A multi-threaded implementation of a PromQL Query Engine based on the Volcano/Iterator model.
The project is currently under active development.
The engine intends to have full compatibility with the original engine used in Prometheus. Since implementing the full specification will take time, we aim to add support for most commonly used expressions while falling back to the original engine for operations that are not yet supported. This will allow us to have smaller and faster releases, and gather feedback on regular basis. Instructions on using the engine will be added after we have enough confidence on its correctness.
The following table shows operations which are currently supported by the engine
| Type | Supported | Priority |
|---|---|---|
| Binary expressions | Full support | |
| Histograms | Full support | |
| Subqueries | Full support | |
| Aggregations | Full support except for count_values |
Medium |
| Aggregations over time | Full support except for quantile_over_time with non-constant argument |
Medium |
| Functions | Full support except for holt_winters and predict_linear with non-constant argument |
Medium |
At the beginning of a PromQL query execution, the query engine computes a physical plan consisting of multiple independent operators, each responsible for calculating one part of the query expression.
Operators are assembled in a tree-like structure with every operator calling Next() on its dependants until there is no more data to be returned. The result of the Next() function is a column vector (also called a step vector) with elements in the vector representing samples with the same timestamp from different time series.
This model allows for samples from individual time series to flow one execution step at a time from the left-most operators to the one at the very right. Since most PromQL expressions are aggregations, samples are reduced in number as they are pulled by the operators on the right. Because of this, samples from original timeseries can be decoded and kept in memory in batches instead of being fully expanded.
In addition to operators that have a one-to-one mapping with PromQL constructs, the Volcano model also describes so-called Exchange operators which can be used for flow control and optimizations, such as concurrency or batched selects. An example of an Exchange operator is described in the Intra-operator parallelism section.
Since operators are independent and rely on a common interface for pulling data, they can be run in parallel to each other. As soon as one operator has processed data from an evaluation step, it can pass the result onward so that its upstream can immediately start working on it.
Parallelism can also be added within individual operators using a parallel coalesce exchange operator. Such exchange operators are indistinguishable from regular operators to their upstreams since they respect the same Next() interface.
One challenge with the streamed execution model is knowing how much memory to allocate in each operator for each step.
To work around this issue, operators expose a Series() method which returns the labels for all time series that they will ever produce (for all Next() calls). Operators at the very bottom of the tree, like vector and matrix selectors, have this information since they are responsible for loading data from storage. Other operators can then call Series() on the downstream operator and pre-compute all possible outputs.
Even though this might look like an expensive operation, its cost is identical to just one evaluation step. Knowing sizes of input and output vectors also allows us to:
- allocate memory very precisely by properly sizing vector pools (see section bellow),
- use arrays instead of maps for indexing data, leading to faster execution times due to having less allocations and using index-based lookups, and
- use tight loops in operators by eliminating conditional statements associated with maps.
Since time series are decoded one step at a time, vectors between execution execution steps can be recycled manually instead of relying on the garbage collector. Each operator has its own pool that it uses to allocate new step vectors and send results to its upstream. Whenever the upstream operator is finished with processing a step vector, it will return that vector to the pool of its downstream so that it can be reused again for subsequent steps.
There are currently no mechanisms to apply memory limits to queries within the engine. This is a highly desirable feature, and we would like to explore ways in which we can support it.
The current implementation uses goroutines very liberally which means the query will use as many cores as possible. Limiting the number of cores which a query can use is not yet implemented but we would eventually like to have support for it.
Each PromQL query is initially treated as a declarative (logical) plan and is optimizes before execution. The engine currently supports several optimizers, some of which are enabled by default and others need to be explicitly opted-into. Optimizers implement the Optimizer interface and all implementations can be found in the logicalplan package.
THe engine can be extended through custom optimizers which can be used injected at instantiation. These optimizers can be used to either rearrange the logical nodes into a new plan, or to inject new nodes altogether.
It is also possible to modify the actual execution of a query by injecting a node implementing the UserDefinedOperator interface. This node type has a MakeExecutionOperator method which can be used to control which execution operator should be instantiated for the logical node.
The engine supports a distributed mode where aggregations can be delegated to multiple remote engines, each responsible for an independent dataset. This mode is currently implemented through an optimizer which rewrites a query as a combination of multiple remote and one local aggregation. For example, when two remote engines are available, a query like:
sum(rate(http_request_total[4m]))
would be rewritten as
sum(
coalesce(
sum(rate(http_request_total[4m])) # remote engine 1
sum(rate(http_request_total[4m])) # remote engine 2
)
)
The inner aggregations are forwarded to remote engines and the global result is completed in memory.
