The OFFICIAL... Neosd MicroSD Card thread...

[hardyhell]

hey guys.

so there is no sd card winner and it depends on the multi cart itself?

thanks

 

[mastamuzz]

apparently no in my case but you are free to try different cards what may be good for me may not be good for you, I followed some e-myths about SD cards the zeroed format then fill your card with 4 gig files and erase then format U and then format fat32 you may want to try out with a card straight from the box and then do the myths and see if anything changes for you.

 

[Razoola]

hey guys.

so there is no sd card winner and it depends on the multi cart itself?

thanks

Yes, this is correct in some cases. At the moment not enough is known but some users of the NeoSD (MVS and AES) are 20% slower than others (loading kof2003) and it has been confirmed it is not releated to the MVS or AES hardware or the brand of SD card used. If your kof2003 loads in around 3:20 or slower very likly your neosd is one of those affected if your sure yout using a fast sdcard.

 

[Heinz]

While Im not helping I have to say 2-3mins is a lot faster than

1. The money I would earn to pay for said game and
2. Faster than shipping anywhere in the world.

If its under 4 mins then its a good deal.

**Fixed by Rot... Cuz Lachlan types like convict...**

 

[mastamuzz]

Soap fell from the keyboard?

And back to square one if 4 minutes bothers someone then go back to teh emus, or multi carts cause you are still discovering the NEO library.

 

[Gyrian]

Here are my cheapo micro SD times, for what it's worth.

NeoSD AES 1.05
Dane 8gb HC class 4 micro SD card

3:22 KOF2003 flashing time | 0:37 erase | 2:45 flash
0:24 Magician Lord flashing time | 0:03 erase | 0:21 flash

I get a fair amount of variability in erase times. Flashing times are consistent, I timed both games @ 2:45 & 0:21 multiple times each.

 

[Razoola]

3:22 KOF2003 flashing time | 0:37 erase | 2:45 flash
0:24 Magician Lord flashing time | 0:03 erase | 0:21 flash

I get a fair amount of variability in erase times. Flashing times are consistent, I timed both games @ 2:45 & 0:21 multiple times each.

This 0:37 erase time does not make sence. From what I have been told during testing this issue, the expected erase time should be about 0:50 in ideal conditions, and that should still allow the entire flash process to complete well under 3:00. Someone is either telling porkies or something else is happening on top as well.

These tests are going to have to be done again to get any meaning full information now if your sure your times are correct Gyrian. Not with maglord though, just with kof2003, flashing the game twice and only taking info from the 2nd flash. Giving the initial erase time, following flash time and final total time (both added together).

For me it is as follows.

0:52 + 2.24, total 3:25

Given the times neosd told me he gets and what neodev told me about the expected initial erase time for kof2003, it would mean neosds total time of 2:42 would be made up of 0:50 + 1:52 = 2:42. It would be intresting to see if this initial 0:50 erase time is actually accurate or can be used to determine if you have a NeoSD affected by this problem.

 

[Razoola]

Just spoken to neodev again about this and it appears what I was told before is not totally accurate. A 0:52 erase cycle during the flashing process is not normal and slower than the expected norm. Bear in mind this erase cycle is not reliant on the sdcard either. The expected erase time in an ideal situation should be 37 seconds but given that is rare, a couple of seconds slower is also very acceptable. Will be intresting to see if others affected by slow flashing times also have this much slower than expected initial erase time.

 

[Gyrian]

Appreciate your reply, Raz.

I should've dug deeper on erase time once I made the distinction. My flash time measured consistently, and I was about ready to post once I realized that Rot had requested total time. My total for KOF2003 surprised me a bit, and my impression of erase times was one of variability based upon trying out a number of games.

I've verified that I was mistaken about this, having tested again for KOF2003 erase time specifically. Five tests got me between 35s to 38s. Clearly there is a technical reason here, but it would sure make for a tangible difference if NeoSD could start by overwriting instead.

 

[neodev]

Appreciate your reply, Raz.

