Nintendo 64

[night shell]

Right here

That's a static screen now

[hollow eagle]

Thanks Ill try that!

[neat sierra]

How could I have forgotten about the safety dance rom, it’s amazing (that it exists)

[vast delta]

there are people having issues with Mario 64

[night shell]

I'll run again and then try the turbo mode

[vapid hawk]

what i do is use an alternate ini thats identical to my main one except with vsync 1 on

then i switch to that for the small number of games that do tons of switching

dino crisis/silent hill/crono cross etc

[night shell]

Second run, crash at nearly the same spot

It's a good job I like this song

[kind mountain]

Uhoh have we found a new test

[deft matrix]

I figured I would try a few more games. F-Zero X goes right to a black screen, Wave Race 64 does the same but does load if I reset the core and then crashes later. So for me at least it isn't just the larger ROMs

[night shell]

The best Test

yup, video played to completion on turbo core

[vast delta]

wdym uhoh. HELL YEA we found a new test

tests > games

[mortal panther]

Where's Bad Apple when you really need it? 😏

[night shell]

...I'm literally just testing that and it does seem to crash 😅

[vast delta]

i'm trying to make sense of the pll config registers to see if I can modify memtest, but won't be able to take an in-depth look before sunday

[night shell]

I8LZRLEVideo.N64 I think crashes

Devo-TimeOutForFun.N64 also crashes - definitely think @wanton sun is onto something - the bigger ROMs do crash after a certain point

[hollow eagle]

Ok so I was already on vsync_adjust=1, so I kept that the same. I removed rpak from the N64-database.txt RE2 lines, and now its working perfectly with no black screens. Tested playing for 10 mins.

[deft matrix]

Sounds like your display is fussy then

[worn delta]

Is this a music video? Does it definitely work on DE-10 nano? I know some videos don't work properly

[night shell]

I have no idea, I don't have a DE-10 nano to test. But When using the turbo core, they don't seem to crash

[iron wren]

What part of the body is a fussy

[worn delta]

Also without posting a direct link, where do you find these music videos for N64 like Safety Dance? I have not come across them in packs

[night shell]

https://github.com/PeterLemon/N64/ I found them here

[deft matrix]

Foot puss-

[night shell]

\Video\I8LZRLEVideo\I8LZRLEVideo.N64
\Video\MenWithoutHats-SafetyDance\MenWithoutHats-SafetyDance.N64
\Video\Devo-TimeOutForFun\Devo-TimeOutForFun.N64

All 3 of these freeze at some point during their runtime on the normal N64 core, but run fine on the turbo core

(On my Mister Pi stack)

\Video\Pixar-Geri'sGame\Pixar-Geri'sGame.N64 seemed to work fine, but that's a slightly smaller file

[worn delta]

We could do with a GutHub issue for this one, it does come up periodically (specifically Safety Dance). Anyone know why there are issues with these running if they work on real hardware?

[night shell]

For record, BigBuckBunny.N64 does not even boot - but that's expected, given its size

[hearty python]

Many Neo Geo Games do the same...

[deft matrix]

I did check the KoF 2k3 ROM (89MB) and it ran okay

[hollow eagle]

Just tested all three on my Mister Pi stack on the normal N64 core, all worked fine no freezes.

[night shell]

and you're not having any issues with any of the games, are you?

[hollow eagle]

Correct.

[vast delta]

the old jotego sdram test core ran fine for you at 48mhz as well, right?

[hollow eagle]

Thats right. I ran it for 10 mins without issue.

[night shell]

seems like you're a lucky one!

[kind mountain]

So far you have the only fully functioning stack

[hearty python]

No blinking LED ?

[hollow eagle]

No blinking LED, and no sounds made.

[hearty python]

Strange

[wanton sun]

actually i was wrong, only 1.5mbyte of the safety dance would be on the second chip

[wintry lark]

taki ram - blinking led. resident evil 2 on n64 core crashes quick. switch to a 2.9 spare stick i had, no crash

swapped back and forth 6x now. replicates easily

[deft matrix]

did you say you had a de10-nano based setup too?

[wintry lark]

i have 2 other misters.

[deft matrix]

right, if you don't mind would you try the taki sdram module in one?

[karmic sequoia]

do i need to test the men without hats to enter this column

[wintry lark]

that i can do soon no problem.

[kind mountain]

Oh yeah you have a valid 48mhz test too eh

Maybe we have 2 champions

[wanton sun]

ignore the safety dance unless you enjoy hearing it 🤣

[wintry lark]

my 48mhz test on both the 2.9 and taki ram blinks LED almost immediately.

[wanton sun]

but i can say the safety dance plays fully until the end(black screen) on my mister with no hang

[karmic sequoia]

if your friends don't dance then they're no friends of mine

[wanton sun]

with the known flickering boxes on screen

[wintry lark]

my n64 core is 24/08/18.

re2 fails after the nintendo logo with taki ram.

read 48 fails with taki ram. immeidately. 2.9 kit fails after about 30s

[knotty zealot]

taki ram rhymes with tacked on

[neat sierra]

Maybe takis factory has trouble getting the real ICs just like we do when we order from Ali. Try cleaning the label on the ram with some 99% IPA and see if it comes off 😜

[wintry lark]

swapped back to 2.9 ram. re2 works perfect. i think the mem is a concern

[knotty zealot]

not for me, i don't have the taki ram from the chinese menu

i have misteraddons quality samurai steel

[neat sierra]

They fold every ram chip 100 times

[worn delta]

Any idea what causes the flickering? Seems an odd one if it isn't like this on real hardware

[neat sierra]

Poor Robert is trying to make his game and everyone is ram this ram that, fix my safety dance 😉

[kind mountain]

But yet Jesusfish tested his taki ram on his de10 and it passes 48mhz test ok

It’s some weird combo of the board and ram together

[wanton sun]

i never tried on hardware, but as far as i tried with emulators, the core is the best so far. Most either crash or show only pink garbage

must be using some really weird things

[neat sierra]

“Most complete version of N64 safety dance 10/10” - Gamespot

[worn delta]

Potentially some of those things that say in the developer guides from the time "do not use"?

[wanton sun]

considering it immidiatly brings up some unimplemented error flags, i would guess that is the case

[wintry lark]

taki ram in a mister addons setup, no crashes with re2. that same ram in the mister pi crashes 100% of the time.

[deft matrix]

same as thejesusfish then, what the hell

[rugged nexus]

so if you swap your Taki Ram and Mister Add ons ram, both MiSTer setups would work perfectly?

[kind mountain]

Yep same results Jesus saw

[wintry lark]

so. taki ram in my mister addons setup. 48 ram test passes. no crashes with the suspect n64 games.

mister addons ram in mister pi. both the taki and mister addons ram fail the 48 ram test. re2 fails with the taki ram but not the MA ram......

[knotty zealot]

this is just what we needed. ram 1, ram 2, ram 3, ali ram, v2.5 ram, v3.0 ram. it works, it doesn't. it works, it doesn't. i have this ram. you have that ram. what is even ram.

all this taki TROUBLE is loaded onto devs to figure shit out

[wintry lark]

i have 2.5 ram, 2.9, 3.0, multiple sticks of each.

[vapid hawk]

i'm just waiting for the return of

[wintry lark]

taki ram works in the MA setup. read 48 passes. re2 doesn't fail. that same ram in the taki setup, fails.

[neat sierra]

Version 1.0 ram is probably worth a fortune in the collectors market

[wintry lark]

mister addons 2.9 mem in the taki setup, re2 hasn't crashed as fast as taki ram, but the read test fails.

[rugged nexus]

we need a flow chart

[wintry lark]

ugh lol

[stone pilot]

I suspect Taki himself is working overtime on it, too. Everyone here is testing and tinkering because they like it, and they're helpful

[wintry lark]

yep. I'm not mad or anything. I'm all for testing and tinkering. hope i can be of some help

[stone pilot]

The fact that the new RAM and the new FPGA need to be combined in order to fail is super weird, and points at sensitivity that I didn't see anyone expect. I think it's also why QMTech and Taki both started with clones: there are most likely plenty of things in the MiSTer stack that basically accidentally work

[wintry lark]

but, the taki board with mister addons ram, read 48 fails. but taki ram and mister addons ram work and pass read 48 in mister addons setup. (but, some claim read 48 isn't valid, so take it for what its worth)

[rich warren]

Taki stack on standard core

• 48mhz: bottom 1/8th of screen red and yellow light flash after 3 minutes, no freeze
• Conker, RE2 run fine
• Devo video: no freeze
• I8 Anime video: no freeze
• Safety Dance: no freeze

[stone pilot]

Ah, my bad

[knotty zealot]

https://tenor.com/view/yuriatomoki-yuria-riku-tomoki-riki-japan-gif-26124719

get a load of THIS

[thorn quiver]

🤔

[twin barn]

Anyone with issues on the Taki clone should first reach out to Taki's support potentially for an RMA in my opinion.

It really sounds like a design flaw that needs to be fixed or a manufacturing inconsistency at this point and not any kind of issue with the design of the core.

If someone is able to address it in the core, see that as a surprising bonus.

[knotty zealot]

have him figure out a glitch if it doesn't work on his ram

if it doesn't happen on the samurai steel ram

[thorn quiver]

Samurai steel ram

What's next, RGB?

[neat sierra]

Someone send a bad ram to another pi user who isn’t having problems

It’s like both ram and board can be borderline and you can’t have two borderline together. Maybe the pins and traces are made of wishes

[knotty zealot]

i wish upon a star

[hearty python]

Do a further test with my second taki ram board. This work also like the first one. No problems in N64, stable at 150 Mhz, but fail also the 48 MHz test. But this time with much more red screen and sound and LED blinking.

