Release Notes for SMCPWR NDIS 2 MAC Driver V2.60 for DOS, WFW, OS/2 and Win95
===========================================================================================
New features and modifications in V2.60:
--------------------------------------------------------
1. The Driver now supports enabling of Extended PCI commands,
(the commands new as of PCI spec 2.1), for those 2104x/2114x
devices which support the commands.
Enabling the commands may increase PCI or CardBus throughput,
depending on the system implementation.
As a default, the driver will enable the maximum possible number of
the three Extended PCI Commands,
Memory Write Invalidate (MWI);
Memory Read Line (MRL);
Memory Read Multiple (MRM);
after verifying that the system implementation (PCI bridge chipset
and BIOS) supports it.
The devices for which the commands will be enabled:
21140A rev >= 21 ,
21143 rev >= 20.
Due to incompatibility problems with certain PCI host bridges
and PCI-to-PCI bridges, the driver currently does not enable
the Memory Write Invalidate (MWI) and Memory Read Line (MRL) commands
simultaneously. The driver will only enable one of the two commands;
by default, MRL will be enabled.
The driver has 3 new keywords (EXT_MWI, EXT_MRM, and EXT_MRL). These
keywords enable and disable the operation of the respective Extended
PCI Commands, to allow user override; in no case will an invalid
combination be allowed.
2. Support for SROM version 4.0 with or without magic packet (TM of AMD)
block. This also fixed a bug in calculating the CRC in the SROM. the
bug caused illegal SROMs to be accepted.
3. Keywords unification with NDIS 3 driver:
Summary of changes in ndis2 keywords:
3.1. the two ndis2 keywords DATA_RATE and SIA_MODE will be united to the
single ndis3 keyword ConnectionType.
Below is a comparison of the ndis3 keywords for wfw 3.11 (as written in
the .inf file), to the ndis2 keywords. note that ndis3 keywords are
translated into numbers in protocol.ini , and ndis2 keywords are remained
as strings in the protocol.ini file.
ndis2 ndis3
----- -----
Autosense Autosense
_10Base2_(BNC) 10Base2 (BNC)
_10Base5_(AUI) 10Base5 (AUI)
_10BaseT 10BaseT
_10BaseT_FD 10BaseT FD
_10BaseT_No_Link_Test 10BaseT No_Link_Test
--- AutoSense No_Nway
--- Reserved
_100BaseTx 100BaseTx
_100BaseTx_FD 100BaseTx FD
_100BaseT4 100BaseT4
In Win95 the ndis2 keywords are identical to those of ndis3, and are
translated to the original keywords in protocol.ini
3.2. keywords of ndis3 that present functionality that isn't relevant to
the ndis2 driver are not unified.
keywords list:
AdapterType, BusNumber, SlotNumber, FunctionNumber,
AdapterCfid, PciCommand,
ExternalSIA, SiaRegister0-3,
MinTransmitBuffers,
NwayProtocol, MiiNway,
MapRegisters.
3.3. some keywords of ndis3 had equivalent keywords in the ndis2 driver,
that have different names. these keywords names are changed in
the ndis2 driver. (keeping the old names in the next releases)
keywords list:
SoftwareCrc,
SnoozeMode,
TransmitThreshold/100 - see PS below.
3.4. some ndis3 keywords don't have ndis2 equivalents, but have odi-client
equivalents. these keywords were added.
keywords list:
BurstLength,
PciLatencyTimer,
CacheLineSize,
MRM_OFF, MRL_OFF, MWI_OFF
3.5. in ndis2 driver, there were keywords for the number of both transmit
and receive buffers - NUM_TX_BUFFERS and NUM_RX_BUFFERS. in ndis3 -
only for receive buffers - ReceiveBuffers. the two keywords
ReceiveBuffers and TransmitBuffers were added to ndis2 driver,
while keeping the old keywords for the meantime.
