1. Field of the Invention
This invention relates to caching where portions of data are stored in slower main memory and are transferred to a faster memory comprising cache structures between one or more processors and the main memory. For a system with a shared Level 2 (L2) cache(s) memory between the processor(s) and the main memory, an individual L1 cache memory of a processor must first communicate to an associated L2 cache memory system or check with other L2 cache memory systems which may possibly contain a copy of a cache line of data at a given cached location prior to, or upon modification or appropriation of a cache line of data at a given cached location, and includes provisions for notifying the L2 cache memory systems upon determining when the data stored in a particular cache memory location has been replaced.
2. Background Art
In multi-processor systems with a secondary shared cache, e.g., a L2 cache memory, the shared L2 cache memory holds cache lines of data beyond the actual residency of the cache memory in subordinate caches, e.g., a Level 1 (L1) cache memory. This is the value of a secondary L2 cache memory; it contains a superset of all subordinate caches. This reduces traffic from the main memory. Hereinafter, as will be well understood by those skilled in the art, an L1 cache memory may be referred to as an L1 cache and an L2 cache memory may be referred to as an L2 cache; and a cache line of data is referred to as a cache line or a line.
When a processor, e.g., processor A, requests a line for store (“EX” or exclusive request) that line is installed in both the L1 and the L2 caches as exclusive to processor A. When another processor, e.g., processor B, requests the line, processing is elongated if the line is held exclusive to processor A.
However, as is often the case, the central processor A is no longer using the line, and, in fact the line has aged out of the L1 cache and is present only in the L2 cache. Nevertheless, the line is still marked in the L2 cache as being exclusive to the central processor A. Thus extra processing, such as cross investigating and cross invalidating, is required to remove the marking of exclusivity of this line to processor A from the L1 cache and the L2 cache.
In many cases as described here, this line no longer exists in the L1 cache. This unnecessarily elongates the fetch request for another central processor B for the same line, reducing performance. Thus a need exists to speed up this process.
The method and system described herein provides caching where portions of data are stored in slower main memory and are transferred to faster memory between one or more processors and the main memory. As a general rule, the cache is such that an individual cache system must communicate to other associated cache systems, or check with such cache systems, to determine if they contain a copy of a given cached location prior to or upon modification or appropriation of data at a given cached location. The cache further includes provisions for determining when the data stored in a particular memory location may be replaced.
By way of background, a given line must exist in an L1 cache corresponding to a central processor (CP) to be used by such central processor. Now, when a given line exists in both the L1 cache and the L2 cache, the state of the given line is said to match in both caches. That is, the given line is either marked exclusive or shared, while being modified to the same marked ownership state in both levels. If a given line ages out of the L1 cache and only resides in the L2 cache, then the given line can no longer be in use by the original central processor. More recently used lines have replaced the given line. However, the L2 cache continues to hold the given line in a state of “Exclusive” to the original central processor.
The method and system in accordance with this invention changes the state of the given line in the L2 cache when the given line is to be replaced in the L1 cache. The state of the given line changes from exclusive to a particular identified processor to exclusive to no processor. Thus, in accordance with this invention, when an exclusive line ages out of a particular processor, the line state in the L2 cache will change from exclusive to the particular identified processor to “none”, and no cross interrogation or invalidation is necessary when the given cache line is to be returned for a request from any L1 cache, thereby improving overall system performance.
In accordance with this invention caching is provided where portions of data are stored in a slower main memory and are transferred to a faster memory comprising cache structures between one or more central processors and the slower main memory. For a system with a shared L2 cache(s) between the processor(s) and the main memory, an individual L1 cache or a processor must first communicate to an associated L2 cache memory systems, or check with such other L2 cache systems, to obtain a copy of a given cached location prior to or upon modification or appropriation of data at a given cached location. The cache further includes provisions for notifying the L2 cache systems upon determining when the data stored in a particular cache memory location has been replaced.
As described herein, when exclusive given line ages out of an L1 cache, the L1 cache sends a signal to the L2 cache, indicating that the exclusive given line no longer exists in the L1 cache. This causes the L2 cache to be updated to indicate that the given line is “disowned.” That is, the ownership of the given line is changed from that particular central processor to no specific processor. i.e. “unowned.”