I should've dug deeper on erase time once I made the distinction. My flash time measured consistently, and I was about ready to post once I realized that Rot had requested total time. My total for KOF2003 surprised me a bit, and my impression of erase times was one of variability based upon trying out a number of games.

I've verified that I was mistaken about this, having tested again for KOF2003 erase time specifically. Five tests got me between 35s to 38s. Clearly there is a technical reason here, but it would sure make for a tangible difference if NeoSD could start by overwriting instead.

Keep in mind that erase time depends on the game size, as the firmware will only erase the required space for the game, so small games have quicker erase time than larger games.

Also, erasing is a mandatory step for flashroms, the data can't just be overwritten, writting to a flashrom can only convert bits from 1 to 0, but in order to convert a bit back to 1, you must erase it, and erasing can only be performed at erase block sizes (128KB for the large flashes in NeoSD).

Also, erasing is an operation that it self timed inside the flashroms, so firmware doesn't have actually any control on the time it takes,It just says "erase sectors from x to y" and then waits for the flash to report it has finished. Just external factors like temperature, voltage stability, component tolerance... affect that time.

 

[Rot]

Hehe... Funny...

My opinion on load times and this thread... is that it's sorta cool to have fast load times... BUT... the main priority of this device is to play games accurately and it sure as hell does that..

SO... unless you're changing out games every 15 minutes... this to me is a non-issue...

SURE, there are some of you who feel they must compare it the the 161-1 carts or the other MD, Snes etc Flashcarts... but all in all... that's dumb..

xROTx

PS. The Neosd has been developed with foresight and the homebrew market is serviced well with this device... don't forget that mofo's...

 

[Razoola]

Your right ROT, this does not affect the product once the game is flashed. It does however have an effect given some NeoSDs are faster than others and while that is acceptabled if the cause is external it still comes down to wether this is a fault with flash chips on some NeoSDs or something else that can be fixed in firmware. This is quite important given much has been said about flashspeed in the past and no doubt will be said again in the future given there could be a competing product in the future.

The way I see it is like this, a big deal about the 30% speed increase in the new firmware and we were all happy. In turn then, a 20% speed defisit in some NeoSD carts cannot be dismissed if the blame can be placed on the neoSD itsself and not external factors.

Depending on the what the fault is then, does this mean neosd will replace NeoSD cards with this issue for those that want a repleacment?

 

[neodev]

Your right ROT, this does not affect the product once the game is flashed. It does however have an effect given some NeoSDs are faster than others and while that is acceptabled if the cause is external it still comes down to wether this is a fault with flash chips on some NeoSDs or something else that can be fixed in firmware. This is quite important given much has been said about flashspeed in the past and no doubt will be said again in the future given there could be a competing product in the future.

The way I see it is like this, a big deal about the 30% speed increase in the new firmware and we were all happy. In turn then, a 20% speed defisit in some NeoSD carts cannot be dismissed if the blame can be placed on the neoSD itsself and not external factors.

Depending on the what the fault is then, does this mean neosd will replace NeoSD cards with this issue for those that want a repleacment?

Well, I don't see it as a fault in the cart. All electronic devices have tolerances, not all devices are manufactured exactly the same. The flash and erase times given by the manufacturer are mean times, so there are faster and slower flashes, the same way there are speed differences in a SD card when you use it on a PC, even using two cards of the same model, that's normal. I can show you some data, taken from manufacturer manual:

S29GL128P (page 62): Chip erase time: typical 64s. Maximum: 256s . The first erase cycle, that would erase the P rom data, that is 1MB + 8MB is then: 64s for 16MB -> 36s for 9MB, that means, 36s is the typical value, but according to a table, it could go up to 144s, and still it's in spec, so, that's not an issue, that's normal. What the manufacturer doesn't state is how temperature or voltage affects to these values, keep in mind that we are using 3.3V (right in the middle of the spec 3.0V to 3.6V), so, does it increase the flashing time? maybe, it's not explained, maybe in some flash chips it does, maybe in other it doesn't. Who is seeing 37s in flash erase time, are in the "typical" range, and 50s? yes, it's still in range, it's not the typical, but it's well below the maximum.
You can refer to that manual, it's public, just search for S29GL128P in the Cypress page.

