump in clock speed to account for those 10-15% gains, if it even improves at all.

This launch and the way they've gone about hyping it up only to have it come crashing down has really left a bad taste in my mouth. I've gone AMD since the T-bird days because I wanted to support the underdog. Unfortunately, with this Bulldozer fiasco and the various rumors swirling around about how the management inside AMD operates has left me thinking that perhaps I should have gone blue a long time ago.

Doesn't VIA still have an x86 license? Where did you guys go? :D
 
Doesn't VIA still have an x86 license? Where did you guys go?
Click to expand...
They do. Last I heard, they were working on a modular design with long pipelines...oh, wait.... :confused:
 
I think the engineers were already baked when they designed BD.
Click to expand...
I agree. I haven't seen JF-AMD post anything on his blog since BD launched. He used to have a smart mouth. Not anymore. Now he's either sucking it somewhere in a dark corner or getting baked...
 
Valshistixol said:
If the power usage wasn't insane I would agree but I still think a 2500K is the better choice.
Click to expand...

Power usage is really only relevant for people planning on keeping the CPU workload at 100% for hours at a time every day. For example if you were planning to fold with the CPU. For normal computer usage and gaming the additional power draw won't be nearly as much. Most peoples CPU sits idle or at low load the majority of the day and Bulldozer is very efficient under those types of loads. In the HardOCP tests the 8150 system @ 4.6Ghz drawed only 120w compared to the 1100t system @ 4.2Ghz drawing 152w.

It's no different than if you were planning on Bitcoin mining with a couple of AMD 6990's which under full load when overclocked draw a massive amount of power. But for the everyday user who only loads the GPU's when gaming the power draw is deemed acceptable for a high-end gaming machine.
 
Magic drivers.... like the TLB problem, didn't fix everything. A CPU that's not beta would have helped more. I can't see much life in this and I hope AMD has something more powerful in the waiting. There are a lot of jobs at AMD that might not be here next year if this company goes under.
 
wolfman3k5 said:
I agree. I haven't seen JF-AMD post anything on his blog since BD launched. He used to have a smart mouth. Not anymore. Now he's either sucking it somewhere in a dark corner or getting baked...
Click to expand...

Actually, JF-AMD was sent so many nasty messages that he decided it wasn't worth bothering to come onto forums and give us his take on what AMD is doing. As an AMD employee, he obviously wasn't allowed to just come out tell us classified information, but he did try to set any obvious misconceptions straight. Now that enough a$$hole$ have mindlessly and childishly attacked him for no particularly good reason other than that he was providing us with an AMD insider's perspective, we've lost him as a resource. Oh well.
 
I think it sucks that JF-AMD got *attacked* and there's no excuse for that kind of childish behavior, but I'm not so sure his contributions were particularly significant. He wasn't allowed to act as an AMD employee, so his posts were more or less, "Wait until launch", "No numbers before launch", "Not representative".

I know everyone's emotional about the BD launch, but I think most enthusiasts just expected too much. BD performs right in line with what I thought it would. It's just not priced in line with its performance nor is availability anything short of disappointing.
 
I am sorry as well that JF-AMD allegedly received trash emails. I don't know why anyone would send him anything, but I do believe that he started speaking out about Bulldozer because AMD asked him to. Of course neither AMD nor JF-AMD could ever admit to that.

Lets imagine for a second that the company that you're working for had a really revolutionary product, and you knew that, then you would talk very enthusiastically about it. Your statements would be short and to the point, and of course, in the limits that you would be allowed to talk about a future product.

But JF-AMD knew that Bulldozer was shit, yet he still came out and talked, and most of what he said was useless and even patronizing. If this is how AMD is attempting to do damage control, then they are truly pathetic.

I will post some JF-AMD quotes, and then you guys can make up your own mind about the guy:

Q. Why don't you release benchmarks before launch? You could steal so much business away from the other guy?
A. Again, releasing benchmarks before launch will simply stall sales. Believe me, if the competition thinks they are out of position, will they just sit back and say "oh well" or will they react? Handing them benchmarks is simply giving them time to form a strategy. I am not in the business of helping them, they are on their own on this one.
Click to expand...
I have seen at least since 2006 that Intel has released hardware to reviewers and even benchmarks before a product launch. Also, Intel's launch dates are pretty damn accurate.

