#: 5473 S6/Applications 22-Jul-90 22:54:24 Sb: #5454-#Question on PhantomGraph Fm: JOHN TEAGUE 75715,1670 To: Zack Sessions 76407,1524 (X) Zack, Could you check that modpatch script again. When I used it the third line: c 02c3 0c 0d gave a "byte does not match" message. I tried to run phantomgraph anyway, but it locked up when I tried to print. There is 1 Reply. #: 5491 S6/Applications 23-Jul-90 21:25:20 Sb: #5473-Question on PhantomGraph Fm: Zack Sessions 76407,1524 To: JOHN TEAGUE 75715,1670 As Mark mentions, I have a typo in my message. It should be $02ce. Sorry. Zack #: 5477 S6/Applications 23-Jul-90 05:42:25 Sb: #5454-Question on PhantomGraph Fm: Mark Griffith 76070,41 To: Zack Sessions 76407,1524 (X) Zack, That should be $02ce in the third line of your patch for DMPTANDY.DRV. Mark #: 5476 S15/Hot Topics 23-Jul-90 04:44:17 Sb: GFX Speed Comparison Fm: Kevin Darling (UG Pres) 76703,4227 To: all Just for fun, I've run a speed comparison test... what I did was to GET an 80x40 block of gfx, and then PUT it across the screen in a matrix (the actual count of PUTs is close to 1400). All programs were in Basic. The results were: 15Mhz 68070 (MM/1) at 640x208x16-color - 6 seconds (prelim ver) 2Mhz 6809 (Coco3) at 640x192x 4-color - 490 seconds (old grfdrv) 2Mhz 6809 (Coco3) at 640x192x 4-color - 34 seconds (fast grfdrv) 2Mhz 6809 (Coco3) at 640x192x 4-color - 55 seconds (RSDOS Basic) Previous tests drawing lines and circles had also shown the same ratio between the 68K and L-II fast grfdrv... about 5:1. Now note that the MM/1 was moving twice as much data, because it was in 16-color mode vs 4-color on the CoCo. So I reran the fast grfdrv L-II test using 16-color mode (same number of PUTs) to be more fair... 2Mhz 6809 (Coco3) at 320x192x16-color - 54 seconds (fast grfdrv) Now the 68070 shows up as almost 10 times as fast. Starting both tests at the same time really shows this speed diff up... the MM/1 goes splat! splat! splat! onto the screen and finishes, while the coco continues for what seems like an eternity. And I had been so proud of my fast grfdrv Puts . #: 5479 S9/Utilities 23-Jul-90 08:41:56 Sb: #5466-#Reading OS9 Directorys Fm: Pete Lyall 76703,4230 To: George Hendrickson 71071,2003 (X) Right - the FD sector has all the info ABOUT the file (owner, size, dates, attributes). Best approach has been found to open two paths. One to the directory you're reading, and one to the whole RAW disk (i.e. /DD@). This way, when you read the LSN # from the directory, you can seek to the LSN * 256 on the RAW disk path and get right to the file's info. This is much faster than opening each file and doing an SS.FD getstat on it. Pete] There is 1 Reply. #: 5483 S9/Utilities 23-Jul-90 12:15:51 Sb: #5479-Reading OS9 Directorys Fm: George Hendrickson 71071,2003 To: Pete Lyall 76703,4230 (X) Oh I see..thanks for the help. I'll get right on it. Than I hope to make this program a good one. #: 5480 S15/Hot Topics 23-Jul-90 10:04:57 Sb: CoCo/OSK Basic(09) Fm: Ed Gresick 76576,3312 To: Mike Guzzi 76576,2715 (X) Hi Mike! Read your comments on Basic on the MM/1. Did not seem to jive with my experiences so I duplicated your tests. My CoCo is model 3 with 512k memory - basically a 'vanilla' machine. For the 68k tests, I have a PT68 model K4 which uses a 68000 chip, has 4 megs of memory and runs at 16 MHz. OS is version 2.3, Basic is version 2.1. I also had the tests run two 'VME' machines. One machine was a 68010 running at 12 MHZ. The operating system for this machine is Version 2.2 and the version of basic is 1.2. No other tasks were running. The second machine was a 68030 running at 25 MHz. The operating system for this machine is Version 2.3 and the version of basic is 2.3. This machine has a math coprocessor and an indeterminate number of other tasks were active. These are the results (in seconds) (I included your results on the MM/1 for comparison purposes.) CoCo MM/1 PT68K4 68010 68030 Conditions 