An engine using the distributed mode can be created through the NewDistributedEngine function. The user is expected to pass an implementation of RemoteEndpoints which has a single Engines() method. When invoked, Engines() should return all remote engines that can be used for a single query. The Engines() method is called separately for each individual query which allows the RemoteEndpoints implementation to do continuous service discovery and inject engines as they become available.
The interfaces used for remote execution can be found in api package. Note that the RemoteEngine interface has a NewRangeQuery method, similar to the one in the Prometheus v1.QueryEngine interface. It is up to the user of the library to implement this method as they see fit. An example implementation could be to forward the query to an HTTP /api/v1/query_range endpoint of a Prometheus instance. In Thanos, this method is implemented as a gRPC call to a Thanos Querier.
For more details on the overall design, please refer to the proposal in the Thanos project.
If you are interested in the benchmark results captured by continuous benchmark, please check here.
These are the latest benchmarks captured on an Apple M1 Pro processor.
Note that memory usage is higher when executing a query with parallelism greater than 1. This is due to the fact that the engine is able to execute multiple operations at once (e.g. decode chunks from multiple series at the same time), which requires using independent buffers for each parallel operation.
Single core benchmarks
name old time/op new time/op delta
RangeQuery/vector_selector 33.5ms ± 3% 43.4ms ± 3% +29.59% (p=0.008 n=5+5)
RangeQuery/sum 46.6ms ± 1% 34.3ms ± 2% -26.37% (p=0.008 n=5+5)
RangeQuery/sum_by_pod 145ms ± 1% 46ms ± 3% -68.36% (p=0.008 n=5+5)
RangeQuery/topk 46.7ms ± 2% 37.0ms ± 6% -20.79% (p=0.008 n=5+5)
RangeQuery/bottomk 46.9ms ± 1% 35.3ms ± 7% -24.72% (p=0.008 n=5+5)
RangeQuery/rate 65.7ms ± 1% 72.2ms ± 2% +9.88% (p=0.008 n=5+5)
RangeQuery/sum_rate 76.6ms ± 1% 61.9ms ± 1% -19.16% (p=0.008 n=5+5)
RangeQuery/sum_by_rate 180ms ± 3% 74ms ± 7% -58.94% (p=0.008 n=5+5)
RangeQuery/quantile_with_variable_parameter 263ms ± 6% 99ms ± 3% -62.38% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_one_to_one 119ms ± 1% 31ms ± 3% -74.20% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_many_to_one 396ms ± 1% 69ms ± 1% -82.52% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_vector_and_scalar 241ms ± 1% 51ms ± 1% -78.85% (p=0.008 n=5+5)
RangeQuery/unary_negation 35.6ms ± 2% 46.6ms ± 5% +31.00% (p=0.008 n=5+5)
RangeQuery/vector_and_scalar_comparison 205ms ± 3% 57ms ± 4% -72.48% (p=0.008 n=5+5)
RangeQuery/positive_offset_vector 33.2ms ±10% 43.2ms ± 3% +30.20% (p=0.008 n=5+5)
