Post Reply 
newRPL - Updated to build 1510 [official build remains at 1487]
12-15-2023, 04:46 AM
Post: #321
RE: newRPL - Updated to build 1510 [official build remains at 1487]
(11-21-2023 03:15 PM)Francois Lanciault Wrote:  Hi Claudio,

I wonder if you saw this post ? Not that it matters really, but I am intrigued.

François

(10-31-2023 02:55 PM)Francois Lanciault Wrote:  Hi,

I have installed newRPL on my Prime G1. All is fine. However I was a bit surprised to see that, running my usual RPL benchmark, newRPL on the Prime runs at roughly half the speed as it does on the HP-50G.

Considering the Prime processor clock is more then twice the speed than the processor on the 50G and more recent, I wonder what makes newRPL runs more slowly on the prime.

Moreover the Python interpreter on the same Prime runs an implementation of the same benchmark 6X faster than newRPL. Different language I agree, but my impression was that newRPL, being written in C, was very close to the host hardware.

Ideas ?

I don't know which benchmark it is but it's not possible to be slower on the Prime being the same code. My only guess is that perhaps your memory was dirty and a garbage collection was triggered during the test?? A GC cleaning up 32 MB of ram can take a while... if you post your benchmark I can run it on my 50g and Prime and we can compare results.
Find all posts by this user
Quote this message in a reply
12-15-2023, 05:00 AM
Post: #322
RE: newRPL - Updated to build 1510 [official build remains at 1487]
(11-27-2023 12:24 PM)Klaus Wrote:  I had to reinstall the stock ROM on my HP50 after using the NewRPL ROM for a long time. After the installation, I discovered that the HOME directory with all user files, all the libraries I had installed, and all the special settings of the computer were there as originally --- as if by magic. How is that possible? I always thought that when the NewRPL operating system was installed on the HP50, all memory would be rewritten, meaning everything would be deleted. Can someone explain this to me -- would be very grateful!
Greetings Klaus

The 50g stores Port0 and your HOME directory in the higher portions of your Ram. The "lower" parts here means the bottom of the 512k. The first 128kb are used by the Kinpo OS and the Saturn emulator, the second 128k are used for Port1 if I remember correctly, and only then the user visible memory starts. First system variables and then TempOb, and above that the stack. All of that memory so fart can be lost without any consequence. Above the stack is your Home directory and above that Port0.
NewRPL takes 4k pages and assigns it to different areas as needed, but starts taking memory from the bottom so if you used your calc lightly it is entirely possible that you never needed to use the higher portions of ram, and your data stayed there untouched. The TTRM prompt will aggressively try to recover your Home directory so don't be too surprised that it found it.
Find all posts by this user
Quote this message in a reply
12-18-2023, 02:36 PM
Post: #323
RE: newRPL - Updated to build 1510 [official build remains at 1487]
Visual Studio Code (VSC) is the source code editor that is increasingly used in computer programming education.

How complex would it be to create an newRPL plugin in Visual Studio Code?
Find all posts by this user
Quote this message in a reply
12-18-2023, 04:35 PM
Post: #324
RE: newRPL - Updated to build 1510 [official build remains at 1487]
(12-15-2023 04:46 AM)Claudio L. Wrote:  I don't know which benchmark it is but it's not possible to be slower on the Prime being the same code. My only guess is that perhaps your memory was dirty and a garbage collection was triggered during the test?? A GC cleaning up 32 MB of ram can take a while... if you post your benchmark I can run it on my 50g and Prime and we can compare results.

I ran the benchmark 12 times with different starting values (time of benchmark depends on starting value). All results were slower than HP-50G and it was a bit worst at every new iteration of the test (maybe a clue?)

How is the best way to send you the actual program ?

François
Find all posts by this user
Quote this message in a reply
12-27-2023, 03:19 AM (This post was last modified: 12-27-2023 03:20 AM by Claudio L..)
Post: #325
RE: newRPL - Updated to build 1510 [official build remains at 1487]
(12-18-2023 04:35 PM)Francois Lanciault Wrote:  
(12-15-2023 04:46 AM)Claudio L. Wrote:  I don't know which benchmark it is but it's not possible to be slower on the Prime being the same code. My only guess is that perhaps your memory was dirty and a garbage collection was triggered during the test?? A GC cleaning up 32 MB of ram can take a while... if you post your benchmark I can run it on my 50g and Prime and we can compare results.

I ran the benchmark 12 times with different starting values (time of benchmark depends on starting value). All results were slower than HP-50G and it was a bit worst at every new iteration of the test (maybe a clue?)

How is the best way to send you the actual program ?

François

A PM through the forum is fine or an email at my first name at hpgcc3.org.
Find all posts by this user
Quote this message in a reply
Post Reply 




User(s) browsing this thread: 1 Guest(s)