Ok, but then how does Emscripten handles C/C++ code that deals with int64_t/uint64_t values?
On Wednesday, November 2, 2016 at 3:53:34 PM UTC-4, Brion Vibber wrote: > > JavaScript can't natively handle 64-bit integers, and there's no > Int64Array type to read from. > > If you know the value will be in the range of 53 bits, or don't care that > precision will be lost at larger magnitudes, then it can be handled as a > double-precision float on the JS side: > > var low = Module.HEAP32[ptr / 4]; > var high = Module.HEAP32[(ptr + 4) / 4]; > var double = high * Math.pow(2, 32) + low; > > -- brion > > On Wed, Nov 2, 2016 at 12:22 PM, Robert Goulet <robert...@autodesk.com > <javascript:>> wrote: > >> Question about Emscripten HEAP access, is there a way to read 64bit >> integer signed/unsigned values? I see HEAPF64 and HEAP32, etc., but no >> HEAP64? >> >> Thanks! >> >> >> On Wednesday, September 21, 2016 at 5:23:48 PM UTC-4, Robert Goulet wrote: >>> >>> Thanks guys, the code works as expected now. I didn't get back as much >>> performance as I hoped but its faster nonetheless. Cheers! >>> >>> On Wednesday, September 21, 2016 at 5:10:16 PM UTC-4, Charles Vaughn >>> wrote: >>>> >>>> Yeah, I usually end up screwing up the pointer shifts. >>>> >>>> Another thing to keep in mind is if you used packed structs (don't >>>> recommend that for Emscripten built code), you'll need to do something >>>> like >>>> DataView. >>>> >>>> On Wednesday, September 21, 2016 at 1:14:05 AM UTC-7, jj wrote: >>>>> >>>>> You can also go via the HEAP objects, so >>>>> >>>>> event.time = HEAPF64[addr>>3]; >>>>> event.name = Pointer_stringify(HEAP32[addr+8 >> 2]); >>>>> event.thread_id = HEAP32[addr+12 >> 2]; >>>>> event.type = HEAP8[addr + 16]; >>>>> event.core_id = HEAP8[addr + 17]; >>>>> >>>>> 2016-09-21 0:06 GMT+03:00 Charles Vaughn <cva...@gmail.com>: >>>>> >>>>>> If you're struct is packed (or you know the padding points) and >>>>>> contiguous, you can use a DataView ( >>>>>> https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/DataView >>>>>> >>>>>> ) to wrap a slice of the underlying buffer of the various HEAPs. >>>>>> >>>>>> Something like: >>>>>> >>>>>> var view = new DataView(HEAP8.buffer.slice(addr)); >>>>>> event.time = view.getDouble(0, true); >>>>>> event.name = Pointer_stringify(view.getUint32(8, true)); >>>>>> event.thread_id = view.getInt32(12, true); >>>>>> event.type = view.getInt8(16, true); >>>>>> event.core_id = view.getInt8(17, true); >>>>>> >>>>>> >>>>>> On Tuesday, September 20, 2016 at 1:41:56 PM UTC-7, Robert Goulet >>>>>> wrote: >>>>>>> >>>>>>> The JS code looks like this: >>>>>>> >>>>>>> event.time = getValue(addr, 'double'); >>>>>>> event.name = Pointer_stringify(getValue(addr + 8, '*')); >>>>>>> event.thread_id = getValue(addr + 12, 'i32'); >>>>>>> event.type = getValue(addr + 16, 'i8'); >>>>>>> event.core_id = getValue(addr + 17, 'i8'); >>>>>>> >>>>>>> since the different elements of the structure varies in size, I >>>>>>> guess I have to use different HEAP methods? >>>>>>> >>>>>>> On Tuesday, September 20, 2016 at 4:36:47 PM UTC-4, jj wrote: >>>>>>>> >>>>>>>> Directly accessing HEAP32 will definitely be much faster than the >>>>>>>> very generic getValue() function. If you don't need any of the >>>>>>>> genericity >>>>>>>> of what getValue() offers, using HEAP32 is definitely recommended. >>>>>>>> >>>>>>>> 2016-09-20 23:27 GMT+03:00 Robert Goulet <robert...@autodesk.com>: >>>>>>>> >>>>>>>>> Thanks jj, I ended up using getValue on the JS side to get the >>>>>>>>> data from the pointer I pass from C. Is there any performance >>>>>>>>> concerns with >>>>>>>>> this or should I use HEAP32 instead? >>>>>>>>> >>>>>>>>> On Tuesday, September 20, 2016 at 12:33:01 PM UTC-4, jj wrote: >>>>>>>>>> >>>>>>>>>> The src/library_xxx.js files are generally good examples. >>>>>>>>>> >>>>>>>>>> Here's one snippet where C function passes a pointer to an >>>>>>>>>> integer array and length of that array to JS side, and JS code reads >>>>>>>>>> through the array: >>>>>>>>>> https://github.com/kripken/emscripten/blob/master/src/library_openal.js#L329. >>>>>>>>>> >>>>>>>>>> If not using JS code that lives in js-libraries, the i32 {{{ >>>>>>>>>> makeGetValue >>>>>>>>>> }}} can be replaced with a direct HEAP32[pointer >> 2]. >>>>>>>>>> >>>>>>>>>> Another example with filling a struct in JS side: >>>>>>>>>> https://github.com/kripken/emscripten/blob/master/src/library_html5.js#L180 >>>>>>>>>> >>>>>>>>>> and reading the fields from a pointer to a struct: >>>>>>>>>> https://github.com/kripken/emscripten/blob/master/src/library_html5.js#L1728 >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> 2016-09-20 16:45 GMT+03:00 Robert Goulet <robert...