RangeQuery/at_modifier_ 18.7ms ± 3% 18.0ms ± 5% ~ (p=0.095 n=5+5)
RangeQuery/at_modifier_with_positive_offset_vector 17.9ms ± 2% 17.2ms ± 2% -3.78% (p=0.008 n=5+5)
RangeQuery/clamp 252ms ± 7% 62ms ± 4% -75.28% (p=0.008 n=5+5)
RangeQuery/clamp_min 253ms ± 5% 59ms ± 3% -76.75% (p=0.008 n=5+5)
RangeQuery/complex_func_query 455ms ± 2% 68ms ± 3% -85.06% (p=0.008 n=5+5)
RangeQuery/func_within_func_query 265ms ± 3% 89ms ± 3% -66.33% (p=0.008 n=5+5)
RangeQuery/aggr_within_func_query 273ms ± 1% 91ms ± 2% -66.43% (p=0.008 n=5+5)
RangeQuery/histogram_quantile 579ms ± 2% 204ms ± 5% -64.68% (p=0.008 n=5+5)
RangeQuery/sort 299ms ± 1% 43ms ± 2% -85.57% (p=0.008 n=5+5)
RangeQuery/sort_desc 294ms ± 2% 44ms ± 3% -84.97% (p=0.008 n=5+5)
NativeHistograms/selector 620ms ± 1% 662ms ± 5% +6.79% (p=0.008 n=5+5)
NativeHistograms/sum 1.21s ± 7% 1.01s ± 1% -16.42% (p=0.008 n=5+5)
NativeHistograms/rate 4.57s ± 3% 4.49s ± 1% ~ (p=0.310 n=5+5)
NativeHistograms/sum_rate 5.04s ± 1% 4.79s ± 1% -4.99% (p=0.008 n=5+5)
NativeHistograms/histogram_sum 930ms ± 2% 1068ms ± 6% +14.77% (p=0.008 n=5+5)
NativeHistograms/histogram_count 980ms ± 7% 1059ms ± 7% ~ (p=0.095 n=5+5)
NativeHistograms/histogram_quantile 1.20s ± 1% 1.02s ± 4% -14.80% (p=0.008 n=5+5)
name old alloc/op new alloc/op delta
RangeQuery/vector_selector 24.5MB ± 0% 38.1MB ± 0% +55.28% (p=0.008 n=5+5)
RangeQuery/sum 7.13MB ± 0% 10.20MB ± 0% +43.11% (p=0.008 n=5+5)
RangeQuery/sum_by_pod 79.9MB ± 0% 22.7MB ± 0% -71.61% (p=0.008 n=5+5)
RangeQuery/topk 7.38MB ± 0% 12.68MB ± 0% +71.84% (p=0.008 n=5+5)
RangeQuery/bottomk 7.44MB ± 0% 12.72MB ± 0% +71.02% (p=0.029 n=4+4)
RangeQuery/rate 25.6MB ± 0% 41.0MB ± 0% +60.30% (p=0.008 n=5+5)
RangeQuery/sum_rate 8.19MB ± 0% 13.09MB ± 0% +59.73% (p=0.016 n=5+4)
RangeQuery/sum_by_rate 80.7MB ± 0% 25.5MB ± 0% -68.43% (p=0.008 n=5+5)
RangeQuery/quantile_with_variable_parameter 174MB ± 0% 39MB ± 0% -77.60% (p=0.016 n=5+4)
RangeQuery/binary_operation_with_one_to_one 16.5MB ± 0% 21.6MB ± 0% +30.83% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_many_to_one 72.0MB ± 0% 55.8MB ± 0% -22.54% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_vector_and_scalar 39.1MB ± 0% 40.2MB ± 0% +2.80% (p=0.008 n=5+5)
RangeQuery/unary_negation 25.6MB ± 0% 39.4MB ± 0% +54.13% (p=0.008 n=5+5)
RangeQuery/vector_and_scalar_comparison 37.7MB ± 0% 39.9MB ± 0% +5.63% (p=0.008 n=5+5)
RangeQuery/positive_offset_vector 23.0MB ± 0% 36.6MB ± 0% +58.83% (p=0.008 n=5+5)
RangeQuery/at_modifier_ 39.8MB ± 0% 33.1MB ± 0% -16.75% (p=0.008 n=5+5)
RangeQuery/at_modifier_with_positive_offset_vector 39.6MB ± 0% 32.9MB ± 0% -16.83% (p=0.016 n=4+5)
RangeQuery/clamp 39.2MB ± 0% 38.5MB ± 0% -1.69% (p=0.016 n=5+4)
