A cache to speed-up the reading of ROOT datasets.
When writing a TTree, the branch buffers are kept in memory. A typical branch buffersize (before compression) is typically 32 KBytes. After compression, the zipped buffer may be just a few Kbytes. The branch buffers cannot be much larger in case of TTrees with several hundred or thousand branches.
When writing, this does not generate a performance problem because branch buffers are always written sequentially and, thanks to OS optimisations, content is flushed to the output file when a few MBytes of data are available. On the other hand, when reading, one may hit performance problems because of latencies e.g imposed by network. For example in a WAN with 10ms latency, reading 1000 buffers of 10 KBytes each with no cache will imply 10s penalty where a local read of the 10 MBytes would take about 1 second.
The TreeCache tries to prefetch all the buffers for the selected branches in order to transfer a few multi-Megabytes large buffers instead of many multi-kilobytes small buffers. In addition, TTreeCache can sort the blocks to be read in increasing order such that the file is read sequentially.
Systems like xrootd, dCache or httpd take advantage of the TTreeCache in reading ahead as much data as they can and return to the application the maximum data specified in the cache and have the next chunk of data ready when the next request comes.
Yes, some corner cases. For example, when reading only a small fraction of all entries such that not all branch buffers are read.
This class acts as a file cache, registering automatically the baskets from the branches being processed via direct manipulation of TTrees or with tools such as TTree::Draw, TTree::Process, TSelector, TTreeReader and RDataFrame when in the learning phase. The learning phase is by default 100 entries. It can be changed via TTreeCache::SetLearnEntries.
The usage of a TTreeCache can considerably improve the runtime performance at the price of a modest investment in memory, in particular when the TTree is accessed remotely, e.g. via a high latency network.
For each TTree being processed a TTreeCache object is created. This object is automatically deleted when the Tree is deleted or when the file is deleted. The user can change the size of the cache with the TTree::SetCacheSize method (by default the size is 30 Megabytes). This feature can be controlled with the environment variable ROOT_TTREECACHE_SIZE
or the TTreeCache.Size option. The entry range for which the cache is active can also be set with the SetEntryRange method.
The usage of TChain or TEventList have influence on the behaviour of the cache:
The learning phase is started or restarted when:
The learning period is stopped (and prefetching is started) when:
The TTreeCache can optimize its behavior on a cache miss. When miss optimization is enabled (see the SetOptimizeMisses method), it tracks all branches utilized after the learning phase which caused a cache miss. When one cache miss occurs, all the utilized branches are be prefetched for that event. This optimization utilizes the observation that infrequently accessed branches are often accessed together. An example scenario where such behavior is desirable, is an analysis where a set of collections are read only for a few events in which a certain condition is respected, e.g. a trigger fired.
When this mode is enabled, the memory dedicated to the cache can increase by at most a factor two in the case of cache miss. Additionally, on the first miss of an event, we must iterate through all the "active branches" for the miss cache and find the correct basket. This can be potentially a CPU-expensive operation compared to, e.g., the latency of a SSD. This is why the miss cache is currently disabled by default.
A few use cases are discussed below. A cache may be created with automatic sizing when a TTree is used:
In some applications, e.g. central processing workflows of experiments, the list of branches to read is known a priori. For these cases, the TTreeCache can be instructed about the branches which will be read via explicit calls to the TTree or TTreeCache interfaces. In less streamlined applications such as analysis, predicting the branches which will be read can be difficult. In such cases, ROOT I/O flags used branches automatically when a branch buffer is read during the learning phase.
In the examples below, portions of analysis code are shown. The few statements involving the TreeCache are marked with //<<<
If you use RDataFrame or TTreeReader, the system will automatically cache the best set of branches: no action is required by the user.
The TreeCache is automatically used by TTree::Draw. The method knows which branches are used in the query and it puts automatically these branches in the cache. The entry range is also inferred automatically.
The user must enable the cache and tell the system which branches to cache and also specify the entry range. It is important to specify the entry range in case only a subset of the events is processed to avoid wasteful caching.
In the Process function we read a subset of the branches. Only the branches used in the first entry will be put in the cache
In this example, exactly two branches are always used: those need to be prefetched.
In this example, two branches are always used: in addition, some analysis functions are invoked and those may trigger the reading of other branches which are a priori not known. There is no point in prefetching branches that will be used very rarely: we can rely on the system to cache the right branches.
Once the event loop terminated, the number of effective system reads for a given file can be checked with a code like the following:
Another handy command is:
Definition at line 32 of file TTreeCache.h.
