- 09 Mar, 2018 4 commits
-
-
Hana Kim authored
Define a set of color names available in trace viewer https://user-images.githubusercontent.com/4999471/37063995-5d0bad48-2169-11e8-92be-9cb363e21c38.png Change-Id: I312fcbc5430d7512b4c39ddc79a769259bad8c22 Reviewed-on: https://go-review.googlesource.com/99055Reviewed-by:
Heschi Kreinick <heschi@google.com>
-
Hana Kim authored
Also grey out instants that represent events occurred outside the task's span. Furthermore, if the unrelated instants represent user annotation events but not for the task of the interest, skip rendering completely. This helps users to focus on the task-related events better. UI screen shot: https://gist.github.com/hyangah/1df5d2c8f429fd933c481e9636b89b55#file-golang-org_cl_99035 Change-Id: I2b5aef41584c827f8c1e915d0d8e5c95fe2b4b65 Reviewed-on: https://go-review.googlesource.com/99035 Run-TryBot: Hyang-Ah Hana Kim <hyangah@gmail.com> Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Robert Griesemer authored
Strangely enough, the existing implementation used adjusted (by line directives) source positions to determine layout and thus required position corrections when printing a line directive. Instead, just use the unadjusted, absolute source positions and then printing a line directive doesn't require any adjustments, only some care to make sure it remains in column 1 as before. The new code doesn't need to parse line directives anymore and simply ensures that comments with the //line prefix and starting in column 1 remain in that position. That is a slight change from the old behavior (which ignored incorrect line directives, e.g. because they had an invalid line number) but unlikely to show up in real code. This is prep work for handling of line directives that also specify columns (which now won't require much special handling anymore). For #24143. Change-Id: I07eb2e1b35b37337e632e3dbf5b70c783c615f8a Reviewed-on: https://go-review.googlesource.com/99621Reviewed-by:
Matthew Dempsky <mdempsky@google.com>
-
Joe Tsai authored
'"' has special semantic meaning that conflicts with using it as Comma. Change-Id: Ife25ba43ca25dba2ea184c1bb7579a230d376059 Reviewed-on: https://go-review.googlesource.com/99696 Run-TryBot: Joe Tsai <thebrokentoaster@gmail.com> Reviewed-by:
Ian Lance Taylor <iant@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
-
- 08 Mar, 2018 27 commits
-
-
Austin Clements authored
It's a bit mysterious that _defer.sp is a uintptr that gets stack-adjusted explicitly while _panic.argp is an unsafe.Pointer that doesn't, but turns out to be critically important when a deferred function grows the stack before doing a recover. Add a comment explaining that this works because _panic values live on the stack. Enforce this by marking _panic go:notinheap. Change-Id: I9ca49e84ee1f86d881552c55dccd0662b530836b Reviewed-on: https://go-review.googlesource.com/99735 Run-TryBot: Austin Clements <austin@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Austin Clements authored
On all non-x86 arches, runtime.abort simply reads from nil. Unfortunately, if this happens on a user stack, the signal handler will dutifully turn this into a panicmem, which lets user defers run and which user code can even recover from. To fix this, add an explicit check to the signal handler that turns faults in abort into hard crashes directly in the signal handler. This has the added benefit of giving a register dump at the abort point. Change-Id: If26a7f13790745ee3867db7f53b72d8281176d70 Reviewed-on: https://go-review.googlesource.com/93661 Run-TryBot: Austin Clements <austin@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
Keith Randall <khr@golang.org>
-
Austin Clements authored
Everything except for amd64, amd64p32, and 386 currently defines and uses an abort function. This CL makes these match. The next CL will recognize the abort function to make this more useful. Change-Id: I7c155871ea48919a9220417df0630005b444f488 Reviewed-on: https://go-review.googlesource.com/93660 Run-TryBot: Austin Clements <austin@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Austin Clements authored
Currently, throw may grow the stack, which means whenever we call it from a context where it's not safe to grow the stack, we first have to switch to the system stack. This is pretty easy to get wrong. Fix this by making throw switch to the system stack so it doesn't grow the stack and is hence safe to call without a system stack switch at the call site. The only thing this complicates is badsystemstack itself, which would now go into an infinite loop before printing anything (previously it would also go into an infinite loop, but would at least print the error first). Fix this by making badsystemstack do a direct write and then crash hard. Change-Id: Ic5b4a610df265e47962dcfa341cabac03c31c049 Reviewed-on: https://go-review.googlesource.com/93659 Run-TryBot: Austin Clements <austin@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Austin Clements authored
Currently parts of unrecoverable panic handling (notably, printing panic messages) can happen on the user stack. This may grow the stack, which is generally fine, but if we're handling a runtime panic, it's better to do as little as possible in case the runtime is in an inconsistent state. Hence, this commit rearranges the handling of unrecoverable panics so that it's done entirely on the system stack. This is mostly a matter of shuffling code a bit so everything can move into a systemstack block. The one slight subtlety is in the "panic during panic" case, where we now depend on startpanic_m's caller to print the stack rather than startpanic_m itself. To make this work, startpanic_m now returns a boolean indicating that the caller should avoid trying to print any panic messages and get right to the stack trace. Since the caller is already in a position to do this, this actually simplifies things a little. Change-Id: Id72febe8c0a9fb31d9369b600a1816d65a49bfed Reviewed-on: https://go-review.googlesource.com/93658 Run-TryBot: Austin Clements <austin@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Austin Clements authored