RangeQuery/clamp_min 39.1MB ± 0% 38.5MB ± 0% -1.75% (p=0.008 n=5+5)
RangeQuery/complex_func_query 53.8MB ± 0% 40.6MB ± 0% -24.54% (p=0.016 n=5+4)
RangeQuery/func_within_func_query 40.2MB ± 0% 41.1MB ± 0% +2.18% (p=0.008 n=5+5)
RangeQuery/aggr_within_func_query 40.2MB ± 0% 41.1MB ± 0% +2.19% (p=0.008 n=5+5)
RangeQuery/histogram_quantile 47.5MB ± 0% 57.9MB ± 0% +21.88% (p=0.016 n=5+4)
RangeQuery/sort 37.8MB ± 0% 38.1MB ± 0% +0.67% (p=0.008 n=5+5)
RangeQuery/sort_desc 37.8MB ± 0% 38.1MB ± 0% +0.67% (p=0.008 n=5+5)
NativeHistograms/selector 761MB ± 0% 774MB ± 0% +1.72% (p=0.016 n=4+5)
NativeHistograms/sum 943MB ± 0% 931MB ± 0% -1.21% (p=0.008 n=5+5)
NativeHistograms/rate 2.86GB ± 0% 2.87GB ± 0% +0.53% (p=0.029 n=4+4)
NativeHistograms/sum_rate 3.04GB ± 0% 3.03GB ± 0% -0.41% (p=0.016 n=4+5)
NativeHistograms/histogram_sum 786MB ± 0% 775MB ± 0% -1.42% (p=0.008 n=5+5)
NativeHistograms/histogram_count 787MB ± 0% 774MB ± 0% -1.63% (p=0.016 n=5+4)
NativeHistograms/histogram_quantile 942MB ± 0% 932MB ± 0% -1.14% (p=0.008 n=5+5)
name old allocs/op new allocs/op delta
RangeQuery/vector_selector 99.1k ± 0% 111.9k ± 0% +12.96% (p=0.016 n=5+4)
RangeQuery/sum 103k ± 0% 107k ± 0% +3.41% (p=0.016 n=5+4)
RangeQuery/sum_by_pod 598k ± 0% 202k ± 0% -66.28% (p=0.008 n=5+5)
RangeQuery/topk 108k ± 0% 114k ± 0% +5.18% (p=0.008 n=5+5)
RangeQuery/bottomk 109k ± 0% 116k ± 0% +6.20% (p=0.008 n=5+5)
RangeQuery/rate 111k ± 0% 136k ± 0% +22.33% (p=0.016 n=4+5)
RangeQuery/sum_rate 115k ± 0% 131k ± 0% +13.45% (p=0.008 n=5+5)
RangeQuery/sum_by_rate 608k ± 0% 226k ± 0% -62.89% (p=0.008 n=5+5)
RangeQuery/quantile_with_variable_parameter 1.67M ± 0% 0.58M ± 0% -65.23% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_one_to_one 75.3k ± 0% 89.2k ± 0% +18.55% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_many_to_one 637k ± 0% 173k ± 0% -72.86% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_vector_and_scalar 117k ± 0% 116k ± 0% -1.24% (p=0.008 n=5+5)
RangeQuery/unary_negation 111k ± 0% 124k ± 0% +11.82% (p=0.008 n=5+5)
RangeQuery/vector_and_scalar_comparison 105k ± 0% 113k ± 0% +7.22% (p=0.008 n=5+5)
RangeQuery/positive_offset_vector 73.1k ± 0% 86.0k ± 0% +17.69% (p=0.008 n=5+5)
RangeQuery/at_modifier_ 74.1k ± 0% 62.6k ± 0% -15.53% (p=0.000 n=5+4)
RangeQuery/at_modifier_with_positive_offset_vector 68.1k ± 0% 56.6k ± 0% -16.90% (p=0.000 n=5+4)
RangeQuery/clamp 118k ± 0% 116k ± 0% -1.69% (p=0.016 n=5+4)
RangeQuery/clamp_min 117k ± 0% 115k ± 0% -1.75% (p=0.008 n=5+5)
RangeQuery/complex_func_query 136k ± 0% 120k ± 0% -11.97% (p=0.016 n=5+4)
RangeQuery/func_within_func_query 130k ± 0% 137k ± 0% +5.40% (p=0.008 n=5+5)
RangeQuery/aggr_within_func_query 130k ± 0% 137k ± 0% +5.40% (p=0.008 n=5+5)
RangeQuery/histogram_quantile 617k ± 0% 656k ± 0% +6.29% (p=0.016 n=5+4)
