Search

[rampart]

If the Hynix memory chip were to be replaced would it eliminate slowdown? What would be the best memory chip to replace the Hynix with?

[NT]

Those are very good questions.  It's gonna take either someone with good knowledge of RAM chips in general or someone who is brave enough to test replacement chips with their Supercard.

[plflorence]

and someone rich  :lol:
but two thumbs up for anyone who's willing to do it!

[Diablo]

So, the RAM chip in the Supercard is the Hynix HY57V561620CTP-H.  That chip has a clock frequency of 133 MHz.  If you up the clock frequency, it should, theoretically, improve the performance of the Supercard (the ROM can be accessed faster, which is what has been reported as the issue that plagues the Supercard).  Hynix makes the HY57V561620C(L)T(P)-6(I), which is the same family as the Supercard chip, so it is a possible alternative.  Of course, I'd rather put Micron, or even Infineon in there myself.

Also, clock frequency is not the only thing that may need improving.  The refresh rate may be too slow.

------------------------------------------------------------------------
EDIT:  Speaking of which, the stats for the Hynix chip says that it has a cycle time of "8192 refresh cycles / 64ms".  Now, I've been up for 26 hours now, so I may not be reading this right.  However, is that statement not saying that the chip completes 8192 cycles in 64ms?  That would mean it has a cycle rate of around 800ns, right?  That's just way too slow.  Do you think they mean µs?  That would make a cycle rate of 8ns, which makes much more sense.

Now, going off that, a comparable Micron chip has a cycle rate of 6.5ns.  That is, comparably, a substantial increase in performance.  That, more than clock rate, in my opinion, would improve the Supercard's performance issues.

------------------------------------------------------------------------

BTW, did you notice those italics in my first paragraph?  Yeah, all that's guess work.  I, personally, don't think it's the RAM that's the problem: I think it's the coding.

What are the RAM specs of other flash carts that don't suffer from the same slowdown problems that the Supercard does?  I'll check myself in a little while, but feel free to do the same.


EDIT:  From what I can tell, the majority of other flash carts' chips have cycle rates of 110-150ns.  If that is the case, then the Supercard's may, in fact, be 800ns.

If all that information is correct, then I would say that chip is definitely a performance bottleneck.

[NT]

So, the RAM chip in the Supercard is the Hynix HY57V561620CTP-H.  That chip has a clock frequency of 133 MHz.  If you up the clock frequency, it should, theoretically, improve the performance of the Supercard (the ROM can be accessed faster, which is what has been reported as the issue that plagues the Supercard).  Hynix makes the HY57V561620C(L)T(P)-6(I), which is the same family as the Supercard chip, so it is a possible alternative.  Of course, I'd rather put Micron, or even Infineon in there myself.

Also, clock frequency is not the only thing that may need improving.  The refresh rate may be too slow.

------------------------------------------------------------------------
EDIT:  Speaking of which, the stats for the Hynix chip says that it has a cycle time of "8192 refresh cycles / 64ms".  Now, I've been up for 26 hours now, so I may not be reading this right.  However, is that statement not saying that the chip completes 8192 cycles in 64ms?  That would mean it has a cycle rate of around 800ns, right?  That's just way too slow.  Do you think they mean µs?  That would make a cycle rate of 8ns, which makes much more sense.

Now, going off that, a comparable Micron chip has a cycle rate of 6.5ns.  That is, comparably, a substantial increase in performance.  That, more than clock rate, in my opinion, would improve the Supercard's performance issues.

------------------------------------------------------------------------

BTW, did you notice those italics in my first paragraph?  Yeah, all that's guess work.  I, personally, don't think it's the RAM that's the problem: I think it's the coding.

What are the RAM specs of other flash carts that don't suffer from the same slowdown problems that the Supercard does?  I'll check myself in a little while, but feel free to do the same.


EDIT:  From what I can tell, the majority of other flash carts' chips have cycle rates of 110-150ns.  If that is the case, then the Supercard's may, in fact, be 800ns.

If all that information is correct, then I would say that chip is definitely a performance bottleneck.

Nice research.  

I asked an engineer that I work with the other day about this and he said that it's likely that the clock speed would likely be set (can't be changed) in the Supercard firmware.  So, by that theory, bumping up the clock speed with some new chips wouldn't help.  

I'm not sure if lower-latency chips would also be dependent on the Supercard firmware.  If you think of the BIOS on a PC you know that both the clock speed and the latency are set at a particular speed and performance gains from faster memory aren't realized until you adjust your memory clock speed or CAS latency timings.  So who knows....it still needs to be tested.

[Diablo]

Keeping with the PC comparison in mind, the speed settings can also be set to "optimum" or "accelerated".  While I highly doubt that would be the case in the Supercard's situation, if it were, replacing the memory chips (or even CPU!) would result in using the full potential of the updated components.

