Squeak-dev May 2008

squeak-dev@lists.squeakfoundation.org

141 participants
174 discussions

Dumb question about Hydra VM
by Bill Schwab 03 May '08

03 May '08

Hello all, In doing some catch-up reading, I found Igor's notes: http://squeakvm.org/%7Esig/hydravm/devnotes.html and am wondering whether (in particular) the FFIPlugin's support for the new VM means that one can call external functions on a new thread? Just clarifying before I get too enthusiastic. In particular, I am hoping for an equivalent to Dolphin's overlapped call mechanism, which tells the Dolphin VM to make an affected call using one[*] of a pool of native threads, blocking only the calling Process rather than the entire image for the duration of the call. Will Hydra offer similar capability? Bill [*] originally, the threads were chosen by availability, in D6, my understanding is that (more or less) the VM guarantees to make all overlapped calls for a given Process on the same native thread. Wilhelm K. Schwab, Ph.D. University of Florida Department of Anesthesiology PO Box 100254 Gainesville, FL 32610-0254 Email: bschwab(a)anest.ufl.edu Tel: (352) 846-1285 FAX: (352) 392-7029

2 1

Re: [squeak-dev] hydra vm update
by Igor Stasenko 03 May '08

03 May '08

Oops, didn't noticed that message sent only to John. ---------- Forwarded message ---------- From: Igor Stasenko <siguctua(a)gmail.com> Date: 2008/5/3 Subject: Re: [squeak-dev] hydra vm update To: johnmci(a)smalltalkconsulting.com 2008/5/3 John M McIntosh <johnmci(a)smalltalkconsulting.com>: > > > Consider, that my implementation was based on observations of what > > windows VM doing. And its already used a multimedia timer, which was > > set to trigger checking for interrupts each 1 msec (or at minimum time > > periods which OS can provide, but not less than 1mses). And these > > routines created own timer thread, hidden from the eyes of developer. > > So, what i did, i just replaced this thread by own implementation. > > Also, because of using multimedia timer, i seen an ovelap: an > > optimistic ( interruptCheckCounter handling) was doing the same as > > multimedia timer does, which, IMHO, can't be considered as a good > > algorithm. > > > > After refactoring, i got a code which can handle timer with better > > accuracy comparing to old VM. (Read a topic some time back about > > Delays accuracy). > > In fact, i was surprised, when seen, that my implementation provides > > more accurate timers, i expected it to be worser :) > > > > Ok, I'll have to run some benchmarks, when I changed the macintosh VM to > pound the interrupt delay > logic 1000 a second back in the era of 500 Mhz machines the impact on > performance was noticeable. > Maybe today no one cares, maybe the folks chasing the why does opening > windows take 2x as long > can fix that, then mmm we'll consume part of the gain back in overhead to > improve Delay accuracy. > My benchmarks showing opposite: the tinyBenchmarks runs faster with new model than with old checkForInterrupts. And Delay accuracy is improved. Maybe this can be not true for other platforms, but for Windows, i got gain in both areas, at a cost of implementing own timer thread routine. :) > Maybe I can clock watch before handleEvents() to avoid the overhead of a > timer routine. > > I note we use to have clock watching on each primitive call in ages past, > but removed it since we > found that under certain conditions one could spend % of time just getting > the clock, some vestiges > of that lurk via the non-existant lowres millisecond clock function. Maybe > it's not noticeable now. > > Ya, benchmarking first... > Also note, that HydraVM is targeted for multicore CPUs. In case if timer thread resides on different core than interpreter thread, there will be no scheduler overhead to switch active thread, and this makes delays handling even more accurate. Just run test again on quad-core box with same image: |delay bag| delay := Delay forMilliseconds: 1. bag := Bag new. 1000 timesRepeat:[bag add: [delay wait] timeToRun]. bag sortedCounts HydraVM: a SortedCollection(932->2 67->1 1->3) a SortedCollection(932->2 68->1) a SortedCollection(932->2 68->1) Croquet VM: a SortedCollection(952->2 48->1) a SortedCollection(951->2 48->1 1->4) a SortedCollection(951->2 46->1 3->3) This can be interpreted to one of following: - Hydra delays is more accurate - maybe its having similar accuracy (or even lower), but spends less time to signal semaphore, so we got noticeable increase in numbers with 1 msec results. No wonder, benchmarks on vanilla VM are faster : Croquet VM 1 tinyBenchmarks '485768500 bytecodes/sec; 14611978 sends/sec' '486229819 bytecodes/sec; 14520007 sends/sec' '485768500 bytecodes/sec; 14554360 sends/sec' [ 1 tinyBenchmarks ] timeToRun 5337 5336 5402 HydraVM: 1 tinyBenchmarks '454706927 bytecodes/sec; 13731345 sends/sec' '455516014 bytecodes/sec; 13363453 sends/sec' '456327985 bytecodes/sec; 13210400 sends/sec' '453900709 bytecodes/sec; 13373136 sends/sec' [ 1 tinyBenchmarks ] timeToRun 5692 5770 5796 Not sure, if using timeToRun is fair here, because it using timeToRun in code to determine one of the parameters. This actually shows an overhead of introducing interpreter as argument to each function. However, it would be interesting to add as option to build VM using thread-local storage to minimize impact of introducing multiple interpreter instances. As a bonus we'll have full compatibility with old primitives, because we don't need to pass interpreter instance as argument. But it was a design choice, and we decided to pass interpreter as extra argument instead of using thread-local storage. -- Best regards, Igor Stasenko AKA sig.