RangeQuery/sort 106k ± 0% 112k ± 0% +6.10% (p=0.008 n=5+5)
RangeQuery/sort_desc 106k ± 0% 112k ± 0% +6.10% (p=0.008 n=5+5)
NativeHistograms/selector 9.63M ± 0% 9.64M ± 0% +0.14% (p=0.016 n=4+5)
NativeHistograms/sum 11.1M ± 0% 11.1M ± 0% +0.02% (p=0.008 n=5+5)
NativeHistograms/rate 34.1M ± 0% 34.1M ± 0% +0.08% (p=0.016 n=5+4)
NativeHistograms/sum_rate 35.6M ± 0% 35.6M ± 0% +0.05% (p=0.008 n=5+5)
NativeHistograms/histogram_sum 9.65M ± 0% 9.64M ± 0% -0.04% (p=0.008 n=5+5)
NativeHistograms/histogram_count 9.65M ± 0% 9.64M ± 0% -0.04% (p=0.008 n=5+5)
NativeHistograms/histogram_quantile 11.1M ± 0% 11.1M ± 0% +0.02% (p=0.008 n=5+5)Multi-core (8 core) benchmarks
name old time/op new time/op delta
RangeQuery/vector_selector-8 31.1ms ± 1% 14.7ms ± 1% -52.66% (p=0.008 n=5+5)
RangeQuery/sum-8 49.3ms ± 2% 11.0ms ± 0% -77.74% (p=0.008 n=5+5)
RangeQuery/sum_by_pod-8 138ms ± 4% 15ms ± 0% -89.16% (p=0.016 n=5+4)
RangeQuery/topk-8 47.7ms ± 4% 11.0ms ± 0% -77.03% (p=0.008 n=5+5)
RangeQuery/bottomk-8 48.3ms ± 3% 11.1ms ± 5% -76.95% (p=0.008 n=5+5)
RangeQuery/rate-8 61.4ms ± 2% 21.1ms ± 3% -65.63% (p=0.008 n=5+5)
RangeQuery/sum_rate-8 77.9ms ± 1% 19.0ms ± 4% -75.63% (p=0.008 n=5+5)
RangeQuery/sum_by_rate-8 165ms ± 1% 22ms ± 3% -86.80% (p=0.008 n=5+5)
RangeQuery/quantile_with_variable_parameter-8 234ms ± 3% 25ms ± 1% -89.17% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_one_to_one-8 121ms ± 2% 14ms ± 1% -88.53% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_many_to_one-8 405ms ± 2% 30ms ± 1% -92.49% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_vector_and_scalar-8 245ms ± 2% 20ms ± 0% -91.88% (p=0.008 n=5+5)
RangeQuery/unary_negation-8 32.3ms ± 3% 15.5ms ± 2% -52.10% (p=0.008 n=5+5)
RangeQuery/vector_and_scalar_comparison-8 206ms ± 2% 21ms ± 2% -89.78% (p=0.008 n=5+5)
RangeQuery/positive_offset_vector-8 27.6ms ± 1% 13.9ms ± 4% -49.83% (p=0.008 n=5+5)
RangeQuery/at_modifier_-8 12.6ms ± 2% 10.0ms ± 2% -20.88% (p=0.008 n=5+5)
RangeQuery/at_modifier_with_positive_offset_vector-8 12.0ms ± 3% 9.6ms ± 1% -19.73% (p=0.008 n=5+5)
RangeQuery/clamp-8 246ms ± 4% 31ms ± 4% -87.26% (p=0.008 n=5+5)
RangeQuery/clamp_min-8 251ms ± 4% 27ms ±17% -89.10% (p=0.008 n=5+5)
RangeQuery/complex_func_query-8 480ms ± 5% 38ms ± 5% -92.15% (p=0.008 n=5+5)
RangeQuery/func_within_func_query-8 279ms ± 1% 32ms ± 1% -88.59% (p=0.008 n=5+5)
RangeQuery/aggr_within_func_query-8 274ms ± 6% 32ms ± 2% -88.28% (p=0.008 n=5+5)
RangeQuery/histogram_quantile-8 545ms ± 5% 97ms ± 1% -82.15% (p=0.008 n=5+5)
RangeQuery/sort-8 301ms ± 7% 15ms ± 3% -94.92% (p=0.008 n=5+5)
RangeQuery/sort_desc-8 295ms ± 3% 15ms ± 1% -94.88% (p=0.008 n=5+5)
NativeHistograms/selector-8 417ms ± 3% 217ms ± 3% -47.95% (p=0.008 n=5+5)
