On 1 Jun 2021, Paul Rubin wrote
(in article <877djdfs8h.fsf@nightsong.com>):

> Bill Findlay<findlaybill@blueyonder.co.uk> writes:
> > > > The 20KSLOC compiler ran on a 1.5MIPS machine.
> > > Yes, but 1) 20KSLOC per what unit of time,
>
> Ok, but that's maybe 5x slower than Turbo Pascal, which compiled 1000s
> of LOC per second on machines of that class.

Well that is not what emerged in the conversation I reported.
The details are vague now, the gist was that the 1977 compiler
on comparable machines was several times faster that Turbo.
I see no point in pursuing this further.

--
Bill Findlay

On Tuesday, June 1, 2021 at 6:04:49 PM UTC-5, Bill Findlay wrote:
> On 1 Jun 2021, Paul Rubin wrote
> (in article <877djdf...@nightsong.com>):
> > Bill Findlay<findl...@blueyonder.co.uk> writes:
> > > > > The 20KSLOC compiler ran on a 1.5MIPS machine.
> > > > Yes, but 1) 20KSLOC per what unit of time,
> >
> > Ok, but that's maybe 5x slower than Turbo Pascal, which compiled 1000s
> > of LOC per second on machines of that class.
> Well that is not what emerged in the conversation I reported.
> The details are vague now, the gist was that the 1977 compiler
> on comparable machines was several times faster that Turbo.

May I ask what is meant by "comparable machines"?

Here's why I ask: it can't have been a machine based on the Intel 8088, because that wasn't available until 1979. An elderly embedded engineer I know says that the CPU used in the PC series, at least the early PC's (8088 & 286) was a terrible CPU. He likes to joke how his <1MHz 6809-based "Trash 80 Color Computer" at $500 could run circles around the >4MHz 8088-based IBM at $1500.

So I'm curious if you know the basis of the claim that it was comparable hardware: clock speed, RAM, etc.

> I see no point in pursuing this further.

I can understand, and I apologize if I shouldn't have asked, but I am genuinely curious about what was meant. I hope you don't mind.

On 2021-06-02 16:04, John Perry wrote:
> On Tuesday, June 1, 2021 at 6:04:49 PM UTC-5, Bill Findlay wrote:
>> On 1 Jun 2021, Paul Rubin wrote
>> (in article <877djdf...@nightsong.com>):
>>> Bill Findlay<findl...@blueyonder.co.uk> writes:
>>>>>> The 20KSLOC compiler ran on a 1.5MIPS machine.
>>>>> Yes, but 1) 20KSLOC per what unit of time,
>>>
>>> Ok, but that's maybe 5x slower than Turbo Pascal, which compiled 1000s
>>> of LOC per second on machines of that class.
>> Well that is not what emerged in the conversation I reported.
>> The details are vague now, the gist was that the 1977 compiler
>> on comparable machines was several times faster that Turbo.
>
> May I ask what is meant by "comparable machines"?

Around the end of 80's we used the dhrystone benchmark to compare machines.

> Here's why I ask: it can't have been a machine based on the Intel 8088, because that wasn't available until 1979. An elderly embedded engineer I know says that the CPU used in the PC series, at least the early PC's (8088 & 286) was a terrible CPU. He likes to joke how his <1MHz 6809-based "Trash 80 Color Computer" at $500 could run circles around the >4MHz 8088-based IBM at $1500.

That is my recollection too. I remember that an elderly 1MHz PDP-11
outperformed 12MHz 286. But the instruction set of PDP-11 was eons ahead
of the 286's mess.

> So I'm curious if you know the basis of the claim that it was comparable hardware: clock speed, RAM, etc.

It is simply so that C compilers were garbage that time. C is a
difficult language to compile comparing to Turbo Pascal, especially
using the methods that were popular then.

The only decent C compiler I remember from that time was DEC VAX C.

Even advanced machines like Motorola 68k, HP had horrific C compilers.
[Sun's SPARC C was somewhat better] The first thing to do was to
bootstrap an early GCC on these machines, because the standard compilers
were insufferable. I carried the sources with on a tape (maybe even on a
reel, I do not remember). That was fun.

--
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

"Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> writes:

> Even advanced machines like Motorola 68k, HP had horrific C compilers.

HP salesman: "Our compiler supports ANSI C (except for function
prototypes)"

On 2 Jun 2021, John Perry wrote
(in article<52640622-179a-40d3-a0c1-da113a8984f2n@googlegroups.com>):

> On Tuesday, June 1, 2021 at 6:04:49 PM UTC-5, Bill Findlay wrote:
> > On 1 Jun 2021, Paul Rubin wrote
> > (in article <877djdf...@nightsong.com>):
> > > Bill Findlay<findl...@blueyonder.co.uk> writes:
> > > > > > The 20KSLOC compiler ran on a 1.5MIPS machine.
> > > > > Yes, but 1) 20KSLOC per what unit of time,
> > >
> > > Ok, but that's maybe 5x slower than Turbo Pascal, which compiled 1000s
> > > of LOC per second on machines of that class.
> > Well that is not what emerged in the conversation I reported.
> > The details are vague now, the gist was that the 1977 compiler
> > on comparable machines was several times faster that Turbo.
>
> May I ask what is meant by "comparable machines"?

Machines that ran other programs at about the same speed,
I specifically had in mind the Whetstone and the Ackermann
function benchmarks.

> Here's why I ask: it can't have been a machine based on the Intel 8088,
> because that wasn't available until 1979.

The 20KSLOC/min compiler ran on an ICL 1906S,
which had a Whetstone rating of ~800 KWIPS.

> > I see no point in pursuing this further.
>
> I can understand, and I apologize if I shouldn't have asked, but I am
> genuinely curious about what was meant. I hope you don't mind.

I don't mind, it's just that I thought the lack of concrete
data makes further discussion a bit pointless.

--
Bill Findlay

> I was unable to find it.
> Can you provide the link?
> Or search terms?

"turbo pascal 6 source" -> (2nd link on Google) ->
https://archive.org/details/TP6SrcInst
"Zip" link:
https://archive.org/download/TP6SrcInst/%5BDOS%20Application%5D%20Borland%20-%20Turbo%20Pascal%206.0%20%28Sources%20included%29%20-%20Installation%20Disks%20%5Bhttp---usuarios.lycos.es-scratchupload%5D.zip

On Wednesday, June 2, 2021 at 12:51:14 PM UTC-6, gautier wrote:
> > I was unable to find it.
> > Can you provide the link?
> > Or search terms?
> "turbo pascal 6 source" -> (2nd link on Google) ->
> https://archive.org/details/TP6SrcInst
> "Zip" link:
> https://archive.org/download/TP6SrcInst/%5BDOS%20Application%5D%20Borland%20-%20Turbo%20Pascal%206.0%20%28Sources%20included%29%20-%20Installation%20Disks%20%5Bhttp---usuarios.lycos.es-scratchupload%5D.zip
Thank you.