The previous CL introduced isConstDelta. Use it to simplify the OpSlicemask optimization in the prove pass. This passes toolstash -cmp. Change-Id: If2aa762db4cdc0cd1c581a536340530a9831081b Reviewed-on: https://go-review.googlesource.com/87481Reviewed-by:
-
Austin Clements authored
This adds four new deductions to the prove pass, all related to adding or subtracting one from a value. This is the first hint of actual arithmetic relations in the prove pass. The most effective of these is x-1 >= w && x > min ⇒ x > w This helps eliminate bounds checks in code like if x > 0 { // do something with s[x-1] } Altogether, these deductions prove an additional 260 branches in std and cmd. Furthermore, they will let us eliminate some tricky compiler-inserted panics in the runtime that are interfering with static analysis. Fixes #23354. Change-Id: I7088223e0e0cd6ff062a75c127eb4bb60e6dce02 Reviewed-on: https://go-review.googlesource.com/87480Reviewed-by:Alexandru Moșoi <alexandru@mosoi.ro>
-
Austin Clements authored
This adds a few simple deductions to the prove pass' fact table to derive unsigned concrete limits from signed concrete limits where possible. This tweak lets the pass prove 70 additional branch conditions in std and cmd. This is based on a comment from the recently-deleted factsTable.get: "// TODO: also use signed data if lim.min >= 0". Change-Id: Ib4340249e7733070f004a0aa31254adf5df8a392 Reviewed-on: https://go-review.googlesource.com/87479Reviewed-by:
-
Austin Clements authored
Currently the prove pass uses implication queries. For each block, it collects the set of branch conditions leading to that block, and queries this fact table for whether any of these facts imply the block's own branch condition (or its inverse). This works remarkably well considering it doesn't do any deduction on these facts, but it has various downsides: 1. It requires an implementation both of adding facts to the table and determining implications. These are very nearly duals of each other, but require separate implementations. Likewise, the process of asserting facts of dominating branch conditions is very nearly the dual of the process of querying implied branch conditions. 2. It leads to less effective use of derived facts. For example, the prove pass currently derives facts about the relations between len and cap, but can't make use of these unless a branch condition is in the exact form of a derived fact. If one of these derived facts contradicts another fact, it won't notice or make use of this. This CL changes the approach of the prove pass to instead use *contradiction* instead of implication. Rather than ever querying a branch condition, it simply adds branch conditions to the fact table. If this leads to a contradiction (specifically, it makes the fact set unsatisfiable), that branch is impossible and can be cut. As a result, 1. We can eliminate the code for determining implications (factsTable.get disappears entirely). Also, there is now a single implementation of visiting and asserting branch conditions, since we don't have to flip them around to treat them as facts in one place and queries in another. 2. Derived facts can be used effectively. It doesn't matter *why* the fact table is unsatisfiable; a contradiction in any of the facts is enough. 3. As an added benefit, it's now quite easy to avoid traversing beyond provably-unreachable blocks. In contrast, the current implementation always visits all blocks. The prove pass already has nearly all of the mechanism necessary to compute unsatisfiability, which means this both simplifies the code and makes it more powerful. The only complication is that the current implication procedure has a hack for dealing with the 0 <= Args[0] condition of OpIsInBounds and OpIsSliceInBounds. We replace this with asserting the appropriate fact when we process one of these conditions. This seems much cleaner anyway, and works because we can now take advantage of derived facts. This has no measurable effect on compiler performance. Effectiveness: There is exactly one condition in all of std and cmd that this fails to prove that the old implementation could: (int64(^uint(0)>>1) < x) in encoding/gob. This can never be true because x is an int, and it's basically coincidence that the old code gets this. (For example, it fails to prove the similar (x < ^int64(^uint(0)>>1)) condition that immediately precedes it, and even though the conditions are logically unrelated, it wouldn't get the second one if it hadn't first processed the first!) It does, however, prove a few dozen additional branches. These come from facts that are added to the fact table about the relations between len and cap. These were almost never queried directly before, but could lead to contradictions, which the unsat-based approach is able to use. There are exactly two branches in std and cmd that this implementation proves in the *other* direction. This sounds scary, but is okay because both occur in already-unreachable blocks, so it doesn't matter what we chose. Because the fact table logic is sound but incomplete, it fails to prove that the block isn't reachable, even though it is able to prove that both outgoing branches are impossible. We could turn these blocks into BlockExit blocks, but it doesn't seem worth the trouble of the extra proof effort for something that happens twice in all of std and cmd. Tests: This CL updates test/prove.go to change the expected messages because it can no longer give a "reason" why it proved or disproved a condition. It also adds a new test of a branch it couldn't prove before. It mostly guts test/sliceopt.go, removing everything related to slice bounds optimizations and moving a few relevant tests to test/prove.go. Much of this test is actually unreachable. The new prove pass figures this out and doesn't try to prove anything about the unreachable parts. The output on the unreachable parts is already suspect because anything can be proved at that point, so it's really just a regression test for an algorithm the compiler no longer uses. This is a step toward fixing #23354. That issue is quite easy to fix once we can use derived facts effectively. Change-Id: Ia48a1b9ee081310579fe474e4a61857424ff8ce8 Reviewed-on: https://go-review.googlesource.com/87478Reviewed-by:
