I believe the technical term for the transistor latency(switching states by 50% volt change) is called Propagation Delay, and so the power unused - with the potential to become wasted energy as capacitance/heat - will be the percentage of transistors stopping switching between the two scenarios, then x 50% of the higher power draw scenario(TDP maybe 200watt x 0.5) x the propagation delay(average delay of those transistors)
en.wikibooks.org
Propagation Delay
In all transistors today, there is a certain amount of latency between the time that a control signal is applied and the time that the output is affected. This delay is called propagation delay. These delays can be broken up into two parts.
• tPLH This is the delay time resulting from a LOW to HIGH transition
• tPHL This is the delay time resulting from a HIGH to LOW transition
Keep in mind: These two values may not be the same and are usually defined on the device's datasheet.
Propagation delay can be influenced by capacitance in the circuit. Propagation delay is also a rough indicator of effective speed of the device. For example, a gate with a 5 nanosecond propagation delay will respond much faster than a gate with a 120 nanosecond propagation delay.
Propagation delay is measured at the 50% mark and measures the time elapsed between the input and output signal changes.
en.wikipedia.org
Electronics
…...
Propagation delay increases with operating temperature, as resistance of conductive materials tends to increase with temperature. Marginal increases in supply voltage can increase propagation delay since the upper switching threshold voltage, VIH (often expressed as a percentage of the high-voltage supply rail), naturally increases proportionately.[3] Increases in output load capacitance, often from placing increased fan-out loads on a wire, will also increase propagation delay. All of these factors influence each other through an RC time constant: any increase in load capacitance increases C, heat-induced resistance the R factor, and supply threshold voltage increases will affect whether more than one time constants are required to reach the threshold. If the output of a logic gate is connected to a long trace or used to drive many other gates (high fanout) the propagation delay increases substantially.
Wires have an approximate propagation delay of 1 ns for every 6 inches (15 cm) of length.[4] Logic gates can have propagation delays ranging from more than 10 ns down to the picosecond range, depending on the technology being used.[4]
Exactly how this all measures into the scenario that Cerny described I'm speculating, and how the scenario of the map screen improved by a frame-rate cap - I don't have the games mentioned to experience it first hand - I assume it is because the sync reduces power draw in the propagation delay, which then fits better with the worst case scenario they engineered for - with the Pro - which Cerny said (paraphrasing) was guesstimate, of which the PS5 paradigm shift changes.
You could well be correct, and I may have got the wrong end of the stick, but the algebra of power, clock, switching, wasted energy of heat fits with my understanding - in my mind at least
I think at this point you’re playing pretend for attention and just confusing people genuinely curious to learn more and figure out how these things are working.
The delay I asked for clarification on was how you described power remaining high even after shifting to a lighter work load, which you then went on to bizarrely relate to the distance between the positive and negative terminals of the chip, somehow implying that it takes longer for an electronic wave to propagate along this path than for transistors to cascade and do what they do in one clock tick—something that is ironically proven wrong by you now bizarrely talking about how long it takes a transistor to switch.
What you’re linking to now is just completely unrelated to what you were originally saying and has nothing at all to do with going from a complex scene full of triangles to a simple one with an uncapped frame rate.
Are you that desperate to be seen as an expert on these forums that you don’t care about posting complicated looking gibberish and confusing people that are genuinely trying to learn more?
This isn’t the first time you’ve piggybacked off a post I’ve made with a reply that’s basically nothing more than Star Trek jargon pretending to add more. I’ve generally ignored them, but this time you specifically called me out and are implying I have something wrong when what you’re saying makes no sense at all, and the only motive I can really come up with right now is wanting to be seen as an expert online at any cost.
But even if this nonsense pseudoscientific technobabble was true for why something like a pre-patch Rocket League menu screen consumed so much power ans caused so much noise/heat (despite not coming after any complex scene at all), you’ve failed on multiple occasions now to even acknowledge me asking you to explain how on earth this would persistently apply heat to a chip indefinitely. You ignore it and apparently hit Google for some more jargon that you don’t really understand but hope sounds vaguely like an answer and is complicated enough to throw people off the scent. It’s extremely bizarre and I do not at all get the impression from you that you are interesting in learning about how these things work, or wanting to help others understand, too.
You also tell me I’ve misunderstood the GDC talk but then use a race to idle condition to explain the scenario when clock-speeds increase, despite Cerny explicitly saying that this is not the case, and would be an unproductive scenario to increase clock speeds.
Instead you ignored both of these and try to find anything at all related to transistors and latency, even if it isn’t anything at all to do with your original point regarding latency.
Please do no waste my own or other’s time pretending to be an expert on something you are clearly not. There are other ways to be appreciated in a community that don’t actually do the opposite of helping people understand things for your own ego.
