Path: csiph.com!xmission!news.alt.net!feeder.usenetexpress.com!tr1.iad1.usenetexpress.com!border1.nntp.dca1.giganews.com!nntp.giganews.com!news.iecc.com!.POSTED.news.iecc.com!nerds-end From: gah4 Newsgroups: comp.compilers Subject: Re: Bit swizzling Date: Thu, 10 Sep 2020 10:34:18 -0700 (PDT) Organization: Compilers Central Lines: 28 Sender: news@iecc.com Approved: comp.compilers@iecc.com Message-ID: <20-09-026@comp.compilers> References: <20-09-014@comp.compilers> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Injection-Info: gal.iecc.com; posting-host="news.iecc.com:2001:470:1f07:1126:0:676f:7373:6970"; logging-data="67222"; mail-complaints-to="abuse@iecc.com" Keywords: optimize, hardware Posted-Date: 10 Sep 2020 13:55:01 EDT X-submission-address: compilers@iecc.com X-moderator-address: compilers-request@iecc.com X-FAQ-and-archives: http://compilers.iecc.com In-Reply-To: <20-09-014@comp.compilers> Xref: csiph.com comp.compilers:2597 On Saturday, September 5, 2020 at 9:45:43 AM UTC-7, Rick C. Hodgin wrote: > Are there any algorithms which take a known-at-compile-time sequence > of bitwise operations on an 8-bit to 64-bit quantity, and optimize > them down to their minimal set of operations? > > For example, if I have an 8-bit byte and I want to swizzle the bits > thusly: > > Input: 07 06 05 04 03 02 01 00 > Output: 05 04 07 02 01 03 00 06 There is a lot of work, and many algorithms, for logic optimization, or minimization, that, for example, will find the optimal combination of NAND and NOR gates to evaluate some logical operation. This is used to compile Verilog or VHDL into either FPGA bit files or ASIC masks. On the other hand, logic minimization tends to believe that you can wire anything to anything else. That is, bit swizzle is just wiring. The question you ask doesn't come up. (Later on, routing has to get all the wires to the appropriate place, but that is true in general, not just for this case.) Note that it isn't just adjacent bits, but bits with the same spacing in the input and output, such that you can mask with AND, shift, and combine with OR. I suspect that with the operations usually available: AND, OR, XOR, and shift, it wouldn't be hard to find an algorithm for the optimal case. If you add more operations, maybe allow also add and multiply, it gets interesting.