- Pharo-dev - lists.pharo.org

Re: [Pharo-dev] [ANN] Fuel 1.9.3

by Tudor Girba

Thanks a lot for this effort! Doru On Wed, Feb 26, 2014 at 10:13 PM, Max Leske <maxleske(a)gmail.com> wrote: > Dear fellow Smalltalkers > > Mariano, Martin and myself are happy to announce Fuel version 1.9.3 which > includes the follwing changes to 1.9 (1.9.1 and 1.9.2 were never officially > announced): > > - (feature) the #fuelReplacement selector offers the ability to statically > replace an object (e.g. with nil) during analysis. This can lead to > significantly smaller graphs and improved speed when serializing large > graphs > - (feature) serialize the same graph that was analyzed instead of > retrazing the graph during serialization. This prevents changes in the > graph from happening between analysis and serialization > - (change) store source when serializing CompiledMethod objects. Needed > because Opal does not allow decompilation > - (fix) migrate variables across hierarchy > - (feature) Object>>serializeToFileNamed: shortcut for serializing > arbitrary objects > - (fix) better compression for LargeNegativeIntegers ( > https://pharo.fogbugz.com/f/cases/12052/Fuel-could-store-LargeNegativeInteg… > ) > > Pharo30 already includes most of these changes (integration request: > https://pharo.fogbugz.com/f/cases/13000/Integrate-Fuel-1-9-3) > > This release runs with: > - Pharo: 1.1.1, 1.1.2, 1.2, 1.2.2, 1.3, 1.4, 2.0, 3.0 > - Squeak: 4.1, 4.2, 4.3, 4.4, 4.5 > > The documentation is not up to date yet but we will remedy that situatoin > within the next couple of days ( > http://rmod.lille.inria.fr/web/pier/software/Fuel) > If you encounter any bugs or have suggestions, please contact us on the > Fuel mailing list ( > http://lists.gforge.inria.fr/cgi-bin/mailman/roster/pharo-fuel) or create > a new issue in our bug tracker (https://code.google.com/p/fuel/) > > We want to thank everyone who contributed to this release (by keyboard or > opinion) and also all of those people who actively use Fuel. > > Cheers, > Max > -- www.tudorgirba.com "Every thing has its own flow"

Feb. 27, 2014

Re: [Pharo-dev] [Pharo-users] Pharo 3.0 on Mac OS X 10.9

by Tudor Girba

Nice. Thanks! Doru On Wed, Feb 26, 2014 at 11:52 PM, Sven Van Caekenberghe <sven(a)stfx.eu>wrote: > Pharo 3.0 on Mac OS X 10.9 > > The out of the box experience. Installing and running Pharo 3.0 on a blank > Mac OS X 10.9 account in 40 seconds. > > https://www.youtube.com/watch?v=xxljgugI4c4 > > With 3 extra minutes to demo some networking and image manipulation > interactions. > > Sven > > -- > Sven Van Caekenberghe > Proudly supporting Pharo > http://pharo.org > http://association.pharo.org > http://consortium.pharo.org > > > > > > -- www.tudorgirba.com "Every thing has its own flow"

Feb. 27, 2014

Re: [Pharo-dev] [fun] Sparkline in Pharo

by Tudor Girba

Nice :) Doru On Wed, Feb 26, 2014 at 9:40 PM, Alexandre Bergel <alexandre.bergel(a)me.com>wrote: > Hi! > > I am playing with unicode characters: > > -=-=-=-=-=-=-=-=-=-=-=-= > Gofer new > url: 'http://smalltalkhub.com/mc/abergel/Sparkline/main'; > addPackage: 'Sparkline'; > load. > -=-=-=-=-=-=-=-=-=-=-=-= > > #( 1 2 3 4) spark => 'âââââ > #(-5 10 0) spark => 'âââ' > RxsNode withAllSubclasses spark: #numberOfMethods => 'âââââââââââââ > > > I got the inspiration from https://github.com/holman/spark > > By the way, I would love to have a short tutorial about unicode characters > at esug. > > Cheers, > Alexandre > -- > _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: > Alexandre Bergel http://www.bergel.eu > ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > > > > -- www.tudorgirba.com "Every thing has its own flow"