4. Some bug fixes in the ndis2 statistics.
5. A bug fixed in calling the ReceiveLookAhead routine. If the protocol left
the indications off, and no more packets were received since the call until
the IndicationOn request by the user, the packets which were delayed (received
before the call to ReceiveLookAhead) were not indicated to the upper level,
until another packet was received. This bug was fixed in the current version.
6. Fixed some theoretical problems in code reentrancy of the driver.
7. fixed a big that caused broadcast packets to be accepted even when it should
not be.
Testing performed prior to release
-----------------------------------
All testing described below was carried out on combinations of the
following parameters:
- Devices: Digital Semiconductor chips 21040, 21041, 21140, 21142, 21143.
- Different implementations (boards) of the devices named above, with
various SROM programming.
- Network data rate: 10Mb or 100Mb.
- Local bus: PCI.
- Operating systems: DOS, WFW31, WFW311, OS/2 2.x, OS/2 3.0, Win95.
- Networking OS: Lan Manager(client, server), Lan Server(requester, server),
Windows and Win95 native networking components. All these components
have several different versions.
- Various network setups: Upper protocols (e.g. TCP/IP, NetBEUI), internal
parameters (protocol buffer size, network orientation - connection or
connectionless, and more).
- Various device parameters: Num of buffers, thresholds.
- Serial connections and variations: Tp, Tp full duplex, AUI, BNC.
- Machine-specific parameters: CPU speed, PCI bridge implementations, BIOS
and PCI BIOS implementations, memory, multiple NICs.
NDIS2 conformance:
------------------
MTTOOL conformance test was run on the DOS driver in many different
combinations. All tests were point-to-point configurations using
the different devices (21040, 21041 21140, 21142, 21143) in 10Mb
or 100Mb playing
with driver's parameters (threshold, num of buffers), doing the testing
on different machines (with varying CPU speed and PCI implementation).
Testing was smooth besides:
1. Several tests failed in 100Mb mode due to the high data transfer
rate. We consulted Microsoft on the matter and were granted a waiver.
Stress testing:
---------------
Tests we used are: MTTOOL stress tool, ReadRite, CCtest, WinTest.
WinMCL testing wasn't carried out since
Microsoft granted us a waiver (there are problems with the test
itself).
Some of the stress tests are long (36-96 hours) and some short (2-
24 hours).
Setups:
-------
We tried different setups, parameter combinations, etc. Among these:
- Multiple NICs setups: Two NICs of ours - using the same or different
devices, one of ours and other vendor's NIC. These were tried on
LanServer 3.0, 4.0, WFW 3.1 and 3.11 .
Both automatic installation and manual installation were tried.
- PROTOCOL.INI parameter variations: Num of rcv/tx buffers, Tx threshold,
SW CRC, and more.
- SROM parameter variations: Changing version num, changing default media,
using legacy format boards, coexistence with PROTOCOL.INI values, DS21041
DS21140 and DS21143, different board implementations.
- Serial-connection variations: FD/not FD, TP/AUI/BNC,
Using hard-coded values or autodetected values.
- Autodetection/Autosensing variations: Trying the different devices,
trying several run-time scenarios, trying strange cases (i.e. -
connecting the cable after initialization, pulling it out in the
middle of transactions, sticking two cables in the same time, ...) .
All this in different machines and OS/NOS cominations.
- Removal and reinstallation in different OS/NOS combinations.
- Coexistence with other devices - Networking devices and others.
Release Notes for SMCPWR NDIS 2 MAC Driver V2.41 for DOS, WFW, OS/2 and Win95
===========================================================================================
New features and modifications in V2.41:
--------------------------------------------------------
Full support for Micro-Linear ML6692 MII Phy.
The driver supports both Autosense Nway, and Forcing the Phy to a 10
or 100 MBps medium.
Testing performed prior to release
-----------------------------------
All testing described below was carried out on combinations of the
following parameters:
- Devices: Digital Semiconductor chips 21040, 21041, 21140, 21142, 21143.