[knotty zealot]

next up: advanture taco board

[wintry lark]

im gonna try 2 2.9 ram kits in taki setup.

[mortal panther]

BasketSnake Bundle when?

[wintry lark]

dual ram fails immediately in read 48.

[lean spruce]

wishing for DE25 to release soon

[wintry lark]

if only.

[cloud peak]

Oh no

Are people using jotegos test cores to test their sdram?

Please do not use!

It is using old sdram controller code

[remote hatch]

You can't stop them anyway 😂

[mortal panther]

Like I've said several times already... they have been deprecated on his repo for a reason. They only test the first 32MB anyway.

[cloud peak]

Correct

[raw oriole]

I keep tellin'em

[mortal panther]

It is unreliable to use as 'evidence' of anything. There were plenty of DE10-Nanos with many different SDRAM module combinations where that test core reported failures when it shouldn't have.

That was before his first CPS-2 beta...

So it's been a while.

[native mesa]

DE-10 Mondo

I have a feeling the next batch of misterpis will be delayed after the root cause is determined

[neat sierra]

It’s okay they are distracted from the sus JT test by safety dance

[deft matrix]

uh oh

[kind mountain]

But if passing means everything works then let it add to paired testing lol

[rich warren]

I'm just hopping around the core testing random games and it is so good
👑

[kind mountain]

We seem to have no other test

Other then run re2

[zinc dew]

Reposting this so people are aware of their options.

[native mesa]

Welcome to your personal hell. This is the hobby.

[zinc dew]

Also his Discord - https://discord.com/invite/AX2XdFk

[cloud peak]

Ah too bad

[zinc dew]

Taki is great about following up and responding to people. I suggest joining his Discord and seeing what options are available to you.

And I assume any new information will likely be communicated there first.

[rich warren]

Makes a great background activity while working! This Taki MiSTer is for my kid so I'm making sure everything is working. I've been following the core since the beginning and it still amazes me

[zinc dew]

You’re an awesome dad, that’s such a cool thing to give your kid.

[rich warren]

gonna TapTo it up as well!

[native mesa]

I like the idea of just cycling through different games with different attract screens

Tapto is sick. I printed a bunch of cards, sorry to be off-topic though

I love you Robert

[plush summit]

Cart shaming is my thing, keep TapTo in the news

[kind mountain]

Got mine for my kid too, she’s 2 though so I might as well use it. 🤪

[neat sierra]

It doesn’t though, the known good misteraddons ram was failing the test according to a comment above so probably time to abandon it as a test or view any results very critically

[kind mountain]

Correct but some taki boards have passed it and the de10’s pass it so is it irrelevant? Probably true but why do some pass

[vast delta]

have we seen anyone with good jotego sdram 48mhz testing that is having issues with RE2/CBFD?

[zinc dew]

Because the Jotego test is based upon flaws and incorrect SDRAM logic. So the test is giving you incorrect information.

[vast delta]

sure

[neat sierra]

Oh I’m mistaken thought misteraddons ram in de10 failed the old Jotego memory controller test

But it was misteraddons ram in a mistaki

[vast delta]

but it may very well be anecdotal evidence, if anything

[cloud peak]

If people wants to test their SDRAM there are other cores that pushes the SDRAM heavily

[vast delta]

that something related to lower clock makes itfail

[kind mountain]

I get that but if it follows certain hardware it’s indicating something, no?

[cloud peak]

Neo Geo, CPS2, Outrun

Archie

[neat sierra]

But yeah I still wouldn’t collect and report those tests. Different known flawed memory controller being tested. Could just be noise unless someone verifies what the code is doing

[zinc dew]

It does using logic that’s been found to be problematic and flawed. So the test is effectively providing useless information because the source of verification is incorrect.

[cloud peak]

Or if someone wants to do the ultimate test, run Raizing core

[zinc dew]

Using that memtest doing more harm than good and adding to the confusion.

[vast delta]

it looks like the problem happens when the clocks are not pushed

[zinc dew]

Which comically is also using the outdated and flawed SDRAM logic.

[cloud peak]

Indeed

[vast delta]

hence N64 turbo working

[neat sierra]

Hehe raizing games sometimes fail with True Misters

[zinc dew]

The only available memtest thats accurate and is used by MiSTer vendors is the one included with the official MiSTer repository - https://github.com/MiSTer-devel/MemTest_MiSTer

[cloud peak]

Yes

[hushed nova]

and it's not useful for exercising this issue

[vast delta]

that does not test the clocks the N64 core drives the SDRAM at

[neat sierra]

Yeah you’d need some willing dev to make a new test (or verify Jotegos is maybe valid)

[zinc dew]

Damnit Zakk, you’re like a Ninja. Come out of nowhere to stab me lmao.

[cloud peak]

Also

[hushed nova]

also the raizing core does not use 'outdated' sdram controllers. it has other issues but that's not why

[zinc dew]

Oh ok

[cloud peak]

Do not test with the cores generated by the unstable nightly

[zinc dew]

Thanks for correcting me

Someone else told me that info I swear I’m not making it up lol

@hushed nova with all sincerity, should I stop telling people to avoid using the Jotego memtest?

I don’t want to misdirect anyone

[hushed nova]

that one is old and I don't know what it does or doesn't do. someone that actually knows about sdram timings/refreshes etc would have to verify if it is doing something useful

[zinc dew]

Ok great, I’ll refrain then from commenting any further

Apologies

[cloud peak]

Zakk roms wen

[untold dragon]

Correct me if I'm wrong, but a failure in Jotego's test does not correlate with Misters not functioning correctly, right? All Most Mister Pis seems to fail this test (mine did), but even if they fail, they might be able to run N64 games just fine (again, same with mine). The test isn't actually useful until someone with the proper knowledge can identify it is useful, like Zakk said, or until we actually determine failing the test is a symptom related to and actual Mister functionality issue.

[twin barn]

Jotego had rewritten a lot of his sdram controller after those tests were made years ago.

I didnt realize people were using those old tests lol

They are irrelevant.

[raw oriole]

It's really only come up this last day or two

Someone got really into the old JT ram test and it spread like wildfire, since it almost always shows as a fail on a Taki board

Unfortunate, but people love to see a big red flag

Even if it's completely meaningless

[eternal ravine]

[kind mountain]

Not all. A couple people have had passes so far

[green epoch]

so it's been 48 hours. I'm sure we have a solution, yes? let's have it then

I'm hearing the solution is that you shouldn't play re2 and conker because they are bad games

[neat sierra]

Turbo core no?

[karmic sequoia]

thus far we have two passing reports and in both cases the N64 games worked, so within the current set it appears that passing that test is sufficient to avoid problems but not necessary

[neat sierra]

Seems like the right solution is to reach out to Taki and get a swap out with a mated pair of board & ram that pass the RE2 test but that’s up to him and customers

[karmic sequoia]

and nobody has successfully reproduced an issue with any other core, right?

people have tried PSX, NeoGeo, etc. without problems thus far

[cold crypt]

with the QMtech, soldered SDRAM, conkers plays fine. resident evil 2 freezes when trying to enter an alleyway near the beginning

[green epoch]

aside from the music videos? no

I should try cps 2 to see. what's the most intense cps 2 game? alpha 2?

neo geo kof2003 ran fine

someone said outrun too, so I will test that

[cold crypt]

to be honest, no idea what n64 core it comes with preloaded. 24.08.18

[green epoch]

thats the latest

[hushed nova]

if Robert's hunch about 'bigger' roms being an issue is true, cps2 probably won't trigger the bug either

[worn delta]

Are people with these issues on clone boards taking these issues to the vendors/makers? Realistically this isn't a core issues, at this stage it is clearly a hardware issue. There really isn't anything any of us can do to help fix these problems. The cores work fine on DE-10 Nanos.

[neat sierra]

Garou on neogeo is big right?

[deft matrix]

kof 2k3 is the largest ng rom

89MB

i had no problems with that

[karmic sequoia]

i think we have covered pretty well that Taki at least is aware of the issue and also that "this doesn't work" in a vacuum is largely useless information

[vast delta]

let's remember that sdram clock speed seems to be a factor

Neogeo and CPS2 run it much faster

which correlates to memtest being stable north of 140MHz

[quick light]

I don't think it was necessarily just asking core Devs to help and fix.

But it was trying to collectively find out what isn't working, what is causing the issue and how to reproduce said issue. I guess most of this occurred in this thread considering that's the core the Mister Pi/pi ram didnt like.

[karmic sequoia]

yeah inasmuch as any devs have been involved it is similarly out of curiosity/interest in identifying root cause rather than anyone badgering them for fixes, as far as i've seen

[kind mountain]

Yep no one is expecting devs to jump on fixing anything but we appreciate any insight they can shed so we can try and figure out what is going on.

[green epoch]

yeah the fact that @wanton sun has done anything is above and beyond

devs just know more about what's happening under the hood, so they can help direct testing (if they feel the itch to do it)

it would be helpful to go to taki and say "here is what we think the problem is" rather than "yo, resident evil 2 doesn't work"

[iron wren]

Can’t really expect hardware manufacturers to fix a bug where the cause is unknown and behavior is inconsistent between setups. It’s not even known for sure if qa-mated memory/mr pi would definitively fix the issue.

[random girder]

I feel like nothing is known for sure yet.

[karmic sequoia]

we have a lot more scattered information than we started with but very few strongly-drawn conclusions still

[hushed nova]

there's a consistent enough repo. honestly debugging this probably involves actual hardware debugging

and possibly sitting there with 100s of units trying to figure out if it was a specific defect in assembly/parts etc

[iron wren]

Watch, it’ll be something goofy like the usb dongle taki included causing interference with the taki memory at low speeds

[karmic sequoia]

i did take my taki dongle out and replace it........