Q. Will bulldozer be faster than....?
A. I don't need to finish the question, please read above.
Click to expand...
This one right here said allot about what was about to come... Both Intel and AMD usually "leak" information when they have something good.

Q. I read on xyz site that you were launching on xxxxx?
A. Yeah, and I read on another site that elvis was still alive. The reason I don't comment on date rumors is that there are a limited number of days in the quarter. Once you say no to some, and suddenly say "no comment" or don't answer that one, immediately everyone thinks that is the date. So, no matter how crazy it sounds, you can't answer any of them.
Click to expand...
Nice, eh?

And now for my favorite two:

Q. Is IPC higher on bulldozer? All I care about is IPC.
A. IPC is simply a measurement. What if IPC was 2X what it is today, but clock speed was 500MHz. Is that what you want? You are getting double IPC, right? IPC is only one measure. The people that are telling you IPC is the only thing that matters have an agenda. Taking only one measurement, out of context, is like trying to say that a person's weight is all that matters. I weigh 195. Does that make me fat? Does that make me skinny? It is impossible to say unless you know my height. IPC is like weight - it tells you something in context to other factors, but is meaningless on its own.
Click to expand...
and

Q. What about single threaded performance?
A. See above. Also, if all you care about is single threaded performance, might I recommend a lovely, inexpensive single core processor for your system?
Click to expand...
This guy was obviously nervous and irritated at the same time about an upcoming product that his company was launching and about which he knew that would be an epic fail. Confident people don't try to substitute clever humor with cheap sarcasm, neither do they take cheap shots at potential customers.

You can read the whole thing here: http://www.guru3d.com/news/amd-bulldozer-prelaunch-faq/

By the way, JF-AMD means John Fruehe AMD.
 
Last edited:
like I was saying AMD is too small to have both AMD cpu side and AMD video card side split them up again and let awesomeness like the amd64 and xp thunderbirds come back again
 
wolfman3k5 said:
Thank you Kyle. That looks like a good idea. Since there is only one question per post, would I be able to create 2 or even 3 posts? As many others here that have bought, and still buy AMD products, we all want to see this company succeed. My main concern is IPC with Bulldozer, but I also have 1 or 2 questions about thread optimization.

Thank you
Click to expand...

Post as many questions as you like.
 
mzs_biteme said:
The real question is: will AMD really try to answer all our questions truthfully and with tangible facts, or give us the "JF-AMD" version yet again.....??? :confused:
Click to expand...

I doubt that they would want to transform this into a PR nightmare for themselves. I have seen everything from sarcasm to resentment in this thread, and someone at AMD must have read all of it.
 
http://arstechnica.com/gadgets/news/2011/10/can-amd-survive-bulldozers-disappointing-debut.ars/1

I'd highly advise reading that arstech article. It pretty much sums up how I've felt aboutsome of the concerns regarding the processor's performance.

I have a feeling it'll be one of those chips you'll take out of your closet in 10 years time, dust it off and fire it up to see that it performs extraordinarily well. But for today's software and FPU heavy tasks and the relative slow embrace of cpu+gpu computing for FP, it's too early and therefore underperforms for it's price range.

They really should have marketed the Orochi cores strictly to the server market (coupled with a decrease in clock speed I think it should perform very very well and have relatively low power consumption) and just die-shrunk a Phenom II (maybe APU?) with some minor tweaks and marketed that for the desktop/laptop segment. The Bulldozer sort of looks like a server CPU being marketed for desktops that's tacked on a lot of L2 cache and assumes you'll be carrying a good GPU to help out with FP (and the software to go along with it)

I have to disagree with the arstech article on a couple of points.

1 - windows 8 will not be a saving grace of this chip. Relying on a future version of an operating system to utilize a chip's potential isn't a sound market strategy and doesn't please the majority of users who won't be using that OS -- for years to come, really.

