DM48n

[c3d]

(09-20-2024 09:09 AM)Nigel (UK) Wrote:  

(09-20-2024 01:56 AM)jeanwilson Wrote:  What I'm wondering is, will db50x running on the DM42n suffer the same speed penalty as when running on the DM32? Because currently db48x which runs on the DM42 is currently 2 times faster than db50x on the DM32 (see https://www.hpmuseum.org/forum/thread-97...e%20plenty) which nevertheless benefits from a processor which is supposedly twice as fast! Even if the DM32 and DM42n share the same hardware, do they share the same speed limitation (especially since they will both be based on DMCP5, I believe)? Or is it a matter of principle that makes the DM32 run twice as slowly as the DM42?

Tell us when you can clarify the situation.

The times in the post you link to date from March this year. It was in March that DMCP5 was updated to allow the processor frequency to increase when on USB power. Assuming that the timings were done on USB power the processor frequencies in the test would have been 80MHz for the DM42 and 48MHz for the DM32. If done today on USB power the DM32 frequency would be 160MHz and the results should reflect this.

Since I don’t actually have a DM32 I can’t be sure that this is correct.

Nigel (UK)

I can actually confirm this, though with a caveat: after the upgrade, you cannot unplug the USB while DB48X is running without turning the calculator off, otherwise there is a power loss and you end up rebooting with a blank state. I'll investigate.

The good news is that on USB, the DM32 now runs the NQueens benchmark in 450ms down from 1800ms (and roughly 1300 for the DM42 on USB). Yay! So as long as you remember to switch the calculator off before unplugging, everything is fine!

I'll update the performance benchmarks page for the DM32 after that upgrade.