MiSTer FPGA Forum

Interesting comment from Facebook:

The Mister's Next core is a compromise - it's not flawless.
Because its main memory is based on sdram, the memory access is not fast enough to match the Next's sram based memory. It starts to insert wait states at 14 MHz which makes the system a lot slower. You can hear this with a simple basic program that runs at 28 MHz and beeps the same tone with layer 2 on and off. On a real Next there is no change in tone but on a Mister you will hear the tone drop with layer 2 on. For many current programs, this has not been noticeable so far because (1) earlier versions of the Next were limited to 14 MHz and the cpu was slowed to 7 MHz during the active display portion; now the system can always run at 28 MHz and (2) earlier software was written for 14 or even 7 MHz and current software is not pushing the Next hardware yet.

The Mister also has the Arm handle the ps2, joystick and sd card i/o. The arm is too slow to serve the sd card data at full speed so games will only be able to draw data from sd card at reduced rates. This can be noticeable especially if audio or video is played directly from sd card. The playvid dot command, that plays video with digital music, is a kind of double whammy example as it requires running at the full 28MHz rate and it streams video data at top speed directly from the sd card.

It is possible to solve these issues on the Mister with a special zx next core because you can buy a sram add-on module and an sd i/o card that such a core could access directly as on the Next but I think those things are uncommon in the Mister community so almost all users will use the compromise core instead. I believe most other systems the Mister is implementing don't suffer these shortcomings; it's just the base Mister system's hardware that has insufficient bandwidth in some areas to meet the zx next spec.

Beyond that you're talking about the difference between an actual computer and an fpga near-complete clone. Things are always missing on clones because the clone hardware rarely is a superset of the source hardware. Depending on what's important to the user, those missing items may not matter. For example, the zx next core on other systems will be missing many of: rtc, pi zero peripheral, expansion bus, esp wifi, membrane keyboard, ps2 mouse, ps2 keyboard, two sega joysticks, 2 Mbps serial through joysticks / wifi connector, tape i/o, beeper, vga or hdmi, etc.
Anyway the Mister is a great system and I am not trying to talk it down. I'm only pointing out the zx next core is a compromise and not timing accurate on the base Mister especially and that compromise is going to show more as time passes.

Sorg stated on Facebook the timings on the core are correct and I believe he said the core is a bit of mess anyway if my memory serves me correctly

There has been very few core updates though as he doesnt like fantasy computers

There is also a experimental overclock for the ARM side now

The ZX Next die hards are a funny group if it's not the unobtainable Kickstarter version it's not good enough which is odd considering its a FPGA core

After using the ZX next core on MiSTer and Neptuno, I really can't see what all the fuss is about.

The MiSTer core is a port of this core --> https://gitlab.com/SpectrumNext/ZX_Spectrum_Next_FPGA for reference.

Can you link to the facebook post for this? I can't find it.

Sorgelig has added some significant improvements to this core, for instance adding additional prefetch to the sdram to increase performance, improved the wait states, etc... When is that comment from?

Some thoughts from my perspective:

The sd card transfer speed is not limited by the ARM processor, it's limited by the USB bus on the DE10-Nano, which is USB 2.0 I believe. That limitation is pretty tolerant though, you can potentially stream 720p video off of there smoothly, what exactly is the test that this person is referring to?

There is already switching for various CPU speeds in the core (https://github.com/MiSTer-devel/ZXNext_ ... #L218-L221), all the person that wrote this would have to do is actually try and use the MiSTer core and they would see that. Maybe their comment is very outdated since that option was added back in April 8, 2021.

The sram vs sdram comment seems... odd to me. What usable software relied upon the latency of modern-day sram that couldn't be done by sdram? The difference is pretty minimal I would think for something that ran at only 28MHz, but this is my ignorance showing. SDRAM prefetch was added around the same time as the cpu speed switching to increase performance, so I don't know what it's referring to specifically, or if it's outdated.