4 3 1 3 1 As written (k=1000) 40 12 26 8 k=10000 78 59 23 52 14 k=20000 10 13 8 8 2 k=1000 (Changed integers to real) Note that in all cases the PT68 was faster than the CoCo. Marginally when using real numbers and substantially when using integers. (continued) #: 5481 S15/Hot Topics 23-Jul-90 10:06:18 Sb: #CoCo/OSK Basic(09) Fm: Ed Gresick 76576,3312 To: Mike Guzzi 76576,2715 (X) (continued from preceeding message) I have no explanation for the poor performance of the 68010 machine except that the version of basic09 was an old one. Have no idea as to its relative performace. But, it is interesting that it handled the case of real numbers as well as the PT68. The CoCo had no background tasks running. The PT68 had a 'cron-like' program running as a background task. I don't think this process influenced the tests significantly. And, I think the 1 MHz difference in clock speeds between the MM/1 and PT68 is insignificant when compared to the overall results. To be sure we're running the same tests, I used the basic09 'benchmark' program on page 196 of 'The Complete Rainbow Guide to OS-9'. On the 680x0 machines the line 'SHELL "date,t"' was changed to 'SHELL "date"'. It would be interesting to see a similar test run on Frank Hogg's machines; i.e., the 'tomcat', 'tc70', 'QT', etc. I curious as to the speed differences. I had 'heard' the 68070 is not as fast as the 68000 because of internal architectural differences. Do you know if this is true? Or, is the speed difference due to the version of OSK and Basic used? It might be interesting to continue this testing; i.e., running ML programs on the MM/1 and PT68K. If you'd like to devise some tests and send them to me, I'd be happy to run them. And, if you're nearby (I'm in Delaware) and would like to come by, I'd be happy to let you run any programs and tests you'd like (maybe bring your machine down and run side-by-side tests). Ed There are 2 Replies. #: 5485 S15/Hot Topics 23-Jul-90 15:42:45 Sb: #5481-CoCo/OSK Basic(09) Fm: Kevin Pease 70516,1633 To: Ed Gresick 76576,3312 (X) Ed The version of Basic that Mike is using is pretty old. Also the MM/1 will have the same speed as the TC70 in the MM/1 minimum configuration. The memory on the minimum MM/1 is shared with the video display and is much slower than if it were private. The memory is accesed every 533 ns instead of being accesed every 266 ns if there were no wait states. When the MM/1 is upgraded with the I/O board the speeds will improve significantly. I ran the drystone benchmark on my old PT68k and got around 950 drystones /second with a 12.5 MHz clock. The MM/1 with fast memory on the I/O board gets around 950 drystones - The MM/1 like make has will get around 600 drystones/second. The 68070 processor is definitly slower then the 68000. A 68070 at 15 MHz is the same general speed as a 68000 at 12.5 MHz. The TC70 should show verry close to the same speed. although I don't know much about memory speed on that board. Kevin Pease 70516,1633 #: 5488 S15/Hot Topics 23-Jul-90 19:41:30 Sb: #5481-#CoCo/OSK Basic(09) Fm: Kevin Darling (UG Pres) 76703,4227 To: Ed Gresick 76576,3312 (X) I added a test on a 3-meg MM/1 (2-megs fast RAM for program, 1-meg shared RAM for video) and got these times in seconds (slowest to fastest): 1000 10000 20000 1000 CPU ints ints ints real Computer Basic =========== ==== ==== ==== ==== =========== ===== 2Mhz 6809 - 4 40 78 10 CoCo 15Mhz 68070 - 3 30 59 19 MM/1 (1meg) 2.1 12Mhz 68010 - 3 26 52 8 VMEbus ? 1.2 15Mhz 68070 - 2 20 39 13 MM/1 (3meg) 2.1 16Mhz 68000 - 1 12 23 8 PT68K 2.1 25Mhz 68030 - 1 8 14 2 VMEbus ? 