- Different implementations (boards) of the devices named above, with
various SROM programming.
- Network data rate: 10Mb or 100Mb.
- Local bus: PCI.
- Operating systems: DOS, WFW31, WFW311, OS/2 2.x, OS/2 3.0, Win95.
- Networking OS: Lan Manager(client, server), Lan Server(requester, server),
Windows and Win95 native networking components. All these components
have several different versions.
- Various network setups: Upper protocols (e.g. TCP/IP, NetBEUI), internal
parameters (protocol buffer size, network orientation - connection or
connectionless, and more).
- Various device parameters: Num of buffers, thresholds.
- Serial connections and variations: Tp, Tp full duplex, AUI, BNC.
- Machine-specific parameters: CPU speed, PCI bridge implementations, BIOS
and PCI BIOS implementations, memory, multiple NICs.
NDIS2 conformance:
------------------
MTTOOL conformance test was run on the DOS driver in many different
combinations. All tests were point-to-point configurations using
the different devices (21040, 21041 21140, 21142, 21143) in 10Mb
or 100Mb playing
with driver's parameters (threshold, num of buffers), doing the testing
on different machines (with varying CPU speed and PCI implementation).
Testing was smooth besides:
1. Several tests failed in 100Mb mode due to the high data transfer
rate. We consulted Microsoft on the matter and were granted a waiver.
Stress testing:
---------------
Tests we used are: MTTOOL stress tool, ReadRite, CCtest, WinTest.
WinMCL testing wasn't carried out since
Microsoft granted us a waiver (there are problems with the test
itself).
Some of the stress tests are long (36-96 hours) and some short (2-
24 hours). Following are several sample setups that were used:
1. One server machine (DS21140, PCI, 486DX, TP, WFW3.11, 10Mb).
Three clients (2 DOS 1 OS/2) ran ReadRite stress on the net, all
using random file sizes ( < 512000 bytes ). The test ran for ~48 hrs.
2. One server machine (DS21143 100Mb, WFW3.1, TP, PCI) serving
two 10Mb clients for ReadRite stress, and being a client of a Pentium
machine 100Mb with WinTest. ~40 hours.
3. One server (different combinations). One WFW client with 6 - 12
open stress windows open to the server running CCtest or ReadRite.
Durtaion up to 96 hours.
4. Cyclic: 3 machines in a loop, each as a clint to one machine and server
to another. 10Mb using different stress tests.
Setups:
-------
We tried different setups, parameter combinations, etc. Among these:
- Multiple NICs setups: Two NICs of ours - using the same or different
devices, one of ours and other vendor's NIC. These were tried on
LanServer 3.0, 4.0, WFW 3.1 and 3.11 .
Both automatic installation and manual installation were tried.
- PROTOCOL.INI parameter variations: Num of rcv/tx buffers, Tx threshold,
SW CRC, and more.
- SROM parameter variations: Changing version num, changing default media,
using legacy format boards, coexistence with PROTOCOL.INI values, DS21041
DS21140 and DS21143, different board implementations.
- Serial-connection variations: FD/not FD, TP/AUI/BNC,
Using hard-coded values or autodetected values.
- Autodetection/Autosensing variations: Trying the different devices,
trying several run-time scenarios, trying strange cases (i.e. -
connecting the cable after initialization, pulling it out in the
middle of transactions, sticking two cables in the same time, ...) .
All this in different machines and OS/NOS cominations.
- Removal and reinstallation in different OS/NOS combinations.
- Coexistence with other devices - Networking devices and others.
Release Notes for SMCPWR NDIS 2 MAC Driver V2.40 for DOS, WFW, OS/2 and Win95
===========================================================================================
New features and modifications in V2.40:
--------------------------------------------------------
1. Added support for 21143 chip which includes:
- Force medium for TP, TP full duplex, BNC, AUI, 100BaseTX
and 100Base Tx Full duplex media.
