Updated: 11 December 1998

VAX MACRO and Instruction Set Reference Manual

Vector operate instructions are always executed to completion, even if a vector arithmetic exception occurs. If an exception occurs, a default result is written. The default result is as follows:

• The low-order 32 bits of the true result for integer overflow.
• Zero for floating underflow if exceptions are disabled.
• An encoded reserved operand for floating divide by zero, floating overflow, reserved operand, and enabled floating underflow. For vector convert instructions that convert floating-point data to integer data, where the source element is a reserved operand, the value written to the destination element is UNPREDICTABLE.

The exception condition type and destination register number are always recorded in the Vector Arithmetic Exception Register (VAER) when a vector arithmetic exception occurs. Refer to Section 10.2.3, Internal Processor Registers, for more information.

10.6.3 Vector Processor Disabled

As a result of error conditions or software control, the vector processor signals the scalar processor not to issue any more vector instructions. The vector processor is disabled when this signal is generated and its state is reflected in VPSR<VEN>. Because the scalar and vector processors can execute asynchronously, the scalar processor may not receive this signal immediately. As a result, the scalar processor may continue to view the vector processor as enabled and send it vector instructions. Once the scalar processor receives this signal, it will view the vector processor as disabled and will not send it any more vector instructions (including MFVP/MTVP). While the vector processor is disabled, and in the absence of hardware errors, it will complete all pending instructions in its instruction queue including those sent by the scalar processor after the vector processor became disabled.

The vector processor can either disable itself or be disabled by software. The following error conditions cause the vector processor to disable itself:

• Vector arithmetic exception (flagged by VPSR<AEX>)
• Hardware error (flagged by VPSR<IMP> in some implementations)
• On some implementations, receipt of an illegal vector opcode (flagged by VPSR<IVO>)

In these cases, the vector processor clears VPSR<VEN> and flags the error condition by setting the appropriate bit in VPSR. (See Table 10-1.)

Software disables the vector processor by writing a zero into VPSR<VEN> using an MTPR instruction. Once the vector processor is disabled, only software can enable it. The software does this by writing a one to VPSR<VEN> using an MTPR. Recall that after performing an MTPR to VPSR, software must then issue an MFPR from VPSR to ensure that the new state of VPSR will affect the execution of subsequently issued vector instructions. The MFPR will not complete in this case until the new state of the vector processor becomes visible to the scalar processor.

When the vector processor disables itself due to a hardware error, it is implementation dependent whether the vector processor completes any pending vector instruction. However, in this case, the vector processor ensures when it is reenabled that all incompleted instructions have been flushed from the instruction queue.

If the scalar processor attempts to issue a vector instruction after it views the vector processor as disabled, then a vector processor disabled fault occurs. The vector processor disabled fault uses SCB offset 68 (hex). The exception handling software (running on the scalar processor) can then read the vector internal processor registers (IPRs) with MFPR instructions to determine what exception conditions are recorded in the vector processor and if the vector processor is still busy processing other unfinished instructions.

Once the scalar processor views the vector processor as disabled, the only operations that can be issued to the vector processor are MTPR and MFPR to and from the vector IPRs.

10.6.4 Handling Disabled Faults and Vector Context Switching

The following flow outlines the required steps for handling a vector processor disabled fault.

If the new process executing on the scalar processor has a vector instruction to execute, saving and restoring the state of the vector processor---that is, vector context switching---is done as part of handling a subsequent vector processor disabled fault.

If a vector processor disabled fault occurs and the current scalar process is also the current vector process, then software must perform the following procedure:

1. Obtain the vector processor status by reading the VPSR using the MFPR instruction.
2. Perform the following checks to see if any of these conditions caused the vector processor to be disabled. If any of these conditions exist, a decision to not continue this flow may occur.
   1. If VPSR<IVO> is set, then write one to clear VPSR<IVO> using the MTPR instruction, and report an illegal vector opcode error.
   2. If VPSR<IMP> is set, then write one to clear VPSR<IMP> using the MTPR instruction, and report an implementation-specific error.
   3. If VPSR<AEX> is set, then write one to clear VPSR<AEX> using the MTPR instruction, and enter the vector arithmetic exception handler with information in VAER.
3. If the software scalar-context-switch flag is set, indicating that a scalar context switch has been done, then perform the following:
   1. Make sure the vector processor has access to correct P0LR, P0BR, P1LR, and P1BR values.
   2. If any vector translation buffer needs to be invalidated, then write zero into the VTBIA IPR using the MTPR instruction. Vector translation buffer flushing is required if the process was swapped out and the mapping change has not yet been made known to the vector translation buffer.
   3. Clear the software scalar-context-switch flag.
4. Enable the vector processor by writing one to VPSR<VEN> using the MTPR instruction. Ensure the new state of the vector processor becomes visible to the scalar processor by reading VPSR with the MFPR instruction.
5. REI to retry the vector instruction at the time of the vector processor disabled fault. If there is an asynchronous memory management exception pending, it is taken when that vector instruction is reissued to the vector processor.