2.3 This shows the speed increase when not running out of video RAM on the MM/1. It also seems to show up an anomaly. Why did the 68010 do so well using REAL variables? Does the earlier Basic version use smaller data or is just a mistake in your chart? - kev There is 1 Reply. #: 5503 S15/Hot Topics 24-Jul-90 02:59:39 Sb: #5488-#CoCo/OSK Basic(09) Fm: Ed Gresick 76576,3312 To: Kevin Darling (UG Pres) 76703,4227 (X) Hi Kevin! We don't know the reason for the 68010 being so slow. The tests were run several times and we got the same results each time. We *suspect* the problem has to do with the version of basic used. We intend to retry with a later version of basic. Ed There is 1 Reply. #: 5504 S15/Hot Topics 24-Jul-90 03:25:28 Sb: #5503-CoCo/OSK Basic(09) Fm: Kevin Darling (UG Pres) 76703,4227 To: Ed Gresick 76576,3312 Ed - it wasn't that it was slow.... it was that the 68010 was so FAST on the REALs test. That is, not in line with its other results. As to why it was slow overall otherwise, I'd suspect slow memory? #: 5482 S14/misc/info/Soapbox 23-Jul-90 10:14:40 Sb: #5302-Serial Without Tears Fm: Tom Napolitano 70215,1130 To: James Jones 76257,562 (X) Sir, It probably just seems to work. All hardware, like all software, has at least one bug. i.e. The first bugfree program has yet to be written. You've heard the old saw: Every program has at least one bug, and every program can be shortened by one line. Therefore by regression every program can be reduced to one line of code.....that doesn't work. tom n #: 5484 S15/Hot Topics 23-Jul-90 15:41:57 Sb: #5450-MM/1 Conference Fm: Ed Gresick 76576,3312 To: Steve Wegert 76703,4255 (X) I plan on attending (but, that is past my bedtime - yawn) Ed #: 5486 S15/Hot Topics 23-Jul-90 18:02:24 Sb: #5450-MM/1 Conference Fm: Ron Lammardo 75706,336 To: Steve Wegert 76703,4255 (X) Steve, Will Try to make it!! Ron #: 5494 S15/Hot Topics 23-Jul-90 23:07:58 Sb: #5450-MM/1 Conference Fm: Jim Peasley 72726,1153 To: Steve Wegert 76703,4255 (X) Steve; Count me in... at least for a while... Wed. is bowling league nite. ..Jim O_o AACK! =( )= U #: 5513 S15/Hot Topics 24-Jul-90 14:40:10 Sb: #5450-MM/1 Conference Fm: Mike Guzzi 76576,2715 To: Steve Wegert 76703,4255 (X) I will try to make it but i work nights so i may be unable to attend. i may be able to jump in later on but its likely not at all.. so please somebody capture it for me??? thanks Mike #: 5515 S15/Hot Topics 24-Jul-90 16:11:55 Sb: #5450-MM/1 Conference Fm: Zack Sessions 76407,1524 To: Steve Wegert 76703,4255 (X) I'll be there. #: 5487 S6/Applications 23-Jul-90 19:38:29 Sb: #Speller Fm: Wendell Benedetti 72766,2605 To: Bob van der Poel 76510,2203 (X) Bob, Here's a comparison of your spell program and the User's Group Spell that I sent you: File Length Poel Spell User's Spell 77 bytes 54 seconds 6 seconds 675 57 18 1085 58 25 3267 64 47 10800 67 61 14300 70 76 19200 73 82 The User's Group program was faster for files with less then 12k bytes. Your program was faster with files larger than 12K. Wendell There is 1 Reply. #: 5501 S6/Applications 24-Jul-90 01:20:54 Sb: #5487-Speller Fm: Bob van der Poel 76510,2203 To: Wendell Benedetti 72766,2605 (X) Wendell, thanks for the info. I've been doing some more playing with the spell program and have incorporated some of the features form the program you sent me into it. I've managed to double the speed on a 3K and 8K file. I'll post it later this week. #: 5489 S8/BBS Systems/TSMon 23-Jul-90 20:53:01 Sb: BBS Fm: judd terrell 72217,714 To: 73617,3042 Wayne, just an update on Thermal Fusion BBS... Now running On-Line Games... OS-9 and limited non-OS-9 support for CoCo 1/2/3... Look forward to new bbs list when available... Current number for Thermal Fusion, (803) 862-7544... Fountain Inn, SC.... Thanx... Judd... #: 5490 S3/Languages 23-Jul-90 21:22:36 Sb: CLIB Docs Fm: Zack Sessions 76407,1524 To: 76070,41 (X) Mark, Was the popen() and pclose() functions intended to be included in your recent upload of the Krieder Lib docs? If so, then they are missing. Otherwise, NICE job! Zack #: 5492 S8/BBS Systems/TSMon 23-Jul-90 21:29:09 Sb: APBBS Editor question Fm: Zack Sessions 76407,1524 To: 76576,2715 Mike, I am taking a second look ar APBBS. Well, actually my first look, since when I first downloaded it