STM32 chips with 180Mhz and USB is impossible?

No because it will be clearer when we add G30 :wink:

Nope! Then it will be:

The g400 runs at 400MHz
The g120 runs at 120MHz
The g30 runs at 30MHz
The g80 runs at 180MHz —> why isn’t it G180?!

@ Simon from Vilnius - why do you assume that G30 will run @ 30Mhz?

@ iamin - Why not?

Exactly :slight_smile:

Because Gary (or someone) had previously posted that the designation is NOT the clock speed but a relative performance index based on many factors of the feature set of the module.

@ Blue Hair Bob - That’s not how common sense works :slight_smile:

Collapsing multiple dimensions of performance/value down to one arbitrary dimension never works out completely neatly, but it is roughly what they do with car model numbers and appliances and such, so it’s not that weird a concept.

Maybe they should stick with animal names. Does every SOM need to be named after Gus? Perhaps that’s to be his legacy.

Or maybe they should be measured in square millimeters with the ultimate goal being to one day count down to G1!

1 Like

Last time someone did an actual performance comparison, the Cerberus at 168 MHz performed close to the G400, and completely demolished the G120. We must be looking at a lot more than performance here.

Either way, to me, the model numbers are 100% uninteresting. They could call it the G0.75 if they want, it’s still going to be a monster performer.

Wait what? the STM32F405 vastly outperformed the LPC1788?


@ Simon from Vilnius -

Simple thing in C# with G80, it takes 32s 780ms, while Cerberus is 35s126ms, about 7% faster. :smiley:

@ Dat - And the same “simple thing” on G120, G400 and G30 (if not yet too early) takes how long?

@ Mr. John Smith - I think the main difference between Cerb and G120/400 is that cerb uses on chip RAM, which is way faster than the external SDRAM of G120 or G400.
So a pure calculation should be faster on G400 than on Cerb. But when it Comes to RAM Access the Cerb can outperform the other boards easily.

1 Like

@ iamin -

14s 646 on G400,

@ Reinhard Ostermeier - Sweet child of mine you’re right. Aw man, it’s back to the drawing board.

An idea: GHI could provide a chart/summary where it compares its mainboards in terms of (1) “pure calculations” done in managed code, (2) RAM access, (3) ethernet max throughput and so on.

Something like this:

1 Like