Starting a clean slate soon, and Dendou Oni is not ousted...
Dr. Mario
Posts: 331
I have decided to be myself for a bit, just to get my mind in one place, so I may start cleanly on the much-anticipated project, Dendou Oni. I may have to delete my other threads, so I can start fresh on this project, because for some reason, the Octo-core CPU cores (COGs) still appeals to me - also a reminder that I should build a supercomputer. I also had to get better PC hardware, like having the right mouse for circuit board CAD usage (ball mouse is out, since they jump all the time).
Now, I am trying to focus on it a bit more carefully. And, I apologize for discouraging you guys.
Now, I am trying to focus on it a bit more carefully. And, I apologize for discouraging you guys.
Comments
For a good mouse, try a wired or Apple wireless optical mouse with a fine meshed fabric quality mouse pad. It should give the smooth movements for CAD drawings. Also use settings to fine tune it. Load in a high quality brand name driver for a high quality brand name mouse. I'm currently using a Logitech on the laptops and a wireless Apple mouse on the desktop computer. I purchased a mid range priced Logitech and the left and right microswitches have too light of a touch, i.e. half the time they are accidentally pressed. It's best to try before you buy.
It's fine to let the old thread die. Just don't post to them.
I look forward to seeing what your new project is.
Duane
Now, to restart my project (I still have archives of my Dendou Oni project - may try different supercomputing arrangements), I may try exploring the usage of Microchip PIC32MX microcontroller or X-something single-core transputer as an operator to control eight P8X32A to fill their thread request - something like VLIW processor on steroids, only on macroscopic scale. With the PCB CAD, I finally got to ditch my old ball mouse and got Miocrosoft BlueTrack mouse (as it was cheap - $25 at Staples) and I will be doing a slightly different RAM modeling, moving SDRAM to the operator instead, keeping the P8X32A from slowing down - wanting all COGs free to do away with flat memory modeling.
Also, I had high hope for Prop II as it might be a superscalar processor, doing away with the programs quickly, and crunching numbers faster. No Spin? I am fine with it as there is bound to have Spin interpreter update which could occur more often, so no point in burning it onto firmware ROM. Unfortunately, as I knew there's bound to be some delay (like with AMD FX Zambezi - it have been having infancy mortality problems due to some issues with 32nm node process - so little room for mistakes, no wonder the yields have been so low!) - I was correct in a way...
Dr. Mario,
I don't see any issues with using the "X" chip or whatever chip suits your needs, the usual issue is when some people go on and on about the "X" chip in a "X" vs. Prop manner.
So if your project uses "X" or PIC or AVR or whatever works best in conjunction with the Propellers, I think the people here would be interested in the project.
C.W.
I (and I think others) just don't think it appropriate to denegrade the propeller against other micros on these threads. All micros have their place.
If another micro forms a function together with the propeller chip, IMHO that is a different matter, provided the "other" chip is not the focus of the thread to the detriment of the propeller.
For instance, jazzed and I both have ATTiny projects that are a "peripheral" to the propeller. IMHO there is nothing wrong with that discussion, provided we do not go overboard in promoting the ATTiny on this forum. So, our use is just like using any 74xxXXX or I2C chip etc with the propeller.
Also, after digging down on the dirty details of the original transputer itself (not XS1, but before that), I decided to simplify the hardware requirement, moving the SDRAM onto the operator CPU so I don't bog any P8X32A chips down doing compulsory obsessive jobs doing otherwise complicated timing and step-by-step RAM transaction - that will be for Prop II later as pin numbers do matter now, as I want to be able to shoot up to 1MB/s (or if I can do away with complicated Manchester-like coding, 20MB/s) through those P8X32A, to be able to cut computational time a bit - won't be impressive but it's a prototype project, though.
I am thinking about going to Optical bus or just use good old Ethernet copper interconnection through root switch, finally linking to a CPU. I might do FE (Fast Ethernet) link as it could be much cheaper to do so to be able to zip the bits throughout the systems.