AMD ATHLON 64 - Manual (Page 10)

10 Introduction Chapter 1
40555 Rev. 3.00 June 2006
Performance Guidelines for AMD Athlon™ 64 and AMD Opteron™
ccNUMA Multiprocessor Systems
bandwidth test, it exercises both of these modes of operation. The test serves as a latency sensitive test
case when the test threads perform read-only operations and as a bandwidth sensitive test when the
test threads carry out write-only operations. The discussion below explores the performance results of
this test, with an emphasis on behavior exhibited when the test imposes high bandwidth demands on
the low level resources of the system.
Additionally, the tests are run in undersubscribed, highly subscribed, and fully subscribed modes. In
undersubscribed mode, there are significantly fewer threads than the number of processors. In highly
subscribed mode, the number of threads approaches the number of processors. In the fully subscribed
mode, the number of threads is equal to the number of processors. Testing these conditions provides
an understanding of the impact of thread subscription on performance.
Based on the data and the analysis gathered from this synthetic test-bench, this application note
presents recommendations to software developers who are working on applications, compiler tool
chains, virtual machines and operating systems. Finally, the test results should also dispel some
common myths concerning identical performance results obtained when comparing workloads that
are symmetrical in all respects except for the thread and memory placement used.
1.1 Related Documents
The following web links are referenced in the text and provide valuable resource and background
information:
[1] http://www.hotchips.org/archives/hc14/3_Tue/28_AMD_Hammer_MP_HC_v8.pdf
[2] http://www.kernel.org/pub/linux/kernel/people/mbligh/presentations/OLS2004-
numa_paper.pdf
[3] http://www.amd64.org/lists/discuss/msg03314.html
[4] http://www.pgroup.com/doc/pgiug.pdf
[5] http://www.novell.com/collateral/4621437/4621437.pdf
[6] http://opensolaris.org/os/community/performance/mpo_overview.pdf
[7] http://www.opensolaris.org/os/community/performance/numa/observability/
[8] http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dllproc/base/
multiple_processors.asp
[9] http://msdn.microsoft.com/library/default.asp?url=/library/en-us/memory/base/
virtualalloc.asp
[10] http://msdn2.microsoft.com/en-us/library/ms186255(SQL.90).aspx
[11] http://www.microsoft.com/technet/prodtechnol/WindowsServer2003/Library/IIS/
529588d3-71bc-45ea-a84b-267914674709.mspx