Benchmarks 71B versus 48GX

[John W Kercheval]

Does anyone have a raw chip speed comparison between the 71B and the 48GX?

[Jake Schwartz]

Quote:Does anyone have a raw chip speed comparison between the 71B and the 48GX?

Somehow I recall the 71 CPU to be on the order of 600KHz whereby the 28C was 1 MHz, the 28S was 2 MHz and the 48SX and GX were 4 MHz. Hopefully, I didn't mess that up :-)

Jake

[Mark Hardman]

(05-02-2014 09:23 PM)John W Kercheval Wrote:  Does anyone have a raw chip speed comparison between the 71B and the 48GX?

From Craig Finseth's invaluable HPDATAbase:

HP-71
Saturn, 4-bit, 640 KHz 1LK7 5 V CMOSC

HP-28C
1LK7 Saturn CPU (1 MHz)

HP-28S
Lewis, 1 MHz

HP-48SX
Clarke (147 pin TAB), 2 MHz

HP-48GX
Yorke (00048-80063, 160 pin QFP), 4 MHz

Mark Hardman

[John W Kercheval]

Thanks the 48GX is a 4 bit machine correct?

[Mark Hardman]

(05-03-2014 12:37 PM)John W Kercheval Wrote:  Thanks the 48GX is a 4 bit machine correct?

Yes. The Clarke and Yorke chips were four bits (nibble) with a 20 bit maximum address.

Mark Hardman

[Christoph Giesselink]

(05-03-2014 01:57 AM)Mark Hardman Wrote:  HP-28C
1LK7 Saturn CPU (1 MHz)

Sorry, this is wrong. The HP-28C use the same LC frequency generator like in the HP-71B, so the CPU strobe frequency is 600-650 KHz. You can peek the 5 nibble =CSPEED content (#4F00Fh) on a HP-28C with an external peek program. The 5 nibble content is the CPU strobe frequency divided by 16. So a content of #09C05h mean: #09C05h -> #39941d * 16 = #639056d -> ~640 KHz.

(05-03-2014 01:57 AM)Mark Hardman Wrote:  HP-48GX
Yorke (00048-80063, 160 pin QFP), 4 MHz

This is also a mistake. Only very early HP48G Series calculators had a strobe frequency of 4 MHz. The frequency was then reduced to a nominal speed of ~3.68 MHz.

[Mark Hardman]

(05-04-2014 10:56 PM)Christoph Giesselink Wrote:  

(05-03-2014 01:57 AM)Mark Hardman Wrote:  HP-28C
1LK7 Saturn CPU (1 MHz)

Sorry, this is wrong.

(05-03-2014 01:57 AM)Mark Hardman Wrote:  HP-48GX
Yorke (00048-80063, 160 pin QFP), 4 MHz

This is also a mistake.

I will pass the corrections back to Craig.

Mark Hardman

[Didier Lachieze]

(05-04-2014 10:56 PM)Christoph Giesselink Wrote:  

(05-03-2014 01:57 AM)Mark Hardman Wrote:  HP-48GX
Yorke (00048-80063, 160 pin QFP), 4 MHz

This is also a mistake. Only very early HP48G Series calculators had a strobe frequency of 4 MHz. The frequency was then reduced to a nominal speed of ~3.68 MHz.

For reference, on my 48GX SN 3443S in the interactive self-test (ON-D) test [A] returns a speed of ~3.93 MHz. (Exit of the self-test is done by ON-C)

[John W Kercheval]

Ok I tried that test is it holding down the D key while starting it? It does not seem to do anything. How do you do the tests?

[Didier Lachieze]

(05-05-2014 11:23 AM)John W Kercheval Wrote:  Ok I tried that test is it holding down the D key while starting it? It does not seem to do anything. How do you do the tests?

With the 48 already ON - press ON and D simultaneously, for more details see the HP48 FAQ section 4.7 What are the different interactive self-tests?

[cutterjohn]

(05-05-2014 12:12 AM)Didier Lachieze Wrote:  

(05-04-2014 10:56 PM)Christoph Giesselink Wrote:  This is also a mistake. Only very early HP48G Series calculators had a strobe frequency of 4 MHz. The frequency was then reduced to a nominal speed of ~3.68 MHz.

For reference, on my 48GX SN 3443S in the interactive self-test (ON-D) test [A] returns a speed of ~3.93 MHz. (Exit of the self-test is done by ON-C)

Funny enough I was reading somewhere the other day(that seemed fairly reliable) that when the 48G series came out that they were having 4MHz yield problems -> binning. The Xs got the 4MHz "good" chips and the other got the not quite able to make 4MHz chips... (c. 3.5-7MHz IIRC)

[John W Kercheval]

Ok got it.

How do we speed it up? What if we overclock it to 8MHZ for example?

Let's talk HP48GX hardware mods!

[HP67]

Probably better to overclock a 50g. They're still making them, and supposedly they are underclocked as it is.

[Didier Lachieze]

(05-05-2014 01:34 PM)John W Kercheval Wrote:  Ok got it.

What speed do you get?

(05-05-2014 01:34 PM)John W Kercheval Wrote:  How do we speed it up? What if we overclock it to 8MHZ for example?

