10MHz, 12.5MHz and Beyond!

Using, learning, programming and modding the Gigatron and anything related.
Forum rules
Be nice. No drama.
PurpleGirl
Posts: 48
Joined: 09 Sep 2019, 08:19

Re: 10MHz, 12.5MHz and Beyond!

Post by PurpleGirl »

Potato Semiconductors only has a few of the logic chips used. I mention them since their chips run around the GHz range.

http://www.potatosemi.com/

It seems they only have the OR gates and a couple of different buffer chips that are used by the Gigatron. It seems that they have none that are used by the accumulator or the ALU. These seem to be where the bottleneck is.
Last edited by PurpleGirl on 09 Oct 2019, 12:52, edited 1 time in total.
at67
Site Admin
Posts: 647
Joined: 14 May 2018, 08:29

Re: 10MHz, 12.5MHz and Beyond!

Post by at67 »

I've been playing with this ROM for the last couple of days and wow, Gigatron12.5MHz here I come.

P.S. I don't get the truncated screen shown on real hardware and in other emulators on my emulation, is this something you solved in the ROM since those pictures were posted?
Attachments
ROMv3y_0.jpg
ROMv3y_0.jpg (376.27 KiB) Viewed 10029 times
User avatar
marcelk
Posts: 488
Joined: 13 May 2018, 08:26

Re: 10MHz, 12.5MHz and Beyond!

Post by marcelk »

That ROM v3y hasn't changed. The programmed refresh rate is for 31.25 kHz at 12.5 MHz. So at that speed, shrink the displayed pixels...

Last week I tested it again on my 15 MHz system after soldering the last color resistors (I had ran out of 820 Ohms). But then I found the game controller doesn't register anything anymore, and the keyboard adapter can't keep up either: it was spitting out multiple keystrokes for every key pressed.
bmwtcu
Posts: 145
Joined: 01 Nov 2018, 12:02

Re: 10MHz, 12.5MHz and Beyond!

Post by bmwtcu »

Hi Marcel, just so I understand... I know that Dave Jones worked on the 4-layer PCB, but is it possible to buy one from you? Thanks!
User avatar
marcelk
Posts: 488
Joined: 13 May 2018, 08:26

Re: 10MHz, 12.5MHz and Beyond!

Post by marcelk »

Simply drop us an email. We still have a few.
Sugarplum
Posts: 93
Joined: 30 Sep 2020, 22:19

Re: 10MHz, 12.5MHz and Beyond!

Post by Sugarplum »

It might not help much, but you could try making a skip-carry adder. Just replace the upper nibble adder with 2 adders with the carry-in line on one tied to Vcc and the other carry-in tied to the ground. Use the carry-out from the low nibble to drive a multiplexer that controls which high-nibble adder goes on the bus. For no carries, the latency should be about the same. For carries, this should reduce the latency by about 10 ns or so.

Beyond the other mods such as a more compact board, using the fastest surface-mounted parts, etc., adding another pipeline stage might be the only feasible option. However, that might cause there to be 2 delay slots and would require code to be rewritten.
Last edited by Sugarplum on 30 Dec 2020, 22:06, edited 1 time in total.
monsonite
Posts: 101
Joined: 17 May 2018, 07:17

Re: 10MHz, 12.5MHz and Beyond!

Post by monsonite »

Sugarplum,

The Gigatron as it stands, is a marvellous piece of minimalist TTL design, backed up with incredible software. We truly have Marcel to thank for his insight and vision.

As a colour, games machine, it spends all of it's time pushing pixels at the VGA screen. Imagine what it could do if unburdened from this task?

With "Fast" TTL I have achieved 12.5MHz, and I have yet to build up the experimental 4-layer pcb, that Marcel kindly sent me. This might just allow us to push the clock up to about 16MHz. The 27C1024 ROM is rated at 45nS, but I believe that some of them will still function with a 30nS access time.

If we remove the need for 1/4 VGA video generation, and thus the need for precise H-sync and V-sync timing in software, we can take over the Bus-In shift register and clock it at whatever rate we wish, and also fit a shift register to the video output, the aim being creating an interface that is compatible with SPI. With SPI we gain access to a huge range of modern peripherals, UARTs, flash memory, LCD displays, video generation hardware etc.

