I am wondering nowadays, how -for example- Intel is expensive but not the most innovative... where peeps seem to uphold... paying 2 much 4 too less.
Good times, bad times... welcome to the silicon lottery and hope u do better... in Vega's.
Cpu's take several ticks, to finish a complete instruction into output. Ram timings are only important, with the first four stats? Then why are all those other timings mentioned... in an average BIOS? Can it be that cheap RAM does already do a good job and the rest is just bonus for a branch-mark?
If DDR works with the two flanks, of a sine wave.. then why can't a CPU input on upper flank and output on down flank, of one tick? Is RISK worth sacrificing CISK? And what can be done with adding FPGA circuitry in a CPU, for doing interesting stuf like 'emulating' legacy chips and such? Things... & ... questions. -just echoing in my mind-
I have seen a lot of topics about over-clocking CPU's and RAM modules. AND-OR-NotAND a lot has been told... ass a lot has been sold. It could make one flip-flop. What I do know is that everything works better... when harmonized. So if 33Mhz (or an even slower base equivalent in KHz) 'crystal clear' is the base frequency where everything stems from; Isn't it appropriate to first start tuning all timings to this base number? And the see how CL timings etc. fit in this model? As it also should be clearly explained by manufacturers how timings are to be harmonised. Or that UEFI BIOS knows how to synchronize stuff automatically. As I don't see how DDR 1600 MHz with very low timings can be beaten by DDR4 3200 MHz with very high latency count. To me it occurs that people pay way to much money, for broken promises. Although DDR4 has some more pins, so if that adds more data-paths, this should be the major advantage. Not the timings, as far as I can C++. Where good drivers can make BIG differences. Do we really need to gamble... on horses and their prowess? Raw power needs to be in a good f-low... always. Not a rough estimation.
Good times, bad times... welcome to the silicon lottery and hope u do better... in Vega's.
Cpu's take several ticks, to finish a complete instruction into output. Ram timings are only important, with the first four stats? Then why are all those other timings mentioned... in an average BIOS? Can it be that cheap RAM does already do a good job and the rest is just bonus for a branch-mark?
If DDR works with the two flanks, of a sine wave.. then why can't a CPU input on upper flank and output on down flank, of one tick? Is RISK worth sacrificing CISK? And what can be done with adding FPGA circuitry in a CPU, for doing interesting stuf like 'emulating' legacy chips and such? Things... & ... questions. -just echoing in my mind-
I have seen a lot of topics about over-clocking CPU's and RAM modules. AND-OR-NotAND a lot has been told... ass a lot has been sold. It could make one flip-flop. What I do know is that everything works better... when harmonized. So if 33Mhz (or an even slower base equivalent in KHz) 'crystal clear' is the base frequency where everything stems from; Isn't it appropriate to first start tuning all timings to this base number? And the see how CL timings etc. fit in this model? As it also should be clearly explained by manufacturers how timings are to be harmonised. Or that UEFI BIOS knows how to synchronize stuff automatically. As I don't see how DDR 1600 MHz with very low timings can be beaten by DDR4 3200 MHz with very high latency count. To me it occurs that people pay way to much money, for broken promises. Although DDR4 has some more pins, so if that adds more data-paths, this should be the major advantage. Not the timings, as far as I can C++. Where good drivers can make BIG differences. Do we really need to gamble... on horses and their prowess? Raw power needs to be in a good f-low... always. Not a rough estimation.
:strip_exif()/u/10111/crop67d5f1a9809eb_cropped.gif?f=community)
Dit topic is gesloten.