[green epoch]

I don't have mine plugged in

but it is sitting near it

[random girder]

As a data point, I’m not using the dongle and RE2 loads.

[karmic sequoia]

maybe it's the bad vibes

but yeah i think Zakk is right that at this point "test a hundred units both to determine frequency and narrow down source issue" is probably the way to go and it might be good for someone to drop a summary of what we've seen into Taki's discord for that purpose

[iron wren]

It can detect your aura and chooses not to let you play re2 if your aura is good (to prevent you from the horror of doing so)

[random girder]

Rip all the people with bad auras.

[hearty python]

The weired thing is, that it seems, that only 2 or 3 games affected, and only on N64 core. So perhaps, just send the new tested rams to the affeced users.

[deft matrix]

I don't think any 64 games are working for me rn :/

[karmic sequoia]

that's part of where the frequency of the issue comes in too. if it is like 20 people out of 2000 it can probably just be treated as a manufacturing issue that needs pre-release testing and RMAing from now on. if it's like 60% of units then that's not really gonna solve the problem. but again the only way to actually get that info is to start with a ton of undifferentiated units and test how many have the issue

[green epoch]

that's the thing that sucks. the fact that there are tiers of malfunction leans more towards hardware

[magic girder]

It would have been a lot funnier if the Taki board only prevented people from playing Superman 64

[quick light]

You probably need to email Taki.

[karmic sequoia]

lolol

[hearty python]

Has some one looked at the taki discord, to see how many really affected there ?

[kind mountain]

It’s mainly just us in a different place

lol

But some people have found their way here that saw my post asking people to test on Reddit and had failures

[magic girder]

Reddit doesn’t work on my DE10, so I suspect that’s not restricted to only MiSTER Pi boards

[iron wren]

Interestingly I just had problems getting re2 to load. Trying to reset with the menu or via selecting the same game again resulted in error code 000040 no matter what. Resetting the device with the digital io board button once resulted in it working perfectly the first try.

[kind mountain]

https://tenor.com/view/wtf-what-the-fuck-the-fuck-is-this-wth-gif-17571836

[deft matrix]

👻

[green epoch]

coooooool

[mossy vector]

The physical button is the real reset. Software reset can't recover from an unstable pll

[random girder]

Did you have any problems with RE2 before?

[deft matrix]

at least I can load up the turbo core to enjoy some N64 while I wonder why my hardware is broken 😌

[iron wren]

I had one crash when all this first started and I wanted to try it out, but reloading the rom worked iirc.

[hearty python]

First crash could also some garbage in the rams left

[random girder]

You have to pull the rom out and blow on it.

[iron wren]

Ah see that’s where I messed up I think

I pulled it out and spit on it

[magic girder]

#spicy-off-topic is over there

[weak hill]

YOURS works fine?

[cold crypt]

i mean... taki aint the only one having issues. qmtech with or without soldered sdram (supposedly more stable) also cannot run re2 on the first try. havent tried resetting etc

not that it actually matters, but interesting to see what's causing it

or well, it runs, but freezes after a bit

[twin barn]

To be honest, no one here recommended anyone buy the qmtech board

[cold crypt]

it seems to perform just as well as the taki board at least, im digging it

lets hope it doesn't explode

[twin barn]

The trace lengths from FPGA to gpio for sdram must not be identical to the de10-nano

[stone pilot]

Interesting that it fails even with soldered SDRAM. I'd have expected that to be the holy grail

[twin barn]

Another possible explanation is that the FPGA SKU is different enough at the margins that there might be timing differences that weren't going to show up in anything but a couple games on the most boundary pushing cores.

[compact depot]

We need that Terasic engineer who showed up at the forums a while ago here

[magic girder]

Surely their response is “we don’t support products that we don’t manufacture”

Sadly, this is just the price you pay for clone boards. Hopefully the issues will be resolved with time, but if you wanted guaranteed performance on day 1, then you should have gone with the Terasic board 🤷‍♂️

[compact depot]

Is there an official statement from Taki yet about what could be the issue?

I think he will need to speak up if he wants batch 2 to be a success

[magic girder]

There was some comment about selling boards and RAM as a pair, which seems… weird, to me.

[compact depot]

Agreed

[magic girder]

Good luck if ever you want to replace your RAM for whatever reason

But perhaps I’m just being overly critical

[paper orchid]

I kinda "get" why he would do that in the sense that you can get a working combination in the hands of people, but it does bring up that issue of "something breaks"

[magic girder]

I think it’s a nice final QC step when you’re selling a bundle, but trying to figure out what’s actually broken and doing a revision would seem the most sensible solution for the long term

[paper orchid]

Right. If I had one of the clone units I'd try to do some digging, though I really don't want to pry my DE-10 out of its case

[kind mountain]

Also he replied he would add the 48mhz test to the plant QC when pairing boards and ram but if that test isn’t valid then it may not do any good. And memtesting at 140-150mhz has proven to be useless as someone who has 125mhz peak results say his N64 works fine

[compact depot]

Wonder how Saturn is doing with that 125Mhz ram

[kind mountain]

Not sure what the answer to those of us with existing setups with problems will be though, I’m not confident another ram module from him will work since the ram we have can be put on a de10 and work fine

[magic girder]

DE10 being more tolerant kind of suggests that the FPGA board has some issue, I would have thought

Maybe some bad GPIO interface somehow

[kind mountain]

May have to suck it up and get a ram module with alliance ram

[magic girder]

Well, if it were me, I’d keep returning until you get something that works. You paid for it, you shouldn’t have to pay twice.

[paper orchid]

Considering the RAM module itself has available files for the PCB (Github), I'm less suspicious of it since technically speaking you could just have the PCB made and slap whatever parts on it

[kind mountain]

I agree but I can only send stuff back to china so much lol.

[magic girder]

It’s the principle 😄

[paper orchid]

It maybe could be the specific RAM chips in use, I'd have to check the datasheets and the photo of the Taki RAM again

[kind mountain]

It’s just weird that it works fine when put on a de10 and misteraddons ram works fine on the taki board

[paper orchid]

But, given that module runs on the DE-10 fine and IIRC the speed grade on the Taki RAM is better (slightly), I have doubts it's that

[random girder]

It’s the colour of the solder mask. It’s not the same green. That’s why.

[paper orchid]

Considering my RAM is from MiSTer Addons and it's black, solder mask color definitely affects performance

It's why I always go OSHpark purple

[neat sierra]

That’s why I was guessing earlier that it’s like both are borderline. So borderline clone de10 can work with good ram and good clone de10 can work with borderline clone ram but borderline de10 clone and borderline ram is no go. Like there’s some manufacturing issue in both and it’s a lottery on your combo

[kind mountain]

Likely the right answer

[magic girder]

The FPGA board is the thing that’s changed, so that feels like it probably warrants some investigation

[random girder]

This makes a lot of sense.

[neat sierra]

The ram chips are different as well

[paper orchid]

Right, and it could be anything in that case. Could be trace length, trace width, non-matched lengths, basically have to compare against the DE-10 and get real nitpicky

[magic girder]

And this is why I don’t understand why changes were made, unless it’s to prevent being sued or whatever. More changes = more chances for things to go wrong.

I can’t imagine it’s too hard in China to show someone a board and say “make exactly this” - but I’m not in manufacturing

[paper orchid]

So, that begs another question: what all is the same and what all isn't? I know Terasic has a schematic but I don't recall if they have PCB files

If you don't have the exact PCB layout, recreating it means there will most likely be some differences just due to how the board has to be laid out by hand again

[cerulean elk]

No they can clone 1:1

So long as the BOM parts can be sourced

Only thing that is near impossible to 1:1 clone is ceramics / other designed objects in stone / clay. Because they make a mold of the object to clone and that mold produces a 5-10% smaller object in the end due to the molding procedure

Don’t ask why I know this haha. Esoteric knowledge

[paper orchid]

1:1 is all on the amount of effort you put into it (or whether you can get access to files that were used in design/fab)

[neat sierra]

I mean really as far as we know it’s just a small handful of people out of ~2k and even those it’s very nearly the same even though it’s a different design and different ram chips. That’s pretty impressive. There were always going to be edge cases whether apparent or not because it’s not exactly the same

[paper orchid]

There's that too

Maybe these specific boards just had unfortunate luck at the fab

[neat sierra]

It being so few and only at the extreme case of n64 is great

[paper orchid]

I've ordered PCBs and had 4/5 of them turn out useless

Because the bed of nails they laid the panel on, my boards were spaces out perfectly to the point that a specific trace ended up resting on the nails, caused an area to be damaged in a way that the etching process just took off that spot

These weren't thin traces either!

[neat sierra]

We really live in the future that we can just order pcbs from jcl or pcbway or whatever for some hobbyists design

[cerulean elk]

Yes for all we know it’s 10 units out of 2000 initial units. Which would be half a percent. Which would be well within the real of manufacturing variance and issues therein

I’ve heard of V limited instances of issues

[kind mountain]

I don’t know anything about ram

I know some of this soldering doesn’t look great but I also don’t know anything lol

[paper orchid]

That looks a little suspect

[kind mountain]

Other side of the same chip

[neat sierra]

Little messy, you could touch it up with just an iron/flux

Pin looks almost off the pad in that pic

[kind mountain]

Interesting enough its marked as U2... assuming thats the second ram chip?

[neat sierra]

Could touch a multimeter to the top of the pin and the pad in continuity mode

[paper orchid]

Shot in the dark, @kind mountain are you having issues?

[kind mountain]

Yes.

RE2 won't load

[paper orchid]

The RAM's all the same pinout IIRC so y'know what

Let's grab the datasheet I have for the Alliance RAM and take a look

[neat sierra]

Oh yeah and don’t touch it up like I said if you want to return it just in case they give you guff

[paper orchid]

So those look to be... pins 25, maybe 26, and 30?