@aberu It's here...
Expand the replies on this post:
https://www.facebook.com/groups/specnex ... up_comment

CartoonDonkey wrote: ↑Tue Dec 27, 2022 11:19 pm

@aberu It's here...
Expand the replies on this post:
https://www.facebook.com/groups/specnex ... up_comment

Yeah, I have zero belief in the claim that the sd reads keep playvid from being able to play videos. These are the supported video playback formats for playvid:

// 0 = 320x240 with palette 16.7fps 933 stereo samples @ 15.6kHz (frame size = 155) (layer 2)
// 1 = 320x240 no palette 16.7fps 933 stereo samples @ 15.6kHz (frame size = 154) (layer 2)
// 2 = 256x240 with palette 16.7 fps 1866 stereo samples @ 31.1kHz (frame size = 129) (layer 2)
// 3 = 256x240 no palette 17fps 1866 stereo samples @ 31.1kHz (frame size = 128) (layer 2)
// 4 = 256x192 with palette 25fps 933 mono samples @ 23.3kHz (frame size = 99) (layer 2)
// 5 = 256x192 no palette 25fps 933 mono samples @ 23.3kHz (frame size = 98) (layer 2)

None of these come even remotely close to stressing the 80-240mbps read rates of the microSD slot. Not sure what the moderator of that community is currently smoking, but I do wish he'd share, I've had a bad month.

The rates of reads on the MicroSD slot on the MiSTer can handle enough data to stream uncompressed 1080p60 video I think (on paper, but we don't have a GPU so displaying it would be rough).

If the SD card actually is the problem, running over Ethernet to a NAS can be substantially faster.

And the secondary SD card slot isn't uncommon either, as lots of people have the analog IO board.

LamerDeluxe wrote: ↑Wed Dec 28, 2022 8:49 am

And the secondary SD card slot isn't uncommon either, as lots of people have the analog IO board.

It's included on the digital IO board and nearly all the AIOs I've seen

I'd still love to see a beeper speaker working on the next and other cores. Possibly thru gpio?

The facebook post is from Allen Albright, one of the developers of the original ZX Spectrum Next core. I think he knows what he is talking about. He explains the SD speed issue in detail in an answer to this post. If you are in doubt of his points, just try it out. I don't plan to, as I am sure he is right.

Grumpy-Old-Gamer wrote: ↑Tue Dec 27, 2022 3:51 pm

The ZX Next die hards are a funny group if it's not the unobtainable Kickstarter version it's not good enough which is odd considering its a FPGA core

Not sure what you are talking about. There are several next clones available (n-go, zx-dos) and there is an alternate Next core for the MiSTer that relies on SRAM instead of SDRAM that isn't affected by the speed issue. Allen Albright points this out specifically in his post:

"It is possible to solve these issues on the Mister with a special zx next core because you can buy a sram add-on module and an sd i/o card that such a core could access directly as on the Next but I think those things are uncommon in the Mister community so almost all users will use the compromise core instead."

I don't see any ZX Next user complaining about the core running on alternate hardware, I just see a MiSTer die hard making wild accusations about ZX Next users.

Chris23235 wrote: ↑Wed Dec 28, 2022 6:20 pm

The facebook post is from Allen Albright, one of the developers of the original ZX Spectrum Next core. I think he knows what he is talking about. He explains the SD speed issue in detail in an answer to this post. If you are in doubt of his points, just try it out. I don't plan to, as I am sure he is right.

Grumpy-Old-Gamer wrote: ↑Tue Dec 27, 2022 3:51 pm

The ZX Next die hards are a funny group if it's not the unobtainable Kickstarter version it's not good enough which is odd considering its a FPGA core

Not sure what you are talking about. There are several next clones available (n-go, zx-dos) and there is an alternate Next core for the MiSTer that relies on SRAM instead of SDRAM that isn't affected by the speed issue. Allen Albright points this out specifically in his post:

"It is possible to solve these issues on the Mister with a special zx next core because you can buy a sram add-on module and an sd i/o card that such a core could access directly as on the Next but I think those things are uncommon in the Mister community so almost all users will use the compromise core instead."

I don't see any ZX Next user complaining about the core running on alternate hardware, I just see a MiSTer die hard making wild accusations about ZX Next users.

Just spend some time on some ZX Next groups especially FB and you will see what I mean. If you don't own the official Kickstarter version it's not good enough .

Grumpy-Old-Gamer wrote: ↑Sun Jan 08, 2023 1:20 pm

Just spend some time on some ZX Next groups especially FB and you will see what I mean. If you don't own the official Kickstarter version it's not good enough .