Classes | |
struct | IOPos |
struct | MissCache |
Public Types | |
enum | EPrefillType { kNoPrefill , kAllBranches } |
Public Types inherited from TObject | |
enum | { kIsOnHeap = 0x01000000 , kNotDeleted = 0x02000000 , kZombie = 0x04000000 , kInconsistent = 0x08000000 , kBitMask = 0x00ffffff } |
enum | { kSingleKey = BIT(0) , kOverwrite = BIT(1) , kWriteDelete = BIT(2) } |
enum | EDeprecatedStatusBits { kObjInCanvas = BIT(3) } |
enum | EStatusBits { kCanDelete = BIT(0) , kMustCleanup = BIT(3) , kIsReferenced = BIT(4) , kHasUUID = BIT(5) , kCannotPick = BIT(6) , kNoContextMenu = BIT(8) , kInvalidObject = BIT(13) } |
Public Member Functions | |
TTreeCache () | |
Default Constructor. | |
TTreeCache (TTree *tree, Int_t buffersize=0) | |
Constructor. | |
virtual | ~TTreeCache () |
Destructor. (in general called by the TFile destructor) | |
virtual Int_t | AddBranch (const char *branch, Bool_t subbranches=kFALSE) |
Add a branch to the list of branches to be stored in the cache this is to be used by user (thats why we pass the name of the branch). | |
virtual Int_t | AddBranch (TBranch *b, Bool_t subgbranches=kFALSE) |
Add a branch to the list of branches to be stored in the cache this function is called by the user via TTree::AddBranchToCache. | |
virtual void | Disable () |
virtual Int_t | DropBranch (const char *branch, Bool_t subbranches=kFALSE) |
Remove a branch to the list of branches to be stored in the cache this is to be used by user (thats why we pass the name of the branch). | |
virtual Int_t | DropBranch (TBranch *b, Bool_t subbranches=kFALSE) |
Remove a branch to the list of branches to be stored in the cache this function is called by TBranch::GetBasket. | |
virtual void | Enable () |
virtual Bool_t | FillBuffer () |
Fill the cache buffer with the branches in the cache. | |
const TObjArray * | GetCachedBranches () const |
EPrefillType | GetConfiguredPrefillType () const |
Return the desired prefill type from the environment or resource variable. | |
Double_t | GetEfficiency () const |
Give the total efficiency of the primary cache... defined as the ratio of blocks found in the cache vs. | |
Double_t | GetEfficiencyRel () const |
This will indicate a sort of relative efficiency... a ratio of the reads found in the cache to the number of reads so far. | |
virtual Int_t | GetEntryMax () const |
virtual Int_t | GetEntryMin () const |
virtual EPrefillType | GetLearnPrefill () const |
Double_t | GetMissEfficiency () const |
The total efficiency of the 'miss cache' - defined as the ratio of blocks found in the cache versus the number of blocks prefetched. | |
Double_t | GetMissEfficiencyRel () const |
Relative efficiency of the 'miss cache' - ratio of the reads found in cache to the number of reads so far. | |
Bool_t | GetOptimizeMisses () const |
TTree * | GetTree () const |
Bool_t | IsAutoCreated () const |
virtual Bool_t | IsEnabled () const |
virtual Bool_t | IsLearning () const |
virtual Int_t | LearnBranch (TBranch *b, Bool_t subgbranches=kFALSE) |
Add a branch discovered by actual usage to the list of branches to be stored in the cache this function is called by TBranch::GetBasket If we are not longer in the training phase this is an error. | |
virtual void | LearnPrefill () |
Perform an initial prefetch, attempting to read as much of the learning phase baskets for all branches at once. | |
virtual void | Print (Option_t *option="") const |
Print cache statistics. | |
virtual Int_t | ReadBuffer (char *buf, Long64_t pos, Int_t len) |
Read buffer at position pos if the request is in the list of prefetched blocks read from fBuffer. | |
virtual Int_t | ReadBufferNormal (char *buf, Long64_t pos, Int_t len) |
Old method ReadBuffer before the addition of the prefetch mechanism. | |
virtual Int_t | ReadBufferPrefetch (char *buf, Long64_t pos, Int_t len) |
Used to read a chunk from a block previously fetched. | |
virtual void | ResetCache () |
This will simply clear the cache. | |
void | ResetMissCache () |
Reset all the miss cache training. | |
void | SetAutoCreated (Bool_t val) |
virtual Int_t | SetBufferSize (Int_t buffersize) |
Change the underlying buffer size of the cache. | |
virtual void | SetEntryRange (Long64_t emin, Long64_t emax) |
Set the minimum and maximum entry number to be processed this information helps to optimize the number of baskets to read when prefetching the branch buffers. | |
virtual void | SetFile (TFile *file, TFile::ECacheAction action=TFile::kDisconnect) |
Overload to make sure that the object specific. | |
virtual void | SetLearnPrefill (EPrefillType type=kNoPrefill) |
Set whether the learning period is started with a prefilling of the cache and which type of prefilling is used. | |
void | SetOptimizeMisses (Bool_t opt) |
Start of methods for the miss cache. | |
void | StartLearningPhase () |
The name should be enough to explain the method. | |
virtual void | StopLearningPhase () |