[kind mountain]

I don't think its lifted, I dont have anything to take a closer look anymore

[paper orchid]

A2, A3, and A5 according to the Alliance datasheet

Same on the EtronTech datasheet

[kind mountain]

yeah looks like possibly the last 3 in the lower right and the couple up by the U2

are the ones that are rougher

I dont think anythings lifted

but idk

I'm also not taking an iron to it lol

[neat sierra]

Yeah you pass memtest

So probably not actually an issue but not sure how that tests

[paper orchid]

Honestly, an iron on this wouldn't be too bad

Just use flux

[mossy vector]

Why not?

[kind mountain]

I'd be glad to if I have another on the way lol

[neat sierra]

Yeah iron and flux don’t need any solder you’ll just bridge probably if you don’t have a good tip for fine pitch

[paper orchid]

Yeah, I wouldn't add any more solder

Now, on the topic of flux

https://youtu.be/057I2v6D-1k?t=364 Make sure to use this much

[mossy vector]

Copper solder wick fixes all. It even has flux

[kind mountain]

Got a. Better pic of the top side

[kind mountain]

His shops on my shit list lol

[paper orchid]

Which, Louis Rossmann or Gamers Nexus lol

[kind mountain]

Rossmann

[paper orchid]

He's a bit abrasive

He does know what he's doing though

[kind mountain]

I sent my 2016 macbook pro to him when my wife dropped a rum & coke in it... got it back with no screws in it and crashing after 20 minutes

it was awesome

it also had 0 logins in the logs from when I shipped it to got it back lol

[neat sierra]

Too busy making YouTubes to fix your shit

[kind mountain]

Safety dance crashed

Towards the end

[cerulean elk]

Looks lifted. Seems like there is a shadow

If there’s continuity it’s by a thread

[kind mountain]

for what its worth this had no change for me over stock core, RE2 doesn't load. The core we built the other night with the recommended 500 -> 300 change still loads RE2 though.

[night shell]

I think mine crashed at almost the same frame

Remarkably similar

[kind mountain]

Holy shit

[kind mountain]

Another confirmation of a user unable to load RE2 on taki + taki but being able to load it on de10 + taki and taki + ram from the de10. 🤯

[vast delta]

i may have just modified memtest to include lower frequencies

completely untested, as i can't really turn my MiSTer on right now

[night shell]

https://tenor.com/bblxZ.gif

[vast delta]

you have to use a keyboard or controller

and keep pressing down

hopefully it'll let you go less than 70MHz

you should see 70 -> 65 -> 60 -> 55 and then 1 MHz increments until you get to 45

[night shell]

Well this is interesting

This is compeltely stable at 150, I'll keep incrementing until it stabalises again

[twin barn]

Take that modified test with a grain of salt of course.

[night shell]

Of course

But it'd be good for someone with a working (DE10?) setup to also see what happens

[kind mountain]

Yeah we would need known good to compare against

[night shell]

So, at 65 it's stable, at 60 it starts to error - this build doesn't let me check 61-64

[kind mountain]

@green epoch

[kind mountain]

I can try mine tomorrow, in bed now

My taki, I don’t have a de10 lol

Hit the b button to switch to a single bank and see if that goes to 60?

And then switch to the other

[night shell]

Way ahead of you

Both banks fail below 60, but at 60 only one fails 🤔

[green epoch]

what am I testing

[kind mountain]

Yep one bank is bad as suspected

[kind mountain]

That modified memtest on your real mister

[night shell]

But both fail below 60

[kind mountain]

See what happens down below 70mhz

[green epoch]

ok, not on the taki?

[night shell]

Yeah, would be good to see what happens on a working system

[green epoch]

ok, uploading

[kind mountain]

You can there too

To compare to kushans

But one bank failing at 60 and the other not kind of aligns to the theory of the larger ram games that hit the second chip

[night shell]

Weird question but do we adjust voltage when lowering/raising clock speeds? In the land of PC's that's entirely normal but I have no idea about MiSTer's SD RAM

[night shell]

But isn't the N64 core running at 48Mhz? Where I am seeing failures on both

[kind mountain]

62.5

[night shell]

ohhh, I dunno why I thought it was 48. That's Interesting. Can we make the memtest go between 60 and 65 in 1mhz increments, @vast delta ?

[kind mountain]

I think cause 48 was the jotego test

[night shell]

ah yeah, that'll be it

So @green epoch - I'd be real interested in seeing if you get any errors at 60mhz, on either (or both) RAM chips

[green epoch]

Top is de10, bottom is taki

[vast delta]

A grain? I'd say take it with a hefty bucket of salt

[night shell]

Well!

I just saw a SINGLE bit error at 60Mhz on the "good" RAM chip but didn't get a screenshot because I accidentally knocked the controlller 🤦‍♂️

[vast delta]

I can tomorrow for sure. It looks like it's working on a surface level

[kind mountain]

Can de10 go lower without error?

[night shell]

From these early results, it might not be necessary - seems we have potentially a quick and easy way to test/validate the Mister Pi setup

[green epoch]

Obviously not running them for long enough but you get the idea

[green epoch]

No. Errors at 55

[vast delta]

That's my goal

[kind mountain]

The results so far are correlating to real world

[night shell]

It's better than playing 3mins of a song 😂

[green epoch]

Both my taki chips error at 60

[vast delta]

I will of course test this in my MiSTer tomorrow

[kind mountain]

So your RE2 and conkers don’t load at all on standard core right?

[vast delta]

But I don't have an oscilloscope to check if my frequency calculations are correct

[green epoch]

Yes

[kind mountain]

And me and kushan get to the gore splash on RE2

[night shell]

Interesting. I suspect both mine do as well, but one is significantly flakier than the other. Both immediately error below 60 though.

[kind mountain]

So in interested to see if one of mine is better than the other

Like Kushan’s

[green epoch]

@vast delta safe to let this run on my mister overnight?

[vast delta]

No idea. Assume not

But all I did

[green epoch]

Kk

Tomorrow I guess I should swap ram again?

[vast delta]

Was to include more entries in the array holding the pll config parameters

And adjust the behaviour of the inputs accordingly

[night shell]

Would be good to see what the lower limit of the Taki RAM is on a DE10 and non-Taki RAM on a Mister Pi

[night shell]

There it is!

[vast delta]

If anyone wants to go over the code:

https://github.com/AlmeidaRW/MemTest_MiSTer

Changes here:

https://github.com/AlmeidaRW/MemTest_MiSTer/commit/0588ce63e7c84445ccdae456247be57815e59176

[kind mountain]

These certainly seem to be aligning with expectations

[night shell]

There was such a focus on getting the RAM to clock high (I think 150Mhz was the ceiling?), that peopel overlooked it clocking lower. Is there a spec or anything that gives a range of RAM frequencies expected?

[vast delta]

Mister Pi?

[vast delta]

That's what I'm suspecting

[night shell]

Yeah, Mister Pi, at 60mhz one of my Ram chips errors immediately, the other looked fine but if I leave it long enough errors start to creep in

[kind mountain]

And jesus said both of his error immediately individually

[night shell]

Single bit errors as well, which may not crash a core but who knows what it could do

[kind mountain]

Which could explain why Kushan can get to the first couple splash screens in RE and jesus can’t

[night shell]

Yup, tomorrow if Cheese gets a chance, I'll test what happens at 61+ mhz, I reckon at 61 my "nearly good" chip will be solid and my bad chip will still be bad

[kind mountain]

And I’m intrigued to see if my chips behave like yours since we have similar n64 behavior

If bed wasn’t so comfortable….

[vast delta]

Lol

[night shell]

Wait for FlycingCheese's code to be vetted before doing the below

For anyone following along who wants to test, here's what I think we'd like to see:

1. Get FlyingCheese's copy of memtest that allows for sub-70mhz testing: #1096015979055697940 message

2. Test both chips (Two dashes next to the 3) at 65mhz, it should be stable and you should get no errors.

3. Test both chips at 60Mhz - if you get errors, press A to test an individual chip and see if just one chip faults at 60mhz

4. Test Individual chips at lower mhz until you find which Mhz each individual chip "bottoms out" at

Let each test run for at least 10mins, I was getting errors on one chip at 60Mhz but it took about 10mins to appear.

Would be interested to see results on full Mister Pi setups, DE10 setups, DE10+Taki RAM, Mister Pi+ Non-taki RAM

[vast delta]

I'd say we wait until more experienced people vet the code first

It'd be awesome if Robert could take a quick peek

[night shell]

There, don't want to delete the comment as I cba writing it again, but good point on waiting

I don't mind being cavalier with my own hardware 😛

[kind mountain]

Agree on waiting for vetting however results are trending with expectations so it passes the initial sniff test at least

But yeah not going to blast it on Reddit and FB…. Yet.

[night shell]

Once we've vetted the code and got some data comparing Taki's to non-Taki RAM/Mister's, it'd be good to flag him down as it's something he can investigate on his end

There's an irony here. He originalyl delayed the release of the Mister Pi because these very same RAM sticks were meant to be an exact copy but no accreditation to the designer was given. He delayed the launch to put stickers on them all to give full accreditation. Originally, he had his own design and didn't need to credit anyone.

If these aren't 1:1, then they aren't a clone after all 😂

[vast delta]

This can be way more complicated than that

Are the memory chips exactly the same?

Or are they drop-in replacements?

[kind mountain]

Def not the same brand printed on them. Taki came here a couple days ago to say they were from the same dies

[night shell]

https://twitter.com/TakiUdon_/status/1830812551757857108 he did a good video on the RAM mixup

Taki Udon (@TakiUdon_) on X

We need more hands to swap labels. These need to get sent back to the assembly line to finish this faster.

