Phoronix AMD Bulldozer Linux review

[amdfangirl]

AMD FX-8150 is a competitive product to the Intel Core i5 2500K when dealing with multi-threaded workloads. For single-threaded work and other select tasks, the Bulldozer performance is disappointing.

Fortunately, for Linux users, most open-source software is well multi-threaded. If you are running Gentoo, Arch, or just doing a lot of compiling in general, the AMD Bulldozer CPUs should be able to prove their value very well. Beyond that, with open-source software that you may be building, GCC and Open64 already have Bulldozer (version 1) optimizations in place.

Linux users will be able to take full advantage of the Bulldozer architecture sooner than Microsoft Windows customers, which will primarily see the real potential when Windows 8 is released. In the Linux world, there's still some Bulldozer kernel work that's not yet merged and presumably more compiler/kernel optimizations coming, but we will hopefully see all of that merged and ready in time for next spring when Ubuntu 12.04 LTS, Fedora 17, and other Linux distributions are pushing out their new versions. If you are so inclined, you can always pull the patches yourself, tune your compiler options, and make other tweaks today to take greater advantage of these new AMD processors. The upcoming FX-8150 Linux articles have more revealing information.

Another advantage to the Bulldozer CPUs is that they are unlocked and can be overclocked very easily. Hitting around 4.6GHz on the FX-8150 is a breeze, as one of the upcoming articles illustrates, and banging 5GHz is not out of the question at all. Besides the poor performance with single-thread tasks, another disadvantage of the FX-8150 is the price. The launch price of the FX-8150 retail (without water cooling) is $279 USD. This is roughly $60 USD more than the Intel Core i5 2500K "Sandy Bridge", which frequently was faster under Ubuntu Linux.

http://www.phoronix.com/scan.php?page=article&item=amd_fx8150_bulldozer&num=1

Interesting results. Bulldozer actually wins some here. 'Cept for the crazy pricing.

 

[FALC0N]

Pretty fair assessment. More or less what we saw from windows reviews, without the emotional response.

 

[amdfangirl]

But it is interesting to note this disparity between Linux and Windows in terms of multi-threading. It seems to make a signifigant difference. Mind you this is tested without the Bulldozer patches, so BD does have the potential to perform better.

Shame there was no i7-2600K and Phenom II X4/X6 results.

C'est la vie.

 

[FALC0N]

True.

Perhaps a glimpse into a promising future for the arch.

 

[amdfangirl]

If you read all of the benchmarks, Bulldozer really isn't bad.

Since most of the Linux software is well threaded, it sometimes even beats the 2500K.

Just it really needs a price cut.

Just a little hard with 2B transistors.

I'm very excited about the BD optmised Linux benchies, should be interesting.

Will post as so as my RSS feed tells me to.

 

[de5_Roy]

finished reading the article. thanks for the link amdfangirl

 

[Cazalan]

Good comparison but missing power tests.

Bulldozer isn't that bad it's just inefficient.

 

[farrengottu]

i stopped caring about efficiency a while back. but gaming rigs have that effect on people. i just cant care if my power bill is $4 higher at the end of the month.

i love linux just dont love using it. i like simple and linux isnt.
ive used it on and off over the last 15 years but never stopped using windows.
i see that linux is better than windows 7 at using this many cores but since the windows 8 developer version is free and mostly working right now i would slap that on any FX processor for a 5-10% boost in multi-threaded applications over win 7.
these processors are for people that do a lot of stuff at once. something benchmarks aren't representing well.

 

[EXT64]

That was impressive. I'm starting to get a little more excited about my Linux BD testing I'll be doing soon

 

[dogman_1234]

@EXT64,

Cant wait for the results. You may officially motivate me to do a build if it goes well.

 

[EXT64]

Yep, hopefully things will go well. The BD system price wasn't bad, under $100 for the MB and $220 for the 8120. I would have probably spent $30-50 more on a 2500k and Motherboard, so if this thing really does multithreading well, I may be very happy with this chip. And when it gets cold in a few weeks, I might appreciate the high power consumption as well :lol:

 

[dogman_1234]

:lol:

Good Luck.

P.S. Tell us what your PSU is, if you overclocked, and what you bill was for the time being...if you don't mind?

 

[amdfangirl]

That was impressive. I'm starting to get a little more excited about my Linux BD testing I'll be doing soon

Can't wait for the BD optimised benchmarks.

 

[EXT64]

First set won't be OC'ed - not because I'm worried about temps or power, but I just won't have the time initially to do it properly. When/if I do OC, it'll probably just be a little one (up to about 3.6) so that it is at the same clocks as my i7 970 and a 8150 (just so I can feel like I saved $50).

PSU is an Antec 650W (can't remember exact details). I already have the PSU and have been using it for a while - so I've forgotten some of the details. I'll post full specs when I build later this week/weekend.

 

[mayankleoboy1]

its in linux that BD performs as we all expected. between 2500k and 2600k, beating 2500k in many benchies.

so this means in windows , we definitely need a patch too.

 

[mayankleoboy1]

also, the programs tested were not optimised for bulldozer arch.
the next round of benchies will have compiler optimised apps. so hopefully the performance will be better.

but on linux, BD doesnt look half as bad.

 

[amdfangirl]

I can't wait for the server benchmarks

 

[mayankleoboy1]

these benchies make me almost want to buy a BD.
almost, but not quite.

 

[amdfangirl]

BD multi-core scaling in Linux...

http://www.phoronix.com/scan.php?page=article&item=amd_bulldozer_scaling&num=1

 

[EXT64]

That is really neat. Its amazing how much variability there is between workloads - I think almost all of the processors had at least one benchmark where they had near perfect thread scaling, and another where they fell flat on their face.

Maybe I missed this, but did he disable Turbo for the tests? Just wondering why the quad core mobile part did so badly compared to its single thread (if turbo were on, that would explain).

 

[amdfangirl]

AMD Bulldozer With GCC, Open64, LLVM/Clang Compilers

Open64 Compiler Tuning On AMD Bulldozer FX-8150

 

[jimmysmitty]

But it is interesting to note this disparity between Linux and Windows in terms of multi-threading. It seems to make a signifigant difference. Mind you this is tested without the Bulldozer patches, so BD does have the potential to perform better.

Shame there was no i7-2600K and Phenom II X4/X6 results.

C'est la vie.

I agree. But its where it is. About the level of a 2500K which is still a little sad as AMD is marketing it as a 8 core.

 

[mayankleoboy1]

check the latest benchmarks with cache hit fix.
i was hoping for something miracle in the range of 10-15%.
actual results? 2%.

 

[dogman_1234]

check the latest benchmarks with cache hit fix.
i was hoping for something miracle in the range of 10-15%.
actual results? 2%.

On what unit?

 

[mayankleoboy1]

results with modified kernel = 102% of results with unmodified kernel.
this kernel fix is committed to linux 3.2

these modifications made by AMD engineer(s) to prevent the cache hit issue is BD. apparently had been working on it for months.
in retrospect, foolish of me to even think of a 10-15% improvement. BD is dud-ish.