Registered users of our products can download the present full versions of OptiVec and CMATH from this page. In addition, here our customers who bought through an electronic distribution channel can get the rarely used libraries which are omitted from the electronically distributed versions. In order to reduce the download times, all packages are split into several parts, consisting of 
a) the documentation, 
b) the include-files (C/C++ versions only), and 
c) C++ libraries or Pascal/Delphi units for each combination of memory model and processor supported. As specified below, some rarely needed libraries are not available here, but will be sent by email upon request.

The library and unit ZIP files are password-protected. Our registered users get the passwords by e-mail. If you are a registered user and did not receive the password, please drop us a note at optivec%at%gmx.de (replace %at% by the @ sign)! Consult the documentation of your unzip program about how to uncompress password-protected files. In most cases, you will be automatically asked for the password through a popup dialogue.

You should install the files from this page over any previous version (trial or full) of OptiVec or CMATH; in other words, you should not deinstall your previous version, but simply overwrite it. If you do not have any previous version installed, you have to manually create the directories mentioned in the description below.

If you install the registered version over the Shareware version, be sure to recompile your project(s). (Remember that applications, containing functions of the Shareware version, cease functioning after 90 days and shall not run at all on your client's computers!)

Directly jump to the library selection tables:

New features in v5.2:

• Debug library, detecting violations of vector / matrix boundaries, which would otherwise lead to hard-to-locate memory faults. We strongly recommend to use the debug libraries in the initial development phases of an application using OptiVec. Only after dimension consistency has been checked by the debug library and you can be sure no memory issues remain, you should switch over to the high-performance libraries. Read more.
• CMATH for Delphi:
  From Delphi 2006 on, the operators  +  -  *  /  =  :=  <>  are available. As the older Delphi versions did not allow the implementation of these operators, the CMATH versions for Delphi 4 through 2005 continue to offer only the functions add, mul as a replacement for the operators. To ensure portability, these functions remain available also in CMATH for Delphi 2006 or higher.
• Bug fixes:
  • OptiVec for Delphi: The non-linear fitting functions were improved, mainly with respect to errors in the weighted variants and with respect to stability of the "breakout" operation.
  • OptiVec for Visual C++: Several bugs in VQI_ functions were fixed.

New features in v5.1:

• Multi-processor library also for Delphi. All arithmetic functions are now also optimized for multi-core.Read here for details and considerations about using OptiVec on multi-processor computers.
• P8 version, using the SSE3 instruction set of Core2Duo, AMD64x2 (and the latest series of Pentium 4) for faster complex-number multiplications in a variety of functions.
• You should call V_initMT at the beginning of any multi-thread applications making use of OptiVec functions. Previously, this was necessary only if you used the Multi-Core libraries of OptiVec. Now, V_initMT also performs initializations which avoid previously encountered thread-safety issues inside the memory management functions like V?_vector.
• Calculation of the center of gravity, both for vectors and for matrices:
  V?_centerOfGravityInd,    V?_centerOfGravityV,
  M?_centerOfGravityInd,   M?_centerOfGravityV.
  The V?_centerOfGravityInd variant takes the element indices as axis, whereas V?_centerOfGravityV takes a specified X axis.
• The speed of the Fourier Transform routines was improved once more. This latest tweak makes itself felt mostly for smaller vectors (VF_FFT) and for 2D FFT (MF_FFT).
• Alternative forms of the functions for element rotations. The new forms take the necessary buffer memory as an argument instead of requesting it from the operating system each time they are called:
  V?_rotate_buf,
  M?_Rows_rotate_buf,    M?_Cols_rotate_buf.
• VecObj: integralV is now a member function of the vector Y, whose integral over X is being evaluated. X is passed as argument. Previously, this function was defined the other way round.
• Bug fixes in
  VE_polyinterpol,
  VE_ratinterpol,
  MC?_read (which would occasionally fail if the input numbers were written with braces or brackets)
• C/C++, Multi-Core Libraries only: Bug fixes in:
  V?_powexp,
  V?x_pow2,    V?x_pow10
• P7 libraries only: bug fixes in
  MD_TmulM (large MY with uneven lenY caused an error),
  MD_SVsolve, affecting also MD_solveBySVD,    MD_safeSolve,    VD_linfit, and VD_nonlinfit: failed with non-aligned vectors (e.g., static arrays)

v5.0:

• Multi-processor optimized library, scaling with the number of available processor cores for C/C++.
• v4.4.0 - 4.4.5:
  
• The non-linear data-fitting functions are now thread-safe
   
• Visual C++ version only: compatibility problem with <crtdbg.h> fixed. You can now detect memory leaks by including <crtdbg.h> with the _CRTDBG_MAP_ALLOC option on. If you do so, any calls to the VecObj member function X.free() must be replaced by the synonymous member function X.dealloc();
   
• Reorganization of the Win32 libraries for C++ compilers:
  
  A base library is specific for the compiler version and target configuration.
  The main library contains the functions which are independent from the target configuration, but are optimized for different processors.
   