-
Austin Clements authored
This replaces the open-coded intersection of limits in the prove pass with a general limit intersection operation. This should get identical results except in one case where it's more precise: when handling an equality relation, if the value is *outside* the existing range, this will reduce the range to empty rather than resetting it. This will be important to a follow-up CL where we can take advantage of empty ranges. For #23354. Change-Id: I3d3d75924f61b1da1cb604b3a9d189b26fb3a14e Reviewed-on: https://go-review.googlesource.com/87477 Run-TryBot: Austin Clements <austin@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Austin Clements authored
These aid in debugging. Change-Id: Ieb38c996765f780f6103f8c3292639d408c25123 Reviewed-on: https://go-review.googlesource.com/87476 Run-TryBot: Austin Clements <austin@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
Brad Fitzpatrick <bradfitz@golang.org> Reviewed-by:
-
Austin Clements authored
Change-Id: Ie0934f1528d58d4971cdef726d3e2d23cf3935d3 Reviewed-on: https://go-review.googlesource.com/87475 Run-TryBot: Austin Clements <austin@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
David Chase <drchase@google.com> Reviewed-by:
-
Matthew Dempsky authored
This reverts commit dcac984b. Reason for revert: broke LR architectures (arm64, ppc64, s390x) Change-Id: I531d311c9053e81503c8c78d6cf044b318fc828b Reviewed-on: https://go-review.googlesource.com/99695 Run-TryBot: Matthew Dempsky <mdempsky@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
Austin Clements <austin@google.com>
-
Alberto Donizetti authored
The Newton sqrtInverse procedure we use to compute Float.Sqrt should not allocate a number of times proportional to the number of Newton iterations we need to reach the desired precision. At the beginning the function the target precision is known, so even if we do want to perform the early steps at low precisions (to save time), it's still possible to pre-allocate larger backing arrays, both for the temp variables in the loop and the variable that'll hold the final result. There's one complication. At the following line: u.Sub(three, u) the Sub method will allocate, because the receiver aliases one of the arguments, and the large backing array we initially allocated for u will be replaced by a smaller one allocated by Sub. We can work around this by introducing a second temp variable u2 that we use to hold the Sub call result. Overall, the sqrtInverse procedure still allocates a number of times proportional to the number of Newton steps, because unfortunately a few of the Mul calls in the Newton function allocate; but at least we allocate less in the function itself. FloatSqrt/256-4 1.97µs ± 1% 1.84µs ± 1% -6.61% (p=0.000 n=8+8) FloatSqrt/1000-4 4.80µs ± 3% 4.28µs ± 1% -10.78% (p=0.000 n=8+8) FloatSqrt/10000-4 40.0µs ± 1% 38.3µs ± 1% -4.15% (p=0.000 n=8+8) FloatSqrt/100000-4 955µs ± 1% 932µs ± 0% -2.49% (p=0.000 n=8+7) FloatSqrt/1000000-4 79.8ms ± 1% 79.4ms ± 1% ~ (p=0.105 n=8+8) name old alloc/op new alloc/op delta FloatSqrt/256-4 816B ± 0% 512B ± 0% -37.25% (p=0.000 n=8+8) FloatSqrt/1000-4 2.50kB ± 0% 1.47kB ± 0% -41.03% (p=0.000 n=8+8) FloatSqrt/10000-4 23.5kB ± 0% 18.2kB ± 0% -22.62% (p=0.000 n=8+8) FloatSqrt/100000-4 251kB ± 0% 173kB ± 0% -31.26% (p=0.000 n=8+8) FloatSqrt/1000000-4 4.61MB ± 0% 2.86MB ± 0% -37.90% (p=0.000 n=8+8) name old allocs/op new allocs/op delta FloatSqrt/256-4 12.0 ± 0% 8.0 ± 0% -33.33% (p=0.000 n=8+8) FloatSqrt/1000-4 19.0 ± 0% 9.0 ± 0% -52.63% (p=0.000 n=8+8) FloatSqrt/10000-4 35.0 ± 0% 14.0 ± 0% -60.00% (p=0.000 n=8+8) FloatSqrt/100000-4 55.0 ± 0% 23.0 ± 0% -58.18% (p=0.000 n=8+8) FloatSqrt/1000000-4 122 ± 0% 75 ± 0% -38.52% (p=0.000 n=8+8) Change-Id: I950dbf61a40267a6cca82ae72524c3024bcb149c Reviewed-on: https://go-review.googlesource.com/87659Reviewed-by:
Robert Griesemer <gri@golang.org>
-
isharipo authored
Set up to _S (number of bytes in Uint) bytes at time by using BigEndian.Uint32 and BigEndian.Uint64. The performance improves for slices bigger than _S bytes. This is the case for 128/256bit arith that initializes it's objects from bytes. name old time/op new time/op delta NatSetBytes/8-4 29.8ns ± 1% 11.4ns ± 0% -61.63% (p=0.000 n=9+8) NatSetBytes/24-4 109ns ± 1% 56ns ± 0% -48.75% (p=0.000 n=9+8) NatSetBytes/128-4 420ns ± 2% 110ns ± 1% -73.83% (p=0.000 n=10+10) NatSetBytes/7-4 26.2ns ± 1% 21.3ns ± 2% -18.63% (p=0.000 n=8+9) NatSetBytes/23-4 106ns ± 1% 67ns ± 1% -36.93% (p=0.000 n=9+10) NatSetBytes/127-4 410ns ± 2% 121ns ± 0% -70.46% (p=0.000 n=9+8) Found this optimization opportunity by looking at ethereum_corevm community benchmark cpuprofile. name old time/op new time/op delta OpDiv256-4 715ns ± 1% 596ns ± 1% -16.57% (p=0.008 n=5+5) OpDiv128-4 373ns ± 1% 314ns ± 1% -15.83% (p=0.008 n=5+5) OpDiv64-4 301ns ± 0% 285ns ± 1% -5.12% (p=0.008 n=5+5) Change-Id: I8e5a680ae6284c8233d8d7431d51253a8a740b57 Reviewed-on: https://go-review.googlesource.com/98775 Run-TryBot: Iskander Sharipov <iskander.sharipov@intel.com> Reviewed-by:
-
Matthew Dempsky authored
Instead of creating a new &nodfp expression for every recover() call, or a new nodpc variable for every function instrumented by the race detector, this CL introduces two new uintptr-typed pseudo-variables callerSP and callerPC. These pseudo-variables act just like calls to the runtime's getcallersp() and getcallerpc() functions. For consistency, change runtime.gorecover's builtin stub's parameter type from "*int32" to "uintptr". Passes toolstash-check, but toolstash-check -race fails because of register allocator changes. Change-Id: I985d644653de2dac8b7b03a28829ad04dfd4f358 Reviewed-on: https://go-review.googlesource.com/99416 Run-TryBot: Matthew Dempsky <mdempsky@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
