Missing </a>?Code:<h1><img src="css/icons/openbenchmark-small.png" /> bulldozer</h1><p><strong>1110165-LI-1110131LI64:</strong> AMD FX -8150 Eight-Core testing with a ASUS SABERTOOTH 990FX and NVIDIA GeForce 8800 GT 512MB on Gentoo Base release 2.1 via the Phoronix Test Suite.</p><p><strong>A comparison of:</strong> 1110165-LI-1110131LI64, 1110165-AGY-1110158L96, 1110164-GR-1110131LI17, 1110158-LI-1110131LI42, 1110152-LI-1110131LI59, 1110150-GR-1110131LI69, 1110147-GR-1110131LI32, 1110159-LI-1110131LI24, 1110158-IV-1110131LI29.</p><a name="system_table" /><h2>System Hardware / Software</h2>
ATM its seems that one bulldozer Module is 55% faster then one? thuban core: (AMD was promising 80% per CMT core)
So four Modules == 6.2 cores. (... lets ignore the new instructions for a bit ;P)
Imho .. 8150 is a quad core with 50% more power for everything but single threaded apps. The cache trashing due to missing patches/CMT scheduler finish him off.
Seems someone submitted something "more" optimized (compiler?):
and with 2600K:
but its not a 990FX board ;(
MT performance seems to suffer .. a bit.
I would love to see a run of the benchmark suite using AMD reference compiler Open64 (a forked version of GCC with features from SGI's MipsPro).
It's free and open source just like GCC. Available here http://developer.amd.com/tools/open6...s/default.aspx
As far as I know the stock GCC can not do any optimizions for AMD's new AVX/FMA3/FMA4/XOP instruction sets, you need Open64 for that.
I started with a clean Gentoo install running -march=native -O2 and GCC 4.6.1. I did a trial run at 4.8ghz, and I had it up to 5ghz last night. I don't seem to be hitting any thermal barriers as I ran through the pts/compression benchmark at 4.8ghz and my cpu was at 29C (water cooled).