This is an incredibly complex question, but we can get you going by ignoring a few of the trickier issues, and get you to an "order of magnitude" sort of result.
PV cells output a given power (Watts) that is roughly proportional to the irradiance (Watts/Sq. Meter) they are exposed to, corrected for the angle of incidence against their surface. Temperature and spectral content of the light affect the "curve" describing the relationship between output power and irradiance.
PV modules made for actual utility power generation are tested at 1000 W/sq.m, at 25C cell temperature. This roughly corresponds to the peak expected irradiance at the surface of the Earth in sunny places. Actual daily peaks in the summer tend to be in the 800-1100 W/sq.m in the central US. Horizontal irradiance at the top of Earth's atmosphere averages 1350 W/sq.m, though is a bit higher in January, when the Earth is at it's closest to the sun.
An 80% efficient solar module, exposed to 1000 W/sq. meter, would output approximately 800W instantaneously. This oversimplifies things a bit, but is a truthy way of describing the result.
The problem for your calculations is that irradiance is constantly changing, as is ambient temperature, cloud cover, and other things that have a huge impact on your output. Knowing the peak wattage at any given time isn't helpful, because peak watts don't run our lives - we need to know the energy produced over time, by integrating the changing irradiance, and accounting for efficiency losses and weather conditions, in order to determine the amount of energy produced (typically measured in Kilowatt-hours (kWh) in the US).
This would be very hard, except that NREL (US National Renewable Energy Labs) have done most of that work for you, in their package SAM (System Advisor Model) which is a free application (registration required) that is very powerful, but relatively easy to use.
Below are the results of two quick simulations using a hypothetical 80% efficient module:
The system as simulated:
- Module: 80% efficient, 1.0 m^2, 800W-DC output at standard test conditions.
- Inverter: 97% efficient
- DC and AC losses: Standard losses for DC and AC conversion, NOT including utility transmission losses, ~0.88 including soiling, DC losses and AC losses.
- 1 acre of space (~4000 sq. meters)
- Ground Coverage Ratio of 0.33 (33% of the ground area is occupied by modules)
- Module tilt: 30 degrees
This gives us 1320 modules, for a DC system size of 1.056 MegaWatts (nice round number), using commonly used values for GCR and tilt. The layout is a bit conservative, and assumes we have a large amount of space available, and that maximum annual production is our goal, rather than production any given month.
The cost to install a system like this, using today's inverter, racking and labor rates, and a little margin for the installers and financiers would be around 500,000USD + module cost. Crystalline PV modules today cost about 0.85USD/watt on the utility-scale market. I won't speculate about the cost of hypothetical 80% efficient modules (but I can promise you it won't be as cheap as any researcher claims).
Site 1: Nevada Desert. Meteorological source, NREL TMY2, Las Vegas Station
Net annual energy produced: 1,989,000 kWhs
Site 2: Florida. Meteorological source, NREL TMY2, Tallahassee Station
Net annual energy produced: 1,586,000 kWhs
Site 3: Alaska. Meteorological source, NREL TMY2, Anchorage Station
Net annual energy produced: 1,010,000 kWhs
Boiling that down to a usable "formula" would be simple - each acre of land dedicated to this technology would generate the above values annually. You can see that regional conditions contribute to nearly a 2:1 ratio in expected output. Globally, solar has been the most popular in locations that are politically and financially favorable, rather than environmentally favorable, but long-term plans would favor regions with good environmental conditions.
This simulation uses accepted parameters for estimating the production of realistic systems, but these hypothetical modules make huge assumptions. Experience has shown us that new technologies never perform in the real world like they do in the lab. Thin-film technologies have been in development for over 30 years and are still barely pushing 12.5%, while crystalline silicon is at ~15%, and costs less per watt (due to the enormous manufacturing infrastructure).
I've ignored a HUGE issue with supplying vast quantities of renewable energy to a grid. When you get above 15% distributed penetration on a grid, things get unstable fast. Storage methods need to be in place to ensure that periods of excess production aren't wasted, and energy is still available during times of reduced production. This is a non-trivial problem for large systems.
I have also ignored the transmission problem. Even if you could make all the world's energy in Nevada, you couldn't get it where it needs to go. One of distributed generation's benefits is the ability to generate close to the loads they serve, reducing transmission losses.
I don't think I can attach files, but let me know if it is possible for me to send or attach the .zsam simulation file I used, and I will be happy to provide it.