Daniel Martí <mvdan@mvdan.cc> Reviewed-by:
-
Matthew Dempsky authored
All calls to walkstmt/walkexpr/etc should be rooted from funccompile, whereas transformclosure and fninit are called by main. Passes toolstash-check. Change-Id: Ic880e2d2d83af09618ce4daa8e7716f6b389e53e Reviewed-on: https://go-review.googlesource.com/99418 Run-TryBot: Matthew Dempsky <mdempsky@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Matthew Dempsky authored
We're holding onto the function's complete AST anyway, so might as well grab the exit code from there. Passes toolstash-check. Change-Id: I851b5dfdb53f991e9cd9488d25d0d2abc2a8379f Reviewed-on: https://go-review.googlesource.com/99417 Run-TryBot: Matthew Dempsky <mdempsky@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Matthew Dempsky authored
Simplifies the code somewhat and allows removing Param.Field. Passes toolstash-check. Change-Id: Id854416aea8afd27ce4830ff0f5ff940f7353792 Reviewed-on: https://go-review.googlesource.com/99336 Run-TryBot: Matthew Dempsky <mdempsky@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
Cherry Zhang <cherryyz@google.com>
-
Kunpei Sakai authored
Fixes #24297 Change-Id: I759e88655632fda97dced240b3f13392b2785d0a Reviewed-on: https://go-review.googlesource.com/99575Reviewed-by:
Andrew Bonventre <andybons@golang.org> Reviewed-by:
-
Michael Kasch authored
IANA Zoneinfo does not provide names for all timezones. Some are denoted by a sign and an offset only. E.g: Europe/Turkey is currently +03 or America/La_Paz which is -04 (https://data.iana.org/time-zones/releases/tzdata2018c.tar.gz) Fixes #24071 Change-Id: I9c230a719945e1263c5b52bab82084d22861be3e Reviewed-on: https://go-review.googlesource.com/98157Reviewed-by:
-
Ian Lance Taylor authored
Try to fix the build on ppc64-linux and ppc64le-linux, avoiding: --- FAIL: TestInlinedRoutineRecords (2.12s) dwarf_test.go:97: build: # command-line-arguments runtime.systemstack: nosplit stack overflow 752 assumed on entry to runtime.sigtrampgo (nosplit) 480 after runtime.sigtrampgo (nosplit) uses 272 400 after runtime.sigfwdgo (nosplit) uses 80 264 after runtime.setsig (nosplit) uses 136 208 after runtime.sigaction (nosplit) uses 56 136 after runtime.sysSigaction (nosplit) uses 72 88 after runtime.throw (nosplit) uses 48 16 after runtime.dopanic (nosplit) uses 72 -16 after runtime.systemstack (nosplit) uses 32 dwarf_test.go:98: build error: exit status 2 --- FAIL: TestAbstractOriginSanity (10.22s) dwarf_test.go:97: build: # command-line-arguments runtime.systemstack: nosplit stack overflow 752 assumed on entry to runtime.sigtrampgo (nosplit) 480 after runtime.sigtrampgo (nosplit) uses 272 400 after runtime.sigfwdgo (nosplit) uses 80 264 after runtime.setsig (nosplit) uses 136 208 after runtime.sigaction (nosplit) uses 56 136 after runtime.sysSigaction (nosplit) uses 72 88 after runtime.throw (nosplit) uses 48 16 after runtime.dopanic (nosplit) uses 72 -16 after runtime.systemstack (nosplit) uses 32 dwarf_test.go:98: build error: exit status 2 FAIL FAIL cmd/link/internal/ld 13.404s Change-Id: I4840604adb0e9f68a8d8e24f2f2a1a17d1634a58 Reviewed-on: https://go-review.googlesource.com/99415Reviewed-by:
-
Alberto Donizetti authored
And delete them from the asm_test.go file. Change-Id: I124c8c352299646ec7db0968cdb0fe59a3b5d83d Reviewed-on: https://go-review.googlesource.com/99475 Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
Giovanni Bajo <rasky@develer.com>
-
erifan01 authored
The biggest hot spot of the existing implementation is "load" operations, which lead to poor performance. By unrolling the cycle 4 times and 2 times, and using "LDP", "STP" instructions, this CL can reduce the "load" cost and improve performance. Benchmarks: name old time/op new time/op delta AddVV/1-8 21.5ns ± 0% 21.5ns ± 0% ~ (all equal) AddVV/2-8 13.5ns ± 0% 13.5ns ± 0% ~ (all equal) AddVV/3-8 15.5ns ± 0% 15.5ns ± 0% ~ (all equal) AddVV/4-8 17.5ns ± 0% 17.5ns ± 0% ~ (all equal) AddVV/5-8 19.5ns ± 0% 19.5ns ± 0% ~ (all equal) AddVV/10-8 29.5ns ± 0% 29.5ns ± 0% ~ (all equal) AddVV/100-8 217ns ± 0% 217ns ± 0% ~ (all equal) AddVV/1000-8 2.02µs ± 0% 2.02µs ± 0% ~ (all equal) AddVV/10000-8 20.3µs ± 0% 20.3µs ± 0% ~ (p=0.603 n=5+5) AddVV/100000-8 223µs ± 7% 228µs ± 8% ~ (p=0.548 n=5+5) AddVW/1-8 9.32ns ± 0% 9.26ns ± 0% -0.64% (p=0.008 n=5+5) AddVW/2-8 19.8ns ± 3% 10.5ns ± 0% -46.92% (p=0.008 n=5+5) AddVW/3-8 11.5ns ± 0% 11.0ns ± 0% -4.35% (p=0.008 n=5+5) AddVW/4-8 13.0ns ± 0% 12.0ns ± 0% -7.69% (p=0.008 n=5+5) AddVW/5-8 14.5ns ± 0% 12.5ns ± 0% -13.79% (p=0.008 n=5+5) AddVW/10-8 22.0ns ± 0% 15.5ns ± 0% -29.55% (p=0.008 n=5+5) AddVW/100-8 167ns ± 0% 81ns ± 0% -51.44% (p=0.008 n=5+5) AddVW/1000-8 1.52µs ± 0% 0.64µs ± 0% -57.58% (p=0.008 n=5+5) AddVW/10000-8 15.1µs ± 0% 7.2µs ± 0% -52.55% (p=0.008 n=5+5) AddVW/100000-8 150µs ± 0% 71µs ± 0% -52.95% (p=0.008 n=5+5) SubVW/1-8 9.32ns ± 0% 9.26ns ± 0% -0.64% (p=0.008 n=5+5) SubVW/2-8 19.7ns ± 2% 10.5ns ± 0% -46.70% (p=0.008 n=5+5) SubVW/3-8 11.5ns ± 0% 11.0ns ± 0% -4.35% (p=0.008 n=5+5) SubVW/4-8 13.0ns ± 0% 12.0ns ± 0% -7.69% (p=0.008 n=5+5) SubVW/5-8 14.5ns ± 0% 12.5ns ± 0% -13.79% (p=0.008 n=5+5) SubVW/10-8 22.0ns ± 0% 15.5ns ± 0% -29.55% (p=0.008 n=5+5) SubVW/100-8 167ns ± 0% 81ns ± 0% -51.44% (p=0.008 n=5+5) SubVW/1000-8 1.52µs ± 0% 0.64µs ± 0% -57.58% (p=0.008 n=5+5) SubVW/10000-8 15.1µs ± 0% 7.2µs ± 0% -52.49% (p=0.008 n=5+5) SubVW/100000-8 150µs ± 0% 71µs ± 0% -52.91% (p=0.008 n=5+5) AddMulVVW/1-8 32.4ns ± 1% 32.6ns ± 1% ~ (p=0.119 n=5+5) AddMulVVW/2-8 57.0ns ± 0% 57.0ns ± 0% ~ (p=0.643 n=5+5) AddMulVVW/3-8 90.8ns ± 0% 90.7ns ± 0% ~ (p=0.524 n=5+5) AddMulVVW/4-8 118ns ± 0% 118ns ± 1% ~ (p=1.000 n=4+5) AddMulVVW/5-8 144ns ± 1% 144ns ± 0% ~ (p=0.794 n=5+4) AddMulVVW/10-8 294ns ± 1% 296ns ± 