Feb. 27, 2014

Re: [Pharo-dev] hashMultiply

by Andres Valloud

Hello again... > For what occurs in modern CPUs, well, no. Surprising to see that a mul > would be faster than a shr or shl. How comes? No, not faster. But definitely way faster than >30 cycles not so long ago. > I used to be ok with these things when I was writing demoscene code a > loong time ago but I'd need an extremely serious refresh. You said "demoscene"?!... do you have a handle??? Andres. > As a side note, there is a huge uptake in the BigData/mapreduce/hadoop > environment where Smalltalk is sorely absent. Scala seems to fill the > void on the JVM. > There hashing is quite key, to remap all of the mapping phase results to > the reduce nodes. I am surprised to see that Smalltalk vendors haven't > jumped in that space. > > Phil > > > On Wed, Feb 26, 2014 at 2:04 AM, Andres Valloud > <avalloud(a)smalltalk.comcastbiz.net > <mailto:avalloud@smalltalk.comcastbiz.net>> wrote: > > Hello... > > On 2/25/14 1:17 , phil(a)highoctane.be <mailto:phil@highoctane.be> wrote: > > I am currently reading through the Hashing in Smalltalk book > (http://www.lulu.com/shop/__andres-valloud/hashing-in-__smalltalk-theory-and… > <http://www.lulu.com/shop/andres-valloud/hashing-in-smalltalk-theory-and-pra…>__) > and, my head hurting notwithstanding, there are indeed a ton of > gems in > this system. As he mentions, doing the exercises brings a lot of > extra :-) > > > :) thank you. > > When going to 64-bit, and with the new ObjectMemory scheme, I > guess a > couple of identity hashing functions will come under scrutiny. > > e.g. > > SmallInteger>>hashMultiply > | low | > > low := self bitAnd: 16383. > ^(16r260D * low + ((16r260D * (self bitShift: -14) + (16r0065 * low) > bitAnd: 16383) * 16384)) > bitAnd: 16r0FFFFFFF > > > which will need some more bits. > > > IMO it's not clear that SmallInteger>>identityHash should be > implemented that way. Finding a permutation of the small integers > that also behaves like a good quality hash function and evaluates > quickly (in significantly less time and complexity than, say, Bob > Jenkins' lookup3) would be a really interesting research project. I > don't know if it's possible. If no such thing exists, then getting > at least some rough idea of what's the minimum necessary complexity > for such hash functions would be valuable. > > Looking at hashMultiply as a non-identity hash function, one would > start having problems when significantly more than 2^28 objects are > stored in a single hashed collection. 2^28 objects with e.g. 12 > bytes per header and a minimum of one instance variable (so the hash > value isn't a instance-constant) stored in a hashed collection > requires more than 4gb, so that is clearly a 64 bit image problem. > In 64 bits, 2^28 objects with e.g. 16 bytes per header and a > minimum of one instance variable each is already 6gb, and a minimum > of 8gb with the hashed collection itself. > > Because of these figures, I'd think improving the implementation of > hashed collections takes priority over adding more non-identity hash > function bits (as long as the existing hash values are of good quality). > > Did you look at the Hash Analysis Tool I wrote? It's in the Cincom > public Store repository. It comes in two bundles: Hash Analysis > Tool, and Hash Analysis Tool - Extensions. With everything loaded, > the tool comes with 300+ hash functions and 100+ data sets out of > the box. The code is MIT. > > I had a look at how it was done in VisualWorks; > > > The implementation of hashMultiply, yes. Note however that > SmallInteger>>hash is ^self. > > hashMultiply > "Multiply the receiver by 16r0019660D mod 2^28 > without using large integer arithmetic for speed. > The constant is a generator of the multiplicative > subgroup of Z_2^30, see Knuth's TAOCP vol 2." > <primitive: 1747> > | low14Bits | > low14Bits := self bitAnd: 16r3FFF. > ^16384 > * (16r260D * (self bitShift: -14) + (16r0065 * low14Bits) > bitAnd: 16r3FFF) > + (16r260D * low14Bits) bitAnd: 16rFFFFFFF > > The hashing