AMD FX-8150 Review: From Bulldozer To Zambezi To FX

[Guest]

AMD needs a REAL DESKTOP processor, not some corners cutting SERVER PROCESSORS pretend to be one.

 

[Guest]

Seems like I don't neet to upgrade my i7 1G platform.... AMD is really a kind of joke! I don't believe 8 core cannot beat i7 2600K down in H.264 Encoding....dudie shame!

 

[tandemtruths]

This is very much speculation on my part, as I have neither the true experience and knowledge, nor the equipment to verify what I'm about to say. That being said, I have an inkling as to why such a seemingly beautiful architecture can behave in such mediocrity. I first found out about codename Bulldozer a year ago. At that time when discussing to the design profile, AMD referred to the now integer "cores" as integer units inside a physical core. They mentioned the dual pathway multi-threading this design allowed. AMD talked about how this design came into being after noticing Intel's own "Hyperthreading" used a shared pathway on the transistorized core. Now, Intel's "flagmen" do a pretty good job on those single lane roads, that is/was quite evident. AMD, at the time I had initially read up on Bulldozer, said (almost to say that by accident) operating systems would read these integer units as independent cores even though they share an FSB. Multi-threading works by splitting a task, but where does it go when it is prepared? It must be buffered onto an FSB. This is what my theory (though under-equipped) is: Because of the shared FSB1, the integer "cores" bus speed, and thus communication of tasks, is halved if not more so. These integer units also only comprise only half of the transistors per "module", and so are only meant to process smaller task parts. In essence, one could (if this theory has any bearing) say that the advertized clock speed is only behaving at half of it's nominal value. I also have to consider the unholy amount of FSB2 and 3 that could rectify a portion of this loss, and use of multi-threaded applications that don't require very high clock speeds as the individual tasks aren't very intense. However, that would certainly create a throughput bottleneck. My proposition is then this: If AMD could modify the BIOS in such a way as so that the operating system recognizes each "module" as a core instead. With this, the 8000 would be a dual pathway multi-threaded quad core,(as it physically is) the 6000 a multi-threaded triple core, and the 4000 a dual core etc. I believe this would allow for better use of the higher clock speeds for 'handful of intense task' computing that this genre of desktop cpu is mainly called to do. I believe that it would even cut down on power usage as the number of tasks attempting to fit through a small hole are decreased(the fat guy in a little coat issue). That last part I will explain as it sounds like I just pulled it out of my rear. It's a simple "the cable aint big enough for that amount of amps" argument. What happens when you put 600 amps through a 12 gauge wire? It gets HOT. What is heat exactly? I'll give you a hint, in physics class it was always described as being constant. Give up? Energy! Now, when there is such a clog as what I've theorized, it constricts the flow increasing the electron friction. Now there is only one way to calm an excited electron down when it can't move, that is for the atom to give up some of it's energy, again, heat. Think of tasks as gravel. There are some hefty pieces, and there are are small pieces. Multithreaded tasks are small and can fit readily through the small yet many "pipes" of the current 8 core setup. The larger, more intense pieces, need a little electronic Metamucil to get the ball rolling. Now say you have four larger pipes. These pipes also have lanes that organize your gravel better, and they can get those larger pieces through like bad Mexican food. It would be better to have eight pipes if they were the same size, but that takes real estate and also, especially in this case, money. (hence Intel's 990X) My round about point is, it should loose less to waste heat because of the lack of queuing.

The best thing about my theory, if true(ish), The performance could be downright amazing. If it keeps up with the 1100t true six core in it's current twisted bowel state, just imagine anywhere from 30% - 60% (those I did pull out of my rear....for science) increase in ability for monolithic computing. Continue to bear in mind (AHHH! there's an ursus in my head!) that this was all developed from an external, third, maybe fourth person perspective. So it's just my two cents. However, it could be two cents from the distant past, valued at two million cents! (that's $20,000!) So you never know...

-Andrew

 

[xanagu]

Underdog ALWAYS lose guys

Im staying with The Man

 

[Super_Nova]

'Just be careful with the marketing. FX-8150 goes up against Core i5-2500K and Core i7-2500K'

Guess it's a typo
Anyhow, nice review.

 

[professorprofessorson]

Looking at the benchmarks run from all the reviews across the diff sites, it really looks as if the timing is just off for this processors release. Its true, it doesn't beat the SandyVag cpus much in older software test, but on a lot of the new stuff it does. AMD probably should have worked closer with software companies during this processors development over the past couple of years to help them optimize their titles to use more then 3-4 cores if they are present.