0% +0.48% (p=0.040 n=5+5) AddMulVVW/100-8 2.73µs ± 0% 2.73µs ± 0% ~ (p=0.278 n=5+5) AddMulVVW/1000-8 26.0µs ± 0% 26.5µs ± 0% +2.14% (p=0.008 n=5+5) AddMulVVW/10000-8 297µs ± 0% 297µs ± 0% +0.24% (p=0.008 n=5+5) AddMulVVW/100000-8 3.15ms ± 1% 3.13ms ± 0% ~ (p=0.690 n=5+5) DecimalConversion-8 311µs ± 2% 309µs ± 2% ~ (p=0.310 n=5+5) FloatString/100-8 2.55µs ± 2% 2.54µs ± 2% ~ (p=1.000 n=5+5) FloatString/1000-8 58.1µs ± 0% 58.1µs ± 0% ~ (p=0.151 n=5+5) FloatString/10000-8 4.59ms ± 0% 4.59ms ± 0% ~ (p=0.151 n=5+5) FloatString/100000-8 446ms ± 0% 446ms ± 0% +0.01% (p=0.016 n=5+5) FloatAdd/10-8 183ns ± 0% 183ns ± 0% ~ (p=0.333 n=4+5) FloatAdd/100-8 187ns ± 1% 192ns ± 2% ~ (p=0.056 n=5+5) FloatAdd/1000-8 369ns ± 0% 371ns ± 0% +0.54% (p=0.016 n=4+5) FloatAdd/10000-8 1.88µs ± 0% 1.88µs ± 0% -0.14% (p=0.000 n=4+5) FloatAdd/100000-8 17.2µs ± 0% 17.1µs ± 0% -0.37% (p=0.008 n=5+5) FloatSub/10-8 147ns ± 0% 147ns ± 0% ~ (all equal) FloatSub/100-8 145ns ± 0% 146ns ± 0% ~ (p=0.238 n=5+4) FloatSub/1000-8 241ns ± 0% 241ns ± 0% ~ (p=0.333 n=5+4) FloatSub/10000-8 1.06µs ± 0% 1.06µs ± 0% ~ (p=0.444 n=5+5) FloatSub/100000-8 9.50µs ± 0% 9.48µs ± 0% -0.14% (p=0.008 n=5+5) ParseFloatSmallExp-8 28.4µs ± 2% 28.5µs ± 1% ~ (p=0.690 n=5+5) ParseFloatLargeExp-8 125µs ± 1% 124µs ± 1% ~ (p=0.095 n=5+5) GCD10x10/WithoutXY-8 277ns ± 2% 278ns ± 3% ~ (p=0.937 n=5+5) GCD10x10/WithXY-8 2.08µs ± 3% 2.15µs ± 3% ~ (p=0.056 n=5+5) GCD10x100/WithoutXY-8 592ns ± 3% 613ns ± 4% ~ (p=0.056 n=5+5) GCD10x100/WithXY-8 3.40µs ± 2% 3.42µs ± 4% ~ (p=0.841 n=5+5) GCD10x1000/WithoutXY-8 1.37µs ± 2% 1.35µs ± 3% ~ (p=0.460 n=5+5) GCD10x1000/WithXY-8 7.34µs ± 2% 7.33µs ± 4% ~ (p=0.841 n=5+5) GCD10x10000/WithoutXY-8 8.52µs ± 0% 8.51µs ± 1% ~ (p=0.421 n=5+5) GCD10x10000/WithXY-8 27.5µs ± 2% 27.2µs ± 1% ~ (p=0.151 n=5+5) GCD10x100000/WithoutXY-8 78.3µs ± 1% 78.5µs ± 1% ~ (p=0.690 n=5+5) GCD10x100000/WithXY-8 231µs ± 0% 229µs ± 1% -1.11% (p=0.016 n=5+5) GCD100x100/WithoutXY-8 1.86µs ± 2% 1.86µs ± 2% ~ (p=0.881 n=5+5) GCD100x100/WithXY-8 27.1µs ± 2% 27.2µs ± 1% ~ (p=0.421 n=5+5) GCD100x1000/WithoutXY-8 4.44µs ± 2% 4.41µs ± 1% ~ (p=0.310 n=5+5) GCD100x1000/WithXY-8 36.3µs ± 1% 36.2µs ± 1% ~ (p=0.310 n=5+5) GCD100x10000/WithoutXY-8 22.6µs ± 2% 22.5µs ± 1% ~ (p=0.690 n=5+5) GCD100x10000/WithXY-8 145µs ± 1% 145µs ± 1% ~ (p=1.000 n=5+5) GCD100x100000/WithoutXY-8 195µs ± 0% 196µs ± 1% ~ (p=0.548 n=5+5) GCD100x100000/WithXY-8 1.10ms ± 0% 1.10ms ± 0% -0.30% (p=0.016 n=5+5) GCD1000x1000/WithoutXY-8 25.0µs ± 1% 25.2µs ± 2% ~ (p=0.222 n=5+5) GCD1000x1000/WithXY-8 520µs ± 0% 520µs ± 1% ~ (p=0.151 n=5+5) GCD1000x10000/WithoutXY-8 57.0µs ± 1% 56.9µs ± 1% ~ (p=0.690 n=5+5) GCD1000x10000/WithXY-8 1.21ms ± 0% 1.21ms ± 1% ~ (p=0.881 n=5+5) GCD1000x100000/WithoutXY-8 358µs ± 0% 359µs ± 1% ~ (p=0.548 n=5+5) GCD1000x100000/WithXY-8 8.73ms ± 0% 8.73ms ± 0% ~ (p=0.548 n=5+5) GCD10000x10000/WithoutXY-8 686µs ± 0% 687µs ± 0% ~ (p=0.548 n=5+5) GCD10000x10000/WithXY-8 15.9ms ± 0% 15.9ms ± 0% ~ (p=0.841 n=5+5) GCD10000x100000/WithoutXY-8 2.08ms ± 0% 2.08ms ± 0% ~ (p=1.000 n=5+5) GCD10000x100000/WithXY-8 86.7ms ± 0% 86.7ms ± 0% ~ (p=1.000 n=5+5) GCD100000x100000/WithoutXY-8 51.1ms ± 0% 51.0ms ± 0% ~ (p=0.151 n=5+5) GCD100000x100000/WithXY-8 1.23s ± 0% 1.23s ± 0% ~ (p=0.841 n=5+5) Hilbert-8 2.41ms ± 1% 2.42ms ± 2% ~ (p=0.690 n=5+5) Binomial-8 4.86µs ± 1% 4.86µs ± 1% ~ (p=0.889 n=5+5) QuoRem-8 7.09µs ± 0% 7.08µs ± 0% -0.09% (p=0.024 n=5+5) Exp-8 161ms ± 0% 161ms ± 0% -0.08% (p=0.032 n=5+5) Exp2-8 161ms ± 0% 161ms ± 0% ~ (p=1.000 n=5+5) Bitset-8 40.7ns ± 0% 40.6ns ± 0% ~ (p=0.095 n=4+5) BitsetNeg-8 159ns ± 4% 148ns ± 0% -6.92% (p=0.016 n=5+4) BitsetOrig-8 378ns ± 1% 378ns ± 1% ~ (p=0.937 n=5+5) BitsetNegOrig-8 647ns ± 5% 647ns ± 4% ~ (p=1.000 n=5+5) ModSqrt225_Tonelli-8 7.26ms ± 0% 7.27ms ± 0% ~ (p=1.000 n=5+5) ModSqrt224_3Mod4-8 2.24ms ± 0% 2.24ms ± 0% ~ (p=0.690 n=5+5) ModSqrt5430_Tonelli-8 62.8s ± 1% 62.5s ± 0% ~ (p=0.063 n=5+4) ModSqrt5430_3Mod4-8 20.8s ± 0% 20.8s ± 0% ~ (p=0.310 n=5+5) Sqrt-8 101µs ± 1% 101µs ± 0% -0.35% (p=0.032 n=5+5) IntSqr/1-8 32.3ns ± 1% 32.5ns ± 1% ~ (p=0.421 n=5+5) IntSqr/2-8 157ns ± 5% 156ns ± 5% ~ (p=0.651 n=5+5) IntSqr/3-8 292ns ± 2% 291ns ± 3% ~ (p=0.881 n=5+5) IntSqr/5-8 738ns ± 6% 740ns ± 5% ~ (p=0.841 n=5+5) IntSqr/8-8 1.82µs ± 4% 1.83µs ± 4% ~ (p=0.730 n=5+5) IntSqr/10-8 2.92µs ± 1% 2.93µs ± 1% ~ (p=0.643 n=5+5) IntSqr/20-8 6.28µs ± 2% 6.28µs ± 2% ~ (p=1.000 n=5+5) IntSqr/30-8 13.8µs ± 2% 13.9µs ± 3% ~ (p=1.000 n=5+5) IntSqr/50-8 37.8µs ± 4% 37.9µs ± 4% ~ (p=0.690 n=5+5) IntSqr/80-8 95.9µs ± 1% 95.8µs ± 1% ~ (p=0.841 n=5+5) IntSqr/100-8 148µs ± 1% 148µs ± 1% ~ (p=0.310 n=5+5) IntSqr/200-8 586µs ± 1% 586µs ± 1% ~ (p=0.841 n=5+5) IntSqr/300-8 1.32ms ± 0% 1.31ms ± 0% ~ (p=0.222 n=5+5) IntSqr/500-8 2.48ms ± 0% 2.48ms ± 0% ~ (p=0.556 n=5+4) IntSqr/800-8 4.68ms ± 0% 4.68ms ± 0% ~ (p=0.548 n=5+5) IntSqr/1000-8 7.57ms ± 0% 7.56ms ± 0% ~ (p=0.421 n=5+5) Mul-8 311ms ± 0% 311ms ± 0% ~ (p=0.548 n=5+5) Exp3Power/0x10-8 559ns ± 1% 560ns ± 1% ~ (p=0.984 n=5+5) Exp3Power/0x40-8 641ns ± 1% 634ns ± 1% ~ (p=0.063 n=5+5) Exp3Power/0x100-8 1.39µs ± 2% 1.40µs ± 2% ~ (p=0.381 n=5+5) Exp3Power/0x400-8 8.27µs ± 1% 8.26µs ± 0% ~ (p=0.571 n=5+5) Exp3Power/0x1000-8 59.9µs ± 0% 59.7µs ± 0% -0.23% (p=0.008 n=5+5) Exp3Power/0x4000-8 816µs ± 0% 816µs ± 0% ~ (p=1.000 n=5+5) Exp3Power/0x10000-8 7.77ms ± 0% 7.77ms ± 0% ~ (p=0.841 n=5+5) Exp3Power/0x40000-8 73.4ms ± 0% 73.4ms ± 0% ~ (p=0.690 n=5+5) Exp3Power/0x100000-8 665ms ± 0% 664ms ± 0% -0.14% (p=0.008 n=5+5) Exp3Power/0x400000-8 5.98s ± 0% 5.98s ± 0% -0.09% (p=0.008 n=5+5) Fibo-8 116ms ± 0% 116ms ± 0% -0.25% (p=0.008 n=5+5) NatSqr/1-8 115ns ± 3% 116ns ± 2% ~ (p=0.238 n=5+5) NatSqr/2-8 237ns ± 1% 237ns ± 1% ~ (p=0.683 n=5+5) NatSqr/3-8 367ns ± 3% 368ns ± 3% ~ (p=0.817 n=5+5) NatSqr/5-8 807ns ± 3% 812ns ± 3% ~ (p=0.913 n=5+5) NatSqr/8-8 1.93µs ± 2% 1.93µs ± 3% ~ (p=0.651 n=5+5) NatSqr/10-8 2.98µs ± 2% 2.99µs ± 2% ~ (p=0.690 n=5+5) NatSqr/20-8 6.49µs ± 2% 6.46µs ± 2% ~ (p=0.548 n=5+5) NatSqr/30-8 14.4µs ± 2% 14.3µs ± 2% ~ (p=0.690 n=5+5) NatSqr/50-8 38.6µs ± 2% 38.7µs ± 2% ~ (p=0.841 n=5+5) NatSqr/80-8 96.1µs ± 2% 95.8µs ± 2% ~ (p=0.548 n=5+5) NatSqr/100-8 149µs ± 1% 149µs ± 1% ~ (p=0.841 n=5+5) NatSqr/200-8 593µs ± 1% 590µs ± 1% ~ (p=0.421 n=5+5) NatSqr/300-8 1.32ms ± 0% 1.32ms ± 1% ~ (p=0.222 n=5+5) NatSqr/500-8 2.49ms ± 0% 2.49ms ± 0% ~ (p=0.690 n=5+5) NatSqr/800-8 4.69ms ± 0% 4.69ms ± 0% ~ (p=1.000 n=5+5) NatSqr/1000-8 7.59ms ± 0% 7.58ms ± 0% ~ (p=0.841 n=5+5) ScanPi-8 322µs ± 0% 321µs ± 0% ~ (p=0.095 n=5+5) StringPiParallel-8 71.4µs ± 5% 68.8µs ± 4% ~ (p=0.151 n=5+5) Scan/10/Base2-8 1.10µs ± 0% 1.09µs ± 0% -0.36% (p=0.032 n=5+5) Scan/100/Base2-8 7.78µs ± 0% 7.79µs ± 0% +0.14% (p=0.008 n=5+5) Scan/1000/Base2-8 78.8µs ± 0% 79.0µs ± 0% +0.24% (p=0.008 n=5+5) Scan/10000/Base2-8 1.22ms ± 0% 1.22ms ± 0% ~ (p=0.056 n=5+5) Scan/100000/Base2-8 55.1ms ± 0% 55.0ms ± 0% -0.15% (p=0.008 n=5+5) Scan/10/Base8-8 514ns ± 0% 515ns ± 0% ~ (p=0.079 n=5+5) Scan/100/Base8-8 2.89µs ± 0% 2.89µs ± 0% +0.15% (p=0.008 n=5+5) Scan/1000/Base8-8 31.0µs ± 0% 31.1µs ± 0% +0.12% (p=0.008 n=5+5) Scan/10000/Base8-8 740µs ± 0% 740µs ± 0% ~ (p=0.222 n=5+5) Scan/100000/Base8-8 50.6ms ± 0% 50.5ms ± 0% -0.06% (p=0.016 n=4+5) Scan/10/Base10-8 492ns ± 1% 490ns ± 1% ~ (p=0.310 n=5+5) Scan/100/Base10-8 2.67µs ± 0% 2.67µs ± 0% ~ (p=0.056 n=5+5) Scan/1000/Base10-8 28.7µs ± 0% 28.7µs ± 0% ~ (p=1.000 n=5+5) Scan/10000/Base10-8 717µs ± 0% 716µs ± 0% ~ (p=0.222 n=5+5) Scan/100000/Base10-8 50.2ms ± 0% 50.3ms ± 0% +0.05% (p=0.008 n=5+5) Scan/10/Base16-8 442ns ± 1% 442ns ± 0% ~ (p=0.468 n=5+5) Scan/100/Base16-8 2.46µs ± 0% 2.45µs ± 0% ~ (p=0.159 n=5+5) Scan/1000/Base16-8 27.2µs ± 0% 27.2µs ± 0% ~ (p=0.841 n=5+5) Scan/10000/Base16-8 721µs ± 0% 722µs ± 0% ~ (p=0.548 n=5+5) Scan/100000/Base16-8 52.6ms ± 0% 52.6ms ± 0% +0.07% (p=0.008 n=5+5) String/10/Base2-8 244ns ± 1% 242ns ± 1% ~ (p=0.103 n=5+5) String/100/Base2-8 1.48µs ± 0% 1.48µs ± 1% ~ (p=0.786 n=5+5) String/1000/Base2-8 13.3µs ± 1% 13.3µs ± 0% ~ (p=0.222 n=5+5) String/10000/Base2-8 132µs ± 1% 132µs ± 1% ~ (p=1.000 n=5+5) String/100000/Base2-8 1.30ms ± 1% 1.30ms ± 1% ~ (p=1.000 n=5+5) String/10/Base8-8 167ns ± 1% 168ns ± 1% ~ (p=0.135 n=5+5) String/100/Base8-8 623ns ± 1% 626ns ± 1% ~ (p=0.151 n=5+5) String/1000/Base8-8 5.24µs ± 1% 5.24µs ± 0% ~ (p=1.000 n=5+5) String/10000/Base8-8 50.0µs ± 1% 50.0µs ± 1% ~ (p=1.000 n=5+5) String/100000/Base8-8 492µs ± 1% 489µs ± 1% ~ (p=0.056 n=5+5) String/10/Base10-8 503ns ± 1% 501ns ± 0% ~ (p=0.183 n=5+5) String/100/Base10-8 1.96µs ± 0% 1.97µs ± 0% ~ (p=0.389 n=5+5) String/1000/Base10-8 12.4µs ± 1% 12.4µs ± 1% ~ (p=0.841 n=5+5) String/10000/Base10-8 56.7µs ± 1% 56.6µs ± 0% ~ (p=1.000 n=5+5) String/100000/Base10-8 25.6ms ± 0% 25.6ms ± 0% ~ (p=0.222 n=5+5) String/10/Base16-8 147ns ± 0% 148ns ± 2% ~ (p=1.000 n=4+5) String/100/Base16-8 505ns ± 0% 505ns ± 1% ~ (p=0.778 n=5+5) String/1000/Base16-8 3.94µs ± 0% 3.94µs ± 0% ~ (p=0.841 n=5+5) String/10000/Base16-8 37.4µs ± 1% 37.2µs ± 1% ~ (p=0.095 n=5+5) String/100000/Base16-8 367µs ± 1% 367µs ± 0% ~ (p=1.000 n=5+5) LeafSize/0-8 6.64ms ± 0% 6.65ms ± 0% ~ (p=0.690 n=5+5) LeafSize/1-8 72.5µs ± 1% 72.4µs ± 1% ~ (p=0.841 n=5+5) LeafSize/2-8 72.6µs ± 1% 72.6µs ± 1% ~ (p=1.000 n=5+5) LeafSize/3-8 377µs ± 0% 377µs ± 0% ~ (p=0.421 n=5+5) LeafSize/4-8 71.2µs ± 1% 71.3µs ± 0% ~ (p=0.278 n=5+5) LeafSize/5-8 469µs ± 0% 469µs ± 0% ~ (p=0.310 n=5+5) LeafSize/6-8 376µs ± 0% 376µs ± 0% ~ (p=0.841 n=5+5) LeafSize/7-8 244µs ± 0% 244µs ± 0% ~ (p=0.841 n=5+5) LeafSize/8-8 71.9µs ± 1% 72.1µs ± 1% ~ (p=0.548 n=5+5) LeafSize/9-8 536µs ± 0% 536µs ± 0% ~ (p=0.151 n=5+5) LeafSize/10-8 470µs ± 0% 471µs ± 0% +0.10% (p=0.032 n=5+5) LeafSize/11-8 458µs ± 0% 458µs ± 0% ~ (p=0.881 n=5+5) LeafSize/12-8 376µs ± 0% 376µs ± 0% ~ (p=0.548 n=5+5) LeafSize/13-8 341µs ± 0% 342µs ± 0% ~ (p=0.222 n=5+5) LeafSize/14-8 246µs ± 0% 245µs ± 0% ~ (p=0.167 n=5+5) LeafSize/15-8 168µs ± 0% 168µs ± 0% ~ (p=0.548 n=5+5) LeafSize/16-8 72.1µs ± 1% 72.2µs ± 1% ~ (p=0.690 n=5+5) LeafSize/32-8 81.5µs ± 1% 81.4µs ± 1% ~ (p=1.000 n=5+5) LeafSize/64-8 133µs ± 1% 134µs ± 1% ~ (p=0.690 n=5+5) ProbablyPrime/n=0-8 44.3ms ± 0% 44.2ms ± 0% -0.28% (p=0.008 n=5+5) ProbablyPrime/n=1-8 64.8ms ± 0% 64.7ms ± 0% -0.15% (p=0.008 n=5+5) ProbablyPrime/n=5-8 147ms ± 0% 147ms ± 0% -0.11% (p=0.008 n=5+5) ProbablyPrime/n=10-8 250ms ± 0% 250ms ± 0% ~ (p=0.056 n=5+5) ProbablyPrime/n=20-8 456ms ± 0% 455ms ± 0% -0.05% (p=0.008 n=5+5) ProbablyPrime/Lucas-8 23.6ms ± 0% 23.5ms ± 0% -0.29% (p=0.008 n=5+5) ProbablyPrime/MillerRabinBase2-8 20.6ms ± 0% 20.6ms ± 0% ~ (p=0.690 n=5+5) FloatSqrt/64-8 2.01µs ± 1% 2.02µs ± 1% ~ (p=0.421 n=5+5) FloatSqrt/128-8 4.43µs ± 2% 4.38µs ± 2% ~ (p=0.222 n=5+5) FloatSqrt/256-8 6.64µs ± 1% 6.68µs ± 2% ~ (p=0.516 n=5+5) FloatSqrt/1000-8 31.9µs ± 0% 31.8µs ± 0% ~ (p=0.095 n=5+5) FloatSqrt/10000-8 595µs ± 0% 594µs ± 0% ~ (p=0.056 n=5+5) FloatSqrt/100000-8 17.9ms ± 0% 17.9ms ± 0% ~ (p=0.151 n=5+5) FloatSqrt/1000000-8 1.52s ± 0% 1.52s ± 0% ~ (p=0.841 n=5+5) name old speed new speed delta AddVV/1-8 2.97GB/s ± 0% 2.97GB/s ± 0% ~ (p=0.971 n=4+4) AddVV/2-8 9.47GB/s ± 0% 9.47GB/s ± 0% +0.01% (p=0.016 n=5+5) AddVV/3-8 12.4GB/s ± 0% 12.4GB/s ± 0% ~ (p=0.548 n=5+5) AddVV/4-8 14.6GB/s ± 0% 14.6GB/s ± 0% ~ (p=1.000 n=5+5) AddVV/5-8 16.4GB/s ± 0% 16.4GB/s ± 0% ~ (p=1.000 n=5+5) AddVV/10-8 21.7GB/s ± 0% 21.7GB/s ± 0% ~ (p=0.548 n=5+5) AddVV/100-8 29.4GB/s ± 0% 29.4GB/s ± 0% ~ (p=1.000 n=5+5) AddVV/1000-8 31.7GB/s ± 0% 31.7GB/s ± 0% ~ (p=0.524 n=5+4) AddVV/10000-8 31.5GB/s ± 0% 31.5GB/s ± 0% ~ (p=0.690 n=5+5) AddVV/100000-8 28.8GB/s ± 7% 28.1GB/s ± 8% ~ (p=0.548 n=5+5) AddVW/1-8 859MB/s ± 0% 864MB/s ± 0% +0.61% (p=0.008 n=5+5) AddVW/2-8 809MB/s ± 2% 1520MB/s ± 0% +87.78% (p=0.008 n=5+5) AddVW/3-8 2.08GB/s ± 0% 2.18GB/s ± 0% +4.54% (p=0.008 n=5+5) AddVW/4-8 2.46GB/s ± 0% 2.66GB/s ± 0% +8.33% (p=0.016 n=4+5) AddVW/5-8 2.76GB/s ± 0% 3.20GB/s ± 0% +16.03% (p=0.008 n=5+5) AddVW/10-8 3.63GB/s ± 0% 5.15GB/s ± 0% +41.83% (p=0.008 n=5+5) AddVW/100-8 4.79GB/s ± 0% 9.87GB/s ± 