If a vector processor disabled fault occurs and the current scalar process is not the current vector process, then software must perform the following procedure:

1. Check if there is a current vector process. If there is one, then perform the following procedure:
   1. Wait for VPSR<BSY> to be clear using the MFPR instruction.
   2. Perform the following check to see if this condition caused the vector processor to be disabled. If this condition exists, a decision to not continue this flow may occur.
      1. If VPSR<IMP> is set, then report an implementation-specific error.
      2. If VPSR<IVO> is set, then set a software IVO flag for this process. The illegal vector opcode error is handled when this process next tries to execute in the vector processor.
      3. If VPSR<AEX> is set, then set a software AEX flag for this process, and save vector arithmetic exception state from VAER using the MFPR instruction. Any vector arithmetic exception conditions are handled when this process next tries to execute in the vector processor.
   3. At this point there cannot be a synchronous memory management exception pending. But, if asynchronous memory management handling is implemented, there may be an asynchronous memory management exception pending. Because scalar/vector memory synchronization was required before scalar context switching, all such pending exceptions are known at this time. So, if VPSR<PMF> is set, then perform the following procedure:
      1. Set a software asynch-memory-exception-pending flag for this process.
      2. Store implementation-specific vector state in memory starting at the address in VSAR by writing one to VPSR<STS> using the MTPR instruction.
   4. Reset the vector processor state to clear VAER and VPSR, and enable the vector processor. Writing a one to both VPSR<RST> and VPSR<VEN> using the same MTPR instruction accomplishes this. Ensure the new state of the vector processor becomes visible to the scalar processor by reading VPSR with the MFPR instruction.
   5. Store the current vector (V0--V15) and vector control (VLR, VMR, and VCR) register values using VST and MFVP instructions.
   6. Read the VMAC IPR using the MFPR instruction. This ensures scalar/vector memory synchronization and that all hardware errors encountered by previous vector memory instructions have been reported.
2. Make the current scalar process also the current vector process.
3. Clear the software scalar-context-switch flag.
4. Make sure the vector processor has access to correct P0LR, P0BR, P1LR, and P1BR values, and invalidate any vector translation buffer by writing zero to the VTBIA IPR using the MTPR instruction.
5. Load the saved vector (V0--V15) and vector control (VLR, VMR, and VCR) register values using VLD and MTVP instructions.
6. If the software IMP, IVO, or AEX flags for this process are set, perform the following procedure:
   1. Disable the vector processor by writing zero to VPSR<VEN> using the MTPR instruction. Ensure the new state of the vector processor becomes visible to the scalar processor by reading VPSR with the MFPR instruction.
   2. If set, clear the software IMP flag for this process and finish handling the implementation-specific error. A decision to not continue this flow may occur.
   3. If set, clear the software IVO flag for this process and report an illegal vector opcode error occurred. A decision to not continue this flow may occur.
   4. If set, clear the software AEX flag for this process and enter the vector arithmetic exception handler with saved VAER state. A decision to not continue this flow may occur.
7. If the software async-memory-exception-pending flag for this process is set, perform the following procedure:
   1. Clear the software async-memory-exception-pending flag for this process.
   2. Send the vector processor the memory address that points to implementation-specific vector state for this process by writing VSAR using the MTPR instruction.
   3. Reload the implementation-specific vector state for this process and leave the vector processor enabled by writing one to both VPSR<RLD> and VPSR<VEN> using the same MTPR instruction. From this state, the vector processor determines if VPSR<PMF>, VPSR<MF>, or both need to be set, and does it. Ensure the new state of the vector processor becomes visible to the scalar processor by reading VPSR with the MFPR instruction.
8. REI to retry the vector instruction at the time of the vector processor disabled fault. If there is an asynchronous memory management exception pending, it is taken when that vector instruction is reissued to the vector processor.

This section gives examples of Move from Vector Processor (MFVP) exception reporting that are ensured by the vector processor. The rules used to determine the correct result for each example are found in: the tables of dependencies found in Section 10.5.3.3, the description of MSYNC in Section 10.7.2, and the description of MFVP in Section 10.15.

