557446a11e
Added specializations for grpc_mdelem_create.
...
In several cases, we create grpc mdelem structures using known-static
metadata inputs. Furthermore, in several cases we create a slice on
the heap (e.g. grpc_slice_from_copied_buffer) where we know we are
transferring refcount ownership. In several cases, then, we can:
1) Avoid unnecessary ref/unref operations that are no-ops (for static
slices) or superfluous (if we're transferring ownership).
2) Avoid unnecessarily comprehensive calls to grpc_slice_eq (since
they'd only be called with static or interned slice arguments,
which by construction would have equal refcounts if they were
in fact equal.
3) Avoid unnecessary checks to see if a slice is interned (when we
know that they are).
To avoid polluting the internal API, we introduce the notion of
strongly-typed grpc_slice objects. We draw a distinction between
Internal (interned and static-storage) slices and Extern (inline and
non-statically allocated). We introduce overloads to
grpc_mdelem_create() and grpc_mdelem_from_slices() for the fastpath
cases identified above based on these slice types.
From the programmer's point of view, though, nothing changes - they
need only use grpc_mdelem_create() and grpc_mdelem_from_slices() as
before, and the appropriate fastpath will be picked based on type
inference. If no special knowledge exists for the slice type (i.e. we
pass in generic grpc_slice objects), the slowpath method will still
always return correct behaviour.
This is good for:
- Roughly 1-3% reduction in CPU time for several unary/streaming
ping pong fullstack microbenchmarks.
- Reduction of about 15-20% in CPU time for some hpack parser
microbenchmarks.
- 10-12% reduction of CPU time for metadata microbenchmarks involving
interned slice comparisons.
5 years ago
d38178a27d
Few fixes to the gen_upb_api.sh script.
5 years ago
e45c5f021b
Add check_upb_output to sanity test
5 years ago
db3d8be647
Add MetadataInterface abstraction to LB policy API.
5 years ago
7e367da22a
Added more protos to upb
5 years ago
b1d73a01f1
Removed duplicate static table from hpack table. Removed an or instruction for
...
every usage of static grpc metadata. Inlined hpack table lookups for static
metadata.
This leads to faster hpack parser creation:
BM_HpackParserInitDestroy 5.32µs ± 1% 0.06µs ± 1% -98.91% (p=0.000 n=18+19)
And slightly faster parsing:
BM_HpackParserParseHeader<RepresentativeClientInitialMetadata, OnInitialHeader>
456ns ± 1% 435ns ± 1% -4.74% (p=0.000 n=18+19)
BM_HpackParserParseHeader<MoreRepresentativeClientInitialMetadata,
OnInitialHeader>
1.06µs ± 2% 1.04µs ± 2% -1.82% (p=0.000 n=19+20)
It also yields a slight (0.5 - 1.0 microsecond) reduction in CPU time for
fullstack unary pingpong:
BM_UnaryPingPong<TCP, NoOpMutator, NoOpMutator>/0/512
[polls/iter:3.0001 ] 23.9µs ± 2%
23.0µs ± 1% -3.63% (p=0.002 n=6+6)
BM_UnaryPingPong<TCP, NoOpMutator, NoOpMutator>/0/32768
[polls/iter:3.00015 ] 35.1µs ± 1%
34.2µs ± 1% -2.57% (p=0.036 n=5+3)
BM_UnaryPingPong<MinTCP, NoOpMutator, NoOpMutator>/8/0
[polls/iter:3.00011 ] 21.7µs ± 3%
21.2µs ± 2% -2.44% (p=0.017 n=6+5)
5 years ago
d527c1fbda
Pre-compute static metadata index for hpack_encoder.
...
Originally, hpack_encoder would check if a metadata was static or not by
comparing its pointer to the known static metadata global table and checking if
it was within bounds. This check was performed regardless of if the metadata was
static or not, and is somewhat costly.
Instead, we now pre-compute the static metadata index during code generation
time, and store it with static metadata objects. We read that value only if we
are dealing with a static metadata flag (which we know from the storage type of
the grpc_mdelem).