0% +106.12% (p=0.008 n=5+5) AddVW/1000-8 5.27GB/s ± 0% 12.42GB/s ± 0% +135.74% (p=0.008 n=5+5) AddVW/10000-8 5.31GB/s ± 0% 11.19GB/s ± 0% +110.71% (p=0.008 n=5+5) AddVW/100000-8 5.32GB/s ± 0% 11.32GB/s ± 0% +112.56% (p=0.008 n=5+5) SubVW/1-8 859MB/s ± 0% 864MB/s ± 0% +0.61% (p=0.008 n=5+5) SubVW/2-8 812MB/s ± 2% 1520MB/s ± 0% +87.09% (p=0.008 n=5+5) SubVW/3-8 2.08GB/s ± 0% 2.18GB/s ± 0% +4.55% (p=0.008 n=5+5) SubVW/4-8 2.46GB/s ± 0% 2.66GB/s ± 0% +8.33% (p=0.008 n=5+5) SubVW/5-8 2.75GB/s ± 0% 3.20GB/s ± 0% +16.03% (p=0.008 n=5+5) SubVW/10-8 3.63GB/s ± 0% 5.15GB/s ± 0% +41.82% (p=0.008 n=5+5) SubVW/100-8 4.79GB/s ± 0% 9.87GB/s ± 0% +106.13% (p=0.008 n=5+5) SubVW/1000-8 5.27GB/s ± 0% 12.42GB/s ± 0% +135.74% (p=0.008 n=5+5) SubVW/10000-8 5.31GB/s ± 0% 11.17GB/s ± 0% +110.44% (p=0.008 n=5+5) SubVW/100000-8 5.32GB/s ± 0% 11.31GB/s ± 0% +112.35% (p=0.008 n=5+5) AddMulVVW/1-8 1.97GB/s ± 1% 1.96GB/s ± 1% ~ (p=0.151 n=5+5) AddMulVVW/2-8 2.24GB/s ± 0% 2.25GB/s ± 0% ~ (p=0.095 n=5+5) AddMulVVW/3-8 2.11GB/s ± 0% 2.12GB/s ± 0% ~ (p=0.548 n=5+5) AddMulVVW/4-8 2.17GB/s ± 1% 2.17GB/s ± 1% ~ (p=0.548 n=5+5) AddMulVVW/5-8 2.22GB/s ± 1% 2.21GB/s ± 1% ~ (p=0.421 n=5+5) AddMulVVW/10-8 2.17GB/s ± 1% 2.16GB/s ± 0% ~ (p=0.095 n=5+5) AddMulVVW/100-8 2.35GB/s ± 0% 2.35GB/s ± 0% ~ (p=0.421 n=5+5) AddMulVVW/1000-8 2.47GB/s ± 0% 2.41GB/s ± 0% -2.09% (p=0.008 n=5+5) AddMulVVW/10000-8 2.16GB/s ± 0% 2.15GB/s ± 0% -0.23% (p=0.008 n=5+5) AddMulVVW/100000-8 2.03GB/s ± 1% 2.04GB/s ± 0% ~ (p=0.690 n=5+5) name old alloc/op new alloc/op delta FloatString/100-8 400B ± 0% 400B ± 0% ~ (all equal) FloatString/1000-8 3.22kB ± 0% 3.22kB ± 0% ~ (all equal) FloatString/10000-8 55.6kB ± 0% 55.5kB ± 0% ~ (p=0.206 n=5+5) FloatString/100000-8 627kB ± 0% 627kB ± 0% ~ (all equal) FloatAdd/10-8 0.00B 0.00B ~ (all equal) FloatAdd/100-8 0.00B 0.00B ~ (all equal) FloatAdd/1000-8 0.00B 0.00B ~ (all equal) FloatAdd/10000-8 0.00B 0.00B ~ (all equal) FloatAdd/100000-8 0.00B 0.00B ~ (all equal) FloatSub/10-8 0.00B 0.00B ~ (all equal) FloatSub/100-8 0.00B 0.00B ~ (all equal) FloatSub/1000-8 0.00B 0.00B ~ (all equal) FloatSub/10000-8 0.00B 0.00B ~ (all equal) FloatSub/100000-8 0.00B 0.00B ~ (all equal) FloatSqrt/64-8 416B ± 0% 416B ± 0% ~ (all equal) FloatSqrt/128-8 720B ± 0% 720B ± 0% ~ (all equal) FloatSqrt/256-8 816B ± 0% 816B ± 0% ~ (all equal) FloatSqrt/1000-8 2.50kB ± 0% 2.50kB ± 0% ~ (all equal) FloatSqrt/10000-8 23.5kB ± 0% 23.5kB ± 0% ~ (all equal) FloatSqrt/100000-8 251kB ± 0% 251kB ± 0% ~ (all equal) FloatSqrt/1000000-8 4.61MB ± 0% 4.61MB ± 0% ~ (all equal) name old allocs/op new allocs/op delta FloatString/100-8 8.00 ± 0% 8.00 ± 0% ~ (all equal) FloatString/1000-8 10.0 ± 0% 10.0 ± 0% ~ (all equal) FloatString/10000-8 42.0 ± 0% 42.0 ± 0% ~ (all equal) FloatString/100000-8 346 ± 0% 346 ± 0% ~ (all equal) FloatAdd/10-8 0.00 0.00 ~ (all equal) FloatAdd/100-8 0.00 0.00 ~ (all equal) FloatAdd/1000-8 0.00 0.00 ~ (all equal) FloatAdd/10000-8 0.00 0.00 ~ (all equal) FloatAdd/100000-8 0.00 0.00 ~ (all equal) FloatSub/10-8 0.00 0.00 ~ (all equal) FloatSub/100-8 0.00 0.00 ~ (all equal) FloatSub/1000-8 0.00 0.00 ~ (all equal) FloatSub/10000-8 0.00 0.00 ~ (all equal) FloatSub/100000-8 0.00 0.00 ~ (all equal) FloatSqrt/64-8 9.00 ± 0% 9.00 ± 0% ~ (all equal) FloatSqrt/128-8 13.0 ± 0% 13.0 ± 0% ~ (all equal) FloatSqrt/256-8 12.0 ± 0% 12.0 ± 0% ~ (all equal) FloatSqrt/1000-8 19.0 ± 0% 19.0 ± 0% ~ (all equal) FloatSqrt/10000-8 35.0 ± 0% 35.0 ± 0% ~ (all equal) FloatSqrt/100000-8 55.0 ± 0% 55.0 ± 0% ~ (all equal) FloatSqrt/1000000-8 122 ± 0% 122 ± 0% ~ (all equal) Change-Id: I6888d84c037d91f9e2199f3492ea3f6a0ed77b24 Reviewed-on: https://go-review.googlesource.com/77832Reviewed-by:
Vlad Krasnov <vlad@cloudflare.com> Reviewed-by:
-
Lynn Boger authored
When instructions add, and, or, xor, and movd have constant operands in some cases more instructions are generated than necessary by the assembler. This adds more opcode/operand combinations to the optab and improves the code generation for the cases where the size and sign of the constant allows the use of 1 instructions instead of 2. Example of previous code: oris r3, r0, 0 ori r3, r3, 65533 now: ori r3, r0, 65533 This does not significantly reduce the overall binary size because the improvement depends on the constant value. Some procedures show a 1-2% reduction in size. This improvement could also be significant in cases where the extra instructions occur in a critical loop. Testcase ppc64enc.s was added to cmd/asm/internal/asm/testdata with the variations affected by this change. Updates #23845 Change-Id: I7fdf2320c95815d99f2755ba77d0c6921cd7fad7 Reviewed-on: https://go-review.googlesource.com/95135 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Joe Tsai authored
In the situation where a quoted field is necessary, avoid processing each UTF-8 rune one-by-one, which causes mangling of invalid sequences into utf8.RuneError, causing a loss of information. Instead, search only for the escaped characters, handle those specially and copy everything else in between verbatim. This symmetrically matches the behavior of Reader. Fixes #24298 Change-Id: I9276f64891084ce8487678f663fad711b4095dbb Reviewed-on: https://go-review.googlesource.com/99297 Run-TryBot: Joe Tsai <thebrokentoaster@gmail.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Matthew Dempsky authored
This was already done for normal parameters, and the same logic applies for receiver parameters too. Updates #24305. Change-Id: Ia2a46f68d14e8fb62004ff0da1db0f065a95a1b7 Reviewed-on: https://go-review.googlesource.com/99335 Run-TryBot: Matthew Dempsky <mdempsky@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
- 07 Mar, 2018 9 commits
-
-
Ian Lance Taylor authored