Never had any problems. I use a n-go on regular basis and everybody treats it as the same as the kickstarter Next. If it comes to the MiSTer, it is just that it isn't the same when you use the official core, because the SDRAM is not the same as the SRAM resulting in a different behavior of the machine.
In general the Spectrum scene is the most tolerant retro computing scene when it comes to machine varieties that I ever seen. This is mainly because the original Spectrum lived aside so many cloned versions from all over the world.
All the Pentagon, Scorpion, etc. are close but not identical to the original machine. The Next is designed to mimic all these clones (just as the regular Spectrum clone on the MiSTer or the first FPGA implementation of the Spectrum the ZX UNO). This shows to me that the community for these machines is welcoming when it comes to different machines.
I remember one of the main developers being very supportive when it came to answering questions about the machine to the MiST developer of the Next core. The different implementations MiST, MiSTer, n-go even have their own splash screen within the Next firmware. How much more welcoming can you get.
On the Spectrum there is a worldwide community helping out each other, patching programs so that they run on certain machines (more often then not the homebrews are made with something else then the Spectrum in mind and have to be adapted for the Spectrum in order to work).
I never saw anybody looking down on certain Spectrum versions (apart from the people who object the use of modern hardware e.g.FPGA clones in general, but these people are for sure not on the ZX-Next groups).

I do fully agree with aberu. But the facef_ck post seems to be unaware of these details

ZX Next SRAM release for MiSTer: https://github.com/RW-FPGA-devel-Team/ZXNext_Mister

I would recommend to the guy from caralibro facef_ck to make a dual SRAM + SDRAM module and to re-read what he has written because criticism is totally unnecessary. (if he really knows there is a SRAM version, I doubt it)

I have tried the SRAM version as much as I could and it works really well, if you really want to enjoy 100% of the ZXNext in MiSTer the easiest thing is to get the dual memory module. There's no more.

And yes, what kind of shit may be smoking, nothing good

Long before sorgelig ported the ZX Next to SDRAM, benitoss and rampa had already ported it to Altera (MiSTer) using the same SRAM module designed and published by sorgelig.: https://github.com/MiSTer-devel/Hardwar ... SDRAM_SRAM

We use a 32 MB SDRAM module + 2 MB SRAM, I have an N-GO made by ManuFerHi and when I use the SRAM core version for MISTer it works perfectly. ( Comparing MiSTer performance and results) Everyone who knows a little about how an SDRAM instantiation works with a Z80 knows that the only way it works is through wait states.

I think you have to find out a little before raising a post in which a guy from caralibro (facef_ck) says that the SDRAM version for MiSTer has deficiencies. oh! wow great news.

ron wrote: ↑Sun Jan 08, 2023 8:47 pm

I do fully agree with aberu. But the facef_ck post seems to be unaware of these details

ZX Next SRAM release for MiSTer: https://github.com/RW-FPGA-devel-Team/ZXNext_Mister

I would recommend to the guy from caralibro facef_ck to make a dual SRAM + SDRAM module and to re-read what he has written because criticism is totally unnecessary. (if he really knows there is a SRAM version, I doubt it)

I have tried the SRAM version as much as I could and it works really well, if you really want to enjoy 100% of the ZXNext in MiSTer the easiest thing is to get the dual memory module. There's no more.

And yes, what kind of shit may be smoking, nothing good

Long before sorgelig ported the ZX Next to SDRAM, benitoss and rampa had already ported it to Altera (MiSTer) using the same SRAM module designed and published by sorgelig.: https://github.com/MiSTer-devel/Hardwar ... SDRAM_SRAM

We use a 32 MB SDRAM module + 2 MB SRAM, I have an N-GO made by ManuFerHi and when I use the SRAM core version for MISTer it works perfectly. ( Comparing MiSTer performance and results) Everyone who knows a little about how an SDRAM instantiation works with a Z80 knows that the only way it works is through wait states.

I think you have to find out a little before raising a post in which a guy from caralibro (facef_ck) says that the SDRAM version for MiSTer has deficiencies. oh! wow great news.

I am confused here. Do you talk about the post the OP of this thread has quoted?
Have you really read this post? The "guy" as you call him is Allen Albright, he is one of the ZXN developers and he writes:

It is possible to solve these issues on the Mister with a special zx next core because you can buy a sram add-on module and an sd i/o card that such a core could access directly as on the Next but I think those things are uncommon in the Mister community so almost all users will use the compromise core instead. I believe most other systems the Mister is implementing don't suffer these shortcomings; it's just the base Mister system's hardware that has insufficient bandwidth in some areas to meet the zx next spec.

He is fully aware of the SRAM version of the core and he is not talking about the SRAM version at all. He is talking about the SDRAM version. He is explicit stating that the SRAM version isn't affected by this. But maybe you refer to another post, as I said. I am confused.

As an aside... I'm looking to buy a real Spectrum Next if anyone has one for sale.

CartoonDonkey wrote: ↑Mon Jan 16, 2023 9:02 am

As an aside... I'm looking to buy a real Spectrum Next if anyone has one for sale.