- Force medium for media supported by MII Phys such as
TP, TP full duplex, 100BaseTX, 100Base Tx Full duplex and 100BaseT4 media.
- Power-Up autosense and Autosense Nway support for Nics with MII Phys.
- Power-up autosense for Nics using SYM Phy (such as QSI)
while not using the internal TP medium.
- Autosense Nway support for Nics implementing both SYM Phy and internal
TP medium via the Chip's internal Nway machine.
Notice that BNC and AUI media should be manually selected via PROTOCOL.INI
2. Added support to Micro-Linear ML6692 MII Phy.
The driver supports Forcing the Phy to a 10 or 100 MBps medium
(Autosensing is not fully supported yet).
Testing performed prior to release
-----------------------------------
All testing described below was carried out on combinations of the
following parameters:
- Devices: Digital Semiconductor chips 21040, 21041, 21140, 21142, 21143.
- Different implementations (boards) of the devices named above, with
various SROM programming.
- Network data rate: 10Mb or 100Mb.
- Local bus: PCI.
- Operating systems: DOS, WFW31, WFW311, OS/2 2.x, OS/2 3.0, Win95.
- Networking OS: Lan Manager(client, server), Lan Server(requester, server),
Windows and Win95 native networking components. All these components
have several different versions.
- Various network setups: Upper protocols (e.g. TCP/IP, NetBEUI), internal
parameters (protocol buffer size, network orientation - connection or
connectionless, and more).
- Various device parameters: Num of buffers, thresholds.
- Serial connections and variations: Tp, Tp full duplex, AUI, BNC.
- Machine-specific parameters: CPU speed, PCI bridge implementations, BIOS
and PCI BIOS implementations, memory, multiple NICs.
NDIS2 conformance:
------------------
MTTOOL conformance test was run on the DOS driver in many different
combinations. All tests were point-to-point configurations using
the different devices (21040, 21041 21140, 21142, 21143) in 10Mb
or 100Mb playing
with driver's parameters (threshold, num of buffers), doing the testing
on different machines (with varying CPU speed and PCI implementation).
Testing was smooth besides:
1. Several tests failed in 100Mb mode due to the high data transfer
rate. We consulted Microsoft on the matter and were granted a waiver.
Stress testing:
---------------
Tests we used are: MTTOOL stress tool, ReadRite, CCtest, WinTest.
WinMCL testing wasn't carried out since
Microsoft granted us a waiver (there are problems with the test
itself).
Some of the stress tests are long (36-96 hours) and some short (2-
24 hours). Following are several sample setups that were used:
1. One server machine (DS21140, PCI, 486DX, TP, WFW3.11, 10Mb).
Three clients (2 DOS 1 OS/2) ran ReadRite stress on the net, all
using random file sizes ( < 512000 bytes ). The test ran for ~48 hrs.
2. One server machine (DS21143 100Mb, WFW3.1, TP, PCI) serving
two 10Mb clients for ReadRite stress, and being a client of a Pentium
machine 100Mb with WinTest. ~40 hours.
3. One server (different combinations). One WFW client with 6 - 12
open stress windows open to the server running CCtest or ReadRite.
Durtaion up to 96 hours.
4. Cyclic: 3 machines in a loop, each as a clint to one machine and server
to another. 10Mb using different stress tests.
Testing went smooth besides:
1. One machine failed during several stress tests. The same
machine failed similarly when we tried other NICs on it, and we
therefore dropped the matter.
Setups:
-------
We tried different setups, parameter combinations, etc. Among these:
- Multiple NICs setups: Two NICs of ours - using the same or different
devices, one of ours and other vendor's NIC. These were tried on
LanServer 3.0, 4.0, WFW 3.1 and 3.11 .
Both automatic installation and manual installation were tried.
- PROTOCOL.INI parameter variations: Num of rcv/tx buffers, Tx threshold,
SW CRC, and more.