This is the counterpart of StartLearningPhase() and can be used to stop the learning phase. | |
virtual void | UpdateBranches (TTree *tree) |
Update pointer to current Tree and recompute pointers to the branches in the cache. | |
Public Member Functions inherited from TFileCacheRead | |
TFileCacheRead () | |
Default Constructor. | |
TFileCacheRead (TFile *file, Int_t buffersize, TObject *tree=nullptr) | |
Creates a TFileCacheRead data structure. | |
virtual | ~TFileCacheRead () |
Destructor. | |
virtual void | AddNoCacheBytesRead (Long64_t len) |
virtual void | AddNoCacheReadCalls (Int_t reads) |
virtual void | Close (Option_t *option="") |
Close out any threads or asynchronous fetches used by the underlying implementation. | |
virtual Int_t | GetBufferSize () const |
virtual Long64_t | GetBytesRead () const |
virtual Long64_t | GetBytesReadExtra () const |
TFile * | GetFile () const |
virtual Long64_t | GetNoCacheBytesRead () const |
virtual Int_t | GetNoCacheReadCalls () const |
Int_t | GetNseek () const |
Int_t | GetNtot () const |
Long64_t | GetPrefetchedBlocks () const |
virtual TFilePrefetch * | GetPrefetchObj () |
virtual Int_t | GetReadCalls () const |
virtual Int_t | GetUnzipBuffer (char **, Long64_t, Int_t, Bool_t *) |
virtual Bool_t | IsAsyncReading () const |
virtual Bool_t | IsEnablePrefetching () const |
virtual void | Prefetch (Long64_t pos, Int_t len) |
Add block of length len at position pos in the list of blocks to be prefetched. | |
virtual Int_t | ReadBufferExt (char *buf, Long64_t pos, Int_t len, Int_t &loc) |
virtual Int_t | ReadBufferExtNormal (char *buf, Long64_t pos, Int_t len, Int_t &loc) |
Base function for ReadBuffer. | |
virtual Int_t | ReadBufferExtPrefetch (char *buf, Long64_t pos, Int_t len, Int_t &loc) |
prefetch the first block | |
virtual void | SecondPrefetch (Long64_t, Int_t) |
virtual void | SecondSort () |
Sort buffers to be prefetched in increasing order of positions. | |
virtual void | SetEnablePrefetching (Bool_t setPrefetching=kFALSE) |
Set the prefetching mode of this file. | |
virtual void | SetSkipZip (Bool_t=kTRUE) |
virtual void | Sort () |
Sort buffers to be prefetched in increasing order of positions. | |
virtual void | WaitFinishPrefetch () |
Public Member Functions inherited from TObject | |
TObject () | |
TObject constructor. | |
TObject (const TObject &object) | |
TObject copy ctor. | |
virtual | ~TObject () |
TObject destructor. | |
void | AbstractMethod (const char *method) const |
Use this method to implement an "abstract" method that you don't want to leave purely abstract. | |
virtual void | AppendPad (Option_t *option="") |
Append graphics object to current pad. | |
virtual void | Browse (TBrowser *b) |
Browse object. May be overridden for another default action. | |
ULong_t | CheckedHash () |
Check and record whether this class has a consistent Hash/RecursiveRemove setup (*) and then return the regular Hash value for this object. | |
virtual const char * | ClassName () const |
Returns name of class to which the object belongs. | |
virtual void | Clear (Option_t *="") |
virtual TObject * | Clone (const char *newname="") const |
Make a clone of an object using the Streamer facility. | |
virtual Int_t | Compare (const TObject *obj) const |
Compare abstract method. | |
virtual void | Copy (TObject &object) const |
Copy this to obj. | |
virtual void | Delete (Option_t *option="") |
Delete this object. | |
virtual Int_t | DistancetoPrimitive (Int_t px, Int_t py) |
Computes distance from point (px,py) to the object. | |
virtual void | Draw (Option_t *option="") |
Default Draw method for all objects. | |
virtual void | DrawClass () const |
Draw class inheritance tree of the class to which this object belongs. | |
virtual TObject * | DrawClone (Option_t *option="") const |
Draw a clone of this object in the current selected pad for instance with: gROOT->SetSelectedPad(gPad) . | |
virtual void | Dump () const |
Dump contents of object on stdout. | |
virtual void | Error (const char *method, const char *msgfmt,...) const |
Issue error message. | |
virtual void | Execute (const char *method, const char *params, Int_t *error=0) |
Execute method on this object with the given parameter string, e.g. | |
virtual void | Execute (TMethod *method, TObjArray *params, Int_t *error=0) |
Execute method on this object with parameters stored in the TObjArray. | |
virtual void | ExecuteEvent (Int_t event, Int_t px, Int_t py) |
Execute action corresponding to an event at (px,py). | |
virtual void | Fatal (const char *method, const char *msgfmt,...) const |
Issue fatal error message. | |
virtual TObject * | FindObject (const char *name) const |
Must be redefined in derived classes. | |
virtual TObject * | FindObject (const TObject *obj) const |
Must be redefined in derived classes. | |
virtual Option_t * | GetDrawOption () const |
Get option used by the graphics system to draw this object. | |