By way of background, despite improvements in technology, microprocessors are still much faster than main memory. Memory access time is increasingly the bottleneck in overall application performance. As a result, an application might spend a considerable amount of time waiting for data. This not only negatively impacts the overall application performance, but the application cannot benefit much from a processor clock-speed upgrade either.
One way to overcome this problem is to insert a hierarchy of small high-speed buffer memory between the central processor and main memory. Such a buffer is generally referred to as cache memory, or cache for short. Such a hierarchy of buffer or cache is illustrated by an L1 cache 113 and an intermediate L2 cache 121 in
The system 101 first copies the data needed by the central processor 111 from main memory 131 into the L2 cache 121, and then from the L2 cache 121 to the L1 cache 113 and from there into a register (not shown) in the processor 111. Storage of results moves data in the opposite direction. First the system copies the data from the central processor 111 into the L1 cache 113, and then from the L2 cache 121. Depending on the cache architecture details, the data is then immediately copied back to the memory 131 (write-through), or deferred (write-back). If an application needs the same data again, data access time is reduced significantly if the data is still in the L1 cache 113 and L2 cache 121 or only the L2 cache 121. To further reduce the cost of memory transfer, more than one element of data is loaded into the cache. The unit of transfer of data is called a cache block or cache line, (referred to herein as a line). Access to a single data element brings an entire line into the cache. The line is guaranteed to contain the element requested.
Latency and bandwidth are two metrics associated with caches and memory. Neither of them is uniform, but each of them is specific to a particular component of the memory hierarchy. The latency is often expressed in central processor cycles or in nanoseconds, while the bandwidth is usually given in megabytes per second or gigabytes per second.
In practice the latency of a memory component is measured as the time it takes to fetch the first portion of a unit of transfer (typically a cache line). As the speed of a component depends on its relative location in the hierarchy, the latency is not uniform. As a rule of thumb, it is safe to say that latency increases when moving from the L1 cache 113 to the L2 cache 121 to the main memory 131.
Some of the memory components, the L1 cache 113 for example, may be physically located in the central processor 111 as shown in
On some microprocessors, the integrated (on-chip) caches, such as the L1 cache 113, do not always run at the speed of the central processor. They operate at a clock rate that is an integer quotient (½, ⅓, and so forth) of the central processor clock.
Cache components external to the processor do not usually, or only partially, benefit from a processor clock upgrade. Their latencies are often given in nanoseconds. Main memory latency is almost always expressed in nanoseconds.
Bandwidth is a measure of the asymptotic speed of a memory component. This number reflects how fast large bulks of data can be moved in and out of the memory component. Just as with latency, the bandwidth is not uniform. Typically, bandwidth decreases the further one moves the memory component away from the central processor 111.
If the number of steps in a data fetch can be reduced, latency is reduced.
When an exclusive given line ages out of the L1 cache 313, the L1 cache 313 sends a signal to the L2 cache 321, indicating that the given line no longer exists in the L1 cache 313. This causes the L2 cache 321 to be updated to indicate that the given line is “disowned.” That is, the ownership is changed from owned by the particular central processor to “unowned.”
Consider the cache shown in
The cache index selects the tag to be tested to be see if the desired block is in the L2 cache 321. The next step is index selection. In the case of direct mapping, the data is read and sent to the central processor 311 in parallel with the tag being read and checked.
After reading the tag from the L2 cache 321, it is compared to the tag portion of the block address from the central processor 311. To be sure that the tag and the data line both contain valid information, the validity data (as exclusive to a central processor or “unowned”) must be read and verified. If the cache data is valid, the central processor 311 is signaled to load the data from the L2 cache 321.
In the case of a write, the data is also written to a write buffer, for writing to main memory.
In the case of a cache miss, the L1 cache 313 sends a stall signal to the central processor 311, and a read to L2 cache 321 and then to the main memory 331. Reading from the main memory 331 replaces a selected line in the caches e.g., typically the least recently used line, including the data, the tags, and the validity information.