- SROM parameter variations: Changing version num, changing default media,
using legacy format boards, coexistence with PROTOCOL.INI values, DS21041
DS21140 and DS21143, different board implementations.
- Serial-connection variations: FD/not FD, TP/AUI/BNC,
Using hard-coded values or autodetected values.
- Autodetection/Autosensing variations: Trying the different devices,
trying several run-time scenarios, trying strange cases (i.e. -
connecting the cable after initialization, pulling it out in the
middle of transactions, sticking two cables in the same time, ...) .
All this in different machines and OS/NOS cominations.
- Removal and reinstallation in different OS/NOS combinations.
- Coexistence with other devices - Networking devices and others.
Release Notes for SMCPWR NDIS 2 MAC Driver V2.30 for DOS, WFW, OS/2 and Win95
===========================================================================================
New features and modifications in V2.30:
-----------------------------------------
1. The driver will activate the devices 21140A, 21142 as default in Run-mode.
Only when the Keyword "SNOOZE=ENABLE" put in the PROTOCOL.INI file the
driver will activate the devices 21140A, 21142 in Power-saving mode.
2. Corrected potential hang of receive procces detected during testing.
Testing performed prior to release
-----------------------------------
All testing described below was carried out on combinations of the
following parameters:
- Devices: Digital Semiconductor chips 21040, 21041, 21140, 21142.
- Different implementations (boards) of the devices named above, with
various SROM programming.
- Network data rate: 10Mb or 100Mb.
- Local bus: PCI.
- Operating systems: DOS, WFW31, WFW311, OS/2 2.x, OS/2 3.0, Win95.
- Networking OS: Lan Manager(client, server), Lan Server(requester, server),
Windows and Win95 native networking components. All these components
have several different versions.
- Various network setups: Upper protocols (e.g. TCP/IP, NetBEUI), internal
parameters (protocol buffer size, network orientation - connection or
connectionless, and more).
- Various device parameters: Num of buffers, thresholds.
- Serial connections and variations: Tp, Tp full duplex, AUI, BNC.
- Machine-specific parameters: CPU speed, PCI bridge implementations, BIOS
and PCI BIOS implementations, memory, multiple NICs.
NDIS2 conformance:
------------------
MTTOOL conformance test was run on the DOS driver in many different
combinations. All tests were point-to-point configurations using
the different devices (21040, 21041 21140 or 21142) in 10Mb
or 100Mb playing
with driver's parameters (threshold, num of buffers), doing the testing
on different machines (with varying CPU speed and PCI implementation).
Testing was smooth besides:
1. Several tests failed in 100Mb mode due to the high data transfer
rate. We consulted Microsoft on the matter and were granted a waiver.
Stress testing:
---------------
Tests we used are: MTTOOL stress tool, ReadRite, CCtest, WinTest.
WinMCL testing wasn't carried out since
Microsoft granted us a waiver (there are problems with the test
itself).
Some of the stress tests are long (36-96 hours) and some short (2-
24 hours). Following are several sample setups that were used:
1. One server machine (DS21140, PCI, 486DX, TP, WFW3.11, 10Mb).
Three clients (2 DOS 1 OS/2) ran ReadRite stress on the net, all
using random file sizes ( < 512000 bytes ). The test ran for ~48 hrs.
2. One server machine (DS21142 100Mb, WFW3.1, TP, PCI) serving
two 10Mb clients for ReadRite stress, and being a client of a Pentium
machine 100Mb with WinTest. ~40 hours.
3. One server (different combinations). One WFW client with 6 - 12
open stress windows open to the server running CCtest or ReadRite.
Durtaion up to 96 hours.
4. Cyclic: 3 machines in a loop, each as a clint to one machine and server
to another. 10Mb using different stress tests.
Testing went smooth besides:
1. One machine failed during several stress tests. The same
machine failed similarly when we tried other NICs on it, and we
therefore dropped the matter.
