What is Westmere Processor? (Explained)

What is Westmere Processor?

The term Westmere refers to the processor and System on a Chip (SoC) that is built typically on a 32 nm process and characteristically comes with 2^nd generation high-K metal gate transistors.

The Westmere architecture is mainly designed for desktop and server processors. Technically, in this specific architecture, the processors experienced a die-shrink and high-K-layer shrink with respect to the earlier 45 nm manufacturing process.

KEY TAKEAWAYS

• The Westmere architecture refers to the 32 nm processors and SoCs launched on January 7, 2010 and is the successor to Nehalem and the predecessor of Sandy Bridge.
• With the Westmere architecture, the processors have much enhanced virtualization latency and capabilities.
• Encryption and decryption is further enhanced with this architecture in the processors that have a few new sets of instructions and translation support for 1GB hugepages.
• The Westmere CPUs gate off and shut down the unused cores saving their state in the cache. This helps in reducing power consumption by a significant margin.
• These processors are available with different features and under different brand names such as Pentium, Celeron, Intel Core and Xeon and are good for use in desktop and server systems.

Understanding Westmere Processor

The Westmere microarchitecture for desktop and server processors was released in 2010 with some notable improvements and a process shrink in comparison to Nehalem, its predecessor.

In this specific design, the significant changes are as follows:

• Gate pitch is reduced to 112.5 nm from 180 nm
• Interconnect pitch is reduced to 112.5 nm from 160 nm
• SRAM bit cell (HD) is reduced to 0.148 µm² from 0.346 um²
• SRAM bit cell (LP) is reduced to 0.171 um² from 0.382 µm²

It also supports an SRAM bit cell (HS) of 0.199 um².

A few other significant changes in the design, features, and technology, with respect to the Nehalem architecture, are summarized as follows:

• Native six-core Gulftown and ten-core Westmere EX processors
• A new set of instructions that offers three times higher encryption and decryption rates of Advanced Encryption Standard (AES) processes
• Seven new AES instruction sets or AES-NI, where six of them implement the AES algorithm and one implements the PCLMULQDQ or CLMUL instruction set for Carry-Less Multiplication in data compression and cryptography
• Integrated graphics are available in dual core Arrandale and Clarkdale processor packages only
• Enhanced virtualization latency and capabilities with VMX Unrestricted mode support, which allows for 16-bit guests to run real mode and big real mode
• Support for hugepages of 1 GB in size

One significant aspect of the Westmere processors is that they can shut down and gate off the unused cores.

This helps reduce the power consumption of the CPUs.

Though the Nehalem CPUs can do this as well, the Westmere processors can do it in a much better way.

This is because they can even gate off those regions of the processor that are not related to central processing but with other aspects such as:

• L3 cache
• Memory control
• Bus controller and others

In fact, the Westmere processors can power gate everything, as opposed to the Nehalem processors, which can power gate individual cores.

This offers the significant benefit of reducing power draw during idle state.

Brand

Ideally, for the desktop and mobile processors, Westmere was branded the 1^st generation Intel Core CPUs. The different brands and some of their specific features and support are as follows:

• Intel Celeron – It is designed as an entry-level budget processor with two cores and IGP and ECC support.
• Intel Pentium – It is designed as a budget-level processor with two cores and IGP and ECC support.
• Intel Core i3 – It is designed to offer low-end performance with two cores and HT, IGP, and ECC support.
• Intel Core i5 – It is designed to offer mid-range performance with two cores and HT, IGP, TBT, and ECC support.
• Intel Core i7 – It is designed to offer high-end performance. The mobile processors come with two cores and HT, IGP, TBT, and ECC support.
• Intel Core i7 – It is designed for desktop processors to offer high-end performance and comes with six cores and HT and TBT support.
• Intel Core i7 EE – It is designed to offer high-end performance in enthusiast-class CPUs with six cores and HT and TBT support.
• Intel Xeon E7 – It is designed for use in high-end servers and comes with a number of cores ranging from 6 to 10, along with HT, TBT, and ECC support.

General Specifications

Here are some general physical and technical specifications of the Westmere processors that are good to know:

• It was launched on January 7, 2010
• Its predecessor is Nehalem
• Its successor is Sandy Bridge
• The maximum CPU clock rate achieved ranges between 1.06 GHz and 3.46 GHz
• It supports the x86-64 architecture
• The 8-way set associative Level 1 data cache measures 32 KB and supports 64 B per line
• The 4-way set associative Level 1 instruction cache measures 32 KB
• The 8-way set associative Level 2 cache measures 256 KB and supports 64 B per line
• The shared Level 3 cache size may range anywhere between 2 MB and 30 MB
• The load buffers can load up to 48 items
• The store buffers can store up to 32 items
• The Line Fill Buffers (LFB) support up to 10 lines
• The RS can store 36 items
• The Reorder Buffer (ROB) can hold up to 128 items
• The L1 data cache latency is 4 cycles
• The L2 data cache latency is 10 cycles
• The cross penalty within the 64-bytes range is of 4 cycles
• The cross penalty range within the 4096-bytes range is of 20 cycles
• The penalty for branch misprediction is of 15 or 16 cycles
• The number of transistors may range anywhere between 382 million and 2600 million
• The number of cores may range anywhere between 2 and 4 in regular processors and from 4 to 10 in Intel Xeon processors
• The integrated GPU in larger dies may operate within a frequency range of 533 MHz and 900 MHz
• The different types of sockets supported are LGA 1156, LGA 1366, LGA 1567, and rPGA988A.

However, there are a few exceptions to the general specifications and support. For example:

• The Thermal Design Power or TDP includes the integrated Graphics Processing Unit, if present.
• The Clarkdale processors come with 16 Peripheral Component Interconnect Express or PCIe 2.0 lanes. These lanes can be used in different configurations such as 1×16 or 2×8.
• The Clarkdale and Arrandale processors come with the 32 nm dual core processor Hillel as well as the 45 nm integrated graphics device Ironlake, which supports switchable graphics.
• Only a few specific types of high-end CPUs support AES-NI and 1 GB hugepages.

Instruction Sets and Extensions Support

The Westmere processors support quite a few different extensions of IA-32 and x86-64 instructions as follows:

• AES-NI or Advanced Encryption Standard New Instructions
• CLMUL or Carry-less Multiplication
• MMX or MultiMedia eXtensions
• SSE or Streaming SIMD Extensions, along with all its variants such as, SSE2, SSE3, SSSE3, SSE4, SSE4.1, and SSE4.2
• VT-x and VT-d or Virtualization extensions

Translation Lookaside Buffer

The Westmere processors also come with Translation Lookaside Buffers (TLBs) for data and instructions. The specifications of the TLBs are as follows:

• The first-level data TLB or DTLB supports 64 entries for a 4 KB page size and 32 entries for a 2 MB page size.
• The first-level instruction TLB or ITLB supports 128 entries for a 4 KB page size and 7 entries for each logical core for a 2 MB page size.
• The second-level unified TLB or STLB supports 512 entries for a 4 KB page size and none for 2 MB or larger page sizes.

Westmere Processor vs Sandy Bridge

• The Westmere processors are older in comparison to the Sandy Bridge processors, being the predecessor and launched a year earlier.
• The maximum CPU clock speed achieved by the Westmere processors ranges between 1.06 GHz and 3.46 GHz, but in comparison, the cores of the Sandy Bridge processors can achieve a slightly higher clock speed ranging between 1.60 GHz and 3.60 GHz.
• The Level 3 cache size of the Westmere processors can range anywhere from 2 MB to 30 MB. On the other hand, the Level 3 cache size of the Sandy Bridge processors, in comparison, can range anywhere between 1 MB and 8 MB when shared, 10 MB and 15 MB in Extreme, and 3 MB and 20 MB in Xeon processors.
• Though both of these processors support almost the same types of Instruction Set Architecture (ISA) extensions, the Westmere processors support IA 32 and x86-64 instructions, while the Sandy Bridge processors support x86 and x86-64 instructions and TXT extensions.
• The number of transistors in the Westmere processors ranges anywhere between 382 million and 2600 million in a 32 nm chip. On the other hand, the number of transistors on a 32 nm Sandy Bridge processor may range anywhere between 504 million and 2.27 billion.
• The number of cores in the Westmere processors may range anywhere from 4 to 10 in the Xeon processors and 2 to 4 in others. On the other hand, the number of cores in the Sandy Bridge may range anywhere from 2 to 8 in Xeon, 4 to 6 in Extreme, and 1 to 4 in other variants.
• The GPU in the larger Westmere processors typically operates at a speed ranging between 533 MHz and 900 MHz. The different HD GPUs of Sandy Bridge processors operate at different speeds, such as 650 to 1100 MHz, 650 to 1250 MHz, 650 to 1350 MHz, and 850 to 1350 MHz for normal, HD 2000, HD 3000, and P3000 GPUs, respectively.
• The types of sockets supported by the Westmere processors include LGA 1156, LGA 1366, LGA 1567, and rPGA988A. On the other hand, the Sandy Bridge processors support LGA 1155 in desktops and workstations, LGA 2011 in high-end servers, LGA 1356 in low-end, dual-processor servers, and may even support BGA 1023, BGA 1224, and BGA 1284.
• The successor to the Westmere processors is Sandy Bridge itself, but the successors to the Sandy Bridge processor are the Ivy Bridge (Tick) processors and the Haswell (Tock) CPUs.

Conclusion

The Westmere CPUs, as you can see from the specs and features of them reveal, are the worthy successors of the Nehalem-based processors, especially the six-core Xeon processors, in more than one way.

They offer faster AES encryption, a steady performance per core ratio, and better scalability, among other benefits.