book version has: > > multiplication > "Computes self times 1664525 mod 2^38 while avoiding overflow into a > large integer by making the multiplication into two 14 bits > chunks. Do > not use any division or modulo operation." > | lowBits highBits| > > lowBits := self bitAnd: 16r3FFF. > highBits := self bitShift: -14. > ^(lowBits * 16r260D) > + (((lowBits * 16r0065) bitAnd: 16r3FFF) bitShift: 14) > + (((highBits * 16r260D) bitAnd: 16r3FFF) bitShift: 14) > bitAnd: 16rFFFFFFF > > So, 16384 * is the same as bitShift: 14 and it looks like done once, > which may be better. > > > It should be a primitive (or otherwise optimized somehow). At some > point though that hash function was implemented for e.g. ByteArray > in Squeak, I thought at that point the multiplication step was also > implemented as a primitive? > > Also VW marks it as a primitive, which Pharo does not. > > > In VW it is also a translated primitive, i.e. it's executed directly > in the JIT without calling C. > > Keep in mind that the speed at which hash values are calculated is > only part of the story. If the hash function quality is not great, > or the hashed collection implementation is not efficient and induces > collisions or other extra work, improving the efficiency of the hash > functions won't do much. I think it's mentioned in the hash book > (I'd have to check), but once I made a hash function 5x times slower > to get better quality and the result was that a report that was > taking 30 minutes took 90 seconds instead (and hashing was gone from > the profiler output). > > Would we gain > some speed doing that? hashMultiply is used a lof for identity > hashes. > > Bytecode has quite some work to do: > > 37 <70> self > 38 <20> pushConstant: 16383 > 39 <BE> send: bitAnd: > 40 <68> popIntoTemp: 0 > 41 <21> pushConstant: 9741 > 42 <10> pushTemp: 0 > 43 <B8> send: * > 44 <21> pushConstant: 9741 > 45 <70> self > 46 <22> pushConstant: -14 > 47 <BC> send: bitShift: > 48 <B8> send: * > 49 <23> pushConstant: 101 > 50 <10> pushTemp: 0 > 51 <B8> send: * > 52 <B0> send: + > 53 <20> pushConstant: 16383 > 54 <BE> send: bitAnd: > 55 <24> pushConstant: 16384 > 56 <B8> send: * > 57 <B0> send: + > 58 <25> pushConstant: 268435455 > 59 <BE> send: bitAnd: > 60 <7C> returnTop > > > If this is a primitive instead, then you can also avoid the overflow > into large integers and do the math with (basically) > > mov eax, smallInteger > shr eax, numberOfTagBits > mul eax, 1664525 "the multiplication that throws out the high bits" > shl eax, 4 "throw out bits 29-32" > shr eax, 4 > lea eax, [eax * 2^numberOfTagBits + smallIntegerTagBits] > > Please excuse trivial omissions in the above, it's written only for > the sake of illustration (e.g. it looks like the 3 last instructions > can be combined into two... lea followed by shr). Also, did you see > the latency of integer multiplication instructions in modern x86 > processors?... > > I ran some experiments timing things. > > It looks like that replacing 16384 * by bitShift:14 leads to a small > gain, bitShift (primitive 17) being faster than * (primitive 9) > > > Keep in mind those operations still have to check for overflow into > large integers. In this case, large integers are not necessary. > > Andres. > > The bytecode is identical, except send: bitShift instead of send: * > > multiplication3 > | low | > > low := self bitAnd: 16383. > ^(16r260D * low + ((16r260D * (self bitShift: -14) + (16r0065 * low) > bitAnd: 16383) bitShift: 14)) > bitAnd: 16r0FFFFFFF > > > > [500000 timesRepeat: [ 15000 hashMultiply ]] timeToRun 12 > [500000 timesRepeat: [ 15000 multiplication ]] timeToRun 41 (worse) > [500000 timesRepeat: [ 15000 multiplication3 ]] timeToRun 10 > (better) > > It looks like correct for SmallInteger minVal to: SmallInteger > maxVal > > Now, VW gives: [500000 timesRepeat: [ 15000 hashMultiply ]] > timeToRun > 1.149 milliseconds > > Definitely worth investigating the primitive thing, or some NB > Asm as > this is used about everywhere (Collections etc). > > Toughts? > > Phil > > >