virtual const char * | GetIconName () const |
Returns mime type name of object. | |
virtual const char * | GetName () const |
Returns name of object. | |
virtual char * | GetObjectInfo (Int_t px, Int_t py) const |
Returns string containing info about the object at position (px,py). | |
virtual Option_t * | GetOption () const |
virtual const char * | GetTitle () const |
Returns title of object. | |
virtual UInt_t | GetUniqueID () const |
Return the unique object id. | |
virtual Bool_t | HandleTimer (TTimer *timer) |
Execute action in response of a timer timing out. | |
virtual ULong_t | Hash () const |
Return hash value for this object. | |
Bool_t | HasInconsistentHash () const |
Return true is the type of this object is known to have an inconsistent setup for Hash and RecursiveRemove (i.e. | |
virtual void | Info (const char *method, const char *msgfmt,...) const |
Issue info message. | |
virtual Bool_t | InheritsFrom (const char *classname) const |
Returns kTRUE if object inherits from class "classname". | |
virtual Bool_t | InheritsFrom (const TClass *cl) const |
Returns kTRUE if object inherits from TClass cl. | |
virtual void | Inspect () const |
Dump contents of this object in a graphics canvas. | |
void | InvertBit (UInt_t f) |
virtual Bool_t | IsEqual (const TObject *obj) const |
Default equal comparison (objects are equal if they have the same address in memory). | |
virtual Bool_t | IsFolder () const |
Returns kTRUE in case object contains browsable objects (like containers or lists of other objects). | |
R__ALWAYS_INLINE Bool_t | IsOnHeap () const |
virtual Bool_t | IsSortable () const |
R__ALWAYS_INLINE Bool_t | IsZombie () const |
virtual void | ls (Option_t *option="") const |
The ls function lists the contents of a class on stdout. | |
void | MayNotUse (const char *method) const |
Use this method to signal that a method (defined in a base class) may not be called in a derived class (in principle against good design since a child class should not provide less functionality than its parent, however, sometimes it is necessary). | |
virtual Bool_t | Notify () |
This method must be overridden to handle object notification. | |
void | Obsolete (const char *method, const char *asOfVers, const char *removedFromVers) const |
Use this method to declare a method obsolete. | |
void | operator delete (void *ptr) |
Operator delete. | |
void | operator delete[] (void *ptr) |
Operator delete []. | |
void * | operator new (size_t sz) |
void * | operator new (size_t sz, void *vp) |
void * | operator new[] (size_t sz) |
void * | operator new[] (size_t sz, void *vp) |
TObject & | operator= (const TObject &rhs) |
TObject assignment operator. | |
virtual void | Paint (Option_t *option="") |
This method must be overridden if a class wants to paint itself. | |
virtual void | Pop () |
Pop on object drawn in a pad to the top of the display list. | |
virtual Int_t | Read (const char *name) |
Read contents of object with specified name from the current directory. | |
virtual void | RecursiveRemove (TObject *obj) |
Recursively remove this object from a list. | |
void | ResetBit (UInt_t f) |
virtual void | SaveAs (const char *filename="", Option_t *option="") const |
Save this object in the file specified by filename. | |
virtual void | SavePrimitive (std::ostream &out, Option_t *option="") |
Save a primitive as a C++ statement(s) on output stream "out". | |
void | SetBit (UInt_t f) |
void | SetBit (UInt_t f, Bool_t set) |
Set or unset the user status bits as specified in f. | |
virtual void | SetDrawOption (Option_t *option="") |
Set drawing option for object. | |
virtual void | SetUniqueID (UInt_t uid) |
Set the unique object id. | |
virtual void | SysError (const char *method, const char *msgfmt,...) const |
Issue system error message. | |
R__ALWAYS_INLINE Bool_t | TestBit (UInt_t f) const |
Int_t | TestBits (UInt_t f) const |
virtual void | UseCurrentStyle () |
Set current style settings in this object This function is called when either TCanvas::UseCurrentStyle or TROOT::ForceStyle have been invoked. | |
virtual void | Warning (const char *method, const char *msgfmt,...) const |
Issue warning message. | |
virtual Int_t | Write (const char *name=0, Int_t option=0, Int_t bufsize=0) |
Write this object to the current directory. | |
virtual Int_t | Write (const char *name=0, Int_t option=0, Int_t bufsize=0) const |
Write this object to the current directory. | |
Static Public Member Functions | |
static Int_t | GetLearnEntries () |
Static function returning the number of entries used to train the cache see SetLearnEntries. | |
static void | SetLearnEntries (Int_t n=10) |
Static function to set the number of entries to be used in learning mode The default value for n is 10. | |
Static Public Member Functions inherited from TObject | |
static Long_t | GetDtorOnly () |
Return destructor only flag. | |
static Bool_t | GetObjectStat () |
