A CPC is Born! GX4128

[Executioner]

Well, the c64 had 0.9 MHz. Actually it runs thereby in the KHz region.

I think it is actually 1MHz, but the processor is disabled during certain parts of the display in order to fetch sprite data, making it effectively slower.

[TFM]

That may well be :-)
At the c64 datasheet they say "@ 0.985 MHz (PAL version)". So roundet 1 MHz, but i prefer to do runding by cutting numbers ... 0.9 then 

[Badstarr]

Well, that little fraction of MHz may make a fair bit of difference! As I'm sure most people know, the Z80 in the GX4000 is clocked slightly lower than a standard CPC and I can report that while playing FLUFF the slowdown that occurs at various points during the game (you know when there are a fair few sprites going and the "loose a life" sequence where Fluff bounces up and down) seems at least subjectively a fair bit more exaggerated on the GX. I have played the game on both my 464Plus and the GX and it seems a little more sure footed on the 464. I haven't run both at the same time, not quite sure how I would be able to do a move for move comparison to see the difference really to see if its just my imagination. So this made me think is there anything that would work as a sort of benchmark tool for the CPC? I guess it would be kind of interesting to do some sort of benchmarking now that there is the opportunity.

[TFM]

Well take both machines, and start both at the same time, and press both joypads in the same (up right?) direction, ... wait for differences...

[Badstarr]

Just thought I'd give this a bit of a bump as I have now managed to get a PCW keyboard working with my GX4128, not only that but I have replaced most of the keys with the ones from my dead 464 keyboard the remaining ones will be painted to 464 colour scheme for now I have just coloured them in with sharpies (lol) to get an idea of how it will look...

[Bryce]

Really? Which one could that be, except the 256 colors? (I'm asking seriously, like usual, would appreciate some short comments).

I was speaking from a hardware expansion point of view: 5V available at the joystick ports. analogue inputs on the joystick port, fully address decoded cartridge port, S-video signals available. All things that the (Classic) CPC didn't have.

@Badstarr: Crazy colours, it looks like a seasick Enterprise 128

Bryce.

[Badstarr]

Haa! Good description! I might get some paint today and sort those keys out if I can find some that match well enough. I will have to finish the epoxy on the DEL and SHIFT key. I was going to keep the white casing but I think I might paint it black so I guess the GX will have to follow this scheme too. Certainly unique!

[Gryzor]

it looks like a seasick Enterprise 128

Loltastic.

[Badstarr]

Well here is my progress so far, I couldn't resist painting up the unfinished DEL and SHIFT key to see what it all looks like. Not too bad really, I ran out of black paint so the space bar is half base coated, and the really nice satin sheen I managed to get was slightly ruined by a bit of splutter when the can was running out (kinda gives it a more industrial look which I'm starting to think might look pretty good actually, at least from some angles lol)


[attach=2]


I think I will be able to live with it when its finished. I think it's quite an improvement on the old 6128 keyboard idea I had...

[TotO]

Sorry, but all look dirty.
Better to buy a 6128Plus and use it like that, instead off killing/hacking all those poor hardware... 

[Gryzor]

Wow, this looks very neat. Like a modern-date update!! Wish I had a PC keyboard like that

[Badstarr]

Thanks Gryzor! I was just thinking the very same thing! The GX4128 keyboard is in front of me now so I have to reach over to get to my Mac keyboard to type this and I have written most of this reply on the wrong keyboard lol! Shame it isn't connected to the Mac typing fruitlessly into the wrong keyboard gets a little tedious! I do have a black USB keyboard that I might have a go at 464-ing!


And yup, I haven't given the keys a good clean yet (but I think its a vast improvement on the PCW keyboard it was 24 hrs ago lol!) and DEL and SHIFT are not fully finished I just got a little giddy when I placed the newly painted case over the keyboard unit and thought that it does sorta look convincingly 464ish in the flesh. The black/blank keys are intensionally  the same colour as the case as, I don't want to consider them as really being part of the keyboard, they actually send the joystick commands so they can be kinda useful if a joystick aint handy but they aren't laid out very ergonomically lol!.

[TotO]

Sure, it's WIP... So, I'll hope that the final result will make me lie.

[Badstarr]

Just one last thing I haven't really killed any hardware. The 6128 that I borrowed the keyboard from was dead (cheap on ebay always worth it for spares) and now I have repaired it so it got its keyboard back. So that's one more living 6128 in the world. The 464 that donated its keys to the PCW keyboard has a thoroughly calcified membrane and half its ICs are dead. Its cassette unit is missing several vital components too, that's the way it was sold to me. Since I only ever bought it to use its keyboard for my project I feel it fulfilled its purpose better than a landfill at any rate.


Granted I half killed a GX4000 when I started this project but I used parts from that to repair another GX4000 which was a brick until that point, (but very cheap on ebay as a result). The PCW keyboard came from a computer recyclers, and several times its ebay auction expired with no bids and I assume as they weren't selling a PCW as well, that the keyboard had no real future in front of it. Now it's a pretty nifty way of interacting with my GX machine!


I have a 464+ with HxC and 128k so I don't need a 6128+, besides, I have a GX4000 that can do anything a 6128+ can and I have the added bonus of knowing I made it do that   

[TFM]

I have a 464+ with HxC and 128k so I don't need a 6128+, besides, I have a GX4000 that can do anything a 6128+ can and I have the added bonus of knowing I made it do that 

Well, I don't agree here. Think about the frequency of the crystal. That topic was discussed before.