Setups:
-------
We tried different setups, parameter combinations, etc. Among these:
- Multiple NICs setups: Two NICs of ours - using the same or different
devices, one of ours and other vendor's NIC. These were tried on
LanServer 3.0, 4.0, WFW 3.1 and 3.11 .
Both automatic installation and manual installation were tried.
- PROTOCOL.INI parameter variations: Num of rcv/tx buffers, Tx threshold,
SW CRC, and more.
- SROM parameter variations: Changing version num, changing default media,
using legacy format boards, coexistence with PROTOCOL.INI values, DS21041
DS21140 and DS21142, different board implementations.
- Serial-connection variations: FD/not FD, TP/AUI/BNC,
Using hard-coded values or autodetected values.
- Autodetection/Autosensing variations: Trying the different devices,
trying several run-time scenarios, trying strange cases (i.e. -
connecting the cable after initialization, pulling it out in the
middle of transactions, sticking two cables in the same time, ...) .
All this in different machines and OS/NOS cominations.
- Removal and reinstallation in different OS/NOS combinations.
- Coexistence with other devices - Networking devices and others.
Release Notes for SMCPWR NDIS 2 MAC Driver V2.22 for DOS, WFW, OS/2 and Win95
===========================================================================================
New features and modifications in V2.22:
-----------------------------------------
1. Fix bug that prevent the driver from working with IRQ 3,4 and 7.
- The driver now work normally with IRQ's 3,4 and 7.
The driver was tested with the following tests:
+---+-----------------------------------+--------------------------------+
Test Performed Completion Status and remarks
+---+-----------------------------------+--------------------------------+
1. Tested the driver working with IQR's 3,4 and 7 Passed
2. Tested that the driver transmit/receive data using ReadRite test
on various OS/NOS on variou 21040, 21041, 21140, 21142 cards.
Release Notes for SMCPWR NDIS 2 MAC Driver V2.21 for DOS, WFW, OS/2 and Win95
===========================================================================================
New features and modifications in V2.21:
-----------------------------------------
1. Fix the Driver Mii Phy present detection.
The previous Driver was detecting Mii Phy connected to the DS21X4 chip
also in some cases when no Mii Phy was present.
Now the Driver detect Mii Phy only when it is present.
Limitations:
The driver and Nics were tested with the following tests
+---+-----------------------------------+--------------------------------+
Test Performed Completion Status and remarks
+---+-----------------------------------+--------------------------------+
1. Testing SROM parsing
1.1 Testing that driver correctly parses Passed
SROM V1.0
1.2 Testing that driver correctly parses Passed
SROM V3.02
2. Testing that driver correctly Passed
parses User Line-Speed selections
according to the following table:
+--------------------+----------------------++------------------------+
| Line_Speed | DS chip Number || Actual Line_Speed |
| (From PROTOCOL.INI)| || |
+--------------------+----------------------++------------------------+
| None | DS21142 || 10 (MBps) |
| 10 | DS21142 || 10 (MBps) |
| 100 | DS21142 || 100 (MBps) |
+--------------------+----------------------++------------------------+
3. Testing that driver correctly Passed
parses User Connection selections on DS21142 chip board
according to the following table:
+--------------------+----------------------++------------------------+
|Requested Connection| Requested LineSpeed || Selected Connection |
| (From PROTOCOL.INI)| || |
+--------------------+----------------------++------------------------+
| AUTOSENSE | 10, 100 or None || AUTOSENSE |
| BNC | 10 or None || BNC, (10MBps) |
| TP | 10 || 10BaseT |
| TP | 100 || 100BaseTX |
| TP_FULL_DUPLEX | 10 || 10BaseT FULL_DUPLEX |
| TP_FULL_DUPLEX | 100 || 100BaseTX FULL_DUPLEX |
| TP_NLT | 10 || 10BaseT NO_LINK_TEST |
| TP_NLT | 100 || 100BaseTX |
+--------------------+----------------------++------------------------+
NOTE: It is recommended to specified the correct LineSpeed for
every Connection type, because when the LineSpeed isn't
correct for the media the Driver conver the media type.
