Question about memory — ZXNet «hardware.zx»

From Mikhail Tarasov To All 19 December 2005

Hello Mick I continue to mock the computer - I cut off WAIT (in the sense that it’s always on now "1"). So far it works without problems. Although what did this affect and how can I check it? I don't know yet.

From Vlad Semchenko To All 20 December 2005

Hello Mick Mic> I continue to mock the computer - I cut off WAIT (meaning on it Mic> is now always "1"). So far it works without problems. But what is it for Mic> had an effect and I don’t know how to check it yet. Test the number of cycles. At the same time, you will find out the stability of the tire. Maybe You'll find out something else interesting.

From Konstantin Sviridov To All 11 January 2006

Hello caro car> It seems that the second assumption is also correct in Spectrum+2 car> line has 228 bars, and lines have 311. I checked with the , there are indeed 228 bars in the line. As for the error, for , the main thing was to understand that there are more clock cycles, and no less than +2. Moreover, the test was developed by you a little for something else (calculations of fast and slow pages), and fulfilled its purpose great. Thank you very much for that again.

From Kamil Karimov To All 11 January 2006

Hello Conan Con> Judging by the results provided by Kamil, in a branded car Con> +2 test showed: 70848 cycles (in banks), and based on Con> ratio of measures in a line and lines (227*311) there are 70597 of them. That is Con> the error was: + 251 clock cycles. Such an error indicates either my mistake or yours. I'll check tomorrow, but today I can assume that Spectrum+2 does not have 311 lines, and 312 lines, which gives (227*312)=70824 clock cycles - an error of +24 clock cycles. Or there are 311 lines, and the number of clock cycles in a line is 228, which gives (228*312)=70908 - error -60 cycles. Now I checked the test in the ZXMAK emulator, where the number of cycles in frame. As it turned out, the test gives an error of -64 clock cycles :) It looks like the second assumption is also correct in Spectrum+2 there are 228 bars in a line, and 311 lines.

From Alexander Yudin To All 11 January 2006

Hello Costa By the way, this error of 64 clock cycles is only in the test version 1.0 Version 1.1 works correctly down to the beat!

From Alexander Yudin To All 11 January 2006

Hello caro car> As it turned out, the test gives an error of -64 clock cycles :) Strange!? It seems not at minus 64 but at plus 64 as I indicated above: v2_conf2:

From Vitaly Kovalev To All 11 January 2006

Hello Mick Mic> Test is test, but don’t forget I now have a ZX-777 with modifications Mic> (SRAM128 and clock frequency shifted). Maybe with the original one Mic> compare, maybe there is some difference. I also have a 777, completely redone inside and out, and the memory costs sim 1 meter, but it stubbornly showed 228 clock cycles even before installing the sim, although I didn’t wait cut it off, and what’s interesting is that he showed all the company’s multicolors in games that’s right, it’s just that at that time I had a shift in my head that I needed “what would the Pentagon multicolors worked", they worked, only the border effects went crookedly with a shift to the left in each line, it turned out to be an excellent ladder......

From Kamil Karimov To All 11 January 2006

Hello Costa Cos> By the way, this error at 64 clock cycles is only in the test version 1.0 Cos> Version 1.1 works correctly up to the clock! Well, of course, I posted version 1.1, which works without errors. I immediately couldn’t find this test in the forum, so I’m posting it again, especially since it small. File: tstmem11.zip http://zx.pk.ru/attachment.php?attachmentid=2146