[Badstarr]

Granted it's a little slower but I tried a simple benchmarking routine and the difference was negligible, I think I was just imagining a perceived difference in performance when I first tried FLUFF because it seems more or less the same on the 464+. I'm playing Super Cauldron right now on the GX and it's smooth as butter! 

[TFM]

Yeah! Nothing can beat real hardware! 

[Bryce]

The frequency difference between the CPC+ and the GX4000 is so minimal, that I doubt any gamer could tell what machine a game is running on. It's barely humanly noticable.

Bryce.

[Badstarr]

I agree completely. The only thing I noticed when running Fluff on the GX was that when the CPU hits saturation point, ie; when there are lots of sprites on screen and the screen scrolls and everything gets a little choppy. You can notice the affect a little more, but in comparison to my 464+ it's not as pronounced as I perceived it to be. I tend to notice differences in timing more between audio events (so to speak), I'm maybe a little more sensitive to this as I have a degree in sound engineering and matching sounds with visual events is something I had a ridiculous amount of practice doing as a result. I would say having tried Fluff again yesterday on both systems (and this is based purely on intuition) that the lag on the GX when the game stutters is around 150 ms longer than on the 464, just barely perceivable as you say. Its like the lag is stretched out but this depends a lot on what is happening at the time. Platform games as we know depend on timing so lag tends to leap right out. However it seems Prehistorik 2 plays exactly the same on both machines.

[TFM]

The frequency difference between the CPC+ and the GX4000 is so minimal, that I doubt any gamer could tell what machine a game is running on. It's barely humanly noticable.

Bryce.

That's too easy ;-) Games are sometimes encoded in a way that they use up nearly all microseconds. So, if the GX is only for some microseconds too slow, it may miss tha point when the "FRAME" routine starts. That means you will loose a full frame(!). Now, due to this all timing get's messed up and the game will loose any kind of smoothness. To argue with pure speed leads nowhere, you have to consider what the program does. Well, and this is different for everyone. Fluff is one example, but there are others.

[Bryce]

The Frame rate is different too, so it doesn't ever miss a frame.

Bryce.

[TFM]

The Frame rate is different too, so it doesn't ever miss a frame.

Bryce.

Ok, so why is Fluff acting obviously so different?

[Badstarr]

I wish I could think of an easy way to do a side by side comparison between the GX and the CPC. The difficulty is that both machines cant be controlled easily at the same time, my Plus machine loads via HxC, this seems a little quicker than loading from the physical disk that the GX currently loads from. Even recording the video output from each machine separately isn't really a good solution as key presses wont be timed exactly the same and loading times aren't directly comparable. Unless I can rig something up that produces the same pattern of key presses etc, this seems a little over the top when we can say with some certainty that there will be a very slight difference in speed but from a users' point of view they would be hard pushed to tell the difference and any difference is more than likely to be subjective/psychological. I would say it's a bit like after I have washed my car, to me it "feels" better, more responsive etc, but there is no way that simply making it clean and shiny could make any difference to the way it runs, just knowing it looks nicer changes my perception I guess. If I didn't know that the GX was clocked at a reduced rate I doubt I would have noticed anything but knowing it is will have more than likely caused me to over scrutinise any normal slowdown and while I do think that Fluff does seem to have slightly more lag, it is, in real terms, barely perceptible.


@TFM/FS Maybe you could code something for FutureOS that would settle this beyond any doubt? I would be more than happy to try it on my CPC and GX. I have tried a couple of BASIC bench mark routines but in all honesty there is only a minuscule difference, in fact, some don't show any differences at all, I suppose this could be due to the BASIC execution speed or something like that. I guess that my GX system does provide a rather unique opportunity to explore the performance of the GX4000 so we should test it out scientifically. 

[TFM]

Yes the differences are minor, the problems arise when a routine uses up exactly 100% of cpu time, then waits for a FRAME fly, then it's just 1-2 ys too late, so it will wait for the NEXT FRAME-fly-back. That will cause a 50% slow down. But nearly no program uses up 100% cpu time in a FRAME.
I will play around as soon as I have access to my machines.
Well, except Fluff, this discussion is probably more or theoretical nature ;-)
I will think how to check, and try to create a small program for checking....

[Badstarr]

ARRGGHHH!    I was just woken up by a rather loud crash, it seems my cat has attacked my GX! I found it about 6 feet away from the work bench where I left it, looks like a couple of wires have come loose hope its ok, I don't have the nerve to try it out think I will wait until the morning. I cant believe it! The first time I didn't pack it away 


UPDATE...Since I was finding it a little difficult to sleep having found my GX4128 in bits on the floor, I got brave and after a bit of a check that the loose wire weren't causing any shorts and assessing the damage elsewhere. I found that one IC had some broken legs and that seemed to be the extent of the damage. I plugged everything in and got a brief flash of the Amsdos screen then nothing. I unplugged it and happened to glance at a capacitor that I hadn;t noticed was at a right angle, (I assume this took the brunt of the fall) its legs were sort of twisted together, and appeared to be part of the scart connection. I untwisted it and tried again but just the Black Screen of Death. I have tried it now plugged in via the monitor socket and it is working, but no f1 f2 menu which makes snese as the IC that controls this is missing some limbs. I'm sure I can fix it will just take some time to work out where the wires need to go back to and find another 74HC405 to restore the f1 f2 menu.


Not so bad I guess, I think now I can get back to sleep, as for the cat, he is currently howling at the door to get back in grrr!