Examples of Exceptions That Cause MSYNC to Fault

The following examples illustrate which exceptions are ensured by the vector processor to always cause MSYNC to fault:

VVMULF    V1, V1, V2 
VVADDF    V3, V2, V3 
MTVLR     #1 
VSTL      V2, A,  #4 
VVCVTFD           V2, V3 
MSYNC     R0 
      

The MSYNC faults if exceptions occur in the production of V2[0] by the VVMULF or in the storage of V2[0] by the VSTL. MSYNC need not fault if exceptions occur in the production of: V2[1..VLR-1] by the VVMULF, V3[0..VLR-1] by the VVADDF, or V3[0..VLR-1] by the VVCVTFD.

VVADDF    V1, V1, V0 
VLDL      A,  #4, V0 
MSYNC     R0 
      

The MSYNC faults if exceptions occur in the loading of V0[0..VLR-1] from memory. MSYNC need not fault if exceptions occur in the production of V0[0..VLR-1] by the VVADDF.

VVADDF    V1, V1, V2 
VLDL      A,  #4, V1 
MSYNC     R0 
      

The MSYNC faults if exceptions occur in the loading of V1[0..VLR-1] from memory. MSYNC need not fault if exceptions occur in the production of V2[0..VLR-1] by the VVADDF.

VVMULF    V1, V1, V2 
VVGTRF    V2, V3 
VSTL/1    V0, A,  #4 
MSYNC     R0 
      

The MSYNC faults if exceptions occur: in the production of V2[0..VLR-1] by the VVMULF, in the production of VMR<0..VLR-1> by the VVGTRF, or in the storage by the VSTL/1 of elements of V0 for which the corresponding VMR bit is one.

Examples of Exceptions the Processor Reports Prior to MFVP Completion

The following examples illustrate which exceptions the vector processor will report prior to the completion of an MFVP from a vector control register:

VLDL      A, #4, V1 
VVMULF    V1, V1, V2 
MTVLR     #1 
VVGTRF    V2, V3 
MFVMRHI           R1 
MFVMRLO           R2 
      

Unreported exceptions that occur: in the loading of V1[0] from memory by the VLDL, in the production of V2[0] by the VVMULF, and VMR<0> by the VVGTRF are reported by the vector processor prior to the completion of the MFVMRLO. The vector processor need not at that time report any exceptions that occur in the loading of V1[1..63] from memory by the VLDL or in the production of V2[1..63] by the VVMULF. Note that the vector processor need not report any exceptions before completing MFVMRHI.

VVGTRF    V0, V1 
MTVMRLO   #patt 
MFVMRLO           R1 
      

For any value of "i" in the range of 0 to 31 inclusive: the value of VMR<i> delivered by MFVMRLO only depends on the value placed into VMR<i> by the MTVMRLO. As a result, the vector processor need not report exceptions that occur in the production of VMR by the VVGTRF prior to completing the MFVMRLO.

VVMULF/1  V1, V1, V2 
MTVMRLO           #patt 
MFVMRLO           R1 
      

For any value of "i" in the range of 0 to 31 inclusive: the value of VMR<i> delivered by MFVMRLO only depends on the value placed into VMR<i> by the MTVMRLO. As a result, the vector processor need not report exceptions that occur in the production of V2[0..VLR-1] by the VVMULF/1 prior to completing the MFVMRLO.

MTVLR     #64 
VVMULF    V0, V0, V2 
VVGTRF    V0, V2 
MTVLR     #32 
IOTA      #str, V4 
MFVCR     R1 
      

Prior to the completion of the MFVCR, the vector processor must report any exceptions that occurred in the production of V2[0..31] by the VVMULF and VMR<0..31> by the VVGTRF. Note that VCR produced by an IOTA depends only on VMR<0..VLR-1>. Recall that no exceptions can occur in the production of V4[0..VCR-1] by IOTA.

MTVLR     #64 
VLDL      A, #4, V2 
VVGTRF    V0, V1 
VSGTRF/1  #3.0, V2 
MFVMRLO           R1 
      

For any value of "i" in the range of 0 to 31 inclusive: prior to the completion of the MFVMRLO, the vector processor must report any exceptions that occurred: in the loading of V2[i] from memory for which V0[i] is greater than V1[i], in the production of VMR<0..31> by the VVGTRF, and in the production of VMR<0..31> by the VSGTRF/1.

VVMULF    V1, V1, V1 
VSTL      V1, base, #str 
MTVMRLO   base 
MFVMRLO           R1 
      

In this example, the value of VMR<31:0> delivered by MFVMRLO only depends on the value placed into VMR<31:0> by the MTVMRLO -- whether this value is V1[0] or the previous value of the location is UNPREDICTABLE. As a result, the vector processor need not report exceptions that occur in the production of V1 by the VVMULF or in the storage of V1 by the VSTL.

