What has Asus done?? CAS4 vs CAS3 on P5B Deluxe!

[metro.cl]

Well Bachus told me to do some testing cas 3 vs cas4, couse some weirdness is going on asus p5b deluxe, also noticed this last night running pi.


Here are the resutls:

SP8M 3-3-3-8 PC 6400 boot from bios memset as CAS4 (all precharge to activate, CS· to MCH-ODT latency and all precharge refresh were modified with memset all other are from bios)





SP8M 4-4-4-8 PC 6400 boot from bios





SP8M 5-5-5-5-8 PC 6400 boot from bios memset as CAS4 (all precharge to activate, CS· to MCH-ODT latency and all precharge refresh were modified with memset all other are from bios)



So 11 seconds better CAS4 over CAS 3 and CAS5 is 8 seconds better than CAS3 double


Any ideas? a way to fix this? normal?

[Rol-Co]

yep normal, boot 398 c3 and clockgen to 400, run again

[Vikodemous]

How? Who? What?

Have you tried 1M and 32M? I'm wondering what the difference there would be.

yep normal, boot 398 c3 and clockgen to 400, run again

So this has to do with the Boot strap issue. wow

[macci]

How would CL5-5-5 compare there?

The subtimings probably dont have huge impact but would it be possible to set them to identical values also?

[metro.cl]

How would CL5-5-5 compare there?

The subtimings probably dont have huge impact but would it be possible to set them to identical values also?

hehehe will try that also, im testing boot at 398 clockgen to 400 but seems to be the same no diference for CAS3

regards

[bachus_anonym]

yep normal, boot 398 c3 and clockgen to 400, run again

Nope, not really normal. This has started after BIOS 0507 (when 4:5 ratio has been somewhat fixed) and boot @ 398 and clockgen to 400 does not do it. Besides, it's not 398MHz where strap seems to change but 401 BIOS 0507 does not have this problem and CAS3 is indeed a bit faster than CAS4, compared to any other later BIOS.

The subtimings probably dont have huge impact but would it be possible to set them to identical values also?

I think both are identical. Three settings that change according to Memset, are the ones that cannot be set in BIOS. They change with CAS change and are identical on every BIOS I've tried.

[metro.cl]

ok set everything the same for cas3, cas4 and cas5, the 3 settings that changed (all precharge to activate, CS· to MCH-ODT latency and all precharge refresh) were modified in windows with memset so i cant be sure that they are correctly set or windows gets fooled , all other setting are from bios

cas5 is even better than cas3 that is really bad

[Fr3ak]

Do you think its worth trying to get a Asus engeneer comment on this subject? If they dont know why its acting like that, nobody knows...
I dont know if I can run cl3 at all, but I will give it a try tomorrow.

[Holst]

Will WPCRedit work on these boards?

Or am a just 2 years behind the times?

[erwinz]

woah wtf

[metro.cl]

Do you think its worth trying to get a Asus engeneer comment on this subject? If they dont know why its acting like that, nobody knows...
I dont know if I can run cl3 at all, but I will give it a try tomorrow.

dunno mate but would be interesting to have ram working at cas3 on ddr2 mostly because i can get mine at 440 cas3 and at that level performnce sucks.

Also many engeneer from other companies are working to get good ram with CAS3 so we should be able to use it

[ozzimark]

try 4-3-3, just for kicks

[MusicIsMyLife]

I had a performancedrop from running 4-4-4-12 to 3-3-3-8 of ~1 minute in superpi 32m.

E6400 (350x 8)
P5B dlx (bios 0405)
2x 1gb SuperTalent 800 @ 350

Unfortunately I didn´t test 4-3-3-8.

[Tulatin]

Actually, this was a bug i noticed on the P5WD2-E Prem too; when set to CAS 3 @ 800, ram would have a bit of a bandwidth disadvantage to CAS 4; odd, eh?

[ArcTan]

I thought it was just my board?

My 32M runs are about 1min longer running with 3-3-3-8 vs 4-4-4-8 as well

[Giana]

Omg

[toledo]

I thought it was just my board?

My 32M runs are about 1min longer running with 3-3-3-8 vs 4-4-4-8 as well

OMG .

Have you tried 4-4-4-8 vs 4-3-3-8 ? Has anybody tried that ?

[ozzimark]

OMG .

Have you tried 4-4-4-8 vs 4-3-3-8 ? Has anybody tried that ?

that's what i suggested. i'd try it, but i have a p5wd2.. a board that doesn't seem to have these weird problems

[bachus_anonym]

I found a cure and it's so simple - flash to 0711 It works for me, at least... I will post screens in a few minutes but 32M @ 450x9 8-4-4-4.0 gave me 13:08 and now running 8-3-3-3.0 and after 8th loop it's already 3.5s faster...

I gotta say, I start to like those ASUS bios people for listening and fixing glitches lately Now, all we need is to improve 4:5 ratio, so that we can see 476x9 and RAM @ 600MHz

UPDATE: Done!

SPi32M @ 450x9=4050MHz, DDR2-900 8-4-4-4.0 --> 13:08.282

SPi32M @ 450x9=4050MHz, DDR2-900 8-3-3-3.0 --> 12:57.782

Nice

[metro.cl]

I found a cure and it's so simple - flash to 0711 It works for me, at least... I will post screens in a few minutes but 32M @ 450x9 8-4-4-4.0 gave me 13:08 and now running 8-3-3-3.0 and after 8th loop it's already 3.5s faster...

I gotta say, I start to like those ASUS bios people for listening and fixing glitches lately Now, all we need is to improve 4:5 ratio, so that we can see 476x9 and RAM @ 600MHz

UPDATE: Done!

SPi32M @ 450x9=4050MHz, DDR2-900 8-4-4-4.0 --> 13:08.282

SPi32M @ 450x9=4050MHz, DDR2-900 8-3-3-3.0 --> 12:57.782

Nice

please put results, i'll try to check this i already flashed 711

[JuanFlaiter]

That was crappy, good news Bachus found the fix

[camouflage]

What about 975x chipset ???

Also a known bug there ???

[jmax_oc]

I gotta say, I start to like those ASUS bios people for listening and fixing glitches lately Now, all we need is to improve 4:5 ratio, so that we can see 476x9 and RAM @ 600MHz

It's not the subject but where do you manage to go with 4:5 ratio ?
Which is the best bios for it (after 0607 of course) ?

I get problems around 430 FSB (ram 545-550) with 0614

[toledo]

I found a cure and it's so simple - flash to 0711 It works for me, at least... I will post screens in a few minutes but 32M @ 450x9 8-4-4-4.0 gave me 13:08 and now running 8-3-3-3.0 and after 8th loop it's already 3.5s faster...

I thought the bug was related to Cas only , that is , the cas latency was causing problems . Do you mean 3-3-3 with bios 711 is faster than 4-4-4 ?

[zachy]

I thought the bug was related to Cas only , that is , the cas latency was causing problems . Do you mean 3-3-3 with bios 711 is faster than 4-4-4 ?

I've checked 3-3-3-4 vs 4-4-4-4 with same alpha timing 355x9 and guess what? 4-4-4-4 was 17s faster in super pi with 711 bios.