Without the change to cover.go, the new test fails with panic: overlapping edits: [4946,4950)->"", [4947,4947)->"thisNameMustBeVeryLongToCauseOverflowOfCounterIncrementStatementOntoNextLineForTest.Count[112]++;" The original code inserts "else{", deletes "else", and then positions a new block just after the "}" that must come before the "else". That works on gofmt'ed code, but fails if the code looks like "}else". When there is no space between the "{" and the "else", the new block is inserted into a location that we are deleting, leading to the "overlapping edits" mentioned above. This CL fixes this case by not deleting the "else" but just using the one that is already there. That requires adjust the block offset to come after the "{" that we insert. Fixes #23927 Change-Id: I40ef592490878765bbce6550ddb439e43ac525b2 Reviewed-on: https://go-review.googlesource.com/98935 Run-TryBot: Ian Lance Taylor <iant@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: -
Ian Lance Taylor authored
Currently if a profiling signal arrives while executing within a VDSO the profiler will report _ExternalCode, which is needlessly confusing for a pure Go program. Change the VDSO calling code to record the caller's PC/SP, so that we can do a traceback from that point. If that fails for some reason, report _VDSO rather than _ExternalCode, which should at least point in the right direction. This adds some instructions to the code that calls the VDSO, but the slowdown is reasonably negligible: name old time/op new time/op delta ClockVDSOAndFallbackPaths/vDSO-8 40.5ns ± 2% 41.3ns ± 1% +1.85% (p=0.002 n=10+10) ClockVDSOAndFallbackPaths/Fallback-8 41.9ns ± 1% 43.5ns ± 1% +3.84% (p=0.000 n=9+9) TimeNow-8 41.5ns ± 3% 41.5ns ± 2% ~ (p=0.723 n=10+10) Fixes #24142 Change-Id: Iacd935db3c4c782150b3809aaa675a71799b1c9c Reviewed-on: https://go-review.googlesource.com/97315 Run-TryBot: Ian Lance Taylor <iant@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Ian Lance Taylor authored
This normalizes the Linux code to act like other targets. The size argument to the rt_sigaction system call is pushed to a single function, sysSigaction. This is intended as a simplification step for CL 93875 for #14327. Change-Id: I594788e235f0da20e16e8a028e27ac8c883907c4 Reviewed-on: https://go-review.googlesource.com/99077 Run-TryBot: Ian Lance Taylor <iant@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Brad Fitzpatrick authored
Updates #24300 Change-Id: I7752dab67e15a6dfe5fffe5b5ccbf3373bbc2c13 Reviewed-on: https://go-review.googlesource.com/99296Reviewed-by:
-
Vlad Krasnov authored
The lack of proper addMulVVW implementation for arm64 hurts RSA performance. This assembly implementation is optimized for arm64 based servers. name old time/op new time/op delta pkg:math/big goos:linux goarch:arm64 AddMulVVW/1 55.2ns ± 0% 11.9ns ± 1% -78.37% (p=0.000 n=8+10) AddMulVVW/2 67.0ns ± 0% 11.2ns ± 0% -83.28% (p=0.000 n=7+10) AddMulVVW/3 93.2ns ± 0% 13.2ns ± 0% -85.84% (p=0.000 n=10+10) AddMulVVW/4 126ns ± 0% 13ns ± 1% -89.82% (p=0.000 n=10+10) AddMulVVW/5 151ns ± 0% 17ns ± 0% -88.87% (p=0.000 n=10+9) AddMulVVW/10 323ns ± 0% 25ns ± 0% -92.20% (p=0.000 n=10+10) AddMulVVW/100 3.28µs ± 0% 0.14µs ± 0% -95.82% (p=0.000 n=10+10) AddMulVVW/1000 31.7µs ± 0% 1.3µs ± 0% -96.00% (p=0.000 n=10+8) AddMulVVW/10000 313µs ± 0% 13µs ± 0% -95.98% (p=0.000 n=10+10) AddMulVVW/100000 3.24ms ± 0% 0.13ms ± 1% -96.13% (p=0.000 n=9+9) pkg:crypto/rsa goos:linux goarch:arm64 RSA2048Decrypt 44.7ms ± 0% 4.0ms ± 6% -91.08% (p=0.000 n=8+10) RSA2048Sign 46.3ms ± 0% 5.0ms ± 0% -89.29% (p=0.000 n=9+10) 3PrimeRSA2048Decrypt 22.3ms ± 0% 2.4ms ± 0% -89.26% (p=0.000 n=10+10) Change-Id: I295f0bd5c51a4442d02c44ece1f6026d30dff0bc Reviewed-on: https://go-review.googlesource.com/76270Reviewed-by:
-
David du Colombier authored
CL 99175 added TestVetWithOnlyCgoFiles. However, this test is failing on platforms where cgo is disabled, because no file can be built. This change fixes TestVetWithOnlyCgoFiles by skipping this test when cgo is disabled. Fixes #24304. Change-Id: Ibb38fcd3e0ed1a791782145d3f2866f12117c6fe Reviewed-on: https://go-review.googlesource.com/99275Reviewed-by:
-
Elias Naur authored
While working on standalone builds of gomobile bindings, I ran into errors on the form: gcc_darwin_arm.c:30:31: error: ambiguous expansion of macro 'nil' [-Werror,-Wambiguous-macro] /Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS11.2.sdk/usr/include/MacTypes.h:94:15: note: expanding this definition of 'nil' Fix it by undefining nil before defining it in libcgo.h. Change-Id: I8e9660a68c6c351e592684d03d529f0d182c0493 Reviewed-on: https://go-review.googlesource.com/99215 Run-TryBot: Elias Naur <elias.naur@gmail.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Ian Lance Taylor authored
CgoFiles is not included in GoFiles, so we need to check both. Fixes #24193 Change-Id: I6a67bd912e3d9a4be0eae8fa8db6fa8a07fb5df3 Reviewed-on: https://go-review.googlesource.com/99175 Run-TryBot: Ian Lance Taylor <iant@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by:
-
Matthew Dempsky authored
We already require expressions to have already been typechecked before reaching walk. Moreover, all untyped expressions should have been converted to their default type by walk. However, in practice, we've been somewhat sloppy and inconsistent about ensuring this. In particular, a lot of AST rewrites ended up leaving untyped bool expressions scattered around. These likely aren't harmful in practice, but it seems worth cleaning up. The two most common cases addressed by this CL are: 1) When generating OIF and OFOR nodes, we would often typecheck the conditional expression, but not apply defaultlit to force it to the expression's default type. 2) When rewriting string comparisons into more fundamental primitives, we were simply overwriting r.Type with the desired type, which didn't propagate the type to nested subexpressions. These are fixed by utilizing finishcompare, which correctly handles this (and is already used by other comparison lowering rewrites). Lastly, walkexpr is extended to assert that it's not called on untyped expressions. Fixes #23834. Change-Id: Icbd29648a293555e4015d3b06a95a24ccbd3f790 Reviewed-on: https://go-review.googlesource.com/98337Reviewed-by:
-