NativeHistograms/sum-8 897ms ± 1% 271ms ± 2% -69.74% (p=0.008 n=5+5)
NativeHistograms/rate-8 3.76s ± 1% 1.27s ± 2% -66.30% (p=0.008 n=5+5)
NativeHistograms/sum_rate-8 4.24s ± 3% 1.27s ± 4% -70.12% (p=0.008 n=5+5)
NativeHistograms/histogram_sum-8 683ms ± 1% 429ms ± 2% -37.23% (p=0.008 n=5+5)
NativeHistograms/histogram_count-8 681ms ± 1% 423ms ± 1% -37.97% (p=0.008 n=5+5)
NativeHistograms/histogram_quantile-8 903ms ± 3% 268ms ± 2% -70.32% (p=0.008 n=5+5)
name old alloc/op new alloc/op delta
RangeQuery/vector_selector-8 24.5MB ± 0% 38.6MB ± 0% +57.55% (p=0.008 n=5+5)
RangeQuery/sum-8 7.12MB ± 0% 9.89MB ± 0% +39.06% (p=0.008 n=5+5)
RangeQuery/sum_by_pod-8 79.9MB ± 0% 23.7MB ± 0% -70.36% (p=0.008 n=5+5)
RangeQuery/topk-8 7.43MB ± 0% 10.95MB ± 0% +47.39% (p=0.008 n=5+5)
RangeQuery/bottomk-8 7.46MB ± 0% 10.94MB ± 1% +46.61% (p=0.008 n=5+5)
RangeQuery/rate-8 25.6MB ± 0% 41.4MB ± 0% +61.73% (p=0.008 n=5+5)
RangeQuery/sum_rate-8 8.18MB ± 0% 12.67MB ± 0% +54.81% (p=0.008 n=5+5)
RangeQuery/sum_by_rate-8 80.7MB ± 0% 25.3MB ± 1% -68.68% (p=0.008 n=5+5)
RangeQuery/quantile_with_variable_parameter-8 174MB ± 0% 41MB ± 0% -76.60% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_one_to_one-8 16.5MB ± 0% 22.4MB ± 0% +35.63% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_many_to_one-8 72.0MB ± 0% 56.9MB ± 0% -20.99% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_vector_and_scalar-8 39.1MB ± 0% 41.6MB ± 0% +6.39% (p=0.008 n=5+5)
RangeQuery/unary_negation-8 25.6MB ± 0% 40.0MB ± 0% +56.65% (p=0.008 n=5+5)
RangeQuery/vector_and_scalar_comparison-8 37.7MB ± 0% 41.3MB ± 0% +9.45% (p=0.008 n=5+5)
RangeQuery/positive_offset_vector-8 23.0MB ± 0% 37.1MB ± 0% +61.21% (p=0.008 n=5+5)
RangeQuery/at_modifier_-8 39.8MB ± 0% 33.1MB ± 0% -16.67% (p=0.008 n=5+5)
RangeQuery/at_modifier_with_positive_offset_vector-8 39.6MB ± 0% 33.0MB ± 0% -16.74% (p=0.008 n=5+5)
RangeQuery/clamp-8 39.1MB ± 0% 39.1MB ± 0% -0.22% (p=0.008 n=5+5)
RangeQuery/clamp_min-8 39.1MB ± 0% 39.0MB ± 0% -0.27% (p=0.008 n=5+5)
RangeQuery/complex_func_query-8 53.8MB ± 0% 41.9MB ± 0% -22.09% (p=0.008 n=5+5)
RangeQuery/func_within_func_query-8 40.2MB ± 0% 41.7MB ± 0% +3.68% (p=0.008 n=5+5)
RangeQuery/aggr_within_func_query-8 40.2MB ± 0% 41.6MB ± 0% +3.66% (p=0.008 n=5+5)
RangeQuery/histogram_quantile-8 47.5MB ± 0% 60.5MB ± 0% +27.26% (p=0.008 n=5+5)
RangeQuery/sort-8 37.8MB ± 0% 38.6MB ± 0% +2.12% (p=0.008 n=5+5)
RangeQuery/sort_desc-8 37.8MB ± 0% 38.6MB ± 0% +2.12% (p=0.016 n=4+5)
NativeHistograms/selector-8 761MB ± 0% 775MB ± 0% +1.90% (p=0.008 n=5+5)
NativeHistograms/sum-8 940MB ± 0% 933MB ± 0% -0.67% (p=0.008 n=5+5)
NativeHistograms/rate-8 2.86GB ± 0% 2.88GB ± 0% +0.61% (p=0.008 n=5+5)