As is though, due to what trend was shown in all the reviews, I bet as newer game titles will be released, you will start to see this cpu rise in the performance charts, and that is the performance you are going to end up paying for. Honestly though, its not like the performance it posted on the older software was that bad as is. I do feel though they should have waited another 6 months or so to a year before releasing this tech. Knowing the Phenom II x6 was having performance issues when coping with certain titles that did not use past 3-4 cores very well was a clear sign that the SOFTWARE world was not yet caught up to the hardware one. Should be a lesson hard learned for AMD now, if you want the software to run best on your tech, you need to pander to the software houses during development, and not just for graphics card optimizations, but cpu ones also.

 

[techpops]

After reading through AMD's reactions to the bad press, I'm now left with the idea that AMD are just trying to make as much cash as possible from the low end of the market, maybe pick off a few gamers that aren't really informed and as we move into the next year and the year after, we'll see AMD focus almost all its limited resources on the server market where the real money is. Gamers kiss goodbye to the AMD you knew and loved, hello serious enthusiasts, small businesses and enterprise.

I'm in need of the server performance but don't have the cash for it so I'm hopeful AMD will give us a great solution for the C32 server boards that continue to undercut the Xeon solutions while giving us more performance than the measly 2.6ghz Opterons. A 12 core setup isn't cheap but its way cheaper than a similar Xeon setup. 2 8 core CPU's with a similar clock speed to an FX-8150 at just a little more than the 2.6ghz Opterons would be a fantastic deal and really put AMD on the map. My worry is that they won't be able to get the clock speed high enough and we'll end up with the same situation were in on the client side computing. In other words, an underpowered Opteron at 2.6ghz against the client side desktop 1100T processors at 3.2ghz.

My fingers are crossed here. Any idea when the server CPU's are going to get reviewed/benched and priced?

 

[sonofliberty08]

[citation][nom]Why_Me[/nom]What's AMD suppose to do? Do a better marketing scheme for a shoddy second rate product? The benchmarks don't lie. BD has been benched all across the net in the past few days and all of them say the say thing. BD is a flop. Now if AMD wants to market a total flop, then let them have at it.[/citation]
AMD should set the level and price of FX-8 to compete with the i5, then the story will be different, the midrange and entry level market still have competition, it will be benefit for most of the consumer out there, all you intel fanboys still can get the midrange and low end intel chip with reasonable price, intel fanboys only have to burn their stupid good money on the i7 if they got too much money to burn

 

[sonofliberty08]

[citation][nom]tandemtruths[/nom]This is very much speculation on my part, as I have neither the true experience and knowledge, nor the equipment to verify what I'm about to say. That being said, I have an inkling as to why such a seemingly beautiful architecture can behave in such mediocrity. I first found out about codename Bulldozer a year ago. At that time when discussing to the design profile, AMD referred to the now integer "cores" as integer units inside a physical core. They mentioned the dual pathway multi-threading this design allowed. AMD talked about how this design came into being after noticing Intel's own "Hyperthreading" used a shared pathway on the transistorized core. Now, Intel's "flagmen" do a pretty good job on those single lane roads, that is/was quite evident. AMD, at the time I had initially read up on Bulldozer, said (almost to say that by accident) operating systems would read these integer units as independent cores even though they share an FSB. Multi-threading works by splitting a task, but where does it go when it is prepared? It must be buffered onto an FSB. This is what my theory (though under-equipped) is: Because of the shared FSB1, the integer "cores" bus speed, and thus communication of tasks, is halved if not more so. These integer units also only comprise only half of the transistors per "module", and so are only meant to process smaller task parts. In essence, one could (if this theory has any bearing) say that the advertized clock speed is only behaving at half of it's nominal value. I also have to consider the unholy amount of FSB2 and 3 that could rectify a portion of this loss, and use of multi-threaded applications that don't require very high clock speeds as the individual tasks aren't very intense. However, that would certainly create a throughput bottleneck. My proposition is then this: If AMD could modify the BIOS in such a way as so that the operating system recognizes each "module" as a core instead. With this, the 8000 would be a dual pathway multi-threaded quad core,(as it physically is) the 6000 a multi-threaded triple core, and the 4000 a dual core etc. I believe this would allow for better use of the higher clock speeds for 'handful of intense task' computing that this genre of desktop cpu is mainly called to do. I believe that it would even cut down on power usage as the number of tasks attempting to fit through a small hole are decreased(the fat guy in a little coat issue). That last part I will explain as it sounds like I just pulled it out of my rear. It's a simple "the cable aint big enough for that amount of amps" argument. What happens when you put 600 amps through a 12 gauge wire? It gets HOT. What is heat exactly? I'll give you a hint, in physics class it was always described as being constant. Give up? Energy! Now, when there is such a clog as what I've theorized, it constricts the flow increasing the electron friction. Now there is only one way to calm an excited electron down when it can't move, that is for the atom to give up some of it's energy, again, heat. Think of tasks as gravel. There are some hefty pieces, and there are are small pieces. Multithreaded tasks are small and can fit readily through the small yet many "pipes" of the current 8 core setup. The larger, more intense pieces, need a little electronic Metamucil to get the ball rolling. Now say you have four larger pipes. These pipes also have lanes that organize your gravel better, and they can get those larger pieces through like bad Mexican food. It would be better to have eight pipes if they were the same size, but that takes real estate and also, especially in this case, money. (hence Intel's 990X) My round about point is, it should loose less to waste heat because of the lack of queuing. The best thing about my theory, if true(ish), The performance could be downright amazing. If it keeps up with the 1100t true six core in it's current twisted bowel state, just imagine anywhere from 30% - 60% (those I did pull out of my rear....for science) increase in ability for monolithic computing. Continue to bear in mind (AHHH! there's an ursus in my head!) that this was all developed from an external, third, maybe fourth person perspective. So it's just my two cents. However, it could be two cents from the distant past, valued at two million cents! (that's $20,000!) So you never know...-Andrew[/citation]
as what i said, on the marketing side, they should just count 1 BD module as 1 core with 2 threads, and name the FX-8 as FX-4 (quad cores 8 threads), FX-6 as FX-3(triple cores 6 threads) and FX-4 as FX-2(dual cores 4 threads), set the price of the dual cores 4 threads FX-2 to compete with the i3, and quad cores 8 threads FX-4 to compete with i5, and we will hv the different view at here, so they won't fail that hard, and they can still have times to polish up their fresh new architecture and release another 6 cores 12 threads CPU later to compete with the i7 and their old Phenom II X6