Get status of object stat flag. | |
static void | SetDtorOnly (void *obj) |
Set destructor only flag. | |
static void | SetObjectStat (Bool_t stat) |
Turn on/off tracking of objects in the TObjectTable. | |
Protected Attributes | |
Bool_t | fAutoCreated {kFALSE} |
! true if cache was automatically created | |
TObjArray * | fBranches {nullptr} |
! List of branches to be stored in the cache | |
TList * | fBrNames {nullptr} |
! list of branch names in the cache | |
Long64_t | fCurrentClusterStart {-1} |
! Start of the cluster(s) where the current content was picked out | |
Bool_t | fEnabled {kTRUE} |
! cache enabled for cached reading | |
Long64_t | fEntryCurrent {-1} |
! current lowest entry number in the cache | |
Long64_t | fEntryMax {1} |
! last entry in the cache | |
Long64_t | fEntryMin {0} |
! first entry in the cache | |
Long64_t | fEntryNext {-1} |
! next entry number where cache must be filled | |
Int_t | fFillTimes {0} |
! how many times we can fill the current buffer | |
Bool_t | fFirstBuffer {kTRUE} |
! true if first buffer is used for prefetching | |
Long64_t | fFirstEntry {-1} |
! save the value of the first entry | |
Long64_t | fFirstMiss {-1} |
! set to the event # of the first miss. | |
Bool_t | fFirstTime {kTRUE} |
! save the fact that we processes the first entry | |
Bool_t | fIsLearning {kTRUE} |
! true if cache is in learning mode | |
Bool_t | fIsManual {kFALSE} |
! true if cache is StopLearningPhase was used | |
Long64_t | fLastMiss {-1} |
! set to the event # of the last miss. | |
Bool_t | fLearnPrefilling {kFALSE} |
! true if we are in the process of executing LearnPrefill | |
std::unique_ptr< MissCache > | fMissCache |
! Cache contents for misses | |
Int_t | fNbranches {0} |
! Number of branches in the cache | |
Long64_t | fNextClusterStart {-1} |
! End+1 of the cluster(s) where the current content was picked out | |
Int_t | fNMissReadMiss {0} |
Number of blocks read and not found in either cache. | |
Int_t | fNMissReadOk {0} |
Number of blocks read, not found in the primary cache, and found in the secondary cache. | |
Int_t | fNMissReadPref {0} |
Number of blocks read into the secondary ("miss") cache. | |
Int_t | fNReadMiss {0} |
Number of blocks read and not found in the cache. | |
Int_t | fNReadOk {0} |
Number of blocks read and found in the cache. | |
Int_t | fNReadPref {0} |
Number of blocks that were prefetched. | |
Bool_t | fOneTime {kFALSE} |
! used in the learning phase | |
Bool_t | fOptimizeMisses {kFALSE} |
! true if we should optimize cache misses. | |
EPrefillType | fPrefillType |
Whether a pre-filling is enabled (and if applicable which type) | |
Bool_t | fReadDirectionSet {kFALSE} |
! read direction established | |
Bool_t | fReverseRead {kFALSE} |
! reading in reverse mode | |
TTree * | fTree {nullptr} |
! pointer to the current Tree | |
Protected Attributes inherited from TFileCacheRead | |
Bool_t | fAsyncReading |
Bool_t | fBIsSorted |
Bool_t | fBIsTransferred |
Int_t * | fBLen |
[fBNb] | |
Int_t | fBNb |
Int_t | fBNseek |
Int_t | fBNtot |
Long64_t * | fBPos |
[fBNb] | |
Long64_t * | fBSeek |
[fBNseek] | |
Int_t * | fBSeekIndex |
[fBNseek] | |
Int_t * | fBSeekLen |
[fBNseek] | |
Int_t * | fBSeekPos |
[fBNseek] | |
Int_t | fBSeekSize |
Long64_t * | fBSeekSort |
[fBNseek] | |
Int_t * | fBSeekSortLen |
[fBNseek] | |
char * | fBuffer |
[fBufferSize] buffer of contiguous prefetched blocks | |
Int_t | fBufferLen |
Current buffer length (<= fBufferSize) | |
Int_t | fBufferSize |
Allocated size of fBuffer (at a given time) | |
Int_t | fBufferSizeMin |
Original size of fBuffer. | |
Long64_t | fBytesRead |
Number of bytes read for this cache. | |
Long64_t | fBytesReadExtra |
Number of extra bytes (overhead) read by the readahead buffer. | |
Bool_t | fEnablePrefetching |
reading by prefetching asynchronously | |
TFile * | fFile |
Pointer to file. | |
Bool_t | fIsSorted |
True if fSeek array is sorted. | |
Bool_t | fIsTransferred |
True when fBuffer contains something valid. | |
Int_t * | fLen |
[fNb] Length of long buffers | |
Int_t | fNb |
Number of long buffers. | |
Long64_t | fNoCacheBytesRead |
Number of bytes read by basket to fill cached tree. | |
Int_t | fNoCacheReadCalls |
Number of read calls by basket to fill cached tree. | |
Int_t | fNseek |
Number of blocks to be prefetched. | |
Int_t | fNtot |
Total size of prefetched blocks. | |
Long64_t * | fPos |
[fNb] start of long buffers | |
TFilePrefetch * | fPrefetch |
! Object that does the asynchronous reading in another thread | |
Long64_t | fPrefetchedBlocks |
Number of blocks prefetched. | |
Int_t | fReadCalls |
Number of read calls for this cache. | |
Long64_t * | fSeek |
[fNseek] Position on file of buffers to be prefetched | |
Int_t * | fSeekIndex |
[fNseek] sorted index table of fSeek | |
Int_t * | fSeekLen |
[fNseek] Length of buffers to be prefetched | |
Int_t * | fSeekPos |
[fNseek] Position of sorted blocks in fBuffer | |
Int_t | fSeekSize |
