+retroclouds Posted October 24, 2021 Share Posted October 24, 2021 Would it be possible to access (read and write) the original 16K VDP memory RAM from the F18a GPU? Guess not, at least not for writing. But asking anyway, you never know. Would be a cool for a future F18a firmware revision to access the 16K VDP memory on the main board as a second bank of VDP memory. Quote Link to comment Share on other sites More sharing options...
Tursi Posted October 24, 2021 Share Posted October 24, 2021 No, it's one way. The VDP is a slave device and has no access to the rest of the system. It would not be possible to use system memory as VDP RAM, the access time through the CPU is too slow. But, if you wanted the GPU to be get memory from the host system, you could do this in your software. The F18A updater code I wrote actually does this... the GPU does its work till it needs more data, then it writes to a specific address to ask the host system for specific data. The 9900 is just sitting in a loop waiting for this request, and when it gets the request, it fetches the desired data and copies it into VDP RAM, then tells the GPU it's ready. 4 Quote Link to comment Share on other sites More sharing options...
Stuart Posted October 24, 2021 Share Posted October 24, 2021 15 hours ago, retroclouds said: Would be a cool for a future F18a firmware revision to access the 16K VDP memory on the main board as a second bank of VDP memory. The F18a hasn't got the pins fitted to connect to the VRAM on the main board, so no firmware revision could make that happen. 1 Quote Link to comment Share on other sites More sharing options...
+OLD CS1 Posted October 24, 2021 Share Posted October 24, 2021 1 hour ago, Stuart said: The F18a hasn't got the pins fitted to connect to the VRAM on the main board, so no firmware revision could make that happen. As well, ISTR Matt saying there are no more GPIO pins available to even make it physically possible. 1 Quote Link to comment Share on other sites More sharing options...
Tursi Posted October 25, 2021 Share Posted October 25, 2021 I see I mis-read the original intent... it makes more sense to me now. But, the other challenges @Stuart and @OLD CS1 note still exist. 1 Quote Link to comment Share on other sites More sharing options...
+FarmerPotato Posted October 25, 2021 Share Posted October 25, 2021 Doing it all in 16K is one of the creative limitations with the F18A! You want to use all the features—3bpp, separate sprite pats, two scrolling pages, BML! But you have to allocate budget 16K and make compromises. I find that it gets me thinking, and the trade offs are still fun, more fun than being stuck with 9918A. 1 Quote Link to comment Share on other sites More sharing options...
matthew180 Posted October 27, 2021 Share Posted October 27, 2021 As noted, not possible. The primary reason is the lack of physical pins on the original F18A (also already mentioned). I did think about it though, when I was doing the original board design. But the original DRAM is slow, and it would be hard to incorporate all the F18A features with the memory speed limitations of the system DRAMs. The original 9918A did a pretty good job with what it had to deal with, and one of the main reasons the F18A can do what it does is the faster memory access (which is proving to be a complication for the MK2 with its external SRAM). Also, because the F18A does not use the system DRAM, it can be used to "fix" a system that had failing DRAM. If the F18A could also access the host DRAM, people would write software to expect it to be available, and that might no be the case if a host's DRAM was going bad. The MK2 could do it, but I decided to use the extra pins for other purposes like audio input / output, the USB and JTAG interfaces, etc. 6 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.