IBM 86884RX Installation Guide - Page 51

dedicated to the CPUs the 64 MB XceL4 Server Accelerator Cache.

Page 51 highlights

The parallelism is there, but it is not as obvious to the processor and more work has to be done by the hardware before it can be utilized. Itanium 2 supports parallelism on multiple levels. Instruction-level parallelism (ILP) is the ability to execute multiple bundles (three instructions in a bundle) at the same time. The Itanium 2 micro-architecture can deliver faster performance by executing multiple bundles per clock cycle. Parallelism, both at the instruction level and at the SMP system level, permits more efficient use of virtually all system resources to enable improved scalability. The Itanium processor's instruction-level parallelism helps ensure the scalability necessary to manage large data warehouses. Note: Performance should no longer be measured by just the speed in MHz, but also by the degree of parallelism that the processor achieves. Large memory addressability Another key advantage is that 64-bit operating systems can support far more physical memory than a 32-bit operating system. The theoretical limit for directly addressable memory was 4 GB in 32-bit architectures and is now 264 or 16 Exabytes. The Itanium 2 processor memory subsystem has a three-level cache structure consisting of first-level instruction cache, first-level data cache, second-level cache (L2), and third-level cache (L3). In addition, to ensure the processors are optimally used, the x450 has a fourth-level (L4) system cache dedicated to the CPUs (the 64 MB XceL4 Server Accelerator Cache). The increased physical memory includes the following benefits for applications: - Each application can support more users. For a comparison of the number of maximum connected users in SAP for various hardware platforms (including Itanium 2), visit the following link: http://www.sap.com/benchmark/ - Each application has better performance. Increased physical memory allows more applications to run simultaneously and remain completely resident in the system's main memory. This reduces or eliminates the performance penalty of swapping pages to and from disk. - Each application has more memory for data storage and manipulation. Databases can store more of their data in the physical memory of the system. Data access is faster because disk reads are not necessary. - Applications can manipulate large amounts of data easily and more reliably. Video composition and modeling for scientific and financial Chapter 2. Positioning 37

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
  • 31
  • 32
  • 33
  • 34
  • 35
  • 36
  • 37
  • 38
  • 39
  • 40
  • 41
  • 42
  • 43
  • 44
  • 45
  • 46
  • 47
  • 48
  • 49
  • 50
  • 51
  • 52
  • 53
  • 54
  • 55
  • 56
  • 57
  • 58
  • 59
  • 60
  • 61
  • 62
  • 63
  • 64
  • 65
  • 66
  • 67
  • 68
  • 69
  • 70
  • 71
  • 72
  • 73
  • 74
  • 75
  • 76
  • 77
  • 78
  • 79
  • 80
  • 81
  • 82
  • 83
  • 84
  • 85
  • 86
  • 87
  • 88
  • 89
  • 90
  • 91
  • 92
  • 93
  • 94
  • 95
  • 96
  • 97
  • 98
  • 99
  • 100
  • 101
  • 102
  • 103
  • 104
  • 105
  • 106
  • 107
  • 108
  • 109
  • 110
  • 111
  • 112
  • 113
  • 114
  • 115
  • 116
  • 117
  • 118
  • 119
  • 120
  • 121
  • 122
  • 123
  • 124
  • 125
  • 126
  • 127
  • 128
  • 129
  • 130
  • 131
  • 132
  • 133
  • 134
  • 135
  • 136
  • 137
  • 138
  • 139
  • 140
  • 141
  • 142
  • 143
  • 144
  • 145
  • 146
  • 147
  • 148
  • 149
  • 150
  • 151
  • 152
  • 153
  • 154
  • 155
  • 156
  • 157
  • 158
  • 159
  • 160

Chapter 2. Positioning
37
The parallelism is there, but it is not as obvious to the processor and more
work has to be done by the hardware before it can be utilized.
Itanium 2 supports parallelism on multiple levels. Instruction-level parallelism
(ILP) is the ability to execute multiple bundles (three instructions in a bundle)
at the same time. The Itanium 2 micro-architecture can deliver faster
performance by executing multiple bundles per clock cycle. Parallelism, both
at the instruction level and at the SMP system level, permits more efficient
use of virtually all system resources to enable improved scalability.
The Itanium processor's instruction-level parallelism helps ensure the
scalability necessary to manage large data warehouses.
±
Large memory addressability
Another key advantage is that 64-bit operating systems can support far more
physical memory than a 32-bit operating system. The theoretical limit for
directly addressable memory was 4 GB in 32-bit architectures and is now 2
64
or 16 Exabytes.
The Itanium 2 processor memory subsystem has a three-level cache
structure consisting of first-level instruction cache, first-level data cache,
second-level cache (L2), and third-level cache (L3). In addition, to ensure the
processors are optimally used, the x450 has a fourth-level (L4) system cache
dedicated to the CPUs (the 64 MB XceL4 Server Accelerator Cache).
The increased physical memory includes the following benefits for
applications:
Each application can support more users.
For a comparison of the number of maximum connected users in SAP for
various hardware platforms (including Itanium 2), visit the following link:
Each application has better performance. Increased physical memory
allows more applications to run simultaneously and remain completely
resident in the system's main memory. This reduces or eliminates the
performance penalty of swapping pages to and from disk.
Each application has more memory for data storage and manipulation.
Databases can store more of their data in the physical memory of the
system. Data access is faster because disk reads are not necessary.
Applications can manipulate large amounts of data easily and more
reliably. Video composition and modeling for scientific and financial
Note:
Performance should no longer be measured by just the speed in
MHz, but also by the degree of parallelism that the processor achieves.