B is worse. Not only is efficacy lower , there is no combustion advantage if the mean power is the same.
From my experience on power Blue Lasers for cutting and burn print on wood, what matters is the average power which means averaged over the response time of the material.
Most materials are passive yet non-linear in that as they carbonize, they absorb more optical power and accelerate temperature rise at that threshold.
An active material I would consider is where the chemical reaction of photo energy triggers an immediate ionization and spontaneous molecular reaction faster than the time interval of the pulse period. THis assumes a linear absorption until blue absorption rises then triggers some hyptothetical spontaneous reaction (fusion).
So from the phenolic, epoxy, metallic, & cellulose materials that I have tried used so far, none have shown the latter fusion property and all roughly like the former property. If you do plan on making a fusion reactor, let me know. ;)
This means the peak energy is not what triggers faster temperature rise but rather the time duration. dynamic absorption rate, and average power density (optimal focus).
Not only that, the efficacy of power LEDs and Laser LEDs continually declines with rising current above lasing current from I²ESR losses and rising temperature. Although a well designed substrate for heat transfer will support fairly flat efficacy up to 50% of rated power , it measurably declines above this.
I think your only hope is to design nitrogen or liquid closed loop cooling like that used on overclocked CPU's if you wan to overdrive the Laser LEDs or invest in superior optics to focus the beam to a smaller dot, commensurate with Z axis stability.
Also Lens aberration causes the optical laser density to reduce significantly from precision expensive glass optics used on gas laser cutters.
I found that Laser printing on wood was extremely slow compared to an ink-jet printer due to the thermal response time of raising the temperature to carbonization levels. I tried changing gantry speeds, pulsed over current and various PWM levels and nothing worked any faster than continuous full rated power at a constant baseplate temperature with forced air cooling.
I recall instrumenting power absorption for Zirc-Monel Steel Diffusion Bonding during my late 70's work at Bristol Aerospace and even with 100kW available power to the surface, the best solution was continuous power, controlling the feed rate, cooling rate and power levels. Although here the plasma ionization was rather hot and probably resonating, this was not a laser rather 1~5V AC at very high currents. There were significant sparks of optical IR,UV and Xray emissions from sputtered diffusion of 3 surfaces between the giant solid copper roller contacts, so you might say it was pulsed from contact noise and high pressure gas emissions but this was an example of fusion welding or Diffusion bonding when it is done right. but the plasma reaction was an active surface of combustion to assist with generating the heat to diffuse 3 layers of material in a pipe joined together into one solid material so it could handle 1k atmospheres of pressure in a nuclear reactor without leaking.