Best place to look for one would be the fb group, there are regulary people selling their Nexts for reasonable prices.

An alternative would be an n-go, identical to the original KS1 next (with some improvements, e.g. no HDMI backpower and a power switch), but the case is not as nice as the next case (even if the keyboard is better with mechanical keys).

Thanks, yeah well I did own an N-Go and while it was nice - I really like the styling of the real Next. Sold my N-Go and it paid for my MiSTer and then some. So I enjoy the Next core on that. But I think I still like the idea of real hardware for the Next itself.

Tried the FB group...and actually got someone who was willing to sell to me. But when I suggested he send photos of the unit and set up a PayPal Merchant account to make the sale to me with... (So that both parties are protected.) He only wanted a friends and family payment (no fee) and became immediately insulted that I didn't trust him, (A total stranger to me) & rescinded his offer to sell me his Next. I threw in a "nice try!" in my last message to him, legitimately thinking it was a scammer. He went on to the FB forum to tell everyone how saddened he was that no one trusts one another any longer and how deeply offended he was etc. (I guess they're a tight knit community and I didn't really research the guy to see if he was a real person or a scammer.) They've had scammers advertise there before, in fact very recently! Anyway, I apologized profusely to the guy, but not before everyone in the thread jumped on me claiming that my attitude was a travesty and that I "didn't deserve" a Next anyway. Nothing like being judged by a bunch of people you barely know, for trying to take precautions in order to not get scammed out of hundreds of clams. I mean my god. If he met me in person, we already have a lot of the same likes. I'm sure we'd chill out and play some friggin Archon or Speedball II, share a pint and have a great time. I found it completely insane how quickly he rushed to judge me for it all. A friend of mine after I told this story to him: "Effin' Speccy users!" -lol

Anyway, I've made my amends there. I'll keep trying. :-/

Could be alts trying to perpetuate fraud.

Any community that gets upset over you trying to protect yourself from people you don't know isn't worth joining. They've gotten too insular.

CartoonDonkey wrote: ↑Thu Jan 19, 2023 1:08 am

Thanks, yeah well I did own an N-Go and while it was nice - I really like the styling of the real Next. Sold my N-Go and it paid for my MiSTer and then some. So I enjoy the Next core on that. But I think I still like the idea of real hardware for the Next itself.

Tried the FB group...and actually got someone who was willing to sell to me. But when I suggested he send photos of the unit and set up a PayPal Merchant account to make the sale to me with... (So that both parties are protected.) He only wanted a friends and family payment (no fee) and became immediately insulted that I didn't trust him, (A total stranger to me) & rescinded his offer to sell me his Next. I threw in a "nice try!" in my last message to him, legitimately thinking it was a scammer. He went on to the FB forum to tell everyone how saddened he was that no one trusts one another any longer and how deeply offended he was etc. (I guess they're a tight knit community and I didn't really research the guy to see if he was a real person or a scammer.) They've had scammers advertise there before, in fact very recently! Anyway, I apologized profusely to the guy, but not before everyone in the thread jumped on me claiming that my attitude was a travesty and that I "didn't deserve" a Next anyway. Nothing like being judged by a bunch of people you barely know, for trying to take precautions in order to not get scammed out of hundreds of clams. I mean my god. If he met me in person, we already have a lot of the same likes. I'm sure we'd chill out and play some friggin Archon or Speedball II, share a pint and have a great time. I found it completely insane how quickly he rushed to judge me for it all. A friend of mine after I told this story to him: "Effin' Speccy users!" -lol

Anyway, I've made my amends there. I'll keep trying. :-/

You realise that setting up a merchant account would not only result in a fee the seller has to pay to PP but also make the sale a business transaction making the transaction taxable for the seller?

Of course. Problem?

• deleted by me, no need in derailing the thread to much -

ron wrote: ↑Sun Jan 08, 2023 8:47 pm

...

We use a 32 MB SDRAM module + 2 MB SRAM, I have an N-GO made by ManuFerHi and when I use the SRAM core version for MISTer it works perfectly. ( Comparing MiSTer performance and results) Everyone who knows a little about how an SDRAM instantiation works with a Z80 knows that the only way it works is through wait states.

...

I find it surprisingly hard to find information on how to get/buy/make an SRAM card for the MiSTer. Any pointers on how to find/create one are appreciated!

https://www.atari-forum.com/viewtopic.php?t=32920

https://github.com/MiSTer-devel/Hardwar ... 5b901adcb1