 

[massimiliano]

Just curious, would that be possible to get a tool testing these parameters automatically generating a report?

 

[Rot]

Hehe...

You see Raz... I view this perceived issue as how i view the forum... there's range of opinions and posts i will accept and then there's crossing the line...

The device works... tolerance 20%... that's fine...

Not every pcb will have the exact same width of soldering... not ever pcb will have exctly the same pin length...

As long as the device works and the customers are happy... why worry....

It's like I said how I view the forum... there's a few things that you may not be happy with... but why complain... as long as most of you can play...

xROTx

PS. I look at the big picture... SURE there's the minutae... and i take note of them as well.. BUT in the main....

I'm happy with the way things are with the Neosd device...

EDIT: THURSDAYS is my babysitting Grandkiddy day... just sayin'... best not try to stress me out here...

 

[Razoola]

Well, I don't see it as a fault in the cart. All electronic devices have tolerances, not all devices are manufactured exactly the same. The flash and erase times given by the manufacturer are mean times, so there are faster and slower flashes, the same way there are speed differences in a SD card when you use it on a PC, even using two cards of the same model, that's normal. I can show you some data, taken from manufacturer manual:

S29GL128P (page 62): Chip erase time: typical 64s. Maximum: 256s . The first erase cycle, that would erase the P rom data, that is 1MB + 8MB is then: 64s for 16MB -> 36s for 9MB, that means, 36s is the typical value, but according to a table, it could go up to 144s, and still it's in spec, so, that's not an issue, that's normal. What the manufacturer doesn't state is how temperature or voltage affects to these values, keep in mind that we are using 3.3V (right in the middle of the spec 3.0V to 3.6V), so, does it increase the flashing time? maybe, it's not explained, maybe in some flash chips it does, maybe in other it doesn't. Who is seeing 37s in flash erase time, are in the "typical" range, and 50s? yes, it's still in range, it's not the typical, but it's well below the maximum.
You can refer to that manual, it's public, just search for S29GL128P in the Cypress page.

Yes you are right in what you are saying but you are forgetting the following that has been brought up in the thread already. As I understand how it was written in this thread Yosh41 has tried the AES NeoSD and MVS NeoSD (with converter) on the same AES system. Same AES, same PSU, same room tempreture. The MVS NeoSD was 20% faster because his AES neoSD seems to have this issue. In his test you must also agree something other than the AES, PSU or room tempreture is causing the flash chips on his AES NeoSD to act differently.

Its difficult to say in his case that it is not a fault given it seems at the moment that the majority of people do not have this issue. If one on the other hand we were to take the chip specs into account and as the reason for this, one would more expected to see slower times as the norm, and affecting far more NeoSDs than is.

 

[Gyrian]

While I do like to explore these questions we're discussing (why I ultimately contributed my data points), I do want to make it clear that none of this has made one lick of difference to my enjoyment of my NeoSD. These times might read long on paper, but I haven't found the process to be a bother.

Given the options, I do value how the design was laid out. The behavior that accurately reflects a real cart once flashed is great. For my part, it's also kind of a plus that it's better used one game at a time. I find the vast libraries one tends to fill these flash carts with sort of paralyzing sometimes.

 

[neodev]

Yes you are right in what you are saying but you are forgetting the following that has been brought up in the thread already. As I understand how it was written in this thread Yosh41 has tried the AES NeoSD and MVS NeoSD (with converter) on the same AES system. Same AES, same PSU, same room tempreture. The MVS NeoSD was 20% faster because his AES neoSD seems to have this issue. In his test you must also agree something other than the AES, PSU or room tempreture is causing the flash chips on his AES NeoSD to act differently.

