89C51ED2
in [Design]
|
From: MooseFET on 10 Sep 2008 22:49 On Sep 10, 6:35 pm, krw <k...(a)att.bizzzzzzzzzz> wrote: > In article <48C7E34F.2D624...(a)hotmail.com>, > rabbitsfriendsandrelati...(a)hotmail.com says... > > > > > > > krw wrote: > > > > rabbitsfriendsandrelati...(a)hotmail.com says... > > > > krw wrote: > > > > > > There is nothing wrong with PL/M, other than there is hasn't been > > > > > support for it for a quarter century. > > > > > What would it need support for ? > > > > Bugs (the OS variety, if nothing else). I don't use orphanware on > > > new projects. I don't need to add risk to projects. > > > Which bugs would those be ? The product is so mature it's untrue. > > Try reading. I can't take the risk of bugs cropping up in > unsupported software. If there are no bugs in the compiler (don't > believe it) Windows will install them. PLM-86 had a nasty bug that only appeared if you turned on the optimizer. If you declared a variable as double and then by mistake applied a double typecast to it, the optimizer would do a mismatched set of pushes and pops. It meant that your code would leap off into J- space at the ret. It was very hard to find.
From: MooseFET on 10 Sep 2008 22:52 On Sep 10, 8:21 am, Eeyore <rabbitsfriendsandrelati...(a)hotmail.com> wrote: > MooseFET wrote: > > krw <k...(a)att.bizzzzzzzzzz> wrote: > > > rabbitsfriendsandrelati...(a)hotmail.com says... > > > > krw wrote: > > > > > > You don't do subroutines or interrupts? > > > > > declare procedure XYZ interrupt(1) using 1; (register bank 1) MAIN defaults to > > > > register bank zero. > > > > > code > > > > > end; > > > > > For example. The interrupt number defines the int source. > > > > Return address? USING only declares the register bank, it doesn't > > > set it. > > > I think PLM always loads the PSW in an interrupt routine. > > That sounds right to me. It gets into the interrupt routine quite quickly.. > > > It has been a while since I looked at the resulting code form PLM but it rarely > > contains outright mistakes. > > Never seen one. > > > It sometimes does an unneeded operation or two. > > Possibly. I've timed a moderately complex interrupt that included some 16 bit math and > compares and it could complete it in 80us with a 12MHz cystal. That's the 6/12 cycle original > 80C51 design too. Use a new one @ 33MHz and the 2 cycle operations and it would be 5 us or so. Why stop at 33MHz. Use the Silabs 100MIPS 8051F120 and you'd be done before you knew you'd started. BTW: You really only get about 50 to 75 MIPS out of them.
From: MooseFET on 10 Sep 2008 22:57 On Sep 10, 6:36 pm, krw <k...(a)att.bizzzzzzzzzz> wrote: > In article <48C7E40F.1F3FA...(a)hotmail.com>, > rabbitsfriendsandrelati...(a)hotmail.com says... > > > > > > > krw wrote: > > > > rabbitsfriendsandrelati...(a)hotmail.com says... > > > > krw wrote: > > > > > > You don't do subroutines or interrupts? > > > > > declare procedure XYZ interrupt(1) using 1; (register bank 1) MAIN defaults to > > > > register bank zero. > > > > > code > > > > > end; > > > > > For example. The interrupt number defines the int source. > > > > Return address? USING only declares the register bank, it doesn't > > > set it. > > > Uh ? Bloody well does unless you have a different understanding of 'does' to me. > > Not when I was using Intel's software. USING was only a directive > to get the compiler to point R0-7 to the right place. It didn't > actually load the pointers selecting the bank. You remember wrong. USING tells the compiler what to assume is in the PSW.4, PSW.3 bits. In PLM it most likely also results in a move into the PSW. In the ASM51, you could run with no declared bank by not putting a USING into the code. This is handy when you want to make code that is bank independent. My 32 bit math library is bank independent so it can be used in interrupts etc. > > -- > Keith
From: Robert Baer on 11 Sep 2008 02:36 Eeyore wrote: > > krw wrote: > > >>rabbitsfriendsandrelations(a)hotmail.com says... >> >>>krw wrote: >>> >>> >>>>There is nothing wrong with PL/M, other than there is hasn't been >>>>support for it for a quarter century. >>> >>>What would it need support for ? >> >>Bugs (the OS variety, if nothing else). I don't use orphanware on >>new projects. I don't need to add risk to projects. > > > Which bugs would those be ? The product is so mature it's untrue. > > Graham > What did Atmel say about their AFU of that part?
From: Eeyore on 11 Sep 2008 05:11
krw wrote: > rabbitsfriendsandrelations(a)hotmail.com says... > > krw wrote: > > > rabbitsfriendsandrelations(a)hotmail.com says... > > > > krw wrote: > > > > > > > > > There is nothing wrong with PL/M, other than there is hasn't been > > > > > support for it for a quarter century. > > > > > > > > What would it need support for ? > > > > > > Bugs (the OS variety, if nothing else). I don't use orphanware on > > > new projects. I don't need to add risk to projects. > > > > Which bugs would those be ? The product is so mature it's untrue. > > Try reading. I can't take the risk of bugs cropping up in > unsupported software. If there are no bugs in the compiler (don't > believe it) Windows will install them. Ir runs under DOS. Graham |