# HP Forums

Full Version: Why is the 19BII faster than the 17BII(+)?
You're currently viewing a stripped down version of our content. View the full version with proper formatting.

I'm just asking out of curiosity. I definitely don't want to raise the 19BII prices, especially now that my 19BII's battery door latch probably is somewhere in a vacuum cleaner bag...

But I really want to know: Why is the 19BII faster than the 17BII.

The first example is taken from the calculatorstore website:

A=SIGMA(I:1:5000:1:1)

HP 17bII original: 81 seconds

HP 17bII+ gold: 78 seconds

HP 17bII silver:172 seconds

HP 19BII: 73 seconds

Not such a big difference, but anyway...

The second example is from, well, hpmuseum.org and much more surprising:

N-Queens

HP 17BII+: 6m25s

HP 17BII: 3m29s

HP 19BII*: 1m14s

*Using the 17BII solver equation from said web page

How is that possible? Does anybody know? According to HP, the 17BII and the 19BII should be equally fast (15x faster than the 12c).

The 17BII+ has entirely different innards, so any similarity or dissimilarity of its speed to the 17BII is purely coincidence.

The "15x faster than the 12C" is just a generalization, and not representative of any particular function or program. The 12C innards (either old or new) are also entirely different than any of the 17B or 19B models.

The 17BII (non-plus) and 19BII have architecturally similar hardware, and most likely share some firmware. However, differences in the firmware could easily result in some functions taking similar time and others not.

The Saturn microprocessor on the HP-17BII runs at 640 kHZ, as on the HP-42S, I guess. How fast is the clock on the HP-19BII? 1 MHz as on the HP-28C/S or 2 MHz? I wasn't able to find this information anywhere. Thanks in advance.

According to here, the 17B(II) runs at 1MHz, same as 19B(II), 27S, 28S, and 42S.

Here is what Cyrille wrote me in March of 2008, comparing the 17bii with the 17bii+ (I think it is also valid for comparing the 19bii with the 17bii+):

Quote:
The 17bii used a CPU that was (1) designed to do math, (2) was programmed in assembly (or system RPL), and (3) has full access to all the memory on the device. The 17bii+ CPU is programmed in C, which is less effective, and in addition that CPU is really not designed to run C code, making for much more inefficient programs. In addition, the CPU cannot address all of the available memory; therefore all the equations and also temporary equation evaluation data is off-loaded in external block access memory requiring thousands of off-chip memory accesses and swaps that are devastating for the speed.

I think that about says it all, thanks Cyrille.

On that sigma from 1 to 5000, I get:

```17b - 90 seconds
17bii - 94 seconds
19b - 86 seconds
17bii+ (gold) - 170 seconds
```

Edited: 3 May 2012, 9:10 a.m.

On the High End Pioneer Series you can read out the current frequency with the internal memory browser.

Press <ON>+<SK4> to enter the memory browser.
Press <SK1> to read the current speed in Hz.

The HP17BII and the HP19BII use the same CPU. Both have the same multiplier value in the Rate Control register, so both running with the same clock speed. I haven't checked if the firmware is changing the Rate Control register value during calculations, but I haven't watched such behavior in any HP firmware.

But some years ago, I compared HP28S with HP42S entry points. Both are RPL machines and should share many internal functions. But I found an entry point (I can't remember which one) doing exactly the same, use the same registers for entry and exit and the implementation in Saturn assembler was completely different! I don't know anything about the development history, but most of the HP28S implementations were used in the following HP48S calculator. So Eric's assumption about different code realizations resulting in different speed maybe become true. An answer could also give the profiler in the Emu42 debugger.

Christoph

Thank you, that's what I was looking for!

Thank you for the testing the HP-19BII with this unexpected result.