Twitter · Sep 3, 2024, 3:38 AM

[twin barn]

they are not exactly the same

they are enentech or something, not alliance

taki says they are both made by micron exactlyt he same but branded different, but i'd be skeptical of that. however, that being said, it doesn't explain why a real de10-nano + taki-sdram combo works

feels like either some combined signal integrity issue somewhere in general when both the clone board and the taki sdram are used i guess from what ya'll are reporting here. i'm on vacation so can't test my own clone board yet

[night shell]

I bet it's DNS

[twin barn]

it always is

[kind mountain]

It’s not dns
There’s no way it’s dns
It was dns

My favorite haiku

[vast delta]

Fwiw, this ewhatever company may even be getting lower quality dies from micron

Was the ISSI chip ever looked at?

[twin barn]

exactly, and the last time someone used an alternative chip other than alliance we had lots of random issues with those so people stopped

[static sky]

This is a shot in the dark, but has anybody with a problematic memory module tried removing the sticker to see if that makes a difference? Does it go over any of the traces?

[rugged nexus]

do you all think Taki has insurance for this if the RAM, AV Pro, and MiSTer Pi all need replacing? The ram seems finicky, AV Pro has bad analog audio and also doesn't display certain things on CRTs, and not sure if the MiSTer Pi has any issues but seems like it interacts differently with Taki ram than official De10 Nano does

[twin barn]

Taki did say he would handle returns and stuff like that if there are problems

his website has a contact form or email or something like that

[kind mountain]

You know…. I was just thinking that. 🤣

[night shell]

I don't think it touches any traces

[rugged nexus]

what about the Quality seal sticker

[twin barn]

that kind of adhesive should not have enough electrical conductivity to bridge anything like that

if it did, you'd probably see burns in there if it overlapped a power line

[kind mountain]

Those solder joints on the right of that chip don’t give me confidence tho

[twin barn]

are those vias on the pads...?

[kind mountain]

Certainly look like it

[twin barn]

the official design does have that

but you don't usually see that on the solder job

hrm yeah i'd ask him some questions, ask him for some support

[night shell]

I'd be very surprised if there was a problem with the soldering or the sticker - surely that'd cause more problems at high frequencies rather than low?

[twin barn]

seems kinda odd

yeah, and it would cause problems if a taki sdram was used on a real de10-nano

[vast delta]

ISSI IS42S16320F-7TL looks to be drop-in compatible. ISSI is very reputable

[static sky]

It wouldn't have to be a perfect conductor to make a potential difference to the signal integrity. Removing the sticker and repeating the test is easy to do and costs nothing.

[twin barn]

it's an enentech i thought

[vast delta]

No, i know, i'm just looking at other potential chips

[twin barn]

or whatever that alt brand is

[vast delta]

If alliance is hard to come by

[twin barn]

i wouldn't use anything but alliance

it's not, you just gotta pay 2x the price and not cheap out as a manufacturer

official design calls for one of these two options:

https://www.mouser.com/ProductDetail/913-AS4C32M16SB-6TIN
https://www.mouser.com/ProductDetail/Alliance-Memory/AS4C32M16SB-7TIN?qs=sGAEpiMZZMti5BT4iPSEnReBopsBxV3sVReadJBJNZQ%3D

[vast delta]

I understand, but second sourcing is very important, especially for legacy memory like this

[hushed nova]

still, not failing on a normal de10. so it's clearly good enough there

[paper orchid]

So just looking at the datasheets, the timings are the same but the EtronTech chips appear to consume more power in general

[twin barn]

right, i'll test when i get back at some point as well, since i'm really skeptical about the whole thing, something doesn't add up as to what is reported with these issues

[vast delta]

Issi has an automotive qual version of this chip too. Meaning it will be around for a long, long time

And quality control is way more strict

[paper orchid]

"Branded different" gets me raising a brow since even compensating for the speed rating the Alliance chips are lower power

[twin barn]

can you purchase them in single quantities? it's a diy design and all, i haven't looked at the issi alternative specifically you are referring to

[vast delta]

Taki would access to it at the very least through distribution

[green epoch]

@acoustic vortex if you’re interested, there is a new mem test where you can do low clock speeds. I’d be curious how low yours can go without erroring since you seem to have the worst of this.

[vast delta]

Wow, mouser has the auto qual ISSI chip for cheaper than the Alliance

[paper orchid]

I have an older 2.5 RAM board sitting around doing nothing

If you have the mouser link I could snag a couple chips to mess around

(I keep needing to pull the chips off anyways and get 3.0 boards so I can reassemble)

[kind mountain]

If the etrontech requires more power and they work fine on the DE10 and not on the taki board. Could it be a power related issue?

[paper orchid]

When I say more power I mean that in the smallest sense

Like, miliamps

Though the Alliance having 4mA in some cases where the other is, like, 20

[twin barn]

Could be a typo or a copy paste that is erroneous, etc... never know.

[vast delta]

Automotive:
https://www.mouser.com/ProductDetail/ISSI/IS45S16320F-7TLA1?qs=N2Tevpb%252BvoYqSY2DZ%252Biiyg%3D%3D&srsltid=AfmBOoobrcfx2ZIIlPEbUSF5gk8nq6haXwMrKo9vBmAd5KAki9eCOIso&_gl=1*4asbca*_ga*MTAwMjY1NzUzMC4xNzI3MzIxMzA3*_ga_15W4STQT4T*MTcyNzMyMTMwNi4xLjEuMTcyNzMyMTMxNC41Mi4wLjA.

Commercial:
https://www.mouser.com/ProductDetail/ISSI/IS42S86400F-7TL?qs=N2Tevpb%252Bvoayo0XsM%2Fq2jg%3D%3D

[wanton sun]

With all these results, my current assumption is that the IO timing(sample point) is shifting, based on the clock. Could be the traces to the second chip are slightly longer or shorter so that it hits it earlier.
We could of course shift the clock by some nanoseconds to resolve this as the low clock does allow for a shift without any issues

[paper orchid]

A lot of the values are different, but notably across all the speed grades there's 20mA vs. 4mA

Typo would be surprising. Could be completely wrong values too

[vast delta]

This is a big indication that the die isn't exactly the same

[vast delta]

Would this still apply even if he was using the official pcb design for the sdram board?

[hushed nova]

it could be the traces on the actual clone board

who knows

[wanton sun]

the combination of base board and sdram board together makes up for the path delay

[twin barn]

Yeah if the traces were slightly different on both that would explain it if it was a cumulative thing

All it takes is the combined added length going over a threshold

[vast delta]

I recall him stating that his sdram board is unmodified

So the path delay in theory would only be present in the main board

[wanton sun]

maybe we should ask if he looked after them being the same length. Considering how many different sdram controllers are in different mister cores, i wouldn't be surprised if the borders for working trace lengths are very small

[twin barn]

Maybe the caps are cheaper and have less tight tolerances?

[vast delta]

True

[twin barn]

Stole that thought from Dave shadoff

[vast delta]

Or the voltage regulator not being the same

[wanton sun]

it's relativly simple: add some IO delay on the clock pin and see if it improves. Could even run a test that displays the working clock shift boundaries or calibrate for each board at startup

[twin barn]

Need an eye core 😉

Jk

[wanton sun]

yes, that's what I meant 🙂

[hushed nova]

we still don't even know how many units this actually impacts

[paper orchid]

Alright, found a different datahsheet. The one I had was industrial

[hushed nova]

and it feels a bit....not good if cores have to start compensating for clone design deficiencies

[wanton sun]

considering it hits these boards near the 60mhz line, most are probably right at the edge

[twin barn]

Did MiST have to do this when the clones started to come out? :p

[vast delta]

Especially because we don't know what other future issues may be lurking too

[twin barn]

Had anyone with the qmtech boards tested the sdram the same way as earlier?

[vast delta]

I have only seen it with Jotego's 48mhz tests so far

[paper orchid]

OK, that makes me feel better. Same values between the two brands. That being said, the higher speed grade of the Taki RAM also requires a bit more current, but the same could be said of using a higher speed grade of the Alliance RAM

[wanton sun]

did anyone try other cores yet that also need low clock?

https://github.com/MiSTer-devel/Wiki_MiSTer/wiki/Cores-that-use-SDRAM

the ones below 60mhz and sdram required=yes might be worth a test, but i think it could be difficult for the computer ones

[paper orchid]

Time to break out the CoCo2

[twin barn]

Sega Master system would be a good test

[iron wren]

I’ve done a fair bit of gbc and it’s been fine, though it’s considerably higher than n64

[twin barn]

Not sure if Atari7800 is still clocked that low since kitrinx rewrote it

[paper orchid]

So hold up

This is the N64 but it's...63

[vast delta]

Could this issue be related not just to the clock, but also the amount of sdram used?

[wanton sun]

thatis actually wrong, it's 62.5 🙂

[paper orchid]

Watara Supervision is one more than that, so it's truly the real N64

[twin barn]

Rounding error

[paper orchid]

I feel duped

[iron wren]

Japed, even

[twin barn]

Try the zaxxon arcade core

[woeful grove]

I think 7800 is 57mhz

[vast delta]

@wanton sun where do you initialize the clock for the sdram in your code? N64.sv?