2 - I truly don't believe GloFo will be able to make the chips with a stock clock of over 4ghz and equal or lower TDP figures. Subsequently, the improvements to the performance of the chip should rely on restructuring the layout rather than selling a 8250 at 4.0ghz stock. Increasing stock speeds will make for even worse wattage figures and to boot, the overclock will likely be abysmal and the # of chips that fail, too, should be quite high.

3 - They're sort of stuck in between a rock and a hard place. I have no doubt that GPU and CPU computing, along with multithreaded workloads are the future of computing. So where does this leave them with future projects? If they tune Piledriver to be more effective and efficient at FPU than how will this impact Fusion? Low power consumption figures are looking more and more attractive for many of us and a longer pipeline with higher stock clocks is flat-out counter intuitive when trying to decrease the power consumption figures. So if they tune Piledriver in such a way that you shorten the pipeline and decrease the stock clocks (with FP optimization as one of the primary goals) would you need to dump the Orochi cores altogether? Would it be advisable to start fresh or tweak the Bulldozer design? If the latter is chosen, what kind of IPC increase% is possible? What's the time frame for either direction? Considering Intel will be selling their Ivy Bridge chips in late Q1/Q2 2012, you'd have to think that AMD really is in a very rough spot (essentially 3 generations behind: core i3/5/7 first gen, sandy bridge and then Ivy).

Frankly, I think they should have gone through with the Bulldozer/Interlagos as a single server chip. The Llano and Trinity APUs should look more like a die-shrunk 32nm Phenom II. With an APU that performs well per-clock on FP and great video card performance in that single package AMD would have a real winner, much like the Llano. As the gaming industry matures (and we see better scaling with threading and more "fusion" between the GPU and CPU. Consoles finally dumping DX9 wouldn't hurt) they could truly brand themselves as #1 in gaming. We were looking for a home run but what we got was a present-day dud but maybe a future success. AMD didn't do much to quell the rumor mill either and in fact they did quite a bit to contribute to it in a way that ended up hurting their processors even more
 
Last edited:
With the limited time, limited availability, competition to publish results for Bulldozer, I tend to think the strengths if any have not been exposed. Performance per chip size and power looks to be a missed at stock speeds compair to the competition in general. Still more dedicated testing and analysis of this new design I think is needed. Multi-tasking with multiple programs running, (using some smart program configuration manually for modules/cores used). Multi-GPU rigs up to four tested, will Bulldozer feed those GPUs?

Plus some more in debt performance tweaking tests to maximize performance. In the end Bulldozer looks like it has only one way to go - up. I could be wrong though :.
 
noko said:
With the limited time, limited availability, competition to publish results for Bulldozer, I tend to think the strengths if any have not been exposed. Performance per chip size and power looks to be a missed at stock speeds compair to the competition in general. Still more dedicated testing and analysis of this new design I think is needed. Multi-tasking with multiple programs running, (using some smart program configuration manually for modules/cores used). Multi-GPU rigs up to four tested, will Bulldozer feed those GPUs?

Plus some more in debt performance tweaking tests to maximize performance. In the end Bulldozer looks like it has only one way to go - up. I could be wrong though :.
Click to expand...

Most of the CPU Die Area is taken by cache, followed by the memory controller and Hyper Transport Links. The least amount of the CPU Die Area is taken ... not surprisingly ... by the CPU. Bulldozer is not a smart design, from the looks of it, it's more of a hack. I could be wrong though...
 
