diff options
Diffstat (limited to 'devtools/PMELib.htm')
| -rw-r--r-- | devtools/PMELib.htm | 206 |
1 files changed, 206 insertions, 0 deletions
diff --git a/devtools/PMELib.htm b/devtools/PMELib.htm new file mode 100644 index 0000000..7c5e06c --- /dev/null +++ b/devtools/PMELib.htm @@ -0,0 +1,206 @@ +<html> + +<head> +<meta http-equiv="Content-Language" content="en-us"> +<meta http-equiv="Content-Type" content="text/html; charset=windows-1252"> +<meta name="GENERATOR" content="Microsoft FrontPage 4.0"> +<meta name="ProgId" content="FrontPage.Editor.Document"> +<title>New Page 1</title> +</head> + +<body> + +<p align="center"><b><font size="5">PMELib 1.0</font></b></p> +<h2>Introduction</h2> +<p><font size="3"> This library is an interface to the Performance Monitoring +Events (PME) that are available in the Pentium P5, P6 and P4 processors. +There are 18 counters that let you gather information about what the processor is +going during execution. They are described in these manuals:</font></p> +<blockquote> + <p><a href="http://developer.intel.com/design/pentium4/manuals/253665.htm"><font color="#008000" size="3">IA-32 + Intel Architecture Software Developer's Manual Volume 1: Basic Architecture</font></a></p> + <p><a href="http://developer.intel.com/design/pentium4/manuals/253666.htm"><font color="#008000" size="3">IA-32 + Intel Architecture Software Developer's Manual Volume 2A: Instruction Set + Reference, A-M</font></a></p> + <p><a href="http://developer.intel.com/design/pentium4/manuals/253667.htm"><font color="#008000" size="3">IA-32 + Intel Architecture Software Developer's Manual Volume 2B: Instruction Set + Reference, N-Z</font></a></p> + <p><a href="http://developer.intel.com/design/pentium4/manuals/253668.htm"><font color="#008000" size="3">IA-32 + Intel Architecture Software Developer's Manual Volume 3: System Programming + Guide</font></a></p> + <p><font size="3"> <font color="#008000" size="3"> </font><font size="3" color="#000000">See + chapter 15, Appendix A and Appendix B</font></font></p> +</blockquote> +<p><font size="3">Pentium 4 documentation is available <a href="http://developer.intel.com/design/Pentium4/documentation.htm">here</a></font></p> +<p><font size="3">This library is an extension to the utilities from the Game Developer's +Magazine <a href="http://www.gamasutra.com/features/wyatts_world/19990528/pentium3_08.htm">article</a> +by Robert Wyatt in May 1998. The library is now in a class and included +the Pentium 4 processor. Only Intel is currently supported and +tested. </font></p> +<p> </p> +<p><font size="3">Send feedback to <a href="mailto:[email protected]">[email protected]</a></font></p> +<p> </p> +<p> </p> +<h2><b>Installation</b></h2> +<p><font size="3"> Window NT and Windows XP are supported and tested. +Win98 may work, however.</font></p> +<p><font size="3"> </font></p> +<p><font size="3"> You need to install a driver, set some registry settings +and reboot. If you installed the GDPerf.sys from the GD magazine article, +you can skip the installation step. It uses the same driver. Let me know +if you have problems.</font></p> +<p><font size="3"> In the <i>Installation</i> +directory</font></p> +<p><font size="3"> copy +GDPerf.sys to the window driver directory</font></p> +<p><font size="3"> +copy GDPerf.sys <a href="file:///C:/windows/system32/drivers">C:\windows\system32\drivers</a></font></p> +<p><font size="3"> +there is a batch file as an example</font></p> +<p><font size="3"> Run the +PMELib.reg file to set the registry settings</font></p> +<p><font size="3"> Reboot</font></p> +<p> </p> +<h2><b>Configuring PMELib</b><font size="3"> </font></h2> +<p><font size="3">For P5 and P6 processors (anything before Pentium 4), You use the same +interfaces that are described in the Game Developer article. They have +just been incorporated in to the PMELib as is.</font></p> +<p><font size="3"><font color="#000000">In the Pentium 4, are 18 performance +monitoring counters and more than 40 Events Modes that can be captured. +Each Mode has a bit mask that indicates which tests to perform. These are +described in Appendix A of </font><font color="#008000" size="3"><a href="http://developer.intel.com/design/pentium4/manuals/253668.htm">IA-32 +Intel Architecture Software Developer's Manual Volume 3: System Programming +Guide</a> </font><font size="3" color="#000000"> Each of these event +mode has a class dedicated to it. The event modes are listed below.</font></font></p> +<blockquote> + <p><font size="3">Set PerfTest2 for an example.</font></p> + <p><font size="3"><b>Step 1) Choose an Event Mode class</b></font></p> + <p><font size="3"> Example:</font></p> + <p><font size="3"> + Event_branch_retired event;</font></p> + <p> </p> + <p><font size="3"><b>Step 2) Set the Event Mask for the selected Event + Mode</b></font></p> + <p><font size="3"> Example:</font></p> + <p><font size="3"> event.eventMask->MMNP = 1;</font></p> + <p><font size="3"> event.eventMask->MMTP = 1;</font></p> + <p> </p> + <p><b><font size="3">Step 3) Set the privilege level to capture data + from with the <i>SetCaptureMode</i> method</font></b></p> + <blockquote> + <p><font size="3">OS_Only, <font COLOR="#008000">// ring 0, driver + level only + </font></font></p> + <p><font size="3">USR_Only, <font COLOR="#008000">// app level, privilege + levels 1 2 and 3 + </font></font></p> + <p><font size="3">OS_and_USR, <font COLOR="#008000">// all levels 0, 1, 2 + and 3 + </font> + </font></p> + </blockquote> + <p><font size="3"> Optionally, you + can enable tagging in the <i>SetCaptureMode</i> method.