4. Testing Connection Selection
This test tests for driver to correctly Initializes the PHY and DC21X4
Internal registers according to the connection type being used.
4.1 DS21142 with National DP83840 10/100 PHY
Testing that driver Correctly initializes Passed
the PHY and internal registers when
TP connection is selected.
Testing that driver Correctly initializes Passed
the PHY and internal registers when
TP_FULL_DUPLEX connection is selected.
Testing that driver Correctly initializes Passed
the PHY and internal registers when
TP_NO_LINK_TEST connection is selected.
Testing that driver Correctly initializes Passed
the PHY and internal registers when
AUTOSENSE connection is selected and with
Different Nway advertisement.
Testing that driver rejects Passed
BNC connection.
Testing that driver rejects Passed
AUI connection.
Testing that driver Correctly initializes Passed
the PHY and internal registers when
100BaseTX connection is selected.
Testing that driver Correctly initializes Passed
the PHY and internal registers when
100BaseTX_FULL_DUPLEX connection is selected.
Testing that driver rejects Passed
100BaseT4 connection
4.2 DS21142 with internal (10BaseT) PHY
Testing that driver Correctly initializes Passed
its internal registers when
TP connection is selected.
Testing that driver Correctly initializes Passed
its internal registers when
TP_FULL_DUPLEX connection is selected.
Testing that driver Correctly initializes Passed
its internal registers when
TP_NO_LINK_TEST connection is selected.
Testing that driver Correctly initializes Passed
its internal registers when
AUTOSENSE connection is selected and with
Different Nway advertisement.
Testing that driver Correctly initializes Passed
its internal registers when
BNC connection is selected.
Testing that driver rejects Passed
AUI connection.
Testing that driver rejects Passed
100BaseTX connection is selected.
Testing that driver rejects Passed
100BaseTX_FULL_DUPLEX connection is selected.
Testing that driver rejects Passed
100BaseT4 connection
5. Testing Power-Up Autosense.
Testing Power-Up Autosense using Passed
DS21142 and National DP83840
10/100 PHY.
Testing Power-Up Autosense using Passed
DS21142 and EB142 Nic.
6. Testing Dynamic Autosense.
Testing Dynamic Autosense using Passed
DS21142 and National DP83840
10/100 PHY.
7. Testing the driver on OS/2 SMP 2.11 on machine with Dual Pentium CPU's.
The SMP machine was tested as server with Two DS21140 Adapters
one working in 100M bit/sec and the other in 10M bit/sec.
The test we use was ReadRite of 5 100M bit/sec Clients and
3 10M bit/sec Clients runing on the Server disk.
The Clients were runing ReadRite under OS/2 2.1, OS/2 3.0 (WARP)
WFW3.1 and WFW3.11.
The test was runing for 24 hours without any problems.
Testing performed prior to release
-----------------------------------
All testing described below was carried out on combinations of the
following parameters:
- Devices: Digital Semiconductor chips 21040, 21041, 21140, 21142.
- Different implementations (boards) of the devices named above, with
various SROM programming.
- Network data rate: 10Mb or 100Mb.
- Local bus: PCI.
- Operating systems: DOS, WFW31, WFW311, OS/2 2.x, OS/2 3.0, Win95.
- Networking OS: Lan Manager(client, server), Lan Server(requester, server),
Windows and Win95 native networking components. All these components
have several different versions.
- Various network setups: Upper protocols (e.g. TCP/IP, NetBEUI), internal
parameters (protocol buffer size, network orientation - connection or
connectionless, and more).
- Various device parameters: Num of buffers, thresholds.
- Serial connections and variations: Tp, Tp full duplex, AUI, BNC.
- Machine-specific parameters: CPU speed, PCI bridge implementations, BIOS
and PCI BIOS implementations, memory, multiple NICs.
