AMD Bulldozer aka FX series CPU Reviews

[Goku]

Review List:

Hard OCP - CPU tasks - Gaming Performance
Hardware Canucks
Guru 3D
Hardware Heaven

My previous opinion is still the same. I am underwhelmed by the performance. Gaming performance is so so and even loses to the former 1100T (AMD six core CPU) at times. There is times in highly multi threaded programs it comes out ahead of the 2600K or matches it, but this doesn't really matter to gamers. I still don't see myself recommending these unless we see them get a good price cut in the coming months.

[Traak]

Thanks again, Goku. Once again you cut past the hype (instead of cut and paste the hype) to give us the real-world scoop.

[cain marko]

it's fake, the cpuz shot lists it as a code name "bulldozer".....ummmmm not in reality
it would be called zacate. or if was server part interlagos.

sorry it should say zambezi.....
any way architecture has never been used as a code to identify a CPU always the release name...
my conclusion is that these results are fake.

[Goku]

Originally Posted by cain marko

it's fake, the cpuz shot lists it as a code name "bulldozer".....ummmmm not in reality
it would be called zacate. or if was server part interlagos.

sorry it should say zambezi.....
any way architecture has never been used as a code to identify a CPU always the release name...
my conclusion is that these results are fake.

No no. Zacate is their entry level APU aka E-350.

I would love these benchmarks to be wrong. Either way we find out in two days.

[cain marko]

Originally Posted by Goku

No no. Zacate is their entry level APU aka E-350.

I would love these benchmarks to be wrong. Either way we find out in two days.

that why i edited and said sorry should say zambezi
but in any case these results are crap ES(engineering sample) or faked, engineering samples are missing a ton of code to inhibit full performance, for all we know they could have half the cores shut down, but still show all of them.
AMD isn't about to lose a chip to the competition and have them prepare for release of that chip. there is such a thing called espionage.
just wait for AMD or another reputable site to release results with final silicon, it's the only way we are going to get the truth.
as far as i am concerned this was another donamhaiber stunt to get views.
trolling you might say....

[BorisBlade]

I saw some of the early silicon tests about a month ago and they were fairly close to the 2500/2600, and those were early chips so my guess is that if those tests showed em pretty weak, that they are prob fake. No point in them releasing this stuff if its worse than a phenom II, those things were already pretty bad.

[duomaxwl]

Will the new ones run on AM3 socket? My gut says no, but I keep seeing people saying that they can via a BIOS update from the manufacturer. Not sure what to believe, so I figure'd I'd ask.

[cain marko]

Originally Posted by duomaxwl

Will the new ones run on AM3 socket? My gut says no, but I keep seeing people saying that they can via a BIOS update from the manufacturer. Not sure what to believe, so I figure'd I'd ask.

they will work in some boards, gigabyte said no for theirs for sure, saying they became unstable on their boards.
personally i would just go with a am3+ socket if your going to go BD, I have a feeling amd has been keeping the lid tight on these for a reason and it isn't because of poor performance.
JFAMD wrote:http://www.overclock.net/amd-cpus/11076 ... h-faq.html
To get actual performance, you need:

Final production silicon
Final processor microcode<--- most major part of real test scores
Final system BIOS
Final OS optimizations<-- not being released till chip released
Final drivers
An app compiled with the latest flags
A person who understands the app and configures the test properly

[Goku]

Reviews are out. Looks the preview was spot on just about. Oh well.

[Traak]

You know, if people would produce dual-socket motherboards that were similar to mainstream gaming boards, BUT with two sockets, AMD's would be viable, again.

But, if the choices are "AMD or performance" then, what's a gamer to do?

But, dual-socket boards are listed as "server motherboards" and their feature sets are not up to date. None have dual PCIeX16, for example, that I've found. SATA III? USB 3.0?

Nah. Multiple sockets are for servers, not for gamers. How about dual-socket gaming boards?

I guess the cost would be somewhere around ridiculous.

So, Intel wins again. I wonder if Intel was readying a few upgrades to their CPU line to shoot the Bulldozers down in flames if they actually stood out. If so, no need to unveil them yet. When your competitors yesterday stuff beats your today stuff it's bad. When YOUR yesterday stuff also beats your today stuff, then... how bad is that?

[cain marko]

not really spot on it's being necked like crazy, this is a monster waiting for windows scheduling to get right.

the proof hits in one webs review showing one core fully loaded(100%) and the rest sitting below 50%. windows scheduling is screwing this processor bad.
they hit it with 2 tests at once and the things performance showed through.

[Goku]

Doubtful. If there was that big of a fix amd would of waited longer for a release. Everyone thinks thus CPU is no good now no magic patch is going to change that. Maybe piledriver will make them better like Deneb did but I am not holding my breath.

[Vancha]

Yeah, the excuses are coming thick and fast, and all of them weak. The popular one seems to be claiming BD was made for Windows 8, which'll be released after Piledriver...

I'm a member of the green camp to the point of being unable to tell you anything about Intel CPUs, but this is undeniably a failure on AMD's part.

[Traak]

http://www.techspot.com/review/452-a...us/page12.html

Another review. I liked all the graphs.

What I got out of it, for the most part, is that, considering most software, especially games, are just starting to take advantage of more than two processor cores, the eight-core top-line AMD processors are not able to flex an advantage over the four-core Intel processors, which are old, at 9 months, already.

Having lots of cores is great. If anyone needed them for anything.

How about a CPU that either combines or divides threads so IT decides what goes to what cores, instead of the game or other software?

Further, how about some tests that show the performance with one core enabled for that software, then two, then three, on up to eight, so we can see where the performance ceases to increase?

The AMD processor, from the reviews I've seen, well, just sucks compared to the Intel i7. I would like to know why.

From HARDOCP's review page. Civilization V game. I quote:

We also looked at CPU utilization on the AMD FX-8150 to see how well the game was utilizing the CPU. We noticed that all 8 cores were being utilized while we played the game, each core carried a load, none laid flat on the graph. We also noted that the highest peak total CPU usage while gaming was 76%. This indicates that the CPU is being used well in the game as far as utilization goes. There is simply no denying that the Intel equivalent CPUs are superior for performance in this game at this time.



Strange. We have only 76% CPU peak utilization. EIGHT CORES. And yet, the framerates drop, at some points, to ZERO FPS.

This obviously points to a bottleneck in the system. How can you have ZERO frames per second with only about 3/4 of the CPU's capabilities being utilized? Something's wrong, somewhere. The actual things that do the actual computing, the CPU cores, are being bottlenecked by something, somewhere, on both the Intel and AMD side.

What, to me, is also strange, is that the Intel CPU scored higher than the AMD. Why strange? Well, shouldn't the AMD be huffing and puffing, at 100 percent utilization? Why was it not at peak effort?

Well, the answer, again, is that something else is clogging up their system. It isn't the CPU cores. They are the fastest things in the system, in this case.

No, it has to be the RAM, motherboard, VPU, or storage. It can also be the northbridge/southbridge right on the actual CPU die, if it isn't on the motherboard. Something is blocking the CPU's from performing to their full potential, and it isn't the CPU cores.

[cain marko]

the software issue is already catching up, other issue amd was losing way to much ground with the tweaked athlon cores.
so this was a viable resort as i see it, as we continue intel is going to either have to increase core size and adding cores as they go, and keep trying to shove another thread through it.
this is absolutely new architecture and can be tweaked in how many ways to fit the needs.
the results in BF3 were proof that it was massivly threading the FX.
the test i saw with the fully loaded core and the rest sitting at less than 50% thats not an excuse thats the reality of the chip not getting scheduled.
the second test i saw today was actually very good with it running 1866MHz ram. where it outperformed the 2500k and was sitting at or above a 2600k.
this is actually proves this is no dud from AMD and does have a good future.