Adjustments to K performance information

1 files changed, 9 insertions, 3 deletions

diff --git a/implementation/kclaims.md b/implementation/kclaims.md
index 20d40646..1eee6966 100644
--- a/implementation/kclaims.md
+++ b/implementation/kclaims.md
@@ -10,11 +10,13 @@ The points I argue here are narrow. To some extent I'm picking out the craziest
 
 On that note, it's possible *I've* made mistakes, such as incorrectly designing or interpreting benchmarks. If you present me with concrete evidence against something I wrote below, I promise I'll revise this page to include it, even if I just have to quote verbatim because I don't understand a word of it.
 
+This page has now been [discussed](https://news.ycombinator.com/item?id=28365645) on Hacker News, and I have amended and added information based on comments there and elsewhere.
+
 ## The fastest array language
 
 When you ask what the fastest array language is, chances are someone is there to answer one of k, kdb, or q. I can't offer benchmarks that contradict this, but I will argue that there's little reason to take these people at their word.
 
-The reason I have no measurements is that every contract for a commercial K includes an anti-benchmark clause. For example, Shakti's [license](https://shakti.com/download/license) says users cannot "distribute or otherwise make available to any third party any report regarding the performance of the Software benchmarks or any information from such a report". As I would be unable to share the results, I have not taken benchmarks of any commercial K. Or downloaded one for that matter. Shakti could publish benchmarks; they choose to publish a handful of comparisons with database software and none with array languages or frameworks. I do run tests with [ngn/k](https://codeberg.org/ngn/k), which is developed with goals similar to Whitney's K; the author says it's slower than Shakti but not by too much.
+The reason I have no measurements is that every contract for a commercial K includes an anti-benchmark clause. For example, Shakti's [license](https://shakti.com/download/license) says users cannot "distribute or otherwise make available to any third party any report regarding the performance of the Software benchmarks or any information from such a report". This ["DeWitt Clause"](https://en.wikipedia.org/wiki/DeWitt_Clause) is common in many databases. It takes much more work to refute bad benchmarks than to produce them so perhaps in the high-stakes DBMS world it's not as crazy as it seems—even though it's a terrible outcome. As I would be unable to share the results, I have not taken benchmarks of any commercial K. Or downloaded one for that matter. So I'm limited to a small number of approved benchmarks that have been published, which focus on performance as a database and not a programming language. I also run tests with [ngn/k](https://codeberg.org/ngn/k), which is developed with goals similar to Whitney's K; the author says it's slower than Shakti but not by too much.
 
 The primary reason I don't give any credence to claims that K is the best is that they are always devoid of specifics. Most importantly, the same assertion is made across decades even though performance in J, Dyalog, and NumPy has improved by leaps and bounds in the meantime—I participated in advances of [26%](https://www.dyalog.com/dyalog/dyalog-versions/170/performance.htm) and [10%](https://www.dyalog.com/dyalog-versions/180/performance.htm) in overall Dyalog benchmarks in the last two major versions. Has K4 (the engine behind kdb and Q) kept pace? Maybe it's fallen behind since Arthur left but Shakti K is better? Which other array languages has the poster used? Doesn't matter—*they* are all the same but *K* is better.
 
@@ -28,9 +30,13 @@ I have a suspicion that users sometimes think K is faster than APL because they
 
 Popular APL and J implementations interpret source code directly, without even building an AST. This is very slow, and Dyalog has several other pathologies that get in the way as well. Like storing the execution stack in the workspace to prevent stack overflows, and the requirement that a user can save a workspace with paused code and resume it *in a later version*. But the overhead is per token executed, and a programmer can avoid the cost by working on large arrays where one token does a whole lot of work. If you want to show a language is faster than APL generally, this is the kind of code to look at.
 
-K is designed to be as fast as possible when interpreting scalar code, for example using a [grammar](https://k.miraheze.org/wiki/Grammar) that's much simpler than [BQN's](../spec/grammar.md) (speed isn't the only benefit of being simpler of course, but it's clearly a consideration). It succeeds at this, and K interpreters are very fast, even without bytecode compilation in advance.
+K's design is well-suited to interpreting scalar code because of its simplicity: for example, it has only one kind of user-defined function and doesn't allow lexical closures. Its [grammar](https://k.miraheze.org/wiki/Grammar) is much simpler than [BQN's](../spec/grammar.md), although [it seems](https://news.ycombinator.com/item?id=6118565) that K uses bytecode compilation (ngn/k certainly does), making this irrelevant after one quick compilation pass. As a result, K interpreters can be very fast.
+
+But K still isn't good at scalar code! It's an interpreter (if a good one) for a dynamically-typed language, and will be slower than compiled languages like C and Go, or JIT-compiled ones like Javascript and Java. A compiler generates code to do what you want, while an interpreter (including a bytecode VM) is code that reads data (the program) to do what you want. Once the code is compiled, the interpreter has an extra step and *has* to be slower. This is why BQN uses compiler-based strategies to speed up execution, first compiling to bytecode and then usually post-processing that bytecode. Compilation is fast enough that it's perfectly fine to compile code every time it's run.
+
+## Parallel execution
 
-But K still isn't good at scalar code! It's an interpreter (if a good one) for a dynamically-typed language, and will be slower than compiled languages like C and Go, or JIT-compiled ones like Javascript and Java. A compiler generates code to do what you want, while an interpreter is code that reads data (the program) to do what you want. Once the code is compiled, the interpreter has an extra step and *has* to be slower. This is why BQN uses compiler-based strategies to speed up execution, first compiling to bytecode to make syntax overhead irrelevant and then usually post-processing that bytecode. Compilation is fast enough that it's perfectly fine to compile code every time it's run.
+As of 2020, Q supports [multithreaded primitives](https://code.kx.com/q/kb/mt-primitives/) that can run on multiple CPU cores. I think Shakti supports multi-threading as well. Oddly enough, J user Monument AI has also been working on their own parallel [J engine](https://www.monument.ai/m/parallel). So array languages are finally moving to multiple cores (the reason this hasn't happened sooner is probably that array language users often have workloads where they can run one instance on each core, which is easier and tends to be faster than splitting one run across multiple cores). It's interesting, and a potential reason to use K or Q, although it's too recent to be part of the "K is fastest" mythos. Not every K claim is a wild one!
 
 ## Instruction cache