1 0

hydra vm update
by John M McIntosh 03 May '08

03 May '08

some observations (a) In the past we would call checkForInterrupts. One of the interesting things with that routine was that it used interruptCheckCounter to modulate how often it was called. Thus in the bytecode longUnconditionalJump /* begin internalQuickCheckForInterrupts */ if ((foo->interruptCheckCounter -= 1) <= 0) We tried only to call checkForInterrupts every 1 millisecond or so by incrementing, decrementing interruptCheckCounter to some large number. See my note on the squeak mailing list "VM tuning results and a question or two?" November 16, 1999 11:09:04 PM PST (CA) We would of course set that to a negative number if we wanted a more instant response. However now the HydraCode promptly does a return from interp on each jump backwards at some unknown cost, I'd guess this affects looping to some interesting value? (b) Buried in checkForInterrupts was code to thump TheTimerSemaphore when needed and to watch for millisecond clock rollover. But now it seems the platform implementer is now require to build a pthread based routine to mange the next wakeup time, and or do something clever elsewhere to watch when the millisecond clock rolls over and/or exceeds the next expected wakeup time. However that implies each implementation will have the same behaviour, where as before it was magically handled by the VM and not requiring lots of independently written operating system timer magic. As of now I don't have a workable vm since I need to implement this timer thread logic and one hopes that will solve the where we lockup the carbon event handler as something fails to process the draw window request at window activation properly. -- = = = ======================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com = = = ========================================================================

2 2

Multithreading and service initialization
by Norbert Hartl 02 May '08

02 May '08

Hi, my tests of the new rST Code gave me 6 out of 14 tests failing. It took me a while to narrow the problem. I built a vm which prints debug messages of the Socket- Plugin. I grabbed a piece of the log to show what I think went wrong: 5031 0 socketStatus(7) -> 0 5031 0 socketStatus(7) -> 0 5031 0 connectTo(7) 5031 0 connect() => -1 5033 2 connectHandler(7, 0x80ab158, 4) 5033 0 connectHandler: error 111 (Connection refused) 5033 0 notify 7 conn 5033 0 socketStatus(7) -> 0 5038 5 socketStatus(7) -> 0 5040 2 create(8) -> 80ab268 5040 0 socketStatus(8) -> 0 5040 0 socketStatus(8) -> 0 5040 0 listenOnPortBacklogSize(8, 10) 5040 0 socketStatus(8) -> 1 The problem appears that the client connects before the listen socket is bound (and accepting). The listen socket is created in a new thread (ApplicationService>>start). But the exection seems to be delay after the client connect call. This error is raised every time I tried it. What would be the best method to have multithreading and aio (I think the problems can be here, too) but to ensure a certain state is reached before continuing a dependent part of the application? I tried to fix it by delaying the client and this works but this not a problem solution. Norbert

1 0

Class>>deepCopy???
by Michael van der Gulik 02 May '08

02 May '08

Hi all. Why am I not meant to make a deep copy of a class? From the source for TraitBehavior>>deepCopy (which in 3.9 is the method that is invoked when you invoke #deepCopy on a class): deepCopy "Classes should only be shallowCopied or made anew." ^ self shallowCopy Ditto applies for Class>>copy. Surely, a shallow copy of a class is less useful than a deep copy because any modifications to the copy would also modify the original? Are there other reasons for copying a class other than to make a modifiable copy? Gulik. -- Michael van der Gulik <mikevdg(a)gmail.com>

1 0

Re: Debugging slow browsers
by Jerome Peace 02 May '08

02 May '08

Debugging slow browsers from thread: [squeak-dev] Squeak 3.9 alpha image wanted Hi Matthias, With your benchmark I started a search among my collection of 3.9 alphas. The slowdown started at update 6685..6686 and can be seen in the 6686 image. It is absent in the 6684 image. Details are on mantis: 2084 Each overlaping browser slows down screen updates considerably. Caching the image would speed things up. http://bugs.squeak.org/view.php?id=2084 Your guess seems likely to be correct. The translucent gradient was intoduced at this point. It probably requires everything below the browser to be redrawn. The pane-splitters were also introduced here. They proably add to the complexity of redrawing. Good luck in hunting down the solution. Yours in curiosity and service, --Jerome Peace ____________________________________________________________________________________ Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now. http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ

2 1

Re: [squeak-dev] Re: Undoable environment ?
by Frank Urbach 02 May '08

02 May '08

Hi Sophie, you can start Sophie in the normal way. After the startup is proceed go to the File-Menue and klick on exit. The sophie-system will be closed and you are in the squeak-system below sophie. There you find a package called Sophie-Undo. HTH Frank -------- Original Message -------- Subject: [squeak-dev] Re: Undoable environment ? (30-Apr-2008 14:49) From: itsme213 <itsme213(a)hotmail.com> To: frank.urbach(a)schmees.com > "John M McIntosh" <johnmci(a)smalltalkconsulting.com> wrote > > > There is an undo system that impara wrote for Sophie. You should consider > > it, > > Where will I find this? I downloaded Sophie and don't see image + sources. > > Thanks. > > > > > Edelstahlwerke Schmees GmbH Geschäftsleitung Clemens Schmees Sitz D-01796 Pirna Handelsregister Dresden HRB 54 E-Mail: es.la(a)schmees.com WEB: www.schmees.com Diese E-Mail enthält vertrauliche und/oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser Mail ist nicht gestattet. This e-mail may contain confidental and/or privileged information. If you are not intended recipient or have received this e-mail in error, please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.

3 3

Squeak 3.9 alpha image wanted
by Matthias Berth 01 May '08

01 May '08

Hello, can anyone tell me were to find the earliest available 3.9 alpha image? I want to identify the cause of the slowdown of browsers when you have many of them opened. http://article.gmane.org/gmane.comp.lang.smalltalk.squeak.general/108359 I am using the benchmark from Andreas. So far I see the problem in the current squeak-dev 3.9.1, Squeak3.9b-7029.image. Browser speed is OK in Squeak3.8-6665.image (Squeak3.8-6665-basic.zip). Cheers Matthias PS: Benchmark results on my machine: "Run this several times, do not close browser windows." win := Browser openBrowser topView. win extent: 600@350. br := win model. br setClassOrganizer. br systemCategoryListIndex: 2. World doOneCycle. Transcript cr; show: [1 to: 10 do:[:i| br classListIndex: i. World doOneCycle. ]] timeToRun. sq3.9.1-7075dev08.04.1.image 1880 1591 1420 1860 2550 3341 3262 3821 4396 4802 Squeak3.9-final-7067.image 07-Nov-2006 11:07 1001 1410 1842 2353 2794 3273 3809 4440 Squeak3.9b-7029.image 04-May-2006 20:54 1161 1793 2421 3063 3565 4004 4556 4987 Squeak3.8-6665.image Squeak3.8-6665-basic.zip 15-Jun-2005 942 918 930 945 975 985 1040 1076

5 9

Adobe opens (a little) the Flash specs
by Damien Pollet 01 May '08

01 May '08

http://www.adobe.com/aboutadobe/pressroom/pressreleases/200804/050108AdobeO… -- Damien Pollet type less, do more [ | ] http://people.untyped.org/damien.pollet

2 1

Squeak 3.9 alpha image wanted
by Jerome Peace 01 May '08

01 May '08

[squeak-dev] Squeak 3.9 alpha image wanted Hi Matthias, For debugging the most useful to image use is: http://ftp.squeak.org/images_with_all_changes/Squeak3.9a.from3.0.zip This is the final 3.8 image (6665) just after it has been bumped. It also has a complete version history from 3.0 onward. The last 3.9 alpha image before the changes were compressed and new sources produced is (7053). I think that got purged from the ftp site. Someone might know where you can find a version of it. I've also planted several issues on mantis around this topic. See: 6876 Why is my UI slow? http://bugs.squeak.org/view.php?id=6876 2084 Each overlaping browser slows down screen updates considerably. Caching the image would speed things up. http://bugs.squeak.org/view.php?id=2084 and Matthew Fulmer has added: 6890 PluggableListMorph is slow http://bugs.squeak.org/view.php?id=6890 I highly recommend you use mantis to describe your findings and solutions. Mantis helps track things and document them for the future. Then refer to your reports in your posts to this list. Yours in curiosity and service, --Jerome Peace *** >Matthias Berth matthias.berth at googlemail.com >Thu May 1 15:49:13 UTC 2008 > >Hello, > > >can anyone tell me were to find the earliest available 3.9 alpha image? > >I want to identify the cause of the slowdown of browsers when you have >many of them opened. > > http://article.gmane.org/gmane.comp.lang.smalltalk.squeak.general/108359 > >I am using the benchmark from Andreas. So far I see the problem in the >current squeak-dev 3.9.1, Squeak3.9b-7029.image. Browser speed is OK >in Squeak3.8-6665.image (Squeak3.8-6665-basic.zip). > >Cheers > >Matthias > ><details snipped> *** ____________________________________________________________________________________ Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now. http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ

1 0

Jump to page: