Post by thread:[PIC]: DAW how it works (BCD)

Hi to all, I did some test using DAW instruction on MPLAB ICE2000 and this is the result. Movlw 0x0F ; w reg has d=15: B=00001111 DAW ; do decimal adjustments Movwf BCD_REG ; BCD_REG has d=21 b=00010101 C=0 Second digit (F) 05 First digit (0) + 01 Result 015 ; 0 means C=0: 1 is bits 4,5,6,7 and 5 is bits 0,1,2,3 Now lets deal with higher numbers Movlw 0xEF ; w reg has d=239: b=11101111 DAW ; do decimal adjustments Movwf BCD_REG ; BCD_REG has d=85 h=55 b=10101010 Second digit (F) 15 ; decimal First digit (E) + 14 ; decimal Result 155 ; 1 means C=1: 5 is bis 4,5,6,7 and 5 is bits 0,1,2,3 Andre Abelian -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email spam_OUTlistservTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

Thanks for setting that up and testing, it's appreciated. I don't know how hard it is to set up but you might try the following; DC= 1 C = 1 W = 32 (This would be the result of adding 99 + 99) Then execute the 'DAW The result should be (according to Microchip for the 18F series) DC= 1 C = 0 W= 98 Using a 18F452, I arrived at the results (as above) that Microchip said you should get. I found a question nearly identical to mine in the Microchip Developers conference and their answer was as above. I still don't know why but I asked. Maybe I'll get a response in a day or two. asp.microchip.com/webboard/wbpx.dll/~devtools Search for DAW Thanks again Mike Jones {Original Message removed}

Mike, 99+99=198 in decimal if Wreg=198 after DEW your Wreg is 98 C=1 DC=0 Andre Abelian -----Original Message----- Thanks for setting that up and testing, it's appreciated. I don't know how hard it is to set up but you might try the following; DC= 1 C = 1 W = 32 (This would be the result of adding 99 + 99) Then execute the 'DAW The result should be (according to Microchip for the 18F series) DC= 1 C = 0 W= 98 Using a 18F452, I arrived at the results (as above) that Microchip said you should get. I found a question nearly identical to mine in the Microchip Developers conference and their answer was as above. I still don't know why but I asked. Maybe I'll get a response in a day or two. asp.microchip.com/webboard/wbpx.dll/~devtools Search for DAW Thanks again Mike Jones ----- Original Message ----- From: "Andre Abelian" <.....engelecKILLspam@spam@EARTHLINK.NET> To: <PICLISTKILLspamMITVMA.MIT.EDU> Sent: Tuesday, February 04, 2003 2:46 PM Subject: [PIC]: DAW how it works (BCD) {Quote hidden}> Hi to all, > > I did some test using DAW instruction on MPLAB ICE2000 > and this is the result. > > Movlw 0x0F ; w reg has d=15: B=00001111 > DAW ; do decimal adjustments > Movwf BCD_REG ; BCD_REG has d=21 b=00010101 C=0 > > > Second digit (F) 05 > First digit (0) + 01 > Result 015 ; 0 means C=0: 1 is bits 4,5,6,7 and > 5 is bits > 0,1,2,3 > > > Now lets deal with higher numbers > > Movlw 0xEF ; w reg has d=239: b=11101111 > DAW ; do decimal adjustments > Movwf BCD_REG ; BCD_REG has d=85 h=55 b=10101010 > > > Second digit (F) 15 ; decimal > First digit (E) + 14 ; decimal > Result 155 ; 1 means C=1: 5 is bis 4,5,6,7 and > 5 is bits > 0,1,2,3 > > > Andre Abelian > > -- > http://www.piclist.com#nomail Going offline? Don't AutoReply us! > email .....listservKILLspam.....mitvma.mit.edu with SET PICList DIGEST in the body > -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email EraseMElistservspam_OUTTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservspam_OUTmitvma.mit.edu with SET PICList DIGEST in the body

Andre, My apologies, I should have been a little more clear. When I said W=32 I was thinking, but not writing, packed BCD 32, or 0x32. I'm not really sure how that should be written. However, the status bits DC and C should both be set to '1' before the DAW in order to simulate the results an addition (ADDWFC or similar) of two packed BCD numbers, 99.+ 99 in this case. BTW I heard from Microchip already, not much but they said they will look into it. Not sure what that means but it's a start. Mike Jones ----- Original Message ----- From: "Andre Abelian" <@spam@engelecKILLspamEARTHLINK.NET> To: <KILLspamPICLISTKILLspamMITVMA.MIT.EDU> Sent: Tuesday, February 04, 2003 3:59 PM Subject: Re: [PIC]: DAW how it works (BCD) {Quote hidden}> Mike, > > If Wreg = d'32' that is 0x20 > After DAW every thing remains the same d'32' and C=0 DC=0 > The only time DC will be 1 when goes over d'15' and > C will be 1 if your number goes over 99. > > Andre Abelian > > > > {Original Message removed}

part 1 4939 bytes content-type:text/plain; (decoded 7bit) Let me try this again. I probably wasn't clear enough in explaining what I was really getting at. This is probably nothing new to most but it may help some avoid at least one pitfall. What I finally learned is that, for example, if you are adding two multi digit decimal numbers in packed BCD format (such as 9990 + 9990) and are using DAW to adjust the results after each 'add', you need to save the carry after each add and restore it after each DAW (or at least before the next 'add'). If you don't, the final results may not be what you expect. In re-reading the description of the DAW opcode, it only says that the C bit is affected, not how or why. Apparently the C bit is simply a result of the additions that took place when performing the adjust, not a result of the previous addition (like ADDWFC) as you 'might' have liked or expected. In an attached file are a couple of tested examples that, I hope, illustrate the above. BTW Example 1 was confirmed by Microchip in a response to a question I found on their web site. The original question was "Why is this Binary to BCD Conversion" macro not working correctly?". www.piclist.org/techref/postbot.asp?by=time&id=piclist/2002/12/24/164 925a I applied a rather brute force method of saving and restoring the carry bit as per the above and it now works as expected. Now I need to find a more elegant way to save and restore the carry than with the cycle eating brute-force method I tested the 'fix' with. Mike Jones ----- Original Message ----- From: "Andre Abelian" <RemoveMEengelecTakeThisOuTEARTHLINK.NET> To: <spamBeGonePICLISTspamBeGoneMITVMA.MIT.EDU> Sent: Tuesday, February 04, 2003 4:52 PM Subject: Re: [PIC]: DAW how it works (BCD) {Quote hidden}> Mike, > > 99+99=198 in decimal > if Wreg=198 > after DEW > your Wreg is 98 C=1 DC=0 > > Andre Abelian > > > > {Original Message removed} part 2 1363 bytes content-type:text/plain; (decoded 7bit) (In order for this to be readable you may need to change to a fixed width font such as courier). Example 1: Add two packed BCD numbers 99+99 (0x99 + 0x99) OPCODE Register Value DC C --------------- -------- --------- -- - WREG 1001 1001 FileX 1001 1001 x 0 ADDWFC FileX,W WREG 0011 0010 1 1 DAW WREG 1001 1000 1 0 ^ Note the carry -----------^ Example 2: Add two packed BCD numbers 70+70 (0x70 + 0x70) OPCODE Register Value DC C --------------- -------- --------- -- - WREG 0111 0000 FileX 0111 0000 x 0 ADDWFC FileX,W WREG 1110 0000 1 1 DAW WREG 0100 1000 0 1 ^ Note the carry -----------^ Notes: This was tested on a 18F452. FileX is an arbitrary register in Bank 0 -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics part 3 2 bytes -

I am pretty sure you are going to have to 'or' the carry from the ADDWFC with the one from the DAW. You're second text case added 0x70 to 0x70 and then DAW'd The ADDWFC should _NOT_ generate a carry (contrary to the information in your attachment). 0x70 + 0x70 = 0xE0 with _NO_ carry. Then the DAW should fix it up to: 0x40 with a carry. You need that carry into the next byte. What a stupid way to design the DAW instruction. It should have set carry as needed, but never cleared it. This really screws up the performance of my macro, that is for sure. Here is the best I have come up with for saving and merging carry. ADDWFC var,x ; add two bytes RRCF temp,f ; remember the carry DAW ; adjust the value BTFSC temp,7 ; was carrry set? SETC ; yes: set it again Bob Ammerman RAm Systems -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

>Considering long BCD numbers that was DAA or DAW developed for, the carry >never is set when you add two proper BCD numbers. >(99 + 99 = 198). Sorry, this is not true. When you are adding BCD 99 + BCD 99 you are really adding Hex 99 to Hex 99 or decimal 153 + decimal 153, which can obviously generate a carry. >DAW or DAA in 8086 or other mnemotechnics used years ago, even in IBM 360 >worked by the same way. If I remember well, COBOL, PL/1 and early versions >of Pascal used so called long BCD numbers that saves memory and were enough >long for accounting. >Who needs such arithmetics today... BCD is very convenient if entering an displaying decimal values is an important part of the project. >Regards >Igor Bob Ammerman RAm Systems -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Bob, Correct me if I am wrong, but I don't think this will work reliably; > > Here is the best I have come up with for saving and merging carry. > > ADDWFC var,x ; add two bytes > RRCF temp,f ; remember the carry > DAW ; adjust the value > BTFSC temp,7 ; was carrry set? > SETC ; yes: set it again > The RRCF will also shift a bit into the carry bit that may or may not be what you want daw to make a decision on whether to add 6 to the upper nibble or not. This initially looked good but when I tried it I got some really strange results until I figured out what as going on. The following is one instruction cycle longer but seems to work reliably. ADDWFC var,x ; add two bytes MOVFF status, temp ; remember the carry DAW ; adjust the value BTFSC temp,0 ; was carrry set? SETC ; yes: set it again It sure would have been nice if they had made Status Affected: None! Mike Jones > -- > http://www.piclist.com hint: The list server can filter out subtopics > (like ads or off topics) for you. See http://www.piclist.com/#topics > -- http://www.piclist.com hint: To leave the PICList TakeThisOuTpiclist-unsubscribe-requestEraseMEspam_OUTmitvma.mit.edu>