NDIS2 conformance:
------------------
MTTOOL conformance test was run on the DOS driver in many different
combinations. All tests were point-to-point configurations using
the different devices (21040, 21041 21140 or 21142) in 10Mb or 100Mb playing
with driver's parameters (threshold, num of buffers), doing the testing
on different machines (with varying CPU speed and PCI implementation).
Testing was smooth besides:
1. Several tests failed in 100Mb mode due to the high data transfer
rate. We consulted Microsoft on the matter and were granted a waiver.
Stress testing:
---------------
Tests we used are: MTTOOL stress tool, ReadRite, CCtest, WinTest.
WinMCL testing wasn't carried out since
Microsoft granted us a waiver (there are problems with the test
itself).
Some of the stress tests are long (36-96 hours) and some short (2-
24 hours). Following are several sample setups that were used:
1. One server machine (DS21040, PCI, 486DX, TP, WFW3.11, 10Mb).
Three clients (2 DOS 1 OS/2) ran ReadRite stress on the net, all
using random file sizes ( < 512000 bytes ). The test ran for ~48 hrs.
2. One server machine (DS21140 + DS21142 100Mb, WFW3.1, TP, PCI) serving
two 10Mb clients for ReadRite stress, and being a client of a Pentium
machine 100Mb with WinTest. ~40 hours.
3. One server (different combinations). One WFW client with 6 - 12
open stress windows open to the server running CCtest or ReadRite.
Durtaion up to 96 hours.
4. Cyclic: 3 machines in a loop, each as a clint to one machine and server
to another. 10Mb using different stress tests.
Testing went smooth besides:
1. One machine failed during several stress tests. The same
machine failed similarly when we tried other NICs on it, and we
therefore dropped the matter.
Setups:
-------
We tried different setups, parameter combinations, etc. Among these:
- Multiple NICs setups: Two NICs of ours - using the same or different
devices, one of ours and other vendor's NIC. These were tried on
LanServer 3.0, 4.0, WFW 3.1 and 3.11 .
Both automatic installation and manual installation were tried.
- PROTOCOL.INI parameter variations: Num of rcv/tx buffers, Tx threshold,
SW CRC, and more.
- SROM parameter variations: Changing version num, changing default media,
using legacy format boards, coexistence with PROTOCOL.INI values, DS21041
DS21140 and DS21142, different board implementations.
- Serial-connection variations: FD/not FD, TP/AUI/BNC,
Using hard-coded values or autodetected values.
- Autodetection/Autosensing variations: Trying the different devices,
trying several run-time scenarios, trying strange cases (i.e. -
connecting the cable after initialization, pulling it out in the
middle of transactions, sticking two cables in the same time, ...) .
All this in different machines and OS/NOS cominations.
- Removal and reinstallation in different OS/NOS combinations.
- Coexistence with other devices - Networking devices and others.
Release Notes for SMCPWR NDIS 2 MAC Driver V2.20 for DOS, WFW, OS/2 and Win95
===========================================================================================
New features and modifications in V2.20:
-----------------------------------------
1. Added DS21142 device support.
The driver supports DS21142 working with MII PHY or with its internal
SIA port spec.
This driver assumes the following:
- Only one MII PHY exists in Nic.
- If an MII PHY is found and AUTOSENSE operation is selected, then the
driver will perform MII PHY autosensing only.
Download Driver Pack
After your driver has been downloaded, follow these simple steps to install it.
Expand the archive file (if the download file is in zip or rar format).
If the expanded file has an .exe extension, double click it and follow the installation instructions.
Otherwise, open Device Manager by right-clicking the Start menu and selecting Device Manager.
Find the device and model you want to update in the device list.
Double-click on it to open the Properties dialog box.
From the Properties dialog box, select the Driver tab.
Click the Update Driver button, then follow the instructions.
Very important: You must reboot your system to ensure that any driver updates have taken effect.
For more help, visit our Driver Support section for step-by-step videos on how to install drivers for every file type.