Convert bitmap images to scalable vector graphics image for clarity and workability. 1) https://libre-soc.org/3d_gpu/architecture/6600scoreboard/600x-multiple_function_units.png 2) https://libre-soc.org/3d_gpu/architecture/6600scoreboard/600x-dependence_cell_multi_pending.jpg The image multiple_function_units.png has to be split into two, one for read and one for write.
Created attachment 134 [details] multiple_function_units_write svg image SVG version of multiple_function_units_write. Please review.
(In reply to vklr@vkten.in from comment #1) > Created attachment 134 [details] > multiple_function_units_write svg image > > SVG version of multiple_function_units_write. > Please review. ha, that looks really good, do drop it in place of the jpg in the wiki.
Created attachment 135 [details] multiple_function_units_read svg image SVG Image of multiple_function_units_read In Second box(Long Add), wire to OR gate is blue; may be it has to be green?
(In reply to vklr@vkten.in from comment #3) > Created attachment 135 [details] > multiple_function_units_read svg image > > SVG Image of multiple_function_units_read > > In Second box(Long Add), wire to OR gate is blue; may be it has to be green? there are only 3 colours, purple for A green for B orange for X. those colours going in to a Function Unit have to match those going out. so yes, most likely you are correct.
(In reply to vklr@vkten.in from comment #3) > Created attachment 135 [details] > multiple_function_units_read svg image > > SVG Image of multiple_function_units_read > > In Second box(Long Add), wire to OR gate is blue; may be it has to be green? i took a look: you see how "green" (B) goes in to Branch, Increment (1) and Increment (2)? that means that, into the OR gate, a corresponding "green" needs to go *out* of Branch, connected to the OR gate. at present there is only one green out of Increment (1) into the OR gate and another green out of Increment (2) into the same OR gate. the Red (X) likewise goes into all three... ah, and there is one too many blue :)
Created attachment 136 [details] screenshot of multi unit ok so here's a screenshot of the corrections needed, * branch is missing green and red from Branch to the OR gate * yes, Long Add, it should be green to the (2nd) OR gate there's a couple others, do check them. maybe review the write pending one as well, don't draw the lines *through* the Function Units! that's just for illustrative purposes.
Uploaded and changed multiple_function_units.png with two SVG images; one for write and one for read. Also updated relevant page: https://libre-soc.org/3d_gpu/architecture/6600scoreboard/
(In reply to Luke Kenneth Casson Leighton from comment #6) > Created attachment 136 [details] > screenshot of multi unit > > ok so here's a screenshot of the corrections needed, > > * branch is missing green and red from Branch to the OR gate > * yes, Long Add, it should be green to the (2nd) OR gate > > there's a couple others, do check them. maybe review the write pending > one as well, don't draw the lines *through* the Function Units! > that's just for illustrative purposes. Corrected and updated in git(website). Please review.
(In reply to vklr@vkten.in from comment #8) > (In reply to Luke Kenneth Casson Leighton from comment #6) > Corrected and updated in git(website). Please review. look really good. dependence cell one next.
Created attachment 139 [details] svg image of dependence_cell_multi_pending.jpg SVG image of dependence_cell_multi_pending.jpg Please review.
(In reply to vklr@vkten.in from comment #10) > Created attachment 139 [details] > svg image of dependence_cell_multi_pending.jpg > > SVG image of dependence_cell_multi_pending.jpg > > Please review. love the layout, it's very clear. two things which will significantly improve it (see these): https://ftp.libre-soc.org/dependence_cell_multi_pending.jpg https://ftp.libre-soc.org/dependence_cell_multi_pending.svg (1) there are places where the wires cross. because there are no colours, you can't work out which wires "join" and which "cross". could you put "dots" at the crossing points, like in the JPG? (2) once all the dots are there, could you colourise the wires that are all the same signal? (no need to do everything) i made a start on the colours but i messed some of them up, i don't know how. mainly the named inputs, and the named outputs, try to make them REALLY bright primary colours, lots of contrast. i leave it to you to decide the best ones? again, you don't need to do *everything* in colour, but at least everything that has an input name and an output name.
Created attachment 140 [details] Updated colours and connections Updated colours and connections dots. Please review.
Created attachment 141 [details] New correct file. New correct file as attachment. Somehow wrong file was uploaded: id 140. Updated colours and connection dots.
(In reply to vklr@vkten.in from comment #13) > Created attachment 141 [details] > New correct file. > > New correct file as attachment. Somehow wrong file was uploaded: id 140. > > Updated colours and connection dots. brilliant, that looks really good. only thing, the yellow is so close to white, i can't quite make it out. if you can make it a slightly darker yellow, then please go ahead, replace the JPG on 6600scoreboard, and we can close this one as "DONE". nice work veera.
Changed yellow with dark yellow colour for wire lines. Uploaded and changed dependence_cell_multi_pending.jpg with SVG image in git. Also updated relevant page: https://libre-soc.org/3d_gpu/architecture/6600scoreboard/