Allocated size of fSeek. | |
Long64_t * | fSeekSort |
[fNseek] Position on file of buffers to be prefetched (sorted) | |
Int_t * | fSeekSortLen |
[fNseek] Length of buffers to be prefetched (sorted) | |
Static Protected Attributes | |
static Int_t | fgLearnEntries = 100 |
number of entries used for learning mode | |
Private Member Functions | |
TTreeCache (const TTreeCache &)=delete | |
this class cannot be copied | |
Bool_t | CalculateMissCache () |
Calculate the appropriate miss cache to fetch; helper function for FillMissCache. | |
TBranch * | CalculateMissEntries (Long64_t, int, bool) |
Given an file read, try to determine the corresponding branch. | |
Bool_t | CheckMissCache (char *buf, Long64_t pos, int len) |
Check the miss cache for a particular buffer, fetching if deemed necessary. | |
Bool_t | FillMissCache () |
Fill the miss cache from the current set of active branches. | |
IOPos | FindBranchBasketPos (TBranch &, Long64_t entry) |
Given a branch and an entry, determine the file location (offset / size) of the corresponding basket. | |
TTreeCache & | operator= (const TTreeCache &)=delete |
Bool_t | ProcessMiss (Long64_t pos, int len) |
! Given a file read not in the miss cache, handle (possibly) loading the data. | |
Additional Inherited Members | |
Protected Types inherited from TObject | |
enum | { kOnlyPrepStep = BIT(3) } |
Protected Member Functions inherited from TFileCacheRead | |
void | SetEnablePrefetchingImpl (Bool_t setPrefetching=kFALSE) |
TFileCacheRead implementation of SetEnablePrefetching. | |
Protected Member Functions inherited from TObject | |
virtual void | DoError (int level, const char *location, const char *fmt, va_list va) const |
Interface to ErrorHandler (protected). | |
void | MakeZombie () |
#include <TTreeCache.h>
Enumerator | |
---|---|
kNoPrefill | |
kAllBranches |
Definition at line 35 of file TTreeCache.h.
|
privatedelete |
this class cannot be copied
TTreeCache::TTreeCache | ( | ) |
Default Constructor.
Definition at line 305 of file TTreeCache.cxx.
Constructor.
Definition at line 312 of file TTreeCache.cxx.
|
virtual |
Destructor. (in general called by the TFile destructor)
Definition at line 324 of file TTreeCache.cxx.
Add a branch to the list of branches to be stored in the cache this is to be used by user (thats why we pass the name of the branch).
It works in exactly the same way as TTree::SetBranchStatus so you probably want to look over there for details about the use of bname with regular expressions. The branches are taken with respect to the Owner of this TTreeCache (i.e. the original Tree) NB: if bname="*" all branches are put in the cache and the learning phase stopped Returns:
Reimplemented from TFileCacheRead.
Reimplemented in TTreeCacheUnzip.
Definition at line 438 of file TTreeCache.cxx.
Add a branch to the list of branches to be stored in the cache this function is called by the user via TTree::AddBranchToCache.
The branch is added even if we are outside of the training phase. Returns:
Reimplemented from TFileCacheRead.
Reimplemented in TTreeCacheUnzip.
Definition at line 368 of file TTreeCache.cxx.
|
private |
Calculate the appropriate miss cache to fetch; helper function for FillMissCache.
Given an file read, try to determine the corresponding branch.
Given a particular IO description (offset / length) representing a 'miss' of the TTreeCache's primary cache, calculate all the corresponding IO that should be performed.
all
indicates that this function should search the set of all branches in this TTree. When set to false, we only search through branches that have previously incurred a miss.
Returns:
Definition at line 782 of file TTreeCache.cxx.
Check the miss cache for a particular buffer, fetching if deemed necessary.
Given an IO operation (pos, len) that was a cache miss in the primary TTC, try the operation again with the miss cache.
Returns true if the IO operation was successful and the contents of buf were populated with the requested data.
Definition at line 913 of file TTreeCache.cxx.
|
inlinevirtual |
Definition at line 136 of file TTreeCache.h.
Remove a branch to the list of branches to be stored in the cache this is to be used by user (thats why we pass the name of the branch).
It works in exactly the same way as TTree::SetBranchStatus so you probably want to look over there for details about the use of bname with regular expressions. The branches are taken with respect to the Owner of this TTreeCache (i.e. the original Tree) NB: if bname="*" all branches are put in the cache and the learning phase stopped Returns:
Definition at line 578 of file TTreeCache.cxx.
Remove a branch to the list of branches to be stored in the cache this function is called by TBranch::GetBasket.
Returns:
Definition at line 533 of file TTreeCache.cxx.
|
inlinevirtual |
Definition at line 137 of file TTreeCache.h.
|
virtual |
Fill the cache buffer with the branches in the cache.