</font></p> + <p><font size="3"> Example:</font></p> + <p><font size="3"><font COLOR="#0000ff"> + </font>SetCaptureMode(OS_and_USR, TagEnable, 34);</font></p> + <p> </p> + <p><font size="3"><b>Step 4) Optional Configuration</b></font></p> + <p><font size="3"> <b> </b>At this + point you can configure Tagging, Filtering, Overflow and Cascading + options. You can also select one of the legal counters for the selected + Event Mode.</font></p> + <p><font size="3"><b>Step 5) Set the process priority to high</b></font></p> + <blockquote> + <p><font size="3">This reduces the noise from other processes interfering. If you + have an infinite loop in you code and you have these set, you may hang and + need to reboot</font></p> + <p><font size="3">Example:</font></p> + <p><font size="3"> PME * pme = PME::Instance();</font></p> + <p><font size="3"> pme->SetProcessPriority(ProcessPriorityHigh);</font></p> + <font SIZE="2"> + <p> </p> + </blockquote> + </font> + <p><font size="3"><b>Step 6) Start using the counters</b></font></p> + <p><font size="3"> Each Event Mode + counter has the follow ability:</font></p> + <blockquote> + <p><font size="3"> Stop</font></p> + <p><font size="3"> Start</font></p> + <p><font size="3"> Clear - set to + 0</font></p> + <p><font size="3"> Read</font></p> + <p><font size="3"> Write - write a + 64 bit counter value</font></p> + </blockquote> + <p><font size="3"> </font></p> + <p><font size="3"><b>Step 7) Set the process priority to normal</b></font></p> + <p><font size="3"> pme->SetProcessPriority(ProcessPriorityNormal);</font></p> + <h2> </h2> +</blockquote> +<h2><font size="3"><b>Event Modes:</b></font></h2> +<blockquote> + <blockquote> + <p><font size="3">Event_TC_deliver_mode</font></p> + <p><font size="3">Event_BPU_fetch_request</font></p> + <p><font size="3">Event_ITLB_reference</font></p> + <p><font size="3">Event_memory_cancel</font></p> + <p><font size="3">Event_memory_complete</font></p> + <p><font size="3">Event_load_port_replay</font></p> + <p><font size="3">Event_store_port_replay</font></p> + <p><font size="3">Event_MOB_load_replay</font></p> + <p><font size="3">Event_page_walk_type</font></p> + <p><font size="3">Event_BSQ_cache_reference</font></p> + <p><font size="3">Event_IOQ_allocation</font></p> + <p><font size="3">Event_IOQ_active_entries</font></p> + <p><font size="3">Event_FSB_data_activity</font></p> + <p><font size="3">Event_BSQ_allocation</font></p> + <p><font size="3">Event_BSQ_active_entries</font></p> + <p><font size="3">Event_SSE_input_assist</font></p> + <p><font size="3">Event_packed_SP_uop</font></p> + <p><font size="3">Event_packed_DP_uop</font></p> + <p><font size="3">Event_scalar_SP_uop</font></p> + <p><font size="3">Event_scalar_DP_uop</font></p> + <p><font size="3">Event_64bit_MMX_uop</font></p> + <p><font size="3">Event_128bit_MMX_uop</font></p> + <p><font size="3">Event_x87_FP_uop</font></p> + <p><font size="3">Event_x87_SIMD_moves_uop</font></p> + <p><font size="3">Event_TC_misc</font></p> + <p><font size="3">Event_global_power_events</font></p> + <p><font size="3">Event_tc_ms_xfer</font></p> + <p><font size="3">Event_uop_queue_writes</font></p> + <p><font size="3">Event_retired_mispred_branch_type</font></p> + <p><font size="3">Event_retired_branch_type</font></p> + <p><font size="3">Event_resource_stall</font></p> + <p><font size="3">Event_WC_Buffer</font></p> + <p><font size="3">Event_b2b_cycles</font></p> + <p><font size="3">Event_bnr</font></p> + <p><font size="3">Event_snoop</font></p> + <p><font size="3">Event_response</font></p> + <p><font size="3">Event_front_end_event</font></p> + <p><font size="3">Event_execution_event</font></p> + <p><font size="3">Event_replay_event</font></p> + <p><font size="3">Event_instr_retired</font></p> + <p><font size="3">Event_uops_retired</font></p> + <p><font size="3">Event_uop_type</font></p> + <p><font size="3">Event_branch_retired</font></p> + <p><font size="3">Event_mispred_branch_retired</font></p> + <p><font size="3">Event_x87_assist</font></p> + <p><font size="3">Event_machine_clear</font></p> + </blockquote> + </blockquote> +<h2><b>Credits</b></h2> +<p><a href="http://www.gamasutra.com/features/wyatts_world/19990528/pentium3_08.htm"><font color="#008000" size="3">http://www.gamasutra.com/features/wyatts_world/19990528/pentium3_08.htm</font></a></p> +<p><font color="#008000" size="3">Used some tables from <a href="http://user.it.uu.se/~mikpe/">Mikael Pettersson's</a> +<a href="http://user.it.uu.se/~mikpe/linux/perfctr/">pertctf</a> +</font> +</p> +<p align="left"><font color="#008000" size="3">Used the detect code from<a href="http://yotov.org"> +Kamen Yotov's</a> <a href="http://iss.cs.cornell.edu/ia32.htm">ia32lib library</a></font></p> +<p> </p> +<p> </p> + +</body> + +</html> |