NativeHistograms/sum_rate-8 3.04GB ± 0% 3.04GB ± 0% -0.27% (p=0.008 n=5+5)
NativeHistograms/histogram_sum-8 786MB ± 0% 777MB ± 0% -1.24% (p=0.008 n=5+5)
NativeHistograms/histogram_count-8 787MB ± 0% 777MB ± 0% -1.29% (p=0.008 n=5+5)
NativeHistograms/histogram_quantile-8 940MB ± 0% 934MB ± 0% -0.69% (p=0.008 n=5+5)
name old allocs/op new allocs/op delta
RangeQuery/vector_selector-8 98.8k ± 0% 113.6k ± 0% +14.99% (p=0.008 n=5+5)
RangeQuery/sum-8 103k ± 0% 108k ± 0% +5.03% (p=0.008 n=5+5)
RangeQuery/sum_by_pod-8 598k ± 0% 204k ± 0% -65.95% (p=0.008 n=5+5)
RangeQuery/topk-8 109k ± 0% 116k ± 0% +6.76% (p=0.008 n=5+5)
RangeQuery/bottomk-8 110k ± 0% 115k ± 0% +4.46% (p=0.008 n=5+5)
RangeQuery/rate-8 111k ± 0% 138k ± 0% +24.06% (p=0.008 n=5+5)
RangeQuery/sum_rate-8 115k ± 0% 132k ± 0% +14.87% (p=0.016 n=4+5)
RangeQuery/sum_by_rate-8 608k ± 0% 227k ± 0% -62.64% (p=0.008 n=5+5)
RangeQuery/quantile_with_variable_parameter-8 1.66M ± 0% 0.58M ± 0% -64.99% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_one_to_one-8 75.2k ± 0% 93.4k ± 0% +24.19% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_many_to_one-8 637k ± 0% 177k ± 0% -72.24% (p=0.008 n=5+5)
RangeQuery/binary_operation_with_vector_and_scalar-8 117k ± 0% 118k ± 0% +0.65% (p=0.016 n=4+5)
RangeQuery/unary_negation-8 111k ± 0% 126k ± 0% +13.71% (p=0.008 n=5+5)
RangeQuery/vector_and_scalar_comparison-8 105k ± 0% 115k ± 0% +9.37% (p=0.008 n=5+5)
RangeQuery/positive_offset_vector-8 72.9k ± 0% 87.8k ± 0% +20.37% (p=0.029 n=4+4)
RangeQuery/at_modifier_-8 74.0k ± 0% 62.6k ± 0% -15.40% (p=0.008 n=5+5)
RangeQuery/at_modifier_with_positive_offset_vector-8 68.0k ± 0% 56.6k ± 0% -16.77% (p=0.008 n=5+5)
RangeQuery/clamp-8 117k ± 0% 117k ± 0% +0.08% (p=0.008 n=5+5)
RangeQuery/clamp_min-8 117k ± 0% 117k ± 0% ~ (p=0.159 n=4+5)
RangeQuery/complex_func_query-8 136k ± 0% 122k ± 0% -10.31% (p=0.008 n=5+5)
RangeQuery/func_within_func_query-8 129k ± 0% 139k ± 0% +7.04% (p=0.008 n=5+5)
RangeQuery/aggr_within_func_query-8 129k ± 0% 139k ± 0% +7.03% (p=0.008 n=5+5)
RangeQuery/histogram_quantile-8 617k ± 0% 658k ± 0% +6.64% (p=0.008 n=5+5)
RangeQuery/sort-8 105k ± 0% 114k ± 0% +7.98% (p=0.008 n=5+5)
RangeQuery/sort_desc-8 105k ± 0% 114k ± 0% +7.98% (p=0.016 n=4+5)
NativeHistograms/selector-8 9.63M ± 0% 9.64M ± 0% +0.16% (p=0.008 n=5+5)
NativeHistograms/sum-8 11.1M ± 0% 11.1M ± 0% +0.04% (p=0.008 n=5+5)
NativeHistograms/rate-8 34.1M ± 0% 34.2M ± 0% +0.09% (p=0.008 n=5+5)
NativeHistograms/sum_rate-8 35.6M ± 0% 35.6M ± 0% +0.06% (p=0.008 n=5+5)
NativeHistograms/histogram_sum-8 9.65M ± 0% 9.65M ± 0% -0.01% (p=0.008 n=5+5)
NativeHistograms/histogram_count-8 9.65M ± 0% 9.65M ± 0% -0.01% (p=0.008 n=5+5)
NativeHistograms/histogram_quantile-8 11.1M ± 0% 11.1M ± 0% +0.05% (p=0.008 n=5+5)