GoldenEye 007 Nintendo 64 Community, GoldenEye X, Nintendo 64 Games Discussion

[MRKane]

I've read that RAM clockspeed somehow affects the system clock - so if there is faster ram apparently games tested run at a faster speed. Although the poster was rather mum on what the ram was.

I've also seen lots of examples of game corruption due to an overclock, but, back to topic.

I was personally targeting the ram as the latency is HUGE (640 ns), there are a few chips out there that would probably work, but I've no idea where to go for finding an alternative chip
_________________
No Mr. Bond, I expect you to be re-coded!

[zoinkity]

The reason they don't really consider rdram latencies that important is that, to be honest, it has a noticible impact but it isn't a huge gain. As an example, the iQue noticibly runs a little smoother and faster, but not anything like the performance you'd expect from an overclocked system.
_________________
(\_/) Beware
(O.o) ze
(> <) Hoppentruppen!

[MRKane]

Hey! That'd be perfect in my opinion! In saying that, having looked at much of the overclocked work I'm not totally sure that I'd call 90% of it a "performance gain" as the system is running faster, yes, but the games are generally unplayable.

I really should try and track down a iQue and see if there's much that can be done with it, or if the ram could give me a hint in the direction to look for something to swap out.

I did do a heap of research a while ago but frankly gave up on it

EDIT: Will note that I've also read through some of your posts on the matter
_________________
No Mr. Bond, I expect you to be re-coded!

[AL64inthedark]

I can only dream of a way to enhance framerate on Perfect Dark / Goldeneye X without any issue relative to speed or whatever.

Hope some guys will manage to find something, someday.

[MRKane]

So, yet more digging later and running off the theory that to do something like this one should only get to know it better I stumbled across this:

http://www-leland.stanford.edu/class/ee382/MISC/rambus.pdf

Long story short Rambus Ram is pretty special and requires a data feed and latency that's pretty exactly calibrated - while it does have some compensation it's not totally complete - which I think explains the shortfallings when overclocking the system that others have experienced (graphical or performance glitches, seemingly random crashes and instability etc). One user did try crippling the ram and noted that things "improved" graphically, however that's possibly due to it just forcing the data and system frequency back into synchronisation...

Long story short: I think trying to replace the RDRAM with any old chip that has the same pinout but better statistics isn't going to work due to the very nature of the ram itself, and to make things more confusing that latency problem I'm trying to find a solution to might be part of what keeps the system synchronised.

I guess the next question is: Are all expansion packs made equal? Madcat (for instance) was notorious for over-engineering their extensions and they were frequently better than original hardware - is it possible that some of the expansion packs could be packing (excuse the pun) better RAM?
_________________
No Mr. Bond, I expect you to be re-coded!

[Trevor]

From what ive read then yes, simply doing one thing will loose sync and cause the copruption. if you upgraded the RAM then you would also need to overclock the x2 clock (well, whichever number it was) which increases the RAM speed as well as FSB/RCP and initial CPU (Before the CPU internally bumps 1.5x)

So far what Ive seen is that people who overclock the CPU only end up slowing the RAM which puts it out of sync.
The fames dont increase fps but do increase gamespeed/clock.

If the RAM was made better then the CPU could be increased to keep sync...
Actually... I found a wierd chart that upon increasing either RAM or CPU or Both led to failures... I didnt finish reading it.
http://www.assemblergames.com/forums/showthread.php?51656-Mapping-N64-%93Overclockability%94-%96-achieved-3-0x-multiplier-but-not-3-0x-speed

Im confused... hahaha

Trev

P.S. having now read the artical I find it exciting to read that there may be an N64 out there running at 600Mhz RAM 75MHz RCP, 112.5MHz CPU (Or 150 if x2) with no corruption and only requiring the game header changed.
Oh, and 12MB RAM for GEX hahahaha
_________________

[Dylan]

I have an overclocked N64, how do I change the clock speed within the game so PD and GE can be played at the correct speed? (I have a flashcart)

[Trevor]

Well, I beleve the first thing zoink is going to require before telling you is what you overclocked to?
2x or 3x? (i.e. 125MHz or 187.5MHz)
Given all the failures at 3x Im going to guess you are running at 125MHz

Trev
_________________

[MRKane]

This did start more as a "can the RAM latency be reduced" thread, but I just can't find a chip that would be suitable within constraints of clockrate, pinout, speed and also sync (assuming the latency were reduced too much it'd throw it out of sync there again).