Of course, I think you're right (or your buddy's right) that the settings for the chips would already be set in the firmware.  In which case you would not only have to replace the chips, but also flash the firmware with a version that utilizes the speeds of the new components. :cry:

[rampart]

I will replace the chip if I can find a suitable chip for sale. I have not been able to locate one. If anyone has an idea where I can purchase one I would Appreciate it.

[SyNTaXer]

try to get a sample ram chip directly from a semiconductor. it works and it is for free. bye

[NT]

SyNTaXer was able to find a source for the faster chip:

http://www.icsource.com/ICSourceWDL/WebRental.aspx?user=ase&part=HY57V561620CT6&prefix=0&exact=0&stock=0&sort=Date&page=1

[Mc Nasty]

Is it possible that a beginner (as me) can it replace the chip?  
and if it is this way somebody that has already changed it it could provide us an installation diagram (it Approves of dummy's) to make this operation type.  
and for I finish the power it could be increased to my SC? (so that it is a lot but powerful) still putting him a chip but quick that those that here mention...

[SyNTaXer]

trust me, to replace the ram it's best to be an expert in soldering. it did such a replacement years ago on an ipaq, all i can say, "hard work". bye

[NT]

Is it possible that a beginner (as me) can it replace the chip?  
and if it is this way somebody that has already changed it it could provide us an installation diagram (it Approves of dummy's) to make this operation type.  
and for I finish the power it could be increased to my SC? (so that it is a lot but powerful) still putting him a chip but quick that those that here mention...

I wouldn't recommend soldering a chip such as this on your Supercard unless you are experienced at soldering.  There's a lot of pins and they are fairly small.  There's a new software update out now that claims some improvements on how games run, so we should test that and see if it helps any of these slowdown issues.

[keen31]

im new to the supercard scene (in fact i dont even have one yet... ive always thought the idea would be great but didnt know it existed until very recently). i think im going to wait and see if the SC SD has any improvements over the SC CF before i buy either one. anyway...

i was reading all over the board to get as much info as i could to decide if this is what i wanted to buy (vs. a regular flashcart... although i think i was sold on it before i even started). i came across this thread and it really peaked my interest. from the research thats been going on it seems that the SCs weak link is its ram and clock or refresh/cycle speeds on it. the fact that its a hynix chip on there doesnt surprise me that its happening either. im no expert on ram but i would also bet dollars to donuts that the chip has a lot to do with it (i wonder what the bandwidth on the chip is @133mhz... i also wonder if any/all of them are actually running at that speed... it wouldnt be surprising if some/most of the modules are running under speed due to cheap components/manufacturing...) i also think that the coding may have something to do with it as well... but probably not as much as the hardware. you can write perfectly and have it perform like crap on
bad hardware and vice versa if you write garbage and put it on good hardware, so its hard to tell... but i think the heft of the problem lies in the hardware (this is just a guess of course).

what i really wanted to mention was someone's idea of replacing the chip with a better one to see if it could increase performance. while its probably right on the money that it would do nothing if the firmware wasnt written to take advantage of it (i.e. it would be hard clocked at 133mhz) that isnt to say the the refresh/cycle/bandwith/overall performance would not increase. in fact id venture to say that there would be at least a few performance increases with simply putting a better chip on there. i remember reading an article back in the day on replacing/adding ram to your xbox (in fact i just googled a few guides on it)... i dont know the fine details of the xbox using or recognizing it (i do know that xbox linux can take advantage cause its written to) but adding isnt the same as replacing. anyway, i believe the process could be slightly beneficial on its own, but re-coding could possibly make it even more worthwhile.

i wonder if there is anyone in the community that is experienced with this type of technology... enough to attempt a firmware re-write. i dont imagine it would be as simple as writing/modifying say... custom drivers or software, but i would think that with some experience and knowledge a homebrew firmware could be born to perhaps clean up the performance of the current batch of chips and/or attempt to harness/test the performance increases of a new chip... although that is just a guess, i dont know the fine details on writing firmware (if its even possible).

ok, long story short, i think this post has a great idea and i hope someone with the knowhow follows through with some attempts because i think it could lead to some useful info regarding SC performance and maybe initiate others (maybe romman?) to work on either better firmware or better hardware.

that is unless of course the SC SD or that new m3 totally solve all these issues (but F that, i think the current SC has potential and it sounds like a fun project, i just wish i had more to contribute).

i hope others are still interested in this, i hope to see/hear/read more about it since it has a lot of possibility.

[mat]

I know nothing about chips, but SURELY if there was a chip that could make it faster, it would have gone into the supercard design? Rommaster isn't an idiot.

[keen31]

$ is the bottom line... those hynix chips are super cheap(read: crappy), thats why they are in there... they do the job, just not as good as they could. thats the difference between making money and losing money though i suppose, so i guess you cant hold a grudge about it.