For most cases, it is desirable for the vector processor to operate concurrently with the scalar processor so as to achieve good performance. However, there are cases where the operation of the vector and scalar processors must be synchronized to ensure correct program results. Rather than forcing the vector processor to detect and automatically provide synchronization in these cases, the architecture provides software with special instructions to accomplish the synchronization. These instructions synchronize the following:

• Exception reporting between the vector and scalar processors
• Memory accesses between the scalar and vector processors
• Memory accesses between multiple load/store units of the vector processor

Software must determine when to use these synchronization instructions to ensure correct results.

The following sections describe the synchronization instructions.

10.7.1 Scalar/Vector Instruction Synchronization (SYNC)

A mechanism for scalar/vector instruction synchronization between the scalar and vector processors is provided by SYNC, which is implemented by the MFVP instruction. SYNC allows software to ensure that the exceptions of previously issued vector instructions are reported before the scalar processor proceeds with the next instruction. SYNC detects both arithmetic exceptions and asynchronous memory management exceptions and reports these exceptions by taking the appropriate VAX instruction fault. Once it issues the SYNC, the scalar processor executes no further instructions until the SYNC completes or faults.

In beginning the execution of SYNC, the vector processor determines if any previously issued vector instruction has encountered exceptions which have yet to be reported to the scalar processor. If so, the SYNC is faulted; otherwise, the vector processor waits for either of the following conditions to be true:

• A pending or currently executing vector instruction encounters an exception---in which case the SYNC faults
• The vector processor determines that all pending and currently executing vector instructions (including memory instructions in asynchronous memory management mode) will execute to completion without encountering vector exceptions. In that case the SYNC completes.

When SYNC completes, a longword value (which is UNPREDICTABLE) is returned to the scalar processor. The scalar processor writes the longword value to the scalar destination of the MFVP and then proceeds to execute the next instruction. If the scalar destination is in memory, it is UNPREDICTABLE whether the new value of the destination becomes visible to the vector processor until scalar/vector memory synchronization is performed.

When SYNC faults, it is not completed by the vector processor and the scalar processor does not write a longword value to the scalar destination of the MFVP. Also, depending on the exception condition encountered, the SYNC itself takes either a vector processor disabled fault or memory management fault. If both faults are encountered while the vector processor is performing SYNC, then the SYNC itself takes a vector processor disabled fault. Note that it is UNPREDICTABLE whether the vector processor is idle when the fault is generated. After the appropriate fault has been serviced, the SYNC may be returned to through an REI.

SYNC only affects the scalar/vector processor pair that executed it. It has no effect on other processors in a multiprocessor system.

10.7.2 Scalar/Vector Memory Synchronization

Scalar/vector memory synchronization allows software to ensure that the memory activity of the scalar/vector processor pair has ceased and the resultant memory write operations have been made visible to each processor in the pair before the pair's scalar processor proceeds with the next instruction. Two ways are provided to ensure scalar/vector memory synchronization: using MSYNC, which is implemented by the MFVP instruction, and using the MFPR instruction to read the VMAC (Vector Memory Activity Check) internal processor register (IPR). Section 10.7.2.1 discusses MSYNC in detail. Section 10.7.2.2 discusses VMAC in detail.

Scalar/vector memory synchronization does not mean that previously issued vector memory instructions have completed; it only means that the vector and scalar processors are no longer performing memory operations. While both VMAC and MSYNC provide scalar/vector memory synchronization, MSYNC performs significantly more than just that function. In addition, VMAC and MSYNC differ in their exception behavior.

Note that scalar/vector memory synchronization only affects the scalar/vector processor pair that executed it. It has no effect on other processors in a multiprocessor system. Scalar/vector memory synchronization does not ensure that the write operations made by one scalar/vector pair are visible to any other scalar or vector processor. Software can make data visible and shared between a scalar/vector pair and other scalar and vector processors by using the mechanisms described in the VAX Architecture Reference Manual. Software must first make a memory write operation by the vector processor visible to its associated scalar processor through scalar/vector memory synchronization before making the write operation visible to other processors. Without performing this scalar/vector memory synchronization, it is UNPREDICTABLE whether the vector memory write will be made visible to other processors even by the mechanisms described in the VAX Architecture Reference Manual.

Lastly, waiting for VPSR<BSY> to be clear does not guarantee that a vector write operation is visible to the scalar processor.

Copyright © Compaq Computer Corporation 1998. All rights reserved. Legal

4515PRO_033.HTML