 

[Why_Me]

AMD should set the level and price of FX-8 to compete with the i5, then the story will be different, the midrange and entry level market still have competition, it will be benefit for most of the consumer out there, all you intel fanboys still can get the midrange and low end intel chip with reasonable price, intel fanboys only have to burn their stupid good money on the i7 if they got too much money to burn

The problem with that is that the 2500K will mop the floor with that AMD when it comes to over clocking and it will do it for alot cheaper when you figure a $30 - $50 USD cpu cooler will get you to an easy 4.4 - 4.8Ghz with the 2500K. The BD on the other hand is a power hungry hog that runs hot. So unless AMD drastically drops the price of these inferior BD cpu's, I can't see anyone purchasing them unless they either own stock in AMD or maybe they had a recent full frontal lobe lobotomy.

 

[marcelormt]

look how amd changed

http://www.tomshardware.com/reviews/does-amds-athlon-64-x2-6000-have-any-kick-left,1547-3.html

Patrick: The 6000+ is the fastest Athlon 64 X2 dual core processor ever, but what happened to the FX family?

Damon: Patrick, you are right. The X2 6000+ is the fastest AMD64 dual-core processor ever... so why isn't it called FX? To answer that I have to explain what FX is all about... pushing the boundaries of desktop PCs. FX-51 did that right out of the gate, with multiple advantages over other AMD processors, and a clear lead on the competition. Move forward a bit to where AMD put high-performance, native dual-core computing into a single socket with the FX-60. Fast forward again and you see FX pushing new boundaries as "4x4" delivers four high-performance cores with a direct-connect, SLI platform that is ready to be upgraded to 8 cores later this year.

So, with the FX brand now crowning a 4-core (and future 8-core) platform, it seemed logical that new dual-core solutions should reside under the X2 moniker.

 

[Draven35]

[citation][nom]king_solomon[/nom]WHY THE HECK THIS REVIEW HAVEN'T USE MUCH MULTI-THREADED APPLICATION when testing a multi-core CPU? fishy fishy fishy..........[/citation]

Photoshop:multithreaded (not very well, mind you, but...)
After Effects: multithreaded
Premiere: multithreaded
3D Studio Max: multithreaded
Blender: multithreaded
Mainconcept encoder: multithreaded
Handbrake: multithreaded
Visual Studio: multithreaded
PCMark: multithreaded
3DMark: multithreaded
Sandra 2011: multithreaded

I don't see where you can say Tom's wasn't using multithreaded tests.

 

[gsxrme]

How Dare AMD compare that to a 980x in there ad. If you do the math correctly. its cheaper to go Intel now for FPS than AMD. That 980x vs 8150 is a pile. Its more like the 2500k vs 8150

 

[gallovfc]

How about DDR3 1866 next time ? How about overclocked benchmarks against 2600K, 920, 2500K, x6 1100T and x4 980 ?? How about a Multitasking benchmark (like on the 1100T review)? And finally, let's try to preview how it would fare in a server contest. Those 4 FPUs really disappointed me more than the fact that we'll have to wait until Windows 8 comes out to have the real performance test.