The Gigatron will always be reliant on some "external assistance" - or as Marcel said to me in Cambridge last year "Thinking outside of the box" - so why not glue on a $20 Teensy 4.0 and have that provide the interface to the outside world?

There are examples of a Teensy being used as the entire memory and peripheral sub-section to service retro cpus

http://8bitforce.com/

Overkill - yes, but just an upgrade for "Pluggy"

What do you think?
Sugarplum
Posts: 93
Joined: 30 Sep 2020, 22:19

Re: 10MHz, 12.5MHz and Beyond!

Post by Sugarplum »

Yes, the Gigatron is marvelous as it is. And I think the term for the adder mod would be a carry select adder. If one is going for an optimized 4-layer PCB and possibly SMT parts, I think the carry select design would help push for a higher clock rate or improve stability at higher frequencies.

If one wants to keep a 2-stage pipeline, then they would likely want to have more latency parity between the 2 stages. A 3-stage pipeline might be interesting, though that might add another delay slot. So if we can reduce the latency of the control unit and/or the ALU, we could have a faster clock rate. The fetch latency is already less than that of the execution stage, so there is room for a faster clock rate while keeping the 2-stage pipeline.

The shift register and your thread about having a mono mode sound interesting. I wouldn't worry about overkill. I am still planning a way to do some of the things I've had in mind. So overkill can help you gain clarity. It is much like coding. You first get a crude design that works and then you clean it up and optimize it.

I think video mods would be one area to improve performance. And if you have more power due to more dedicated video production, then I/O would be another area to flesh out more.
Dotoro
Posts: 1
Joined: 07 Apr 2021, 20:44

Re: 10MHz, 12.5MHz and Beyond!

Post by Dotoro »

monsonite wrote: 18 May 2018, 11:49
I had chosen to fit sockets throughout the design, so that 74Fxx series of ICs could be fitted. These I sourced on ebay, from various small suppliers, for about £0.50 each in small volume. I also bought some 10nS SRAM, some 45nS OTP ROM, and some cheap adaptor pcbs - so that a SRAM with a SMT package could be fitted into the 28 pin DIL socket on the Gigatron.
Hello i have Gigatron board from your github (Gigatron-CAD)
if i use 74LS logic what crystal should i use - 8mhz?
if i use 74HCT can or should it be 10mhz?
and for 74F can or should i use 12.5mhz? is it stable and was video problem was fixed?
Thank you!)

and can you provide BOM for Gigatron-CAD board, photo maybe,
thanks again!)
at67
Site Admin
Posts: 647
Joined: 14 May 2018, 08:29

Re: 10MHz, 12.5MHz and Beyond!

Post by at67 »

Dotoro wrote: 07 Apr 2021, 20:53 if i use 74LS logic what crystal should i use - 8mhz?
Most people who have overclocked with LS/HCT have gotten to around 8~9MHz.
Dotoro wrote: 07 Apr 2021, 20:53 if i use 74HCT can or should it be 10mhz?
Not sure you can get to 10MHz on HCT without some other changes to the clock circuit and possibly the diode decoding matrix, (there are extensive tests performed by Monsonite and others in this thread).
Dotoro wrote: 07 Apr 2021, 20:53 and for 74F can or should i use 12.5mhz? is it stable and was video problem was fixed?
  • ACT would be better than F for noise immunity, ringing and especially current draw, (assuming it's attainable).
  • The video problem, (squashed pixels), has not been solved as far as I know, it requires a re-write of the ROM native code, there is a ROMv3y.rom that Marcel produced that is a partial solution, (still has squashed pixels).
  • You need a monitor, LCD or display device that can handle the non VGA spec signals.
  • Whether overclocking is stable or not is always dependent on not just the specifications of the components chosen, but of the individual characteristics of the components as well, the noise immunity of the underlying PCB and ringing/transmission line characteristics of the PCB traces.
  • Once you get over 12MHz, it would seem a 4 layer board with individual +5V and GND planes will help with overclocking success substantially, (see the main thread).
Post Reply