I see alot of people here clamoring about wattage and the like of the FX proscessor and alot of them OC anyway. So my question to yall is why would any of us care about power consumption anyway unless we didn't have the PSU power and cooling to OC the chips anyways. I just don't understand at all??? This kind of stuff goes hand in hand doesn't it you have a computer use a dozen fans to cool it LOL and heat your computer room so bad you have to turn on the a/c so you don't sweat then we all complain about power usage and the heat?? :) :);):D Oh by the ways they're all sold out someone must love the FX!!!!! I like mine ok!!! I thought after buying a board and all the parts for a new setup and allready having an 1100 it was sorta stupid for me to buy another 1100 and I'm giving my other setup to my Grandboys for their first rig. So I thought I'd give it a try for my usage it actually feels a little snappier on the desktop on Win7 Ultimate. I think it's the extra memory bandwidth the chip has! Please don't flame me to bad!!! LOL :)
 
Frank_G said:
I see alot of people here clamoring about wattage and the like of the FX proscessor and alot of them OC anyway. So my question to yall is why would any of us care about power consumption anyway unless we didn't have the PSU power and cooling to OC the chips anyways. I just don't understand at all??? This kind of stuff goes hand in hand doesn't it you have a computer use a dozen fans to cool it LOL and heat your computer room so bad you have to turn on the a/c so you don't sweat then we all complain about power usage and the heat?? :) :);):D Oh by the ways they're all sold out someone must love the FX!!!!! I like mine ok!!! I thought after buying a board and all the parts for a new setup and allready having an 1100 it was sorta stupid for me to buy another 1100 and I'm giving my other setup to my Grandboys for their first rig. So I thought I'd give it a try for my usage it actually feels a little snappier on the desktop on Win7 Ultimate. I think it's the extra memory bandwidth the chip has! Please don't flame me to bad!!! LOL :)
Click to expand...

That's great that you are happy with your purchase that is all that really matters at the end of the day regardless what someone on a forum tells you.

And its also good for AMD that they are selling them.

However I still wouldn't buy one for my main desktop machine. I already have bloomfield which uses quite abit of power itself and my next chip I want power consumption to go down not up.
 
mzs_biteme said:
Negative... Shitty GloFo yields are more to blame for FX "shortages", than people buying them out...:eek:
Click to expand...
Well that may be a factor also; but then again AMD might have hand picked a few to package and sell because of bad yields. I hadn't really thought of that possibility till now.
 
Just looked at some of the reviews on Newegg. 64% of the 8150 ratings are five stars? Honestly? That's just BS. Looks like all of the people buying this are already AMD platform owners. Who with an i7 SB or for that matter even a non-SB i7 is going to "upgrade" to BD? The yields can't be that great, nor would I imagine the margins considering the transistor count.
 
Just looked at some of the reviews on Newegg. 64% of the 8150 ratings are five stars?
Click to expand...
If you look at other CPU ratings in Newegg, multiple models in AMD and Intel, it's awash with 5-star ratings. EVERYONE seems to give their chosen CPU a top rating, regardless of model.
 
Newsflash: the 8150 is actually a pretty awesome chip. Just because it doesn't smoke a 2600K in benchmarks doesn't mean that it doesn't do everything those users want -- and more.
 
EuphoricRage470 said:
Newsflash: the 8150 is actually a pretty awesome chip. Just because it doesn't smoke a 2600K in benchmarks doesn't mean that it doesn't do everything those users want -- and more.
Click to expand...

If you consider this awesome:

Having lower IPC performance than your previous gen architecture.

Having "8 cores", 2 billion transistors and a 315mm2 die on a 32nm process while the Phenom II X6 "thuban" 1100T only has 6 cores, 904 million transistors on a 45nm process and yet the 1100T performs on par or better in most games.

Only being able to keep up to a 2500k (32nm 995 million transistors 216mm2 die) in highly multithreaded apps, despite being priced similarly.

All while using 100 Watts more electricity under load than the 2500K, and doubling as a space heater.

Then yes it is awesome!
 
Last edited:
Project 2501 said:
If you consider this awesome:

Having lower IPC performance than your previous gen architecture.

Having "8 cores", 2 billion transistors and a 315mm2 die on a 32nm process while the Phenom II X6 "thuban" 1100T only has 6 cores, 904 million transistors on a 45nm process and yet the 1100T performs on par or better in most games.

Only being able to keep up to a 2500k (32nm 995 million transistors 216mm2 die) in highly multithreaded apps, despite being priced similarly.

All while using 100 Watts more electricity than the 2500K, and doubling as a space heater.