[twin barn]

Thanks, wasn't sure if you did a faster clock during the rewrite or not

[rugged nexus]

this is where we all have to admit we have no sega master system roms ready to go to test

[twin barn]

I do, but I am AFM

[wanton sun]

what do you mean with initialize? it just receives the base bus clock that is also used inside the core
https://github.com/MiSTer-devel/N64_MiSTer/blob/main/N64.sv#L617

[twin barn]

(away from MiSTer)

[vast delta]

I can see none of yall grew up in Brazil then

[rugged nexus]

I am AFM as well lol

[woeful grove]

https://github.com/MiSTer-devel/N64_MiSTer/blob/main/rtl/pll/pll_0002.v#L28 is the pll

[simple olive]

.output_clock_frequency0("62.500000 MHz"),

[vast delta]

I didn't check what's the system clock.. like in memtest for example, it's 50MHz, and you have to leverage the fractional pll to get the clocks like you mentioned here earlier today

[static sky]

Your theory fits with Taki's RAM working on the Terasic DE10. It must be a very "on the edge" type thing though since many of Taki's boards don't exhibit this problem and the production tolerances on PCB's is generally very tight.

[wanton sun]

it's not traces alone. Each chip will have it's own tolerances too

[woeful grove]

if it's 62mhz thats not too agressive

[wanton sun]

e.g. if you have a working spot between -3ns and +3ns for the traces and each chip does +-1ns, then it's no issue. If your window is -0.5ns to 5.5ns, then it does matter

the current sdram design with ddr output fifo was mostly made for the higher clocks to get a consistent result. Discussed that a lot with Sorg 4 years ago or so. As there was never an issue seen at low clocks, this was used everywhere

[woeful grove]

for low clocks i'd look at the refresh before the traces

[wanton sun]

yes i did, made a core that refreshes all the time, no help

[woeful grove]

sorry stepping into an issue many steps behind, of course you'd think of that first 🙂

[wanton sun]

i mostly thought about that, because the turbo core at 80mhz does work and i thought the higher clock speed but same counter means faster refresh

[woeful grove]

could it be some random interference between traces at specific frequencies?

[wanton sun]

in this case, shifting the clock would probably not help

but i still believe that shifting the sample point would help if it's so consistent now also with the memtest at low clocks

especially since the memtest uses different stats than the n64 core, e.g. different latency, different bursts, ...

[vast delta]

Does anyone know if the qmtech clone also uses this etrontech chip?

Their aliexpress store shows alliance chips. Curious to see if that's indeed the case

[acoustic vortex]

I’ll give it a shot tomorrow, thanks for the mention.

[quick light]

(off topic) but even though Taki's board is a clone. It's amazing to see core Devs and other people actually try and determine what the issue is. You couldve just said 'its a clone, that's the issue'. So it's great to see. 🙂

(I don't own a Mister yet, I'm going to try to get the mister pi in batch 2. If this issue isn't resolved by taki by October. Il just get some different ram)

[gritty basalt]

Problem solving is the nerds' bread and butter, and pretty much everyone here is a nerd to varying degrees 🤓

If I had a clone board of my own I'd be contributing to the overall discussion instead of just silently following it like I have been because I'm very interested in understanding what the issues with the MiSTer Pi could be, but money's a bit tight for me at the moment so I can't just buy one to test right now just because. I probably won't have money to spare for a MiSTer Pi until the handheld comes out (which I'm currently expecting to be a couple months or so from now), but for all I know everyone here might have the mystery solved by then, but if not then... well, I'm also a nerd, and as such I like solving problems if I feel it's within my ability to do so.

I want to have a cheaper handheld clone DE-10 I can take with me on the go, so I've got a vested interest in seeing whatever issues the MiSTer Pi is having get resolved, so even if it's just simple debugging on my part I'll gladly do what I can to help everyone out.

Besides, my ability to play my beloved N64 RE2 port on the go seems to be most affected by this issue from what I can see here, and I definitely can't have that!

[deft matrix]

Going back to the modified memtest (for whatever that is worth) each bank immediately fails between 45MHz and 60MHz. 65MHz I've only run very briefly but they haven't immediately failed

[deft matrix]

funny that the design he claims to have been testing for months has alliance modules

https://fxtwitter.com/TakiUdon_/status/1830172720237052120

[thorn quiver]

Any use in looking for voltage drops in the power supply to the ram chips? And/or try different caps

[compact depot]

I did, but I was a Phantom System kid 😛

I have a de10 nano setup that fails the modified ramtest at 45Mhz. Does it mean it will crap out with the cores below, for example?

[twin barn]

Yeah I'd take that test with a grain of salt. Does it fail at 62mhz?

[compact depot]

lemme check

Looking good at 60Hmz here so far

Hmz lol

Yep. 60Mhz seems to be the bottom of it

ramtest mod gives errors starting at 55Mhz and below here

[deft matrix]

I did test the 5200 and 7200 cores and they seemed to run okay

[remote hatch]

no, Alliance chips

[karmic sequoia]

yeah this bug has been a huge nerdsnipe. I'm following eagerly and i don't even have a board with the issues lol

[cold crypt]

my qmtech has alliance sdram as well

[hearty python]

With the new memtest now in mind, is it save to say that there are no problems if the Taki ram passed the 60 MHz test?

[hearty python]

My working taki stack (no problems with n64) does also pass only the 65 MHz, with 60 i got errors on both chips.

[compact depot]

I think that aligns with the same test on a de10 nano so I would say yes?

I would be curious if a taki set passing the 60Mhz fails RE2 or Conker

[hearty python]

Perhaps we need another version of the test core with single steps from 65 to 60 Mhz. N64 core is 62.5 MHz.

So passing the 65 must not automatically mean run the N64 RE2…

[trim bane]

Mine fails with 60hz, fine with 65hz, but all games work just fine

Both chips fail seperately at 60hz, too

[trim bane]

I only played Parsec for 5 mins or so, but that ran with zero issues on the TI -99/4A core

[compact depot]

Maybe this lower frequency test doesn’t tell much?

[rugged nexus]

how do you switch between chips when testing?

[hearty python]

Maybe the partial working ones in N64 have one (first) good 65MHz chip and the second doesn’t pass the 65?

[hearty python]

There is a button for it, I define the s button for it on the keyboard for the core. You see a small beam near the 3 wich toggle then.

[rugged nexus]

ok cool.. i've noticed most people are only posting results for the default selected chip so it would be interesting to know if one chip has different results than the second one

[storm vessel]

Um, so this issue is solely on the Mister Pi or DE10 as well?

[hearty python]

No, only on Mister pi with mister pi ram, other ram from other vendors seems also to solve problems for some on mister pi.

[storm vessel]

Thx. I saw 1000s of messages since last I looked and wasn’t sure if it was applicable to me or not

[hearty python]

This theat was quit silence before mister pi arrived. 😁

[trim bane]

The jotego 48mhz test did have lots of users reporting errors even 4 years ago when it released, on original hardware.

[hearty python]

Yes, but we are now talking over the new modified standard memtest. Mod allow now tests under 70 MHz.

Test is new, and we have to wait for more results.

[trim bane]

I'm aware. Just saying that there used to be issues with low ram speed way back then already too, it's not entirely a new thing with the taki board

[magic girder]

All this talk about memtest got me reading the README again, and reminded me of how confusing I found the documentation.

On the controls, it mentions:
“C - on 128MB module switches between chips.
A - auto mode, detecting the maximum frequency for module being tested. Test starts from maximum frequency. With every error frequency will be decreased.”

Does auto mode also cycle between the chips, or should we use memtest twice and manually flip between chips?

(I naively assumed auto mode handled everything, but I might have been wrong)

[hearty python]

Takis ram is high speed, but there are some issues, with lower MHz. This is not a general issue, seems to be particularly on a small amount of users. But the release was not far away, so we don’t know the real amount yet. When 99% seems to run, then it need time to find out problems and their root causes.

[hearty python]

It will handle both chips, but I don’t know if this is done chip after chip or simultaneously. I guess it’s done not simultaneously, cause it needs some time to decrease at some tests I have done.

[magic girder]

Thank you 🙂

[vast delta]

Cool, another fellow brazuca

I will include more frequencies in a bit

It'd be awesome if anyone here has an oscilloscope to check if the pll formulas are correct

[desert crow]

I'm going to guess there's more than 1 problem going on. A few people said NO n64 games work, those people are having different issues than the people that RE2, and Conker don't work. Those 2 games were problematic and at one time required patches. So I'm not sure those are a good test. Maybe the patch could be improved and would fix it for those people.

[green epoch]

do those cores use the ram chip tho?

[vast delta]

They do

[hearty python]

Could be alone the difference of the chips, two chips fail the 65 MHz test, or only the second…

N64 store the game there and some other things, iirc.

Have forgotten the exact usage for n64. But as Neo Geo is no problem for the mister pi users, that must have something to do with the ram modules are beeing addressed

Some users could also run the turbo core with the 80 MHz.

[sullen basin]

Have just checked, RE2 and Conker boot up on my MiSTer Pi

[desert crow]

if it was just a size problem every game the same size as RE2 probably wouldn't work right?

maybe find a list of all the 64mb games and try those

[zinc dew]

Pokemon Stadium 2

[deft matrix]

I think it's just those 3 besides betas

[zinc dew]

Majora’s Mask?

[kind mountain]

As expected my low speed results look incredibly close to @night shell

No problem at 65 and one chip is worse than the other at 60

Which tracks since Kushan and I can play Pokémon/conkers and RE2 stops for us at the same point on the intro splash screens.

Jesusfish had immediate failures on both his chips at 60 and can’t launch RE2 or conkers at all

Also mine and Kushans safety dance video froze on damn near the identical frame.

[magic girder]

Would if help if some of us with non-Taki based setups also ran the test?

[hearty python]

I can run all, but pass only 65, no chip on my 2 modules pass 60

[zinc dew]

Turbo core works fine for y’all?