This yields slightly faster metadata encoding:
BM_HpackEncoderEncodeHeader<EmptyBatch>/0/16384
[framing_bytes/iter:9 header_bytes/iter:0 ] 34.9ns ± 2%
34.2ns ± 1% -2.04% (p=0.000 n=20+20)
BM_HpackEncoderEncodeHeader<EmptyBatch>/1/16384
[framing_bytes/iter:9 header_bytes/iter:0 ] 34.9ns ± 2%
34.2ns ± 1% -2.01% (p=0.000 n=20+19)
BM_HpackEncoderEncodeHeader<SingleStaticElem>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.6ns ± 0%
49.2ns ± 2% -2.74% (p=0.000 n=18+20)
BM_HpackEncoderEncodeHeader<SingleInternedKeyElem>/0/16384
[framing_bytes/iter:9 header_bytes/iter:6 ] 84.7ns ± 1%
83.5ns ± 1% -1.43% (p=0.000 n=20+20)
BM_HpackEncoderEncodeHeader<SingleInternedElem>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.4ns ± 0%
47.9ns ± 0% -4.83% (p=0.000 n=18+17)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<1, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 51.1ns ± 2%
48.9ns ± 1% -4.32% (p=0.000 n=20+20)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<3, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.8ns ± 2%
48.8ns ± 2% -3.88% (p=0.000 n=19+20)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<10, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.2ns ± 1%
47.9ns ± 0% -4.47% (p=0.000 n=19+16)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<31, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.2ns ± 0%
47.9ns ± 0% -4.46% (p=0.000 n=18+16)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<100, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.2ns ± 0%
47.9ns ± 0% -4.40% (p=0.000 n=19+17)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<1, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.7ns ± 2%
48.8ns ± 2% -3.81% (p=0.000 n=20+20)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<3, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.9ns ± 2%
48.8ns ± 2% -4.05% (p=0.000 n=20+20)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<10, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.1ns ± 0%
48.0ns ± 1% -4.27% (p=0.000 n=17+17)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<31, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.1ns ± 0%
48.0ns ± 1% -4.28% (p=0.000 n=18+17)
BM_HpackEncoderEncodeHeader<SingleInternedBinaryElem<100, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.1ns ± 0%
48.0ns ± 0% -4.33% (p=0.000 n=18+17)
BM_HpackEncoderEncodeHeader<SingleNonInternedElem>/0/16384
[framing_bytes/iter:9 header_bytes/iter:9 ] 91.4ns ± 1%
90.7ns ± 1% -0.79% (p=0.000 n=18+20)
BM_HpackEncoderEncodeHeader<SingleNonInternedBinaryElem<1, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:12 ] 116ns ± 1%
116ns ± 1% -0.46% (p=0.002 n=20+20)
BM_HpackEncoderEncodeHeader<SingleNonInternedBinaryElem<3, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:14 ] 122ns ± 0%
121ns ± 0% -0.69% (p=0.000 n=20+20)
BM_HpackEncoderEncodeHeader<SingleNonInternedBinaryElem<10, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:23 ] 144ns ± 1%
144ns ± 0% -0.23% (p=0.009 n=20+20)
BM_HpackEncoderEncodeHeader<SingleNonInternedBinaryElem<31, false>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:46 ] 232ns ± 0%
232ns ± 1% -0.26% (p=0.021 n=18+19)
BM_HpackEncoderEncodeHeader<SingleNonInternedBinaryElem<1, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:12 ] 92.9ns ± 1%
92.0ns ± 1% -0.97% (p=0.000 n=19+19)
BM_HpackEncoderEncodeHeader<SingleNonInternedBinaryElem<3, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:14 ] 94.0ns ± 1%
92.6ns ± 1% -1.45% (p=0.000 n=20+19)
BM_HpackEncoderEncodeHeader<SingleNonInternedBinaryElem<10, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:21 ] 93.9ns ± 2%
92.8ns ± 1% -1.17% (p=0.001 n=20+19)
BM_HpackEncoderEncodeHeader<SingleNonInternedBinaryElem<100, true>>/0/16384
[framing_bytes/iter:9 header_bytes/iter:111 ] 106ns ± 0%
105ns ± 3% -1.15% (p=0.000 n=18+20)
BM_HpackEncoderEncodeHeader<SingleNonInternedElem>/0/1
[framing_bytes/iter:81 header_bytes/iter:9 ] 355ns ± 1%
354ns ± 0% -0.35% (p=0.015 n=19+20)
BM_HpackEncoderEncodeHeader<RepresentativeClientInitialMetadata>/0/16384
[framing_bytes/iter:9 header_bytes/iter:8.00002 ] 139ns ± 1%
133ns ± 1% -4.46% (p=0.000 n=19+20)
BM_HpackEncoderEncodeHeader<MoreRepresentativeClientInitialMetadata>/0/16384
[framing_bytes/iter:9 header_bytes/iter:16 ] 236ns ± 1%
231ns ± 1% -2.24% (p=0.000 n=20+20)
BM_HpackEncoderEncodeHeader<RepresentativeServerInitialMetadata>/0/16384
[framing_bytes/iter:9 header_bytes/iter:3 ] 73.6ns ± 1%
70.5ns ± 1% -4.14% (p=0.000 n=20+20)
BM_HpackEncoderEncodeHeader<RepresentativeServerTrailingMetadata>/1/16384
[framing_bytes/iter:9 header_bytes/iter:1 ] 50.5ns ± 0%
49.2ns ± 2% -2.60% (p=0.000 n=16+20)
5 years ago
27030f58aa
Renaming data-plane-api to envoy-api.