Reimplemented in TTreeCacheUnzip.
Definition at line 1106 of file TTreeCache.cxx.
|
private |
Fill the miss cache from the current set of active branches.
|
private |
Given a branch and an entry, determine the file location (offset / size) of the corresponding basket.
For the event currently being fetched into the miss cache, find the IO (offset / length tuple) to pull in the current basket for a given branch.
Returns:
Definition at line 708 of file TTreeCache.cxx.
|
inline |
Definition at line 139 of file TTreeCache.h.
TTreeCache::EPrefillType TTreeCache::GetConfiguredPrefillType | ( | ) | const |
Return the desired prefill type from the environment or resource variable.
Definition at line 1794 of file TTreeCache.cxx.
Double_t TTreeCache::GetEfficiency | ( | ) | const |
Give the total efficiency of the primary cache... defined as the ratio of blocks found in the cache vs.
the number of blocks prefetched ( it could be more than 1 if we read the same block from the cache more than once )
Note: This should eb used at the end of the processing or we will get incomplete stats
Definition at line 1817 of file TTreeCache.cxx.
Double_t TTreeCache::GetEfficiencyRel | ( | ) | const |
This will indicate a sort of relative efficiency... a ratio of the reads found in the cache to the number of reads so far.
Definition at line 1841 of file TTreeCache.cxx.
|
inlinevirtual |
Definition at line 144 of file TTreeCache.h.
|
inlinevirtual |
Definition at line 143 of file TTreeCache.h.
|
static |
Static function returning the number of entries used to train the cache see SetLearnEntries.
Definition at line 1866 of file TTreeCache.cxx.
|
inlinevirtual |
Definition at line 146 of file TTreeCache.h.
Double_t TTreeCache::GetMissEfficiency | ( | ) | const |
The total efficiency of the 'miss cache' - defined as the ratio of blocks found in the cache versus the number of blocks prefetched.
Definition at line 1829 of file TTreeCache.cxx.
Double_t TTreeCache::GetMissEfficiencyRel | ( | ) | const |
Relative efficiency of the 'miss cache' - ratio of the reads found in cache to the number of reads so far.
Definition at line 1853 of file TTreeCache.cxx.
|
inline |
Definition at line 138 of file TTreeCache.h.
|
inline |
Definition at line 149 of file TTreeCache.h.
|
inline |
Definition at line 150 of file TTreeCache.h.
|
inlinevirtual |
Definition at line 151 of file TTreeCache.h.
|
inlinevirtual |
Reimplemented from TFileCacheRead.
Definition at line 152 of file TTreeCache.h.
Add a branch discovered by actual usage to the list of branches to be stored in the cache this function is called by TBranch::GetBasket If we are not longer in the training phase this is an error.
Returns:
Reimplemented from TFileCacheRead.
Definition at line 342 of file TTreeCache.cxx.
|
virtual |
Perform an initial prefetch, attempting to read as much of the learning phase baskets for all branches at once.
Definition at line 2242 of file TTreeCache.cxx.
|
privatedelete |
Print cache statistics.
Like:
Reimplemented from TFileCacheRead.
Reimplemented in TTreeCacheUnzip.
Definition at line 1891 of file TTreeCache.cxx.
! Given a file read not in the miss cache, handle (possibly) loading the data.
Process a cache miss; (pos, len) isn't in the buffer.
The first time we have a miss, we buffer as many baskets we can (up to the maximum size of the TTreeCache) in memory from all branches that are not in the prefetch list.
Subsequent times, we fetch all the buffers corresponding to branches that had previously seen misses. If it turns out the (pos, len) isn't in the list of branches, we treat this as if it was the first miss.
Returns true if we were able to pull the data into the miss cache.
Definition at line 855 of file TTreeCache.cxx.
Read buffer at position pos if the request is in the list of prefetched blocks read from fBuffer.
Otherwise try to fill the cache from the list of selected branches, and recheck if pos is now in the list. Returns:
Reimplemented from TFileCacheRead.
Definition at line 2019 of file TTreeCache.cxx.
Old method ReadBuffer before the addition of the prefetch mechanism.
Definition at line 1919 of file TTreeCache.cxx.
Used to read a chunk from a block previously fetched.
It will call FillBuffer even if the cache lookup succeeds, because it will try to prefetch the next block as soon as we start reading from the current block.
Definition at line 1978 of file TTreeCache.cxx.
|
virtual |
This will simply clear the cache.
Reimplemented in TTreeCacheUnzip.
Definition at line 2032 of file TTreeCache.cxx.
void TTreeCache::ResetMissCache | ( | ) |
Reset all the miss cache training.
The contents of the miss cache will be emptied as well as the list of branches used.
Definition at line 689 of file TTreeCache.cxx.
Definition at line 164 of file TTreeCache.h.
Change the underlying buffer size of the cache.
If the change of size means some cache content is lost, or if the buffer is now larger, setup for a cache refill the next time there is a read Returns:
Reimplemented from TFileCacheRead.