[deft matrix]

Yes

[iron wren]

Couldn’t hurt to add more data to the pool

[desert crow]

there's some hacks, some homebrew, zelda debug roms, and DD games are also big

[magic girder]

Alright. I’ll run it in a few hours and let you know. We should compile the data into a spreadsheet or something though, probably

[kind mountain]

Jesusfish ran it on his DE10 last night and his had no problems at 60

Across both chips

[iron wren]

Sure would be nice if the memtest could run at exactly 62.5 for even clearer results

[kind mountain]

This one chip I have is a champ

lol

[hearty python]

Curious

[kind mountain]

This other guy…. Not so much

[magic girder]

Curiouser and curiouser

[zinc dew]

Tell him to pick up the slack!

[neat sierra]

Flick it

[magic girder]

lol slack. I like what you did there

[kind mountain]

https://tenor.com/view/wake-up-gif-8217898333376422901

[zinc dew]

What did I do lol

[magic girder]

slacks. cos timing

But I’m just messing and saying words

[zinc dew]

ohhhhh, yes I meant that. I am very clever, I was just testing you

[magic girder]

testing

[kind mountain]

the weak chip is testing ok so far at 65 by it self.......

[night shell]

You could be right about there being more than one problem. It's work pointing out that those of us having issues with RE2 also can't play until the end of video test roms above a certain size (like safety dance), so there is a suggestion that something is up with the second memory chip at low frequency

[vast delta]

we can't rule out the possibility of my settings to achieve 60MHz not being optimal too

[hearty python]

True, but we See already differences. One chip on a module passes 60 the other not.

[vapid hawk]

are people trying bad apple on neogeo?

thats how we used to test dualram before all these fancy testcores 🙂

[kind mountain]

That too. However a DE10 passing across both chips and me and Kushan having one ok and the other not certainly starts to match up to game performance

[kind mountain]

downloading now

[hearty python]

Perhaps on some modules the speed difference on both chips is to big, how knows.

[neat sierra]

Is there a DOOM / bad apple crossover? Should be.

[kind mountain]

Bad apple is fine but the wiki says neogeo runs ram at 97mhz

[night shell]

I think we're good at 65

[hearty python]

This would explain why Neo Geo is no problem for mister pi

[vast delta]

compiling new version with more frequencies

including 62.5

which will show as 625 in the screen

[night shell]

625mhz!!!

[kind mountain]

YES

[night shell]

Still can't boot RE2 though

[vast delta]

pushing changes to git soon

[hearty python]

Pass you 60?

[night shell]

nope!

At 60, one chip fails immediately, the other eventually starts slipping the odd error - so it's right on the borderline

That's the "good" chip at 60

[night shell]

My guess is that chip will be fine at 62.5, but the other chip will still fail

[kind mountain]

my good chip is good

[night shell]

A single bit error is not necessarily going to cause a crash either, might not even notice it in game

[magic girder]

It would be good if there was an auto mode that counted upwards

For the purposes of this problem, I mean

[hearty python]

True

[vast delta]

Just started running

[ivory laurel]

Memtest 64, that game is hot

[vast delta]

Code changes here:

https://github.com/AlmeidaRW/MemTest_MiSTer/commit/1a331590050ff14a4033ff373ce4f44fbbee6cfe

[vast delta]

the framerate feels the same as most n64 games tbh

[kind mountain]

running 625 now

[vast delta]

still going strong here, will leave it running for the day

but it looks like i will not have issues

[kind mountain]

I got failues at 62

but not 625 so far

[green epoch]

i will test in a little bit

have a bunch of meetings this morning

[kind mountain]

I switched to the weak tit one at 625

[green epoch]

and I'm seeing if I can set a thousand dollars on fire with this 30th anniversary ps5 pro

[kind mountain]

Same

[night shell]

Running 625 as well, no issues so far

[kind mountain]

Not gonna lie tho if they’re going for $3-5k second hand I’m not keeping it lol

[green epoch]

australian ones are on ebay for 10k. not sure if they will sell at that

I wasn't going to buy a ps5 pro. But if I get one of these, then I will use it

I need to hold on to it for my kids' inheritance

[vast delta]

my de10-nano with older zerohimself sdram seems to bottom out at 58MHz

[kind mountain]

I will end up buying a pro anyways

I have too much fomo

I have a Slim with the disc drive already, I will take the drive from that and put in pro then sell the slim as a digital

[green epoch]

I can't justify it without getting this. My ps5 plays games just fine. How many more years are realistically left in this generation. Asking 700 dollars for a half step console seems ambitious.

Asking 1000+ tho...totally fine

[kind mountain]

They got me with Gran Turismo 7 in game ray tracing

gotta have them reflections

[green epoch]

ahh, yeah, I guess if I played gran turismo, I would more into it. that game looked good

[iron wren]

On my mister stack, 60-62 have failures (less the closer to 62.5 I get) then no failures at 62.5 in the first few minutes. Anything higher seems fine too. One chip does slightly worse at 62 than the other, but probably within margin of error.

[kind mountain]

all my friends I race IRL with have a winter GT7 league

so it's heading into that time

My weak ram chip has 134 passes so far at 625 no failures

instantly fails at 062 though

[vast delta]

There are multiple ways of achieving the same clock speed

[kind mountain]

its also passing at 063

[vast delta]

It's just a mathematical operation starting with the base clock which is 50MHz

[night shell]

Ha, I win I guess?

[zinc dew]

y u got a security camera facing you, r you in jail

[night shell]

Security, this is my office which is separate from the house

[zinc dew]

Are you five nights at freddies

[night shell]

I am

[vast delta]

@night shell have you been having N64 issues?

[zinc dew]

You called my bluff, I had nothing lined up after that lol

[kind mountain]

welp...

[green epoch]

not looking good

[night shell]

yup, RE2 doesn't get past the gore warning. That's the only one I've had so far though

[desert crow]

and you're running the newer mister main that autopatchs the roms?

[kind mountain]

Yes we all are

I missed a controller at Walmart too

[rugged nexus]

how do we get this alternate memtest on MiSTer? i only see a rbf file in the _Utility folder on the MiSTer itself

[kind mountain]

Get the file cheese posted above and drop it in there

[rugged nexus]

just leave it as memtest.sv? or do i rename it to .rbf ?

[kind mountain]

Where are you getting .sv from

His is a .rbf

#1096015979055697940 message

[untold dragon]

This file

[rugged nexus]

that github link he posted above

[kind mountain]

Oh that’s source

[green epoch]

You want this @rugged nexus #1096015979055697940 message

[kind mountain]

You need the compiled rbf I linked

[rugged nexus]

oops my bad.. didnt see the rbf link, just saw the github one... thanks

[zinc dew]

Make sure you download the raw file

And if you’re using FileZilla make sure it’s set to binary for transfer

[vast delta]

i took a look at the way quartus calculates 62.5MHz and it's different to mine

let me change it and see it if makes a difference

[topaz otter]

dude these things are all gone

every last one

[kind mountain]

yep

I just got a controller on best buy

[rugged nexus]

I use Winscp and it defaults to binary (at least for me it has) so all good 👍

[topaz otter]

does the PS5 even have any games that aren't remasters of ps3 games

[kind mountain]

we don't play games

[desert crow]

ok the only one with autopatching is MiSTer_20240912 if anyone running a older main tries RE2 they may get a crash since it would be unpatched.

[kind mountain]

we established this, we just buy things and set them up

[topaz otter]

it can't even play N64 games, its useless

at this point the PS5 has less N64 games than the Ique

[deft matrix]

I wonder if a PS5 Pro can run this core

[kind mountain]

yeah we've been through it for days, we're onto too high level of theatens causing it or whatever Scientologists believe

cause we're outta ideas 😛

[green epoch]

someone write a memtest

[kind mountain]

FT: PS 30th Anniversary controller for 1 working MiSTer setup

[vast delta]

god i hate best buy canada

i never got a single pre-order without issues. trying to order a controller too

[desert crow]

don't go off on ps5 in the n64 thread if you want people to read your testing results

[iron dove]

Howlin' mad Murphy

[untold dragon]

Just as a datapoint: I don't have any problems with RE2, Stadium 2, or Conker. The modified memtest fails immediately for any frequency below 62 MHz. I'm using the Mister Pi Mega pack. Still running 62 MHz test.

[topaz otter]

after reading the wildly varying results people are getting with the memtest, I'm beginning to believe that it is almost totally useless

[green epoch]

whomp whomp

[mossy vector]

Stay on topic please

[topaz otter]

on the other hand it does put a cool number on the screen, the person that gets the highest number gets a cookie

[kind mountain]

we're shooting for low numbers round here

https://tenor.com/view/fall-down-golf-gif-7028058033980060670

[cerulean elk]

how many minutes til they are on eBay for thousands

[green epoch]

Too late. Wario was reporting Austrialian models on ebay this morning for 10k

[blissful plaza]

3k already.

[cerulean elk]

gaming cant have nice things

dumbest shit around

[zealous flax]

the standard PS5 pro is $999 in my country :)))

[green epoch]

I just like the color 😦

[zinc dew]

Yeah, that’s a good point. Let’s try to keep this chat focused on N64.

[green epoch]

maybe I'll grab the dual sense elite or whatever it's called if it goes on sale separately

oh they do. I will grab one

anyway, resident evil 2

[neat sierra]

That’s a PSX game!

[kind mountain]

Back to n64 talk

[green epoch]

finally, real gaming

[kind mountain]

I like memtest better

[green epoch]

probably less frustrating

[kind mountain]

Is it tho?