Let's talk HP48GX hardware mods!

This has been done back in the 90's, the performance increase seems to be limited to 30-50% with some side effects: Digitalis Speed UP for HP48G/GX/G+.
See also this post on comp.sys.hp48: Overclocking the HP.

[John W Kercheval]

I get 3.775024

[KimH]

Actually, the later 71B has the refreshed 1LK7 CPU.

According to legend HP always wanted the chip to run a 1MHz Clock, but something went wrong and they ended up with 650KHz which is the 1LF2.

In later years - i think mid 86, they started to mount the 1LK7 on the PCB.

There is no way - that I know - to see the actual CPU version without opening the Case.

If it is 1LK7, the LC circuit can be modified to run at about 1MHz - at about 1.1MHz the CPU starts to become unreliable.

/KimH

(05-03-2014 01:57 AM)Mark Hardman Wrote:  

(05-02-2014 09:23 PM)John W Kercheval Wrote:  Does anyone have a raw chip speed comparison between the 71B and the 48GX?

From Craig Finseth's invaluable HPDATAbase:

HP-71
Saturn, 4-bit, 640 KHz 1LK7 5 V CMOSC

HP-28C
1LK7 Saturn CPU (1 MHz)

HP-28S
Lewis, 1 MHz

HP-48SX
Clarke (147 pin TAB), 2 MHz

HP-48GX
Yorke (00048-80063, 160 pin QFP), 4 MHz

Mark Hardman

[Dave Frederickson]

(06-17-2020 11:43 AM)KimH Wrote:  Actually, the later 71B has the refreshed 1LK7 CPU.

According to legend HP always wanted the chip to run a 1MHz Clock, but something went wrong and they ended up with 650KHz which is the 1LF2.

In later years - i think mid 86, they started to mount the 1LK7 on the PCB.

There is no way - that I know - to see the actual CPU version without opening the Case.

If it is 1LK7, the LC circuit can be modified to run at about 1MHz - at about 1.1MHz the CPU starts to become unreliable.

Did you win the auction for the 1MHz 71B on eBay?

Does the introduction of the 1LK7 correspond to the change is case style or an update from 1BBBB?

Dave

[KimH]

I did not know about the 1MHz auction, but I made a couple of them myself.

The Diagnostic/Service ROM helped me validate how far to push the Clock

I recall that a good indication would be a serial# later than 2623, but opening the case and checking is the best way to determine the CPU version.

/KimH

(06-17-2020 12:19 PM)Dave Frederickson Wrote:  

(06-17-2020 11:43 AM)KimH Wrote:  Actually, the later 71B has the refreshed 1LK7 CPU.

According to legend HP always wanted the chip to run a 1MHz Clock, but something went wrong and they ended up with 650KHz which is the 1LF2.

In later years - i think mid 86, they started to mount the 1LK7 on the PCB.

There is no way - that I know - to see the actual CPU version without opening the Case.

If it is 1LK7, the LC circuit can be modified to run at about 1MHz - at about 1.1MHz the CPU starts to become unreliable.

Did you win the auction for the 1MHz 71B on eBay?

Does the introduction of the 1LK7 correspond to the change is case style or an update from 1BBBB?

Dave

[J-F Garnier]

(06-17-2020 11:43 AM)KimH Wrote:  In later years - i think mid 86, they started to mount the 1LK7 on the PCB.

There is no way - that I know - to see the actual CPU version without opening the Case.

If it's really a 1LK7 then it's the same than used in the 28C and it supports the PC=(A) opcode. It may be a way to detect the CPU version by software. I may try.

Quote:If it is 1LK7, the LC circuit can be modified to run at about 1MHz - at about 1.1MHz the CPU starts to become unreliable.

If it starts to fail at about 1.1MHz, I would not push it to about 1MHz :-)

J-F

[KimH]

Thanks for the input - the Diagnostics says 960k and all tests seem to pass, I have run sequences for about 30 minutes. It's a wrecked unit with Zebra Soldered, Glass Broken, 71B Faceplate lost and so forth - the one I use for that type of stuff, not one that I use for anything else than HW-mods and "Controlling the World".

The PC=(A) could probably be built into a short stub, but I am not good enough - yet - to do that, that is for when the weather turns back into cold and wet

As long as the PC=(A) acts as a NOP in the older CPU, then it would probably not be that hard.

Maybe loading (A) with the address for "Beep" and then PC=(A) it would be a smooth approach

/KimH

(06-17-2020 12:36 PM)J-F Garnier Wrote:  

(06-17-2020 11:43 AM)KimH Wrote:  In later years - i think mid 86, they started to mount the 1LK7 on the PCB.

There is no way - that I know - to see the actual CPU version without opening the Case.

If it's really a 1LK7 then it's the same than used in the 28C and it supports the PC=(A) opcode. It may be a way to detect the CPU version by software. I may try.

Quote:If it is 1LK7, the LC circuit can be modified to run at about 1MHz - at about 1.1MHz the CPU starts to become unreliable.

If it starts to fail at about 1.1MHz, I would not push it to about 1MHz :-)

J-F