AMD Bulldozer FX 8150 CPU Review

Introduction to Bulldozer

Well…AMD was kind enough to send us a package and we found a NEW CPU inside!Since we got the chip might as well do a review of it…lets see how it stacks up against its direct competitor the Intel Core i5 2500K

A few words on Bulldozer

AMD really reinvented the wheel on their Bulldozer. The FX lineup of CPU are different than any other CPU’s out there…now if thats a good thing or a bad thing is question of debate and perception. With Bulldozer AMD tried the “less is more” approach.They put two cores side by side..and then decided what to share between them. At the end they decided to share the Front end , Floating point cores and the L2 cache. Take a look at the diagrams to see what they have done.

Bulldozer Concept and Reality

Now that we know what AMD wanted to do and what they did;lets dig a bit deep into the architecture,the next couple of slides will give you a fair idea about the CPU.

>Here is how it works :

Above two slides show the die of the FX CPU.

Now here’s some info for you so that you guys don’t say it out wrong :

• the die itself is called OROCHI
• the Architecture is called “Bulldozer”
• The code name of the processor we are looking at today is called “Zambezi” and the server variants are called “Interlagos” and “Valencia”

Well..lets hope you wont get it wrong again.

Since Server market is a big playground for AMD power management is a big deal..at-least for them. So Whatever improvement they do on the server side will get ported to the desktop side.Same story with the FX lineup also,the idle frequency is reduced to 1.4Ghz with a matching vcore. The changes between idle and load frequency are instant a one would expect.

Although power is not a big source of concern for us overclockers;for a regular consumer it is a BIG deal as bigger power envelop would ultimately lead to bigger cost.Now.i didn’t have the chance to measure the power output from my system but after looking at some other user experiences i came to know that the average power output from an FX 8150 setup and 2600K setup are as follows :

There is a strong disparity between the two systems..so clearly in terms of efficiency getting a 2600k will make sense to a conservative buyer. They also have a advanced Turbo core,which has a base turbo and an advanced turbo. All cores can exceed their stock frequency (up to 3.9 GHz in this case) if the chip’s TDP isn’t reached.however If a multi-threaded load is applied, chances are it will take up the TDP and the CPU will operate at its base frequency (3.6 GHz on the FX-8150). Light loads will take advantage of the Turbo cores though.With a max turbo of 4.2GHz single threaded to quad threaded applications(majority of software’s and some games too) will experience a nice boost of upto 600MHz depending on the processor.

An overview of the platform

The FX lineup has eight processors spread across seven models (one model has two TDP variants,more on this on the slides). Only Four models was launched at first though:two 8 cores;one six core and one quad core model

Telling of AMD’s cut-throat pricing (and potentially, performance), the top of the line FX-8150 is going to be priced at a very reasonable $245.

With the technical jargon out of the way lets see what AMD has to say in their product slides.The slides are pretty self explanatory.Just some points are worth mentioning:

In the first slide

• The most important comparison in this chart for overclockers and gamers is the fact that AMD has a full 32 lanes of PCIe graphics capability.
• The next most important spec is that all FX processors will have unlocked multipliers. Fun for everybody, regardless of budget!

The next slide pretty much puts the FX-8150 against the 2500K…

Next up is some benchmarking and gaming graphs comparing to its competition.

and some more slides on gaming improvements[both on performance and overall systems cost.

Next up..some info about the benefits of Turbo core(left) and some comparison with an i7 980X based System(right):

Now comes the part we all are excited about..the future!!! We see some heavy construction equipments there;lets hope they get the job done.

You all saw in the slide above that the FX-8150 is a worthy competitor to an i5 2500K but didn’t quite catch its big brother i7 2600K. A 10-15% improvement with Piledriver should close that gap to almost nothing. Intel wont stop though so AMD has to work on the IPC.

So there u have it guys..the Bulldozer in a nutshell.Its a step away from the norm as said earlier..so i urge all of you to not compare with the core i series directly. What AMD did is they have gone for a complete rewrite of the processor as we know normally so its unfair if they are judged on the scales of a existing product.

The Setup

The press KIT:

The chip & Motherboard:

The Setup:

Overclocking

Next up..overclocking and benchmarks(we all love benchmarks….aint it?)

Overclocking the FX is pretty easy and no fuss affair since its an unlocked cpu. Out of the box it booted into 3.6Ghz with no fuss!

With a little bump in the multiplier and voltages i settled for a nice stable overclock of:

and being the nut i am..tried to find the absolute limit i could go on AIR;and i ended up with this:

well…since overcloking the Bulldozer was too FUN;i thought maybe i shuould list the settings i used to OC the FX to different speeds.Here are the setting..milege may vary with your Chip

Our Benchmark Reults

Since the processor is a direct competitor (price wise) to a i5 2500k I will be comparing it with a Core i5 2500k.