a year ago, I never really looked at it. I have it about ready to come up, but am running into a minor glitch. You have some low level IO routines which you only include as a merged packed file. One of these routines, datin, I think, is used by the editor to input each line. Apparently, that routine don't like commas, cuz when I enter a message line which contains a comma (or commas), everything after the first comma is ignored. Is there a fix available for this? Zack #: 5493 S8/BBS Systems/TSMon 23-Jul-90 22:10:23 Sb: Art BBS wanted Fm: Bob Gale 71211,2616 To: All Does anyone know of any art oriented BBS? Any genre/media will do. Doesn't have to be an OS-9 system. Please send me E-Mail. Thanks in advance, ArtBase #: 5495 S10/Tandy CoCo 23-Jul-90 23:08:05 Sb: #5467-#erased file Fm: Jim Peasley 72726,1153 To: Kevin Darling (UG Pres) 76703,4227 (X) Kev; Many thanks. I was so close to the forest, the trees were blocking the view! :-> In my endeavors to fix up the file, I had completely overlooked the FD.SIZ bytes and was concentrating on the FD.SEG bytes. One thing that I wouldn't have gotten tho, is the fact that I needed to copy the repaired file and re-delete the entry. If I get some time this week, I'll write up a procedure file for repairing deleted files and upload it to you for proofreading (and corrections!). The best way for me to learn anything is to try and explain it to someone else! Gracias! ..Jim O_o AACK! =( )= U There are 2 Replies. #: 5497 S10/Tandy CoCo 24-Jul-90 00:15:09 Sb: #5495-erased file Fm: Kevin Darling (UG Pres) 76703,4227 To: Jim Peasley 72726,1153 (X) Great! Someone has needed to do that for quite some time (file on undeleting files - grin). It's the same way with me: if I can get to the point that I can explain something to someone, then I'll remember it better myself. #: 5498 S10/Tandy CoCo 24-Jul-90 00:15:45 Sb: #5495-erased file Fm: Kevin Darling (UG Pres) 76703,4227 To: Jim Peasley 72726,1153 (X) PS: y'know, for small files you could almost write a program to automate a lot of this. Care to try? ;-) #: 5496 S15/Hot Topics 23-Jul-90 23:08:13 Sb: #5470-#Keyboard/trackball Fm: Jim Peasley 72726,1153 To: Jim Williams 72157,3524 (X) Jim; re: $79 keyboard/trackball Forget it! It's chintzy... too light, didn't like the keyboard layout/feel, and the trackball is too close to the keypad and edge of the board. No room for ham-hands like mine. I did play with one trackball for a while tho, and I really liked it. It was a 3 button that had buttons #1 and #2 wrapping around the ball dlmost the full length of the pad, and button #3, the drag-lock centered at the top. Very easy to use and well designed ergonomically. Good control too, using variable cpi. It looked as though the faster yousmoved the ball, the more cpi that is fed to the driver - coarse control; and the slower you moved it, the less cpi. If I remember, it was something like 50-360 cpi. ..Jim O_o AACK! =( )=0 U There is 1 Reply. #: 5517 S15/Hot Topics 24-Jul-90 22:51:00 Sb: #5496-Keyboard/trackball Fm: Jim Williams 72157,3524 To: Jim Peasley 72726,1153 (X) Thanks for the info, I'll consider alternatives to that keyboard, then . --Eet-- Director of Mayhem Extra Terrestrial Imports, Ltd. #: 5499 S12/OS9/68000 (OSK) 24-Jul-90 00:28:40 Sb: #5439-Reading OS9 directorys Fm: Don Kircher 76346,3475 To: Kevin Darling (UG Pres) 76703,4227 (X) Thanks for the reply Kevin, I'll give it a whirl and see how she flys dlk #: 5500 S10/Tandy CoCo 24-Jul-90 01:18:04 Sb: #5464-words Fm: Bob van der Poel 76510,2203 To: PHIL SCHERER 71211,2545 Phil, I just had a look at the source for "words" and it looked fine. So I tried the program, and I too got a 216. What happened is that the compiled program in the archive (as well as the one on my hard drive) is an earlier version of words. All you need to do is either recompile the program, or use the version you have but do a "chd /dd/sys/spell" first. The version you have