Therefore: Dylan, as Trevor said - what's your overclock? I'd love to find out if we can modify the header to get stable results through an overclock (as I've got to wait for an auction to end before I can get another N64 myself!)

Of course the next question is: can a gameshark go up that high as to modify the ROM header?
_________________
No Mr. Bond, I expect you to be re-coded!

[Trevor]

Are you trying to keep to the same system speed?
If so then you need to find 500MHz RAM.
As far as I gathered from reading, the latency isnt a sync issue, its the clocks that are a sync issue.

I would love to see if the guy got the 600-75-115 config working with a header change.

Trev

P.S. I have just noticed an inconsistency in RAM speed quotations.

The Clockrate for official RAM is 250Mhz (500MBits/s)

so, what I think has happened is that somewhere along the line people started quoting Mb/s as Hz...

So, the guy actually has 300MHz RAM (600Mb/s)

Edit: Just noticed I was using Bytes instead of bits. so, just as a guess, RDRAM is ... now... whats the word... hmm... double loaded or something?
Where it loads a bit on the up cycle and another on the down cycle.
_________________


Last edited by Trevor on Sat Nov 15, 2014 8:59 am; edited 2 times in total

[Dylan]

Trevor wrote:

Well, I beleve the first thing zoink is going to require before telling you is what you overclocked to? 2x or 3x? (i.e. 125MHz or 187.5MHz) Given all the failures at 3x Im going to guess you are running at 125MHz Trev

Yeah, I can run it up to 125MHz considering there's a super high failure rate for a NUS-CPU-04 model. I've read there's less failure when attempting to overclock the Funtastic Series consoles that aren't RGB moddable to 187.5Mhz.

[MRKane]

@Trevor: That was something that confused me also! Plus reading up on it I'm not sure that all RDRAM is created the same...and just to make life fun I think the N64 RDRAM is double threaded or synchronous or something nasty like that. Long and short for me is that while I've made a good guess at a chip I could use (see earlier) I cannot find it anywhere I'm open to suggestions and am willing to try!

@Dylan: The clockrate is at 0x4 in the header file. I'm not sure about the number you should replace there, but we have a resident expert (Zoinkity) who will hopefully advise when he next swings by!

You might recognise some names on this thread discussing the matter:
http://www.emutalk.net/threads/54293-n64-card-questions
_________________
No Mr. Bond, I expect you to be re-coded!

[MRKane]

I'm surprised I missed this:
http://www.youtube.com/watch?v=NCbkDgCIKRQ

Information on the crystals for the RCP is horribly sparse online, so I didn't follow it up to much of an extent (besides knowing that different revisions use different crystals and that the crystals are timed to the output) - however the above solves the sync issue with the HMDI card.

EDIT: Just because I can't seem to leave well-enough alone:
http://www.futuretech.blinkenlights.nl/contcpu.html
suggests that the RCP is connected directly to the Main Controller (MC) in a 2/3 ratio, which is kind of contradictory to the above claims...but I'm wondering if that has to do with the board revision. Long story short: Dylan for the win regardless I suppose lol.
_________________
No Mr. Bond, I expect you to be re-coded!

[MRKane]

Can't leave well enough alone...

Was reading through the thread you posted Trevor and found one of those chips was still out there to get:

http://www.ebay.com/itm/4-pcs-TC59R1809HK-TOSHIBA-PDSO-36-2Mx9-RAMBUS-/121367725365

Even more exciting is that the datasheets on this particular chip state that the latency is 10ns running fast (5V) and about 60ns running medium (3.3V) which is a sure improvement of the 640ns on the standard N64 - so that's the test I'll be undertaking when the stuff finally arrives at my door.

It'd be fantastic if there was a 4MB "sister" of this chip so that I could try doing a ramswap with the N64 console, and spare having an expansion pack - anyone with the "know how" on this?
_________________
No Mr. Bond, I expect you to be re-coded!

[Trevor]

Oh WOW, I wish you all the soldering luck

hehe, or... get another 2 4MB chips and do the expansion pak too, giving the N64 16MB of ram for future GE/PD updates.
Im sure Zoink would love to exploite you... I mean your N64 hahaha

Ive just thought... at 16MB of ram, ALL of GE could be loaded ... at once... would it be any faster though? probably not...

that said... the PIF runs at something like 1Mbit/s or something as against the RAM at 500Mbit/s... hmm... anyone care to comment?

Trev
_________________