Details of the test systems:

stock test were done three times and the average was taken into account;the overclocked test(if any) were run only once. Now lets continue to the benchmarks…shall we?? The graphs you see below are made relative to the FX-8150’s performance.

1) AIDA64 CPU SUITE , AIDA64 FPU SUITE (higher is better) :

These tests were run only at stock to give a comparison of how the chips perform under various testing conditions. Seems like AMD’s positioning of this chip against 2500K was right. Both the chips trade blows throughout the CPU tests. FPU tests Doesn’t look right..isn’t? well…although its pretty much keeping up with the competition..if there had been 8 separate FPU cores..it would have killed the 2500K. For whatever reasons.AMD decided to go all out with the CPU cores. NOT the FPU cores.

2) Super PI and Wprime (lower is better)

Not much to say here..AMD gave up on super pi long time ago.The bench uses X86 floating point instruction set(AMD’s Achilles’ heel).so its a given AMD will not be good for Super PI.

As fro Wprime since its a multi threaded benchmark , AMD should blow 2500K out of the water…right? WRONG! AMD would say this is an older benchmark. I would say otherwise. This processor is supposed to be the epitome of multi-core performance. While it still beats its stated competition (2500K remember), benchmarks indicates it lost ground Thuban.

3D benchmarks

Well…2D was a no go…so can Bulldozer pick some point or boints as they call in the bots in 3D benchmarks? the answer is…sadly…NO! The difference is somewhat small between the 2500k and FX-8150 in 3Dmark 11(where CPU plays a very small role) and it was somewhat big in 3DMark Vantage(where CPU plays a big role)..so if i were a 3D loving guy…getting a 2500K makes more sense.

Rendering, Video Conversion and Compression

AMD’s been putting too much hype in Real World performance;they say that’s where they will shine;well,they does.FX-8150 beats the 2500K at stock in cinebench..and when overclocked kicks the daylight out 2500K! So we got a clear winner there in rendering. Lets see what it does when we add some video encoding when rendering? we put the two contenders through X264 benchmark and POV ray benchmark.
FYI pass 1 is only scanning.no encoding so no chance of seeing the performance there but if there encoding involved then AMD happily kicks the 2500K around. So, in real world use BULLDOZER is not bad..its good even..it does what its supposed to do;beat the 2500K.That’s a good thing in my books.

Conclusions
Well folks..that conclude’s my small review on the FX-8150. The CPU is a mixed bag.AMD took a big chance with it and reinvented the wheel with the FX.In the end,it loses in some places,it wins in some places.It wins where it matters , real world performance and it competes nicely with its slated competitor 2500K. So as far as the real world performance goes..its a winner.

NOW , coming to AMD’s main winning formula value for money. Its not priced to topple I7 2600K @ $330;2600K still remains the king of the hill mainstream chip.The Fx-8150 priced at around $260 is an completely unlocked chip and looking at the performance numbers the FX is a clear winner against the I5 2500K priced at around $230.the $30 difference is well worth the money in my opinion.

From an overclockers point of view the CHIP is golden..clocks like a madman..goes places where no sandybridge can go in term of raw cpu power.

In my defense i quote a few CPU frequency submissions from HWBOT :

1) 2600K :
6044.28 MHz
CPU-Z Validator 3.1

2) 2500K :
6009.91 MHz
CPU-Z Validator 3.1

3) FX-8150 :
8585.05 MHz
CPU-Z Validator 3.1

Clocking this chip under air is also fun.much more exciting then Intel sandybridge but maybe that’s just me.

So in the END the FX-8150 gets an “approved” stamp from me and the OCF team.

-Regards ,
Sumon Pathak

UPDATE:
well,i had a detailed discussion about why Bulldozer performance was low due to the thread schedulers incapability in Windows Seven and he agreed to explain all in one small article for the readers.
So guys,please read this excellent article by cilus aka Suryasis in ther link given below:

Originally Posted by Cilus

Before starting the explanation lets be clear with our assumption:-
Let’s consider a Dual module based Bulldozer processor, say FX4100. We have two modules here: M1 and M2. M1 has two integer cores, say C1 and C2 and one Floating Point Core, say F1. Similarly consider M2 has two integer cores C3 and C4 and F2 as the Floating point unit.
In Windows 7 there is no existence of M1 and M2 and it sees four completely independent cores, C1, C2, C3 and C4. It is also not aware about the presence of only two floating point execution units instead of four.