• Bug fixes in:
  M?_mulMT and M?_TmulM (affected matrix with a single column only),
  VLI_not (Pentium 4 only),
  VFx_exp (Pentium III, 4 only),
  MDp_FFTtoC (Pentium III, 4 only),
  MDsym_eigenvalues (Visual C++ version only).
   
• Version 4 introduced:
  P7 version (Pentium 4+). This version extends the benefits of the earlier P6 version to double-precision and integer operations. It is, on the avarage, 20-30% faster than the 486DX/Pentium version, but will not run on CPUs lower than Pentium 4. The scalar functions of CMATH do not profit from the enhanced vector possibilities of the SSE2 command set of the Pentium 4 chip; therefore, no separate P7 version is available for CMATH. For CMATH, use the P6 version instead.
• Mixed floating-point/integer Arithmetic operations like VF_addVI or VD_mulVUS
• The rounding-to-integer functions now employ "silent" saturation instead of generating an error message in the case of overflow.
• Bug fix in the P6 version of MF_LUdecompose. This bug affected also MF_inv and the single-precision data fitting functions of the P6 version.

Please choose which product you want to download / update:

Update OptiVec for Borland C/C++

1. Please download and unzip the documentation, OVDOC.ZIP, or the German version, OVDOCD.ZIP, into the OptiVec root directory on your computer. 
2. Optionally, you may also download the demo programs, DEMOS.ZIP, again into the OptiVec root directory. 
3. Now please download and unzip OVINCL.ZIP, which goes into OPTIVEC/INCLUDE. 
4. Download all needed library ZIP files from the table below, store them in OPTIVEC/LIB and unzip them.

Library Type	General Purpose	Debug	Multi-Processor	Large-Vector
P4: 486DX / Pentium / Athlon	VCF4W	VCF4D	VCF4M	----
P6: Pentium III+	VCF6W	----	VCF6M	VCF6L
P7: Pentium 4+ / Athlon x64	VCF7W	----	VCF7M	VCF7L
P8: AMD64xxx, Core2xxx	VCF8W	----	VCF8M	VCF8L

 
The P6, P7, P8 versions need RAD Studio, BDS, BCB 5 or higher, or BC++ 5.x. Windows XP or Vista is required, as well.
The Large-Vector Versions are optimized for single-processor systems for vectors and matrices larger than the size of the second-level cache of your mainboard. In these versions, the cache is bypassed during load and store operations on vector/matrix elements. A heavy penalty results from their mis-use for smaller vectors or matrices which would have fit into the cache.
The libraries for multi-processor systems run on workstations with discrete processors as well as on multi-core processors like AMD64x2, Core2 Duo and so on.
The libraries for down-compatibility with 386+387 as well as for all 16-bit models and for 286 fossiles are no longer updated. The old versions are, however, still available upon request. We'll send you by email what you need.
 
5. Additionally, you need to download and link either of the following base libraries:

   Configuration	OptiVec base library
   BC runtime library static	VCFS
   BC runtime library as DLL	VCFD

   
    
6. After uncompressing, you can delete all .ZIP files you downloaded.

Update CMATH for Borland C/C++

1. Please download and unzip the documentation, CMDOC.ZIP, or the German version, CMDOCD.ZIP into the CMATH root directory on your computer.
2. Now please download and unzip CMINCL.ZIP, which goes into CMATH/INCLUDE.
3. Download all needed library ZIP files from the table below, store them in CMATH/LIB and unzip them.

Platform	486DX / Pentium / Athlon	Pentium III (only BC++ 5.x, BCB 5 or higher!)
Windows 95/98/ME/NT/2000/XP/Vista	CMATHF4W	CMATHF6W

 
  The libraries for down-compatibility with 386+387 are still available. Upon request, we'll send you by email what you need.
4. Additionally, you need to download and link either of the following base libraries:

   Configuration	CMATH base library
   BC runtime library static	CMATHFS
   BC runtime library as DLL	CMATHFD

    
    
5. After uncompressing, you can delete all .ZIP files you downloaded.

Update OptiVec for Microsoft Visual C++

1. Please download and unzip the documentation, OVDOC.ZIP, or the German version, OVDOCD.ZIP, into the OptiVec root directory on your computer.
2. Optionally, you may also download the demo programs, DEMOS.ZIP, again into the OptiVec root directory. 
3. Now please download and unzip OVINCL.ZIP, which goes into OPTIVEC/INCLUDE.
4. Download all needed library ZIP files from the table below, store them in OPTIVEC/LIB and unzip them.