Feb. 27, 2014

Re: [Pharo-dev] hashMultiply

by Andres Valloud

Hello... On 2/25/14 18:32 , Martin McClure wrote: > Andres and I have an ongoing discussion on this topic. :-) :)... > And #bitShift *should* be faster. Ok, but how much faster? 64x64 bit multiplication on modern x86 is just 8 cycles (!), and there will likely be parallelism done by the CPU anyway... > But test this. In VW, multiplication > is quite a lot faster than bitShift: for some reason. Unless Andres has > fixed this recently. :-) I haven't measured to see what's going on. In general terms though, unlike with multiplication, on top of the overflow check you also have the argument overflow check because on x86 shl eax, 65 is the same as shl eax, 1 which is obviously not the behavior you want. So that means extra cmp, extra jmp, and you also have to deal with positive / negative arguments to select shl or shr/sar so more cmp and jmp, etc... Note though that IIRC the manual says that for SSE registers, doing something like shl xmm0, 129 is the same as xor xmm0, xmm0 (assembler for the sake of illustratio only) which is nice because then you can combine the overflow check with the argument overflow check. The whole SSE thing though is much, much more than just bitShift. Andres.

Feb. 27, 2014

Fwd: [squeak-dev] Fwd: Floats problem with SIXX between Pharo and GemStone

by Nicolas Cellier

---------- Forwarded message ---------- From: Nicolas Cellier <nicolas.cellier.aka.nice(a)gmail.com> Date: 2014-02-27 0:15 GMT+01:00 Subject: Re: [squeak-dev] Fwd: [Pharo-dev] Floats problem with SIXX between Pharo and GemStone To: The general-purpose Squeak developers list < squeak-dev(a)lists.squeakfoundation.org> 2014-02-26 23:38 GMT+01:00 Tobias Pape <Das.Linux(a)gmx.de>: Can we get this into trunk please? > Ie, > > SqNumberPasrser>>exponentLetters > "current comment" > ^'edqEDQ' > > > Best > -Tobias > > You mean modifying Smalltalk syntax? That's a possible option, and it would work in Pharo too. Note that 1.0E2 was previously interpreted as send #E2 to 1.0, so it might change... ...not much thing : the problem is highly hypothetical, we don't use a leading uppercase letter in selectors usually, and we also put a space before a unary or keyword message - one rare exception I encountered was Complex "constants" like 1+2i ... Begin forwarded message: > > *From: *Nicolas Cellier <nicolas.cellier.aka.nice(a)gmail.com> > *Subject: **Re: [Pharo-dev] Floats problem with SIXX between Pharo and > GemStone* > *Date: *26. Februar 2014 23:14:10 MEZ > *To: *Pharo Development List <pharo-dev(a)lists.pharo.org> > *Reply-To: *Pharo Development List <pharo-dev(a)lists.pharo.org> > > > > > 2014-02-26 23:08 GMT+01:00 Nicolas Cellier < > nicolas.cellier.aka.nice(a)gmail.com>: > >> That's exactly for this kind of reasons that I developped an >> ExtendedNumberParser connected to Number class>>readFrom:, to allow a >> variety of formats used in the rest of the world. >> >> It would have been easy to allow upper letter exponents in this parser, >> but the class was considered superfluous and removed from Pharo 3.0... >> >> super easy is just redefine: > > ExtendedNumberParser>>exponentLetters > ^'edqEDQ' > > and then Float readFrom: '1.0000000000000000E-02' works as expected, that > is in Squeak at least... > > > > > > >

