ZX Spectrum Next Kickstarter 2 just launched

Tell us about your other hobbies & none QL related items here :)
User avatar
Peter
QL Wafer Drive
Posts: 1953
Joined: Sat Jan 22, 2011 8:47 am

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by Peter »

mk79 wrote:
Peter wrote:The type of SDRAM used for 68K cores is the same for Mister and Q68, so where should the extra performance come from, especially in highcolour? I don't know how much faster a TG68K.C core for the Mister could be clocked while decently tied to SRDAM, but I doubt it is much, with conservative timings and settings.
There‘s also 1GB of DDR3 RAM and copious amounts of block RAM which can surely be combined into some decent VRAM, freeing SDRAM bandwidth for CPU use.
No, the block RAM is too small for Q68 graphics. DDR3 RAM seems never used for Mister projects, maybe too complex while not providing much improvement in RAS latency compared to SDR.


User avatar
mk79
QL Wafer Drive
Posts: 1349
Joined: Sun Feb 02, 2014 10:54 am
Location: Esslingen/Germany
Contact:

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by mk79 »

Peter wrote:No, the block RAM is too small for Q68 graphics. DDR3 RAM seems never used for Mister projects, maybe for the reason that it does not provide a worthwhile improvement in RAS latency compared to SDR.
I specifically wrote „combined“. The DDR RAM is connected to the ARM core and latency is apparently not good enough for cycle accurate implementations but for VRAM it will do, if necessary by employing its high throughput and use the block RAM as a low latency cache. Case in point, the de10nano can run a decent Linux desktop without the SDRAM extension board, so I‘m sure it would work if I wanted.


User avatar
Peter
QL Wafer Drive
Posts: 1953
Joined: Sat Jan 22, 2011 8:47 am

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by Peter »

mk79 wrote:Case in point, the de10nano can run a decent Linux desktop without the SDRAM extension board, so I‘m sure it would work if I wanted.
That one is using the hardwired cores specifically designed for the chip. I have no time for a closer look, but I'm sure there's a good reason why all Mister projects exceeding block RAM seem to rely on the SDR module. The SDR module is even mentioned as a basic Mister component. Maybe DDR3 support would just be too much work to make sense.


User avatar
mk79
QL Wafer Drive
Posts: 1349
Joined: Sun Feb 02, 2014 10:54 am
Location: Esslingen/Germany
Contact:

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by mk79 »

Peter wrote:That one is using the hardwired cores specifically designed for the chip. I have no time for a closer look, but I'm sure there's a good reason why all Mister projects exceeding block RAM seem to rely on the SDR module. The SDR module is even mentioned as a basic Mister component. Maybe DDR3 support would just be too much work to make sense.
I already wrote that it‘s due to the latency. But IIRC there is a core that uses it to store a CD image or whatever. I‘ve used it myself to store the MD image to free up block RAM, so I know very well what I‘m talking about...


User avatar
Peter
QL Wafer Drive
Posts: 1953
Joined: Sat Jan 22, 2011 8:47 am

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by Peter »

mk79 wrote:I already wrote that it‘s due to the latency.
But in the context of cycle accuracy, which is not relevant for the TG68K.C and others. So if DDR3 was easy to use apart from the ARM CPU, it still makes me wonder why the SDR module isn't saved.
mk79 wrote:But IIRC there is a core that uses it to store a CD image or whatever. I‘ve used it myself to store the MD image to free up block RAM, so I know very well what I‘m talking about...
Aren't those images handled by the ARM CPU, rather than 68K?

Sorry for going off topic - I hope that is okay inside an off-topic topic ;)


User avatar
mk79
QL Wafer Drive
Posts: 1349
Joined: Sun Feb 02, 2014 10:54 am
Location: Esslingen/Germany
Contact:

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by mk79 »

Peter wrote:
mk79 wrote:I already wrote that it‘s due to the latency.
But in the context of cycle accuracy, which is not relevant for the TG68K.C and others. So if DDR3 was easy to use apart from the ARM CPU, it still makes me wonder why the SDR module isn't saved.
Well, most cores didn't originate on MiSTer and thus had no DDR to play with anyway. If you want to read from Mr MiSTer himself: https://www.atari-forum.com/viewtopic.p ... 70#p390270 "Some cores don't need to be very cycle accurate, so they can use DDR3.". As I use the fx68k core with the standard 68k DTACK handling it would be quite trivial to adapt it to DDR and use SDRAM for video. But I guess it makes more sense the other way round.
mk79 wrote:But IIRC there is a core that uses it to store a CD image or whatever. I‘ve used it myself to store the MD image to free up block RAM, so I know very well what I‘m talking about...
Aren't those images handled by the ARM CPU, rather than 68K?
Images are passed from the ARM to the core as a handshaked stream of data. Where the core stores the stream is its business. Redirecting it to DDR wasn't difficult.

Code: Select all

	// ARM -> FPGA download
	output reg        ioctl_download = 0, // signal indicating an active download
	output reg  [7:0] ioctl_index,        // menu index used to upload the file
	output reg        ioctl_wr,
	output reg [24:0] ioctl_addr,         // in WIDE mode address will be incremented by 2
	output reg [DW:0] ioctl_dout,
	input             ioctl_wait,


User avatar
Peter
QL Wafer Drive
Posts: 1953
Joined: Sat Jan 22, 2011 8:47 am

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by Peter »

mk79 wrote:As I use the fx68k core with the standard 68k DTACK handling it would be quite trivial to adapt it to DDR and use SDRAM for video.
That, or even better vice-versa, makes sense if you plan to add highcolour - even if beating the Q68 in terms of speed is not your goal.

By the way, I don't think the Q68 would suffer if someone releases a faster FPGA implementation for a general purpose development kit like the DE10-Nano or whatever. The appeal of the machine is exclusive QL dedication. Expressed in things like using the native CPU for all software tasks, an aesthetical black case, two microdrive-like slots on the front, a stable fixed 68K-only hardware, classic peripherals like QLNET, RS-232, 8-bit bus, and last but not least Sinclair style simplicity. The aim of the Q68 is to be a better QL and nothing else. That's different from "here's a large FPGA and look it can do almost everything".

Which brings me back on topic: I think a QL FPGA core for the ZX Spectrum Next is a cool thing to play with. But the machine is too far from a QL to be exciting for QL purposes. That would require at least a different keyboard. Or ideally a different case, keyboard and PCB with at least Q68 qualities.


User avatar
bwinkel67
QL Wafer Drive
Posts: 1187
Joined: Thu Oct 03, 2019 2:09 am

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by bwinkel67 »

Peter wrote:But the machine is too far from a QL to be exciting for QL purposes. That would require at least a different keyboard. Or ideally a different case, keyboard and PCB with at least Q68 qualities.
I agree with that. Happy they did a second one but too expensive to get it and not use it much :-/


User avatar
Sparrowhawk
Super Gold Card
Posts: 624
Joined: Wed Dec 15, 2010 12:33 pm
Location: @131072
Contact:

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by Sparrowhawk »

When I get mine I'll try to do a review of it running the QL core, if it's of interest to anyone.

It's not due til August 2021, so don't hold your breaths though :) (and KS1 had loads of delays, so who knows when KS2 will actually deliver)


a.k.a. Jean-Yves
User avatar
vanpeebles
Commissario Pebbli
Posts: 2815
Joined: Sat Nov 20, 2010 7:13 pm
Location: North East UK

Re: ZX Spectrum Next Kickstarter 2 just launched

Post by vanpeebles »

I decided against getting the new model ps5 or xbox, and spent the money on one of these instead. Hope I made the right choice :D


Post Reply