Library Type	General Purpose	Debug	Multi-Processor	Large-Vector
P4: 486DX / Pentium / Athlon	OVVC4	OVVC4D	OVVC4M	----
P6: Pentium III+	OVVC6	----	OVVC6M	OVVC6L
P7: Pentium 4+ / Athlon x64	OVVC7	----	OVVC7M	OVVC7L
P8: AMD64xxx, Core2xxx	OVVC8	----	OVVC8M	OVVC8L

 
The Large-Vector Versions are optimized for single-processor systems for vectors and matrices larger than the size of the second-level cache of your mainboard. In these versions, the cache is bypassed during load and store operations on vector/matrix elements. A heavy penalty results from their mis-use for smaller vectors or matrices which would have fit into the cache.
The libraries for multi-processor systems run on workstations with discrete processors as well as on multi-core processors like AMD64x2, Intel Core2 Duo and so on.
5. Additionally, you need to download and link either of the following base libraries:
    

   Visual C++ version	Configuration	OptiVec base library
   Visual C++ 5, 6, 2003	Single-Thread Debug	OVVCSD
   Visual C++ 5, 6, 2003	Single-Thread Release	OVVCSR
   Visual C++ 5, 6, 2003	Multi-Thread Debug	OVVCMTD
   Visual C++ 5, 6, 2003	Multi-Thread Release	OVVCMTR
   Visual C++ 5, 6, 2003	Multi-Thread, MFC as DLL, Debug	OVVCMDD
   Visual C++ 5, 6, 2003	Multi-Thread, MFC as DLL, Release	OVVCMDR
   Visual Studio 2005	Multi-Thread Debug	OVVC8MTD
   Visual Studio 2005	Multi-Thread Release	OVVC8MTR
   Visual Studio 2005	Multi-Thread, MFC as DLL, Debug	OVVC8MDD
   Visual Studio 2005	Multi-Thread, MFC as DLL, Release	OVVC8MDR

    
   
6. After uncompressing them, you can delete all .ZIP files you downloaded. 

Update CMATH for Microsoft Visual C++

1. Please download and unzip the documentation, CMDOC.ZIP, or the German version, CMDOCD.ZIP, into the CMATH root directory on your computer.
2. Now please download and unzip CMINCL.ZIP, which goes into CMATH/INCLUDE.
3. Download all needed library ZIP files from the table below, store them in CMATH/LIB and unzip them.

Platform	486DX/Pentium/Athlon	Pentium III+
Windows 95/98/ME/NT/2000/XP/Vista	CMVC4	CMVC6

 

4. Additionally, you need to download and link either of the following base libraries:
    

   Visual C++ version	Configuration	CMATH base library
   Visual C++ 5, 6, 2003	Single-Thread Debug	CMVCSD
   Visual C++ 5, 6, 2003	Single-Thread Release	CMVCSR
   Visual C++ 5, 6, 2003	Multi-Thread Debug	CMVCMTD
   Visual C++ 5, 6, 2003	Multi-Thread Release	CMVCMTR
   Visual C++ 5, 6, 2003	Multi-Thread, MFC as DLL, Debug	CMVCMDD
   Visual C++ 5, 6, 2003	Multi-Thread, MFC as DLL, Release	CMVCMDR
   Visual Studio 2005	Multi-Thread Debug	CMVC8MTD
   Visual Studio 2005	Multi-Thread Release	CMVC8MTR
   Visual Studio 2005	Multi-Thread, MFC as DLL, Debug	CMVC8MDD
   Visual Studio 2005	Multi-Thread, MFC as DLL, Release	CMVC8MDR

    
   
5. After uncompressing them, you can delete all .ZIP files you downloaded. 

Update OptiVec for Borland Delphi

1. Please download and unzip the documentation, OVDOC.ZIP, or the German version, OVDOCD.ZIP, into the OptiVec for Delphi root directory on your computer.
2. Optionally, you may also download the demo programs, DEMOSD.ZIP, again into the OptiVec root directory. 
3. Download and unzip the needed library ZIP files from the table below. If you previously used the Shareware version, be sure to overwrite the Shareware units with the registered ones downloaded here!