 

[notmyfault]

Yeah, it's a disappointment. An 8-core CPU sounds simply awe inspiring, but I guess that's about as far as it goes. I've been a pretty big fan of AMD and I'll continue to root for them. But, in the background, I'll keep enjoying the power of my i7-2630QM. It's just killer!

 

[ChromeTusk]

The thing is Bulldozer didn't need a new chipset, they could have allowed it to be dropped straight into any Am2+ or AM3 motherboard. I can work that out as they're doing exactly that with the C32 and G34 server boards. But no, you need a super expensive 990FX motherboard to power a Bulldozer CPU that gives you no extra speed over the old 1100T CPU's. You see the problem here? It's an unnecessary leap to a new chipset that seems to be more about trying to get a tiny bit more performance out of benchmarks so Bulldozer doesn't look so bad. So right now you do have no upgrade path buying into anything pre Bulldozer. ... That's not an upgrade path I want to be part of. Intel do switch up motherboards far too often but at least what's out now has some legs and any new board you buy into now has a future over the next few years at least.

When I started rebuilding my main rig, news about an AM3+ socket had not yet come out. I bought an AM3 board with expectations I could upgrade the CPU without any problems. Considering the reviews this week, I am putting off any new builds unless they are needed. Upgrade paths are foggy at this point, IMO.

 

[nagol567]

def getting me a fx4 series soon! cant wait to see some more benchmarks. check this out and scroll down a little less than half way down to see the true benchmarks! http://www.tomshardware.com/forum/315752-28-buying-bulldozer

 

[Guest]

Looks like it will improve with optimization and as more apps support 8 cores. I doubt many games are going to be able to take advantage of that so right now probably a solid quad core is still the way to go. For most this just tells me if you can do so just hold off from replacing your current gear and save your money. It wont be long till we see even faster chips and more software supporting the additional cores. Heck I still have an X26000 and a 9150E that work great. Going to SSD's for the OS makes a huge difference in overall responsiveness and app load times.

 

[gm0n3y]

[citation][nom]the associate[/nom]Waaaahhhhhhhhhhhhh!!!!!!!!Bah, well, been with AMD since my first pc like 8 years ago...Guess I'll be going intel for the first time ever especially since I can get an overkill cpu for just 300 bucks. Hell that's how much I payed for my phenom II 955...[/citation]
I wish I had more thumbs to downrank you with.

 

[Why_Me]

def getting me a fx4 series soon! cant wait to see some more benchmarks. check this out and scroll down a little less than half way down to see the true benchmarks! http://www.tomshardware.com/forum/315752-28-buying-bulldozer

Keep on fooling yourself. BD is a heat producing power hungry hog that doesn't o/c worth a crap without an expensive cooling set up. BD = Big Dud.

http://www.anandtech.com/show/4955/the-bulldozer-review-amd-fx8150-tested <--- AnandTech Review of the FX-8150

 

[spoofedpacket]

I see all this chatter about BD in the server market, but none of my customers have even mentioned interest in it. When I can make a phone call and get an Intel rep out to host a free "lunch-n-learn" event within a week, I can't even get the regional AMD reps to return calls when a major potential buyer has questions or wants some attention from manufacturers. This leaves an impression with buyers and re-sellers alike.

AMD desktop CPUs? I sell to business and government, and tbh, most of the entities I deal with have a policy against purchasing AMD-based systems-- especially when it comes to notebooks.

 

[Guest]

@spoofedpacket

and thats how intel plays the market, not once did you say intel was cheaper or better or more reliable (which they may well be) but hey a free lunch is a free lunch right, AMD just doesn't have the kind of cash to play that kind of game

 

[Why_Me]

@spoofedpacket

and thats how intel plays the market, not once did you say intel was cheaper or better or more reliable (which they may well be) but hey a free lunch is a free lunch right, AMD just doesn't have the kind of cash to play that kind of game

"Cash to play the game"? Your business IQ is a big zero.

 

[Guest]

@Why_Me

dude dont paraphrase me, it should be "AMD just doesn't have the kind of cash to play that kind of game" there's a big difference from what you wrote, and if you think AMD can throw that kind of cash around then your sadly mistaken, yes we all know Intel can shrug off a billion dollar FTC settlement as pocket change, but AMD just doesn't have that kind of financial muscle

talk about business IQ how big do you think AMD is compared to Intel, half their size maybe? maybe closer to 1/100 the size, scale everything down 1/100 and AMD be lucky to buy you a sub for every Joe re-seller, they have to make their budget count which means only catering to the more promising clients like Apple and Nintendo, how about that for business sense

 

[DjEaZy]

http://quinetiam.com/?p=2356