Feb. 26, 2014

Re: [Pharo-dev] iPad version does not work ?[SEC=UNCLASSIFIED]

by Stephen Taylor

On Wed, 2014-02-26 at 08:28 +0100, phil(a)highoctane.be wrote: > You can make it work following the procedure: > http://www.youtube.com/watch?v=7MNsUiCc5FQ&list=UUul18dYQnhOVMfYFg6YSK4A Excellent - I'll watch. > Phil Steve

Feb. 26, 2014

Pharo 3.0 on Mac OS X 10.9

by Sven Van Caekenberghe

Pharo 3.0 on Mac OS X 10.9 The out of the box experience. Installing and running Pharo 3.0 on a blank Mac OS X 10.9 account in 40 seconds. https://www.youtube.com/watch?v=xxljgugI4c4 With 3 extra minutes to demo some networking and image manipulation interactions. Sven -- Sven Van Caekenberghe Proudly supporting Pharo http://pharo.org http://association.pharo.org http://consortium.pharo.org

Feb. 26, 2014

Re: [Pharo-dev] [Glass] Floats problem with SIXX between Pharo and GemStone

by Martin McClure

On 02/26/2014 01:38 PM, Mariano Martinez Peck wrote: > > So, of course " Float readFrom: '1.0000000000000000E-02' " -> 1.0 This is because Pharo only accepts lower-case letters for exponent ($e $d $q). So if you replace E with e Pharo will parse it as you wish. -Martin

Feb. 26, 2014

Re: [Pharo-dev] Floats problem with SIXX between Pharo and GemStone

by Nicolas Cellier

2014-02-26 23:08 GMT+01:00 Nicolas Cellier < nicolas.cellier.aka.nice(a)gmail.com>: > That's exactly for this kind of reasons that I developped an > ExtendedNumberParser connected to Number class>>readFrom:, to allow a > variety of formats used in the rest of the world. > > It would have been easy to allow upper letter exponents in this parser, > but the class was considered superfluous and removed from Pharo 3.0... > > super easy is just redefine: ExtendedNumberParser>>exponentLetters ^'edqEDQ' and then Float readFrom: '1.0000000000000000E-02' works as expected, that is in Squeak at least... > 2014-02-26 22:38 GMT+01:00 Mariano Martinez Peck <marianopeck(a)gmail.com>: > > Hi guys, >> >> When I export SmallDoubles with SIXX in GemStone it generates something >> like this: >> >> <sixx.object sixx.id="15" sixx.name="close" sixx.type="SmallDouble" >> >1.0000000000000000E-02</sixx.object> >> >> This is because "0.01 asString" -> '1.0000000000000000E-02' >> >> Then I want to load this in Pharo. So first problem is that of course >> SmallDouble doesn't exist. So I did: >> >> aSixxShapeChangeStream shapeChangers at: #SmallDouble put: Float. >> But then, SIXX use Float >> readFrom: aStream >> "Answer a new Float as described on the stream, aStream." ^(super >> readFrom: aStream) asFloat >> >> >> So, of course " Float readFrom: '1.0000000000000000E-02' " -> 1.0 >> >> So my 0.01 in gemstone become 1.0 in Pharo.... >> >> Am I doing something wrong or this is expected? What is the best >> workaround? >> >> In GemStone I could print the SmallDouble in the XML with a method like >> >> Float >> xmlRepresentation >> ^ String streamContents: >> [:strm | self printOn: strm base: 10] >> >> Is this the correct approach? If true, I guess I should define: >> >> Float >> sixxContentOn: aStream indent: level context: dictionary aStream >> nextPutAll: self xmlRepresentation >> >> >> Thoughts? >> >> Thanks in advance, >> >> -- >> Mariano >> http://marianopeck.wordpress.com >> > >

Feb. 26, 2014