[vast delta]

this one i'm taking the data as per quartus' megawizard instead of calculating by hand. hopefully this is closer to how the N64 clock is set up

[night shell]

running now

immediate errors

Using your previous version this ran at 625 indefinitely without errors

Looks like a specific chip again

I'll leave it running on the "Good" chip to see if it does eventually error

[ivory laurel]

One working memtest by Christmas

[zealous flax]

both my taki 128mb sticks error out fast, one hits 100+ errors in 1 pass, the other hits 10+ after 10 passes

[vast delta]

I would need to consult with more experienced people to see which one approximates the behaviour of the N64 core

I would believe it's this latest one

[paper orchid]

If I wasn't on a trip right now I'd throw stuff on a scope and try to take a look

Granted I don't have a MiSTer Pi, but I could at least observe a working DE-10

Though, if I had a MiSTer Pi, watch as trying to observe it suddenly fixes the issues rofl

[kind mountain]

Not getting a whole lot of passes here

[green epoch]

I have a meeting in 10 minutes, but then I can jump into testing. Should I even bother testing the pi + taki ram? Or should I head into surgery right away?

[kind mountain]

I would say validate your DE10 setup runs with the new memtest before changing anything

at 625

[green epoch]

ok

[night shell]

yeah that'll be the most useful thing, "working" setups

[zealous flax]

strike that, both modules errors out in the 10,000's , but i can still run RE2 fine(?)
at first it wouldnt boot, or get to the main menu, now it plays fine. Is there any area later in the game that artifacts or crashes?

[kind mountain]

that would require playing RE2

which none of us actually want to do

[zealous flax]

lmao, yeah the 240p - 480i switch from screen to screen is a bit jarring

[untold dragon]

My pi + 625 with this test does not immediately error out. Ran the other one for a while and it did not fail at all either. I'll let this newer version run for a while and see if I get any failures.

[finite anchor]

Set it to 4mb,.. makes it tollerable

Im actally shocked how many games the 'Memory pak' seems to make worse was playing uh.. Hybrid heaven, a game I remember enjoying as a wee-boots and is now ... baffling to me, I dont understand it at all. But yeah. has a hi-res mode at 15fps

Its awful lol

[deft matrix]

I used to preach leaving the jumper pak in tbh

[zealous flax]

https://tenor.com/view/headache-migraine-stressed-out-charlie-day-its-always-sunny-in-philadelphia-gif-5048553

expanding pain

i remember always playing perfect dark in the hi-res mode as a kid

kids dont care if a game runs at 60 or 5 fps, not that perfect dark runs "well" in its normal mode

[finite anchor]

Eh, its not like kids have good taste 😄

[vast delta]

Are you having issues with re2/cbfd?

[untold dragon]

Nope

[brittle mountain]

I'll just throw my test results in the pile too. Seems to run stable down to 62 but anything under that and it errors out instantly. So far I have not encountered any issues while actually using the mister though. Re2 and conker included.

[vast delta]

So it's expected

[untold dragon]

This mimics my testing as well.

[blissful wolf]

hi-res mode with widescreen on a 15inch crt 😈

[kind mountain]

so this 625 is failing for people with Re2 issues and passing for people without

we may be onto something

[brittle mountain]

Didn't spend too much time on re but played a good hour or so of conker.

[mortal panther]

I got in within 28 minutes in queue, and the bundle was already unavailable. Managed to grab a 30th DualSense and the 30th DualSense Edge though, so not a complete waste of time.

[deft matrix]

so my sdram module errors at 63MHz, but not 64MHz. of course 62.5 also errors

[kind mountain]

and this tracks since yours super doesn't work

and ours passes at 63 and sorta works

[snow gulch]

My Qmtech with built in memory fails the test instantly

[cold crypt]

oh lord

i wonder what mine will do

[kind mountain]

and it does or doesnt work with RE2?

[snow gulch]

Sometimes black screen, sometimes run but when the speaker said: "resident evil" to start the game I have black screen

[kind mountain]

so 625 errors and problems with re2

[vast delta]

have you tried the turbo core?

[hearty python]

Interesting, cause QMtech used the original Vendor for the ram chips.

I do a test later, but wich test i should use now, the last with the quartus calc?

[kind mountain]

seems to be the best atm

[snow gulch]

No, where can i find it?

[kind mountain]

pinned messages in this channel

[snow gulch]

Ok

[kind mountain]

https://github.com/MiSTer-devel/N64_MiSTer/blob/main/releases_turbo/N64_80MHz_20240323.rbf

[snow gulch]

Thanks, wait 5 minutes

[snow gulch]

It runs but with a blue line on the right

[green epoch]

Ok, running the 62.5 test on the de10nano to make sure we have a good baseline

[jolly turret]

I have that showing up on my de10-nano as well in some scenes, either blue or yellow line on the side. I'm using a CRT though and it appears to be very fine, as the image might get cropped slightly?

[worn delta]

OK guys, for any discussion about issues on the Taki MiSTer Pi (N64 related, or not) please can we take the chat to
https://discord.com/channels/647909397477195803/1287682359658807317

Lets move these discussions out of here and into this one channel

[paper orchid]

I'd suggest a pin indicating this, but given how Discord on mobile loves making pinned messages super visible </s> I'm not sure how much that would do

[kind mountain]

Pins are much easier to see in an actual channel on mobile vs whatever this is, a thread?

[paper orchid]

Yeah, threads in Discord "forums" don't get the same class of treatment and it's frustrating from a UX perspective

[kind mountain]

took me forever to find pins here when I was on mobile last night

[paper orchid]

Technically IIRC it's not even a thread, that's something entirely different in channels (unless they got rid of that)

(suppose this is a different kind of off-topic so, how about that sin and punishment)

[vapid hawk]

the blue line is normal real hardware does that too

[jolly turret]

Thanks for confirming. I think I made sure to check it on real hardware but couldn't remember anymore.

[snow gulch]

ok thanks

[hearty python]

That is weired, how can i get there without using your link? If i do the normal way, the screen is small on the right (web browser).

[magic girder]

It’s in #1049874847330738286

[hearty python]

weired to use, the link seems to open it normal, but i cant get there without the link

[deft matrix]

or post something

and it'll remain

[hearty python]

Ok, full view work. Thanks

[worn delta]

For people for whom this channel is a regular haunt, please feel free to direct any clone board chat and questions that pop up to the relevant channel:
https://discord.com/channels/647909397477195803/1287682359658807317

Don't need a mod to do that, and be good to get this discussion in the right place and get this channel back to being more on topic.

[hearty python]

Wouldnt it be better to do some subchannel under mister hardware? Like clones, etc.? New users wont found the "hided" stuff...

[iron wren]

They can be directed there by somebody in-the-know at least

[hearty python]

Sure, but if taki sells every month, we have in a short time more Takis as DE10... 😂

[topaz otter]

the numbers mason, what do they mean

[blissful wolf]

man I didn't even think to ask, but are there plans for the N64 core to support DD or is that too much

[topaz otter]

not going to happen the fpga is out of space

some knowledgable person can explain it, I'm dumb

[blissful wolf]

maybe an extremely evil snac adapter for an actual DD lol

[topaz otter]

I write with crayons and sometimes eat them, but basically robert has optimized it as much as possible. There's no more room for any more major features

[jolly turret]

Try to find DD to 64 conversion roms. I play F Zero Expansion Kit this way and it works just fine.

[vast delta]

was about to ask that. F Zero Expansion Kit even has a translation

never tried them in the core. guess i know what i'm doing later

[jolly turret]

I think it's almost exactly the same as the regular F Zero for N64, with the main difference being that the music is in stereo instead of mono.

[vast delta]

there's a track editor too

don't know if the core will save it though, i remember when the everdrive first started supporting these DD conversions and saves were a bit cumbersome

[cold crypt]

qmtech with soldered sdram, 62.5mhz and 65mhz are stable (~30 mins). 62mhz seemed to give an error after 20-30 mins. anything between 45mhz and (including) 61mhz directly errors. 150mhz is stable as well. conker plays fine, re2 seems to freeze at random during gameplay. maybe this data helps someone, idk

[cold crypt]

#mister-hardware message 150mhz here, just for completeness

i hope the overflood of memtest caps aren't too bothersome, i hope they help out someone at least

[iron wren]

Discussion of Memtest at low freq has been moved to https://discord.com/channels/647909397477195803/1287682359658807317 since it seems to (mostly) only affect mister pi setups

[cold crypt]

us qmtechers are forgotten...

[iron wren]

true…

[vast delta]

are you using the latest memtest mod build?

[cold crypt]

yes

this one: #1287682359658807317 message

[kind mountain]

we will allow qmtech people in there

https://tenor.com/view/come-to-me-dr-evil-sad-come-here-gif-5275332

[glossy lake]

https://tenor.com/view/lol-goonies-the-goonies-gif-5760945591659470003

[digital mist]

different behavior observed with my qmtech soldered board. memtest starts showing errors under 64.
on latest N64 stable core, conker boots to the title screen. re2 loops demo endlessly. I haven't actually tried to play either game yet to see if it crashes elsewhere

[cold crypt]

interesting

[hasty musk]

Can you make a version with 155mhz? Id like to check if my ram can reach that. 150 no problem

[deft matrix]

the regular memtest starts at 167, then 160, then 150

[vast delta]

i don't even know if my assumptions are correct yet. i'll wait until someone takes a look at the code before i include any more frequencies. hope you understand

[random girder]

I'm a bit out of the meta. I'm running the new Memtest at 625 and it seems to be passing. Is that where we're at?

[kind mountain]

And we’ve been asked to go over to https://discord.com/channels/647909397477195803/1287682359658807317

[random girder]

Fine! I'll leave.

[gritty basalt]

I'll say! On the educational side of things I'm learning more and more over how FPGA and other types of related hardware works just by watching everyone else talk as they're trying to figure out the issue(s) with the Taki boards. Quite the golden opportunity for me to expand my knowledge a bit!

Anyway, I see the MiSTer Pi discussion is moving out of here, so I'm not gonna say anything more on it here and derail things even further.

[zinc dew]

Perfect Dark > Goldeneye

[kind mountain]

https://tenor.com/view/this-is-a-hot-take-steve-terreberry-this-might-be-controversial-gif-4319274631307110999