is looking for the file "dictionary", not "/dd/sys/spell/dictionary". BTW, I will be uploading a new version of "spell" later this week (as soon as my 2400 baud modem arrives). It is twice as fast again as the one you now have. #: 5502 S3/Languages 24-Jul-90 01:21:14 Sb: #5347-Clib docs Fm: Bob van der Poel 76510,2203 To: Mark Griffith 76070,41 (X) Mark, There really are some too-long lines in the docs. I'm using an 80 column printer in pica. If you are too then something is really strange. Regarding the numbers at the top of the pages, I should have figured out the Unix connection myself. After all (I'm learning bit by bit) if there is something cryptic relating to OS9 then is usually a Unix influence. Just what was the mother-tongue of the folks who created Unix anyway??? Oh, and thanks for the reminder about the index--I guess I can just print out the man.hlp file and stick it in the back of my new manual. Should work fine. I'll leave defdrive() for now--I'm not sure why anyone would ever use that command anyway? Guess it is a hold-over from before /dd conventions? Oh, I just noticed that the description for RAND() is still wrong. Unless I am missing something in the way the function works it does not use a scale factor. It would be nice if it did! Could this be a difference between clib.l and clibt.l? #: 5505 S10/Tandy CoCo 24-Jul-90 11:37:19 Sb: #readmac.pak Fm: Joseph Cheek 76264,142 To: sysop (X) Sorry, it happened again. I uploaded readmac.pak into lib 10 and forgot to say that it requires KD's enhanced GFX2. Could you *Please?* include it in my description? Thank you very much . . . There is 1 Reply. #: 5506 S10/Tandy CoCo 24-Jul-90 12:28:57 Sb: #5505-readmac.pak Fm: Wayne Day 76703,376 To: Joseph Cheek 76264,142 (X) Yep.. we'll take care of it, Joseph. Wayne #: 5507 S9/Utilities 24-Jul-90 12:32:01 Sb: #Reading Directorys Fm: George Hendrickson 71071,2003 To: Pete Lyall 76703,4230 (X) I've been working on my utility and got it to read the directory and print the filenames and the LSN numbers. I set it up to read the directory using TYPE. Here is how I did it. 'TYPE direntry=name:STRING[29]; lsn0,lsn1,lsn2:BYTE' When I tell it to print each filename, after resetting the last character of the filename, and print the LSN's 0-2, this is what it looks like. e.g. FILENAME LSN0=0 LSN1=119 LSN2=203 Which one of those do I use to 'SEEK' to the filenames FD sector and get my info? Do I have to multiply the lsn's together to get that sector number? And if so, what's the formula? (I'm getting closer! This is fun!) OS9 has opened all new doors to me...amazing stuff! There is 1 Reply. #: 5508 S9/Utilities 24-Jul-90 13:58:47 Sb: #5507-Reading Directorys Fm: Pete Lyall 76703,4230 To: George Hendrickson 71071,2003 George - Yep - you're on the trail.... here's the scoop: The LSN is a 3 byte number. Basically, the breakdown is - MSB * 256 * 256 + NSB * 256 + LSB ======================== LSN # In C, there are functions for taking these three bytes and making a LONG INTEGER out of them (and back). Under B09, you may have to use a type REAL, since the result is greater than +/- 32767. Pete #: 5509 S15/Hot Topics 24-Jul-90 14:30:53 Sb: #5402-Is Basic out of date? Fm: Mike Guzzi 76576,2715 To: Kevin Darling (UG Pres) 76703,4227 (X) ~ Well my BBS in Basic09 does need a PEEK for carrier since ACIAPAK has no way to tell it for me. but its not hard-coded so the user may use a pak thats addressed differently then the standard pak. Even though this isn't "clean" code I never have a problem multi-tasking or anything like that. The modem kill switch (type=20) is ucreliable and is a shock to the system. so i let my BBS handle it. Mike #: 5510 S15/Hot Topics 24-Jul-90 14:31:08 Sb: #5409-Is Basic out of date? Fm: Mike Guzzi 76576,2715 To: Mark Griffith 76070,41 (X) ~ Well I too take the stand against "bad programs" and my BBS is clean as it can get with the exception of peeking for carrier.. thats