@autodesk.com> >>>>>>>>>> : >>>>>>>>>> >>>>>>>>>>> Do you have an example of sending pointer into EM_ASM and >>>>>>>>>>> reading it directly from memory? >>>>>>>>>>> >>>>>>>>>>> In my case I am calling EM_ASM close to a thousand times to pass >>>>>>>>>>> engine profiling data to javascript for drawing on the web page, so >>>>>>>>>>> I am >>>>>>>>>>> trying to avoid adding time to the profiling result. If EM_ASM does >>>>>>>>>>> add >>>>>>>>>>> overhead, then I hope to reduce it by calling it only once instead >>>>>>>>>>> of a >>>>>>>>>>> thousand times per frame. I profiled it to about ~2.5ms per frame >>>>>>>>>>> to do >>>>>>>>>>> these thousand calls to EM_ASM, which is a lot if you consider the >>>>>>>>>>> actual >>>>>>>>>>> frame time is <= 17ms. >>>>>>>>>>> >>>>>>>>>>> On Monday, September 19, 2016 at 5:45:21 PM UTC-4, Alon Zakai >>>>>>>>>>> wrote: >>>>>>>>>>>> >>>>>>>>>>>> The most efficient way is to send the pointer into EM_ASM, then >>>>>>>>>>>> do reads directly to memory using the right offsets, but that >>>>>>>>>>>> requires >>>>>>>>>>>> using information about how the data is laid out in memory (on the >>>>>>>>>>>> plus >>>>>>>>>>>> side, the alignment rules are the natural 32-bit ones, with fully >>>>>>>>>>>> aligned >>>>>>>>>>>> doubles). >>>>>>>>>>>> >>>>>>>>>>>> Otherwise multiple calls into EM_ASM adds overhead, but in many >>>>>>>>>>>> cases it wouldn't be noticeable. >>>>>>>>>>>> >>>>>>>>>>>> On Mon, Sep 19, 2016 at 1:57 PM, Robert Goulet < >>>>>>>>>>>> robert...@autodesk.com> wrote: >>>>>>>>>>>> >>>>>>>>>>>>> Hi, >>>>>>>>>>>>> >>>>>>>>>>>>> How do we pass an array of objects to Javascript function from >>>>>>>>>>>>> C? >>>>>>>>>>>>> >>>>>>>>>>>>> Consider the following example: >>>>>>>>>>>>> >>>>>>>>>>>>> struct data { >>>>>>>>>>>>> double a; >>>>>>>>>>>>> int b; >>>>>>>>>>>>> unsigned char c; >>>>>>>>>>>>> }; >>>>>>>>>>>>> >>>>>>>>>>>>> std::vector<data> my_data; >>>>>>>>>>>>> >>>>>>>>>>>>> EM_ASM_ARGS({ >>>>>>>>>>>>> var data_array = ??? >>>>>>>>>>>>> process_data(data_array); >>>>>>>>>>>>> }, my_data); >>>>>>>>>>>>> >>>>>>>>>>>>> Is this possible? I couldn't find any clear documentation >>>>>>>>>>>>> about this topic. >>>>>>>>>>>>> >>>>>>>>>>>>> For the moment I've used the following workaround, but it >>>>>>>>>>>>> doesn't look super efficient: >>>>>>>>>>>>> >>>>>>>>>>>>> for( auto const & i : my_data ) { >>>>>>>>>>>>> EM_ASM_ARGS({ >>>>>>>>>>>>> process_data($0, $1, $2); >>>>>>>>>>>>> }, i.a, i.b, i.c); >>>>>>>>>>>>> } >>>>>>>>>>>>> >>>>>>>>>>>>> Thanks! >>>>>>>>>>>>> >>>>>>>>>>>>> -- >>>>>>>>>>>>> You received this message because you are subscribed to the >>>>>>>>>>>>> Google Groups "emscripten-discuss" group. >>>>>>>>>>>>> To unsubscribe from this group and stop receiving emails from >>>>>>>>>>>>> it, send an email to >>>>>>>>>>>>> emscripten-discuss+unsubscr...@googlegroups.com. >>>>>>>>>>>>> For more options, visit https://groups.google.com/d/optout. >>>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> -- >>>>>>>>>>> You received this message because you are subscribed to the >>>>>>>>>>> Google Groups "emscripten-discuss" group. >>>>>>>>>>> To unsubscribe from this group and stop receiving emails from >>>>>>>>>>> it, send an email to >>>>>>>>>>> emscripten-discuss+unsubscr...@googlegroups.com. >>>>>>>>>>> For more options, visit https://groups.google.com/d/optout. >>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> -- >>>>>>>>> You received this message because you are subscribed to the Google >>>>>>>>> Groups "emscripten-discuss" group. >>>>>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>>>>> send an email to emscripten-discuss+unsubscr...@googlegroups.com. >>>>>>>>> For more options, visit https://groups.google.com/d/optout. >>>>>>>>> >>>>>>>> >>>>>>>> -- >>>>>> You received this message because you are subscribed to the Google >>>>>> Groups "emscripten-discuss" group. >>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>> send an email to emscripten-discuss+unsubscr...@googlegroups.com. >>>>>> For more options, visit https://groups.google.com/d/optout. >>>>>> >>>>> >>>>> -- >> You received this message because you are subscribed to the Google Groups >> "emscripten-discuss" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to emscripten-discuss+unsubscr...@googlegroups.com <javascript:>. >> For more options, visit https://groups.google.com/d/optout. >> > > -- You received this message because you are subscribed to the Google Groups "emscripten-discuss" group. To unsubscribe from this group and stop receiving emails from it, send an email to emscripten-discuss+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