According to the method and system of the invention, when an exclusive given line ages out of an L1 cache, and the L1 cache sends a signal to the L2 cache, this indicates that the given line no longer exists in the L1 cache, thereby causing the L2 cache to be updated to indicate that the given line is “disowned.” That is, the ownership of the given line is changed from the particular central processor to “unowned.”
The embodiment of
Additionally, eliminating cross interrogate processing avoids sending a cross invalidate to an L1 cache 213a or an L1 cache 213b for a given line that is to be replaced in the L2 cache 221. Ordinarily, when a given line ages out of the L2 cache 221, that given line must also be invalidated in the L1 cache 213a or the L1 cache 213b. This maintains a subset rule between the L1 cache 213a or the L1 cache 213b and the L2 cache 221. These two invalidates disrupt normal processing at the L1 cache 213a or the L1 cache 213b. If the given line no longer exists in the L1 cache 213a or the L1 cache 213b, this disruption is unnecessary and impacts performance negatively.
While the method and system have been described with respect to exclusive lines, it is to be understood that the method and system described herein also avoids cross interrogates and cross invalidates for shared lines when one updates an L2 cache directory 221 for shared lines that age out of the L1 cache 211a or the L1 cache 211b.
If a shared cache implementation maintains a separate bit for each central processor to identify which L1 caches “hold the line”, each time a given line ages out of the L1 cache 211a or the L1 cache 211b, the respective “CP holds the line” bit in the L2 cache 221 could be reset. Cross interrogates and cross invalidates need to be performed only to CP's having their “CP (CP) holds line” bit active. If all bits are inactive, all cross interrogates and cross invalidates are avoided. In this way, L2 cache 221 age out replacements avoid unnecessary cross invalidates to the L1 cache 213a or the L1 cache 213b. This too improves performance by avoiding unnecessary disruptions at the L1 cache 213a or the L1 cache 213b.
A further aspect of the invention is a program product comprising computer readable program code to configure and control a computer system. The computer system comprises a plurality of microprocessors, where each microprocessor has an associated L1 cache 213a or an L1 cache 213b, an L2 cache 221 associated to the L1 cache 213a and the L1 cache 213b, and a main memory 231. The program product causes the caches to store portions of data for transfer to and from the microprocessors and the main memory, and further to cause an individual cache communicating to other associated caches to check with the other caches, to determine whether they contain a copy of a given cached location prior to reading data from, reading data to, or modifying cached data. The computer code also determines when the data stored in a particular cache memory location may be replaced.
Specifically, the computer code writes changes to the state of the line in the L2 cache when the given line is to be replaced in the L1 cache, and also changes the state of the given line from exclusive to a particular identified processor to exclusive to no processor.
While the invention has been described with certain preferred embodiments and exemplifications, it is not intended to limit the scope of the invention thereby, but solely by the claims appended hereto.
Number | Name | Date | Kind |
---|---|---|---|
5265232 | Gannon et al. | Nov 1993 | A |
5809526 | Patel | Sep 1998 | A |
5926830 | Feiste et al. | Jul 1999 | A |
5963978 | Feiste et al. | Oct 1999 | A |
6269425 | Mounes-Toussi et al. | Jul 2001 | B1 |
6405322 | Gaither et al. | Jun 2002 | B1 |
6532520 | Dean et al. | Mar 2003 | B1 |
6542926 | Zalewski et al. | Apr 2003 | B2 |
6557084 | Freerksen et al. | Apr 2003 | B2 |
6658539 | Arimilli et al. | Dec 2003 | B2 |
6662275 | Arimilli et al. | Dec 2003 | B2 |
6715040 | Wang et al. | Mar 2004 | B2 |
6745292 | Stevens | Jun 2004 | B1 |
6807619 | Ezra et al. | Oct 2004 | B1 |
6826654 | Arimilli et al. | Nov 2004 | B2 |
6851024 | Wilkes et al. | Feb 2005 | B1 |
6868485 | Conway | Mar 2005 | B1 |
7039765 | Wilkes | May 2006 | B1 |
20040143712 | Armstrong et al. | Jul 2004 | A1 |
Number | Date | Country |
---|---|---|
1093961 | Nov 2002 | CN |
2001-034533 | Sep 2001 | JP |
Number | Date | Country | |
---|---|---|---|
20070174554 A1 | Jul 2007 | US |