Reimplemented in TTreeCacheUnzip.
Definition at line 2064 of file TTreeCache.cxx.
Set the minimum and maximum entry number to be processed this information helps to optimize the number of baskets to read when prefetching the branch buffers.
Reimplemented in TTreeCacheUnzip.
Definition at line 2097 of file TTreeCache.cxx.
|
virtual |
Overload to make sure that the object specific.
Reimplemented from TFileCacheRead.
Definition at line 2119 of file TTreeCache.cxx.
Static function to set the number of entries to be used in learning mode The default value for n is 10.
n must be >= 1
Definition at line 2136 of file TTreeCache.cxx.
|
virtual |
Set whether the learning period is started with a prefilling of the cache and which type of prefilling is used.
The two value currently supported are:
Definition at line 2151 of file TTreeCache.cxx.
Start of methods for the miss cache.
Enable / disable the miss cache.
The first time this is called on a TTreeCache object, the corresponding data structures will be allocated. Subsequent enable / disables will simply turn the functionality on/off.
Definition at line 675 of file TTreeCache.cxx.
void TTreeCache::StartLearningPhase | ( | ) |
The name should be enough to explain the method.
The only additional comments is that the cache is cleaned before the new learning phase.
Definition at line 2161 of file TTreeCache.cxx.
|
virtual |
This is the counterpart of StartLearningPhase() and can be used to stop the learning phase.
It's useful when the user knows exactly what branches they are going to use. For the moment it's just a call to FillBuffer() since that method will create the buffer lists from the specified branches.
Reimplemented in TTreeCacheUnzip.
Definition at line 2178 of file TTreeCache.cxx.
Update pointer to current Tree and recompute pointers to the branches in the cache.
Reimplemented in TTreeCacheUnzip.
Definition at line 2202 of file TTreeCache.cxx.
! true if cache was automatically created
Definition at line 66 of file TTreeCache.h.
|
protected |
! List of branches to be stored in the cache
Definition at line 51 of file TTreeCache.h.
|
protected |
! list of branch names in the cache
Definition at line 52 of file TTreeCache.h.
|
protected |
! Start of the cluster(s) where the current content was picked out
Definition at line 42 of file TTreeCache.h.
! cache enabled for cached reading
Definition at line 63 of file TTreeCache.h.
|
protected |
! current lowest entry number in the cache
Definition at line 40 of file TTreeCache.h.
|
protected |
! last entry in the cache
Definition at line 39 of file TTreeCache.h.
|
protected |
! first entry in the cache
Definition at line 38 of file TTreeCache.h.
|
protected |
! next entry number where cache must be filled
Definition at line 41 of file TTreeCache.h.
|
protected |
! how many times we can fill the current buffer
Definition at line 59 of file TTreeCache.h.
! true if first buffer is used for prefetching
Definition at line 56 of file TTreeCache.h.
|
protected |
! save the value of the first entry
Definition at line 61 of file TTreeCache.h.
|
protected |
! set to the event # of the first miss.
Definition at line 73 of file TTreeCache.h.
! save the fact that we processes the first entry
Definition at line 60 of file TTreeCache.h.
|
staticprotected |
number of entries used for learning mode
Definition at line 65 of file TTreeCache.h.
! true if cache is in learning mode
Definition at line 54 of file TTreeCache.h.
! true if cache is StopLearningPhase was used
Definition at line 55 of file TTreeCache.h.
|
protected |
! set to the event # of the last miss.
Definition at line 74 of file TTreeCache.h.
! true if we are in the process of executing LearnPrefill
Definition at line 68 of file TTreeCache.h.
|
protected |
! Cache contents for misses
Definition at line 105 of file TTreeCache.h.
|
protected |
! Number of branches in the cache
Definition at line 44 of file TTreeCache.h.
|
protected |
! End+1 of the cluster(s) where the current content was picked out
Definition at line 43 of file TTreeCache.h.
|
protected |
Number of blocks read and not found in either cache.
Definition at line 48 of file TTreeCache.h.
|
protected |
Number of blocks read, not found in the primary cache, and found in the secondary cache.
Definition at line 46 of file TTreeCache.h.
|
protected |
Number of blocks read into the secondary ("miss") cache.
Definition at line 50 of file TTreeCache.h.
|
protected |
Number of blocks read and not found in the cache.
Definition at line 47 of file TTreeCache.h.
|
protected |
Number of blocks read and found in the cache.
Definition at line 45 of file TTreeCache.h.
|
protected |
Number of blocks that were prefetched.
Definition at line 49 of file TTreeCache.h.
! used in the learning phase
Definition at line 57 of file TTreeCache.h.
! true if we should optimize cache misses.
Definition at line 72 of file TTreeCache.h.
|
protected |
Whether a pre-filling is enabled (and if applicable which type)
Definition at line 64 of file TTreeCache.h.
! read direction established
Definition at line 62 of file TTreeCache.h.
! reading in reverse mode
Definition at line 58 of file TTreeCache.h.
|
protected |
! pointer to the current Tree
Definition at line 53 of file TTreeCache.h.