5 years ago
0f83755c6e
chttp2 hpack encoder: fast-pathed static md/slice ops
6 years ago
f65208af02
Added slice equality when static fastpath.
6 years ago
7d946633ea
grpc_slice_refcount devirtualization
6 years ago
df3d5bc8f7
Add upb generated code and build rule for lrs protocol.
6 years ago
e959373742
Generate upb code for cds protos and BUILD rule changes
6 years ago
fb15daf8b9
Add generated upb code for endpoints information.
...
Also,
- Update upb submodule for new code generator and update generated
files.
6 years ago
f569cc1b36
Add BUILD rul and generated upb code for protos required to do DiscoveryRequest to xDS server
...
Also,
- cleanup check scripts to look for file extension to exempt upb
generated code.
6 years ago
319fcdf26e
Add a script for generating C code and build rule for protobuf protos.
...
All these changes need to go together to make sense
- changes to use new version of upb in bazel
- allowing includes in build target option
- script for generating c code (upb) for protos
- generated code for example protos
- adding changes for non-bazel builds
- change sanity tests to ignore the generated files.
6 years ago
ac6795a57e
Revert "Changes add a script for generating C code and build rule for protobuf"
...
This reverts commit 62027b7e14
6 years ago
62027b7e14
Changes add a script for generating C code and build rule for protobuf
...
protos
All these changes need to go together to make sense
- changes to use new version of upb in bazel
- allowing includes in build target option
- script for generating c code for protos
- generated code for example build
- adding changes for non-bazel builds
- change sanity tests to ignore the generated files.
6 years ago
fafde8ff21
Adding ads (data-plane-api) method name to wellknown names list.
...
This is the api for getting assignments from xDS server
6 years ago
f85fd026e3
Client-side health checking support.
6 years ago
bf4432b67e
Add comment to specify hpack elements
6 years ago
4772a743e8
Delete hpack lookup table
6 years ago
091f8dd51e
Remove unnecessary callout_is_default checks
6 years ago
2a51d406e8
Minor cleanup to gen_static_metadadta.py
6 years ago
327c514fe7
Some reorganizing to respond to code review comments
6 years ago
5b27941372
Even cleaner implementation that does not modify the grpc_mdelem struct.
6 years ago
0b4e263fdf
Small formatting and style changes
6 years ago
4a9cd9c38b
Fix codegen script to include port_platform
6 years ago
b731178a7c
Fix codegen script to include port_platform
6 years ago
fbe594beeb
Much cleaner approach that should address code review comments as well
6 years ago
009d828341
WIP. Modifying grpc_mdelem to store the static hpack table idnex
6 years ago
fcf533d916
try fix nanopb generation
6 years ago
f3955ba711
Update to use the canonical version of LB proto
7 years ago
bc9573d7ac
back to using quotes for nanopb includes
7 years ago
99dc56758e
more tweaks go output of nanopb gen
7 years ago
00e045dbb2
Fix the inclusion "third_party/nanopb/..." to simply "..."
7 years ago
d8f9fe844f
get rid of premature return in massage_qps_stats
7 years ago
31f4eee720
Clang format and yapf
7 years ago
519d2b1fae
Cleanup stats removal for opt build. Stats will only be collected on debug builds or if GRPC_COLLECT_STATS is defined.
7 years ago
ae3866bf6b
Try removing stats
7 years ago
4454af55e1
Update stats generator
7 years ago
04fb58efbd
Add ALTS code to grpc/core
7 years ago
ae277dd000
Revert "Add ALTS C stack to gRPC core"
7 years ago
eaf6486677
Add ALTS code to grpc/core
7 years ago
718c8341ca
Retry support.
7 years ago
3db2caaf2e
Replace message/deflate,gzip with deflate,gzip
7 years ago
87cd994b04
Upgrade yapf to 0.20.0
...
Upgrade yapf version to 0.20.0 and reformat Python files.
7 years ago
cf1d28b473
Fix windows compile
7 years ago
7a2be20a8a
yapf tools/distrib
7 years ago
5898847ddf
yapf tools/codegen
7 years ago
Arjun Roy
Nicolas "Pixel" Noble
Esun Kim
Mark D. Roth
Vishal Powar
Alexander Polcyn
Vishal Powar
Hope Casey-Allen
Jan Tattermusch
David Garcia Quintas
Yash Tibrewal
ncteisen
Yihua Zhang
Sree Kuchibhotla
Muxi Yan
Mehrdad Afshari