--- lnet/klnds/mxlnd/README.old	2006-12-22 20:35:21.000000000 -0500
+++ lnet/klnds/mxlnd/README	2006-12-22 20:44:25.000000000 -0500
@@ -9,7 +9,7 @@
 MXLND provides support for Myricom's Myrinet Express (MX) communication
 layer in Lustre.
 
-MXLND may be used with either MX-10G or MX-2G. See MX's README for
+MXLND may be used with either MX-10G or Myrinet-2000. See MX's README for
 supported NICs.
 
 Table of Contents:
@@ -32,6 +32,14 @@
 but it has not been tested. MXLND requires Myricom's MX version 1.2.1
 or higher. See MX's README for the supported list of processors.
 
+Because LNET modules are kernel-based, you will need to configure MX (not
+MXLND) this option:
+
+    --enable-kernel-lib
+
+Lustre may configure and build without it, but it will not load and you
+will find output in dmesg regarding undefined objects.
+
 1. Configuring and compiling
 
 MXLND should be already integrated into the Lustre build process. To 
@@ -104,7 +112,8 @@
 
     ntx is the number of total sends in flight from this machine. In actuality,
 MXLND reserves half of them for connect messages so make this value twice as large
-as you want for the total number of sends in flight.
+as you want for the total number of sends in flight. Each send descriptor uses
+a page of kernel memory.
 
     credits is the number of in-flight messages for a specific peer. This is part
 of the flow-control system in Lustre. Increasing this value may improve performance
@@ -122,19 +131,20 @@
     polling determines whether this host will poll or block for MX request com-
 pletions. A value of 0 blocks and any positive value will poll that many times
 before blocking. Since polling increases CPU usage, we suggest you set this to
-0 on client and experiment with different values for servers.
+0 on client and experiment with different values sarting with 0 for servers.
 
 =====================
 II. MXLND Performance
 =====================
 
-On MX-2G systems, MXLND should easily saturate the link and use minimal CPU 
-(5-10% for read and write operations). On MX-10G systems, MXLND can saturate 
-the link and use moderate CPU resources (20-30% for read and write operations).
-MX-10G relies on PCI-Express which is relatively new and performance varies
-considerably by processor, motherboard and PCI-E chipset. Refer to Myricom's
-website for the latest DMA read/write performance results by motherboard. The
-DMA results will place an upper-bound on MXLND performance.
+On MX-2G systems, MXLND should easily saturate the link and use minimal CPU
+(5-10% for read and write operations). On MX-10G systems, MXLND can saturate
+the link and use minimal CPU resources (8-12% for read and write operations)
+with a single client and server.  MX-10G relies on PCI-Express which is
+relatively new and performance varies considerably by processor, motherboard
+and PCI-E chipset. Refer to Myricom's website for the latest DMA read/write
+performance results by motherboard. The DMA results will place an upper-bound
+on MXLND performance.
 
 ============
 III. Caveats
@@ -150,7 +160,7 @@
 2. Multi-homing
 
 At this time, Lustre does not support multi-homing of non-TCP interconnects.
-Thus, a single cannot route between two MX-10G, between two MX-2G, or 
+Thus, a single host cannot route between two MX-10G, between two MX-2G, or
 between MX-10G and MX-2G fabrics.
 
 3. MX endpoint collision