The A64tweaker database

[craig588]

There can be 2 sections, stable and unstable. The reason I only want at least semi stable shots is so people can get ideas on how to tweak their memory, if enough people post shots than this thread can serve as a reference when someone needs to setup X type memory in X configuration, a screenshot than lasts for 5 seconds before crashing doesn't help anyone. Please include the type of ram, the speed in MHz the voltage, and the configuration it is in (EG 2x512, 1x1024, 3x256)

The database:


Memory type | Speed | Configuration | Voltage | Stable | Settings
BH5 | 260MHz | 2x512M | 3.4V | Yes | X (http://www.moominz.myby.co.uk/a64tweaker_260.jpg)
BH5 | 259MHz | 2x256M | 3.55V | Yes | X (http://www.xtremesystems.org/forums/attachment.php?attachmentid=22035)
BH5 | 250MHz | 2X256M | 3.35V | Yes | X (http://www.xtremesystems.org/forums/attachment.php?attachmentid=19914)
BH5 | 233MHz | 2X512M | 3.5V | Yes | X (http://img127.exs.cx/img127/5702/bh5233c.jpg)
BH5 | 260MHz | 1X256M | 3.35V | Maybe | X (http://www.xtremesystems.org/forums/attachment.php?attachmentid=19914)
BH5 | 263MHz | 2x256M | 3.8V | No | X (http://home.student.utwente.nl/r.makkink/41.16.JPG)
OCZ EB | 280MHz | 2X512M | 3.4V | Yes | X (http://s87999156.onlinehome.us/elisha/a64tweaker.JPG)
OCZ EB | 250MHz | 2X512M | 3.2V | Yes | X (http://img76.exs.cx/img76/8061/n2ztimings.jpg)
TCCD | 200MHz | 1X512M | 2.7V | Maybe | X (http://www.xtremesystems.org/forums/attachment.php?attachmentid=19948)

For the stable column:
"Yes" is 5 loops of 3dmark '01 and a 32M super pi run.
"Maybe" means that it is short term, but currently the long term stability is untested.
"No" means that it is benchable, but that is all that the settings are good for.

Your data will not be added unless you include all of the information!
Use the forums attachment feature to include your image, or if it is incredibly large, use www.imageshack.us

[Zanr Zij]

with TCCD I tested, Dynamic Idle cycle counter should set to disable 4 more performance...

[G H Z]

Great idea:)

But what about the performance differences? I think that's what Code Red's Tweaker is all about. Maybe you could include some sandra or everest numbers in this.

And stability is one thing, but speed/performance is really why we're all here. And I for one do not think P95 fit's the bill in finding stability that benchers can live with. I dunno, with all the problems people have running prime on A64 I'm just not sure you'll get good results.

[blinky]

i dont think prime should be considered in an A64 tweaker database

the tweaker is about getting the last few points in 3d and pi benching, so 3dmark stable and pi stable should be the stability tests

running these uber tight timings 24/7 makes so little difference in everyday performance its silly to spend the time worrying about them.

[craig588]

This is acctually supposed to be about helping people find the tightest timings a particular memory stick supports at X speed/voltage/configuration. It's not about performance, or who has the fastest sticks, it's just about making things work with as tight of timings as possible. The increase in the number of timings avalible is rather daunting when going from a P4C or AXP system to a A64 one.

I will use something other than prime though, I made this post initially a month or so ago, before people decided they didn't like prime.

Should the new stability test be 5 loops of 3dmark '01 and a 32M superpi?

[G H Z]

Sounds good:D

It does need to have some performance related measurement however. I could post up any Tweak settings that were stable but what good is it if they are slower than my defaults?

If hope you can get this going as it would give people unfamilar with Code's A64 Tweaker (or A64 BIOS in general), settings they can use that are faster for their particular Ram type.

Just trying to help:)

[Jessfm]

what about a chip code column

[AKRedneck]

Should the new stability test be 5 loops of 3dmark '01 and a 32M superpi?
That seems more like it.... And I don't even think it's necessary to loop all of 3dmark. Just loop lobby high, and if it can pass 75 loops of that, it's relatively stable. I've had it crash after 60 or so loops before. That test alone will turn most memory into egg cookers.
:D

[gocchin]

Mushkin PC3500 Level2 2x256 258.75mhz 3.55v Stable=Yes

[craig588]

Sounds good:D

It does need to have some performance related measurement however. I could post up any Tweak settings that were stable but what good is it if they are slower than my defaults?

The tighest timings ram can use are faster than the defaults. (Unless someone happened to get a very poor stick of ram than can only run the SPD timings) This is just about the tighest possible while still stable, not about any random stable setting. If someone wants to update a particular ram chip in a particular speed range, they can upload their own shot with better timings and I will use that one instead. Saaya also had a really good idea to add a little bit of text to each screen shot informing people of timings that were particularly influential on stablility or performance. Right now none of the shots have that text because the list is still rather short and I'm not sure how many people are interested in it so I don't want to waste my time on something no one cares about.

[craig588]

Mushkin PC3500 Level2 2x256 258.75mhz 3.55v Stable=Yes


Info added, try dropping your TRRD, I have usually been able to run it at 0 with no clock speed hit.

[Elisha]

http://s87999156.onlinehome.us/elisha/a64tweaker.JPG

Trtw setting not stable when set to = 2
but the setting above is benchable well at least 10 loops of lobby high.
this is ocz 3500eb at 8x280mhz @ 3.4v

thanks for the info Craig. i've been tryin to look fo EB timings for a long while.

[blinky]

Should the new stability test be 5 loops of 3dmark '01 and a 32M superpi?sounds good to me

so these are like everyday settings? wheres the database for lowest timings possible for pi 1m, and pifast etc.?

[Jack]

BH-5 | 262,5MHz | 2x 256 | 3,8V | No | pic (http://home.student.utwente.nl/r.makkink/41.16.JPG)

:)

shouldnt you incl SSandra or Everest b/w for example,
than you can compare the different A64Tweaker setting and find out which is best

[craig588]

sounds good to me

so these are like everyday settings? wheres the database for lowest timings possible for pi 1m, and pifast etc.?

Thats where the unstable catagory comes in.

[Trixxy]

BH5 | 260MHz | 2x512M | 3.4V | Yes | pic (http://www.moominz.myby.co.uk/a64tweaker_260.jpg)

looped all game tests 5 times, and did a 32M run of SuperPi to test.

http://www.moominz.myby.co.uk/a64tweaker_260.jpg

[craig588]

Everyones data has been added