Delphi version	Processor compatibility	General Purpose	Debug	Multi-Processor	Large-Vector
Delphi 2006 (BDS2006 or Turbo Delphi 2006) and Delphi 2007 (RAD Studio 2007)	P4: 486DX / Pentium / Athlon	OVD10LB4	OVD10L4D	OVD10L4M	----
	P6: Pentium III+	OVD10LB6	----	OVD10L6M	OVD10L6L
	P7: Pentium 4+ / Athlon x64	OVD10LB7	----	OVD10L7M	OVD10L7L
	P8: AMD64xxx, Core2xxx	OVD10LB8	----	OVD10L8M	OVD10L8L
Delphi 2005	P4: 486DX / Pentium / Athlon	OVD9LIB4	OVD9LB4D	OVD9LB4M	----
	P6: Pentium III+	OVD9LIB6	----	OVD9LB6M	OVD9LB6L
	P7: Pentium 4+ / Athlon x64	OVD9LIB7	----	OVD9LB7M	OVD9LB7L
	P8: AMD64xxx, Core2xxx	OVD9LIB8	----	OVD9LB8M	OVD9LB8L
Delphi 7	P4: 486DX / Pentium / Athlon	OVD7LIB4	OVD7LB4D	OVD7LB4M	----
	P6: Pentium III+	OVD7LIB6	----	OVD7LB6M	OVD7LB6L
	P7: Pentium 4+ / Athlon x64	OVD7LIB7	----	OVD7LB7M	OVD7LB7L
	P8: AMD64xxx, Core2xxx	OVD7LIB8	----	OVD7LB8M	OVD7LB8L
Delphi 6	P4: 486DX / Pentium / Athlon	OVD6LIB4	OVD6LB4D	OVD6LB4M	----
	P6: Pentium III+	OVD6LIB6	----	OVD6LB6M	OVD6LB6L
	P7: Pentium 4+ / Athlon x64	OVD6LIB7	----	OVD6LB7M	OVD6LB7L
	P8: AMD64xxx, Core2xxx	OVD6LIB8	----	OVD6LB8M	OVD6LB8L
Delphi 5	P4: 486DX / Pentium / Athlon	OVD5LIB4	OVD5LB4D	OVD5LB4M	----
	P6: Pentium III+	OVD5LIB6	----	OVD5LB6M	OVD5LB6L
	P7: Pentium 4+ / Athlon x64	OVD5LIB7	----	OVD5LB7M	OVD5LB7L
	P8: AMD64xxx, Core2xxx	OVD5LIB8	----	OVD5LB8M	OVD5LB8L
Delphi 4	P4: 486DX / Pentium / Athlon	OVD4LIB4	OVD4LB4D	OVD4LB4M	----
	P6: Pentium III+	OVD4LIB6	----	OVD4LB6M	OVD4LB6L
	P7: Pentium 4+ / Athlon x64	OVD4LIB7	----	OVD4LB7M	OVD4LB7L
	P8: AMD64xxx, Core2xxx	OVD4LIB8	----	OVD4LB8M	OVD4LB8L


The Large-Vector Versions are for vectors and matrices larger than the size of the second-level cache of your mainboard. In these versions, the cache is bypassed during load and store operations on vector/matrix elements. A heavy penalty results from their mis-use for smaller vectors or matrices which would have fit into the cache.
The units back-compatible to 386+387 are still available upon request: We'll send you by email what you need.
4. After uncompressing them, you can delete all .ZIP files you downloaded. 

Update CMATH for Borland Delphi

1. Please download and unzip the documentation, CMDDOC.ZIP, or the German version, CMDOCD.ZIP, into the CMATH for Delphi root directory on your computer.
2. Download and unzip the needed library ZIP files from the table below. Be sure to overwrite the units in CMATH/LIB4 with the units contained in CMDxLIB4.ZIP, and the units in CMATH/LIB6 with the units in CMDxLIB6.ZIP.

Delphi version	486DX / Pentium	Pentium III+
Delphi 2006 (BDS2006 or Turbo Delphi 2006) and Delphi 2007 (BDS2007	CMD10LB4	CMD10LB6
Delphi 2005	CMD9LIB4	CMD9LIB6
Delphi 7	CMD7LIB4	CMD7LIB6
Delphi 6	CMD6LIB4	CMD6LIB6
Delphi 5	CMD5LIB4	CMD5LIB6
Delphi 4	CMD4LIB4	CMD4LIB6

The units back-compatible to 386+387 are still available upon request: We'll send you by email what you need.
3. After uncompressing them, you can delete all .ZIP files you downloaded.