all! I allow users to imbed the base address in a file so he may use other paks and other ports but I had no choice but to do it that way. sure if I toyed with it i could do other features but ii would be nothing less then dirty. Alot of Alpha's BBS owners complained about it... sure he has some slick features I don't have but the cost is system ram, crashes, and other erratic operations... which would you rather have??? Mike #: 5514 S10/Tandy CoCo 24-Jul-90 15:51:39 Sb: Sound.Master Problems Fm: JIM MCDOWELL 70721,435 To: 76370,1366 I will need more information to be certain as to the problem you are experiencing. Here are some things to check: - SSC and SSCART should be placed on your boot disk (with Config, Ezgen, etc.) To determine if they are working try this: iniz ssc echo hi >/ssc If you get an error these files are not working properly. - Be sure RunB and GFX2 are in memory. Preferably merged together. merge RunB Gfx2 >RunbX attr Runbx e pe load Runbx You can even Runb GFX2 and Sound.Master together.. - Be sure to load Sound.Master into memory. It won't always run correctly by just typing in its name. load Sound.Master sound - Be sure XMODE is in memory, although this shouldn't give an error 43 if missing. From your description it sounds like something is missing in memory that needs to be there so check that first. The program should draw more than just three small screens. It begins with 3 in the upper left corner one in the upper right corner one in the lower left and lower right corner and one in the middle. If it fails before completing this then perhaps one of the draw files are missing. Here is a list of the files included in sound.master: 128 $22 $AEE898 sound 53 $22 $7E9959 cursor 53 $22 $D6120E menu 133 $22 $18D62D filemenu 53 $22 $6C0FF1 process 53 $22 $C4A015 switch 53 $22 $2E6AF4 putregs 52 $22 $1E94CB getname 53 $22 $09D119 getfile 35 $22 $14B9CE tone 29 $22 $F7D2B4 slider 35 $22 $577141 connect 35 $22 $94594E envelope 132 $22 $7980D7 env1 35 $22 $6FF97D noise 129 $22 $469828 ltrbox 132 $22 $623956 spkr 35 $22 $729CE6 menubox 53 $22 $99B7CD pointer 53 $22 $49E1DC pal 14 $22 $2E555B alpha 53 $22 $9EA968 tables 19 $22 $E8C3A5 registers Do an IDENT sound.master -sv to double check this. Its a long reply but I hope it helps. Please let me know if you still have any problems. If you do put together a file with an MDIR and MFREE and IDENTS for sound progs then mail them to me. This with a detailed description will make it easier to pinpoint the A(X. #: 5518 S4/MIDI and Music 24-Jul-90 23:01:29 Sb: #5238-midi help Fm: Lester Hands 70135,430 To: Pete Lyall 76703,4230 The fact that all the software is included is really amazing. Makes me wonder if it is because Microware said "here it is--do what you can with it". Sounds like a wonder machine! #: 5519 S4/MIDI and Music 24-Jul-90 23:05:15 Sb: #5399-midi help Fm: Lester Hands 70135,430 To: Mike Knudsen 72467,1111 Tsk, tsk, Mike! Yeah, my dogs are well fed. Incidentally, the dogs are also doing a much better job of making my time worth while! Perhaps not as elegant as the MM/1 but my AT is supported by a wealth of software. And yes, PC-Lyra has been out for almost 8 months now. Hope to get some advertising going soon. How come I don't see anything in the latest Rainbow for Second City Software? A friend of mine even said that he tried calling them and got a message saying that the number was disconnected! Are you going full steam on the MM/1? Sounds like a dream machine. I just wish that it had been released 5 years ago. Life might be very different now. Lester. #: 5520 S4/MIDI and Music 24-Jul-90 23:08:51 Sb: #5398-midi help Fm: Lester Hands 70135,430 To: Mike Knudsen 72467,1111 No, Mike. The friends who can tell the difference between 96 and 192 ticks per beat were doing it with stock synths. But I heartily agree with you, that some synths are very sloppy when it comes to responding quickly to MIDI input. Press !>