Its difficult to say in his case that it is not a fault given it seems at the moment that the majority of people do not have this issue. If one on the other hand were to take the chip specs into account and as the reason, one would more expected to see slower times affecting far more NeoSDs than is.

Also, AES carts are a bit slower programming because the data goes through the extra FPGA chip in the CHA board and there is a small protocol there, as it controls the C flash data lines, so it having a slower flashing speed in AES is normal.

But as I said, it's not an issue that some carts are slower, their flashes are well in the manufacturer specs. The flashrom manufacturer probably gives a wide range so they don't have to scrap many flash chips that would not fit in a narrower range.

 

[Razoola]

Hehe...

You see Raz... I view this perceived issue as how i view the forum... there's range of opinions and posts i will accept and then there's crossing the line...

The device works... tolerance 20%... that's fine...

Not every pcb will have the exact same width of soldering... not ever pcb will have exctly the same pin length...

As long as the device works and the customers are happy... why worry....

It's like I said how I view the forum... there's a few things that you may not be happy with... but why complain... as long as most of you can play...

xROTx

PS. I look at the big picture... SURE there's the minutae... and i take note of them as well.. BUT in the main....

I'm happy with the way things are with the Neosd device...

EDIT: THURSDAYS is my babysitting Grandkiddy day... just sayin'... best not try to stress me out here...

Yes thats one way of looking at it.

The other way is this. I have reviewed the product and at the moment I would guess and say maybe 1 or 2 percent of NeoSDs sold so far (based on this thread) are 20% slower than the other 98 or 99 percent when it comes to flashing a few games. For most people I fully agree this may be 100% acceptable but its our job (well espicially my job given I reviewed the product) not to hide this fact.

You know how it is, member 'a' may not be happy that he paid the same 400 as member 'b' but his NeoSD is 20% slower. Then goes on to buying SD cards to try and solve something that cannot be fixed unless he tries a different NeoSD cart.

To be honest I wish I did not discover this issue, but unfortunatey with the help of neosd himself I did. Up until that point it was discovered he was 100% sure there could be nothing on the NeoSD that could cause this difference in speed. A few tests and sdcards later its 99% likely to be something on the NeoSD causing this.

 

[Razoola]

Also, AES carts are a bit slower programming because the data goes through the extra FPGA chip in the CHA board and there is a small protocol there, as it controls the C flash data lines, so it having a slower flashing speed in AES is normal.

But as I said, it's not an issue that some carts are slower, their flashes are well in the manufacturer specs. The flashrom manufacturer probably gives a wide range so they don't have to scrap many flash chips that would not fit in a narrower range.

Yes there is a very slight speed difference with the AES card, but its really really low (1% maybe) and not the great difference in speed a couple have had.

Given the wide range the manufacturer gives, the results so far clearly point that is not the cause. Mainly because this would be a more 50/50 split on the people affected, espicially given the number of chips on each NeoSD.

 

[mastamuzz]

I remember the time when there was no NeoSD and also when they came out of the blue with a cart that worked 99% then 100% after msX so why are we getting the underwear in a bunch now? It works it's here and who paid for it enjoys it! Let the NeoSD team surprise us again with an update adding things, forcing shit at this moment is completely stupid.

 

[Razoola]

No one is getting their knickers in a twist , the NeoSD is a great product period.

Its about letting people know this issue may be present on their NeoSD if they start wondering why they can't hit the same flashing times others report.

 

[daithidownunder]

No one is getting their knickers in a twist , the NeoSD is a great product period.

Its about letting people know this issue may be present on their NeoSD if they start wondering why they can't hit the same flashing times others report.

Agreed, it's important to discuss these things in an open forum.

Sent from my HTC 2PST1 using Tapatalk

 

[greatfunky]

A question for neodev comes to my mind : will it be possible to improve again in the future the loading speed or are we at the maximum limit ?

 

[benjiedude]

No one is getting their knickers in a twist , the NeoSD is a great product period.

Its about letting people know this issue may be present on their NeoSD if they start wondering why they can't hit the same flashing times others report.

I appreciate you looking into this and posting it here.