Then yes it is awesome!
Click to expand...

it would be awesome if they cut 100 dollars off retail price on 8150
 
umcpgrad said:
it would be awesome if they cut 100 dollars off retail price on 8150
Click to expand...

I have to agree. If the chip was cheaper I'd definitely consider buying it. Unfortunately, considering the issues with GloFo, yields and the chips it's competing with in that price range it just isn't an attractive offer. If it were on par with the 1075T or 1100T then maybe I'd consider it -- but still not for a gaming rig.

There's some serious issues with the Bulldozer: performance and power consumption. But if the chip were priced to account for those issues I wouldn't have a problem buying it (say ~$150-180 or so). I don't see an immediate price drop even in the realm of possibility
 
Makaveli@BETA said:
And that's why Newegg reviews are useless for the most part.
Click to expand...

Except for ones like this (posted by "Richbo"):

Our company makes cellular image analysis instrumentation and software that takes advantage of the multi-threading capabilities of multi-core CPUs. Since our analysis software runs hundreds of parallel image analysis algorithms that are heavily multi-threaded the CPU is generally maxed out. We’ve been looking for a low cost alternative solution for our customers to replace their current expensive dual-slot Xeon CPU workstations. In early testing with our software my current configuration of a single non-overclocked FX-8150 (ASRock 990FX Extreme 4 motherboard) running Windows 7 64-bit is outperforming our dual-quad core Xeon E5600 workstations by as much as 27%. What makes this even more impressive is that our software code is optimized to use the Intel performance primitives. The results are encouraging considering the BIOS, C++ compilers and Windows scheduler haven’t yet been optimized for the Bulldozer architecture. Once the software catches up this chip should be even more amazing.

Source: http://www.newegg.ca/Product/Product.aspx?Item=N82E16819103960

The more I read, the more suspicious the hate-on for the FX is becoming. The initial reviews called it garbage. Then it was merely "disappointing". Then it was an o.k. chip, but just consumes too much power. Then it could beat the i5 2500k in as many benchmarks as not, but it was still worse. Then it keeps up with the i5 2500k in most benchmarks and wins few. Just recently, it has been smoking the i7 2600k in integer benchmarks and keeping up with it in FP. It seems the benchmarks done on the Asus boards suck, and the benchmarks done on the Gigabyte 990FXA-UD7 seem to show it keeping up or winning against both the i5 2500k and even the i7 2600k:

http://www.kitguru.net/components/c...ion-8-core-review-with-gigabyte-990fxa-ud7/7/

Jesus. By November, the way things are going, the FX-8150 is going to be smoking the i7 990x and the FX-6100 is going to be well ahead of the i7 2600K in all gaming benchmarks. By all appearances, even in it's initial, 'crippled' form, it's a goddam integer monster, and at least on the Gigabyte board, it is beating the i5 2500k handily, and keeping right up with the i7 2600k in Handbrake 9.5:

http://www.kitguru.net/components/c...on-8-core-review-with-gigabyte-990fxa-ud7/16/

Enough of the criticisms of this chip already, until we at least see a comparison of the performance on different motherboards with different bios revisions.
 
Last edited:
The truth is this is exactly the chip I want it seems. I do tons of shit in parallel and couldn't care less about overall peak speed in one thing. My gaming just has to be smooth not amazing, and for what I do (tons of multitasking/virtual machines/multiple game clients) this seems to be the best choice. Cores, cores, cores.
 
OldPueblo said:
The truth is this is exactly the chip I want it seems. I do tons of shit in parallel and couldn't care less about overall peak speed in one thing. My gaming just has to be smooth not amazing, and for what I do (tons of multitasking/virtual machines/multiple game clients) this seems to be the best choice. Cores, cores, cores.
Click to expand...

Indeed - I'm sort of tempted to get one of those to test at work. We churn through disturbing amounts of calculations, so reasonably cheap many-threaded integer performance isn't a bad idea.
(I'll get a 2600K if I upgrade my gaming PC this generation, though.)
 
You must log in or register to reply here.
Back
Top