215 2-2-2-5 stable, but no post at 225 3-4-4-8?

Illah · Sep 14, 2007

Has anyone came across a scenario like this? I'm running 4x512MB of BH-5 and I memtested stable at 215 2-2-2-5 but I have trouble posting at 225MHz no matter what the timings are. In my past experience the gap between stable and no-post was always much larger, especially when loosening timings. Do you think it has to do with using four sticks instead of two?

--Illah

Reefa_Madness · Sep 15, 2007

Have you tried the 225 at 2.5-3-3-8 (CAS 2.5)? With enough voltage I would think that you should be able to run 2-2-2-6, or worst case 2.5-2-2-6.

I've actually never tried running BH-5 at CAS 3 (after all, you use Winbond for tight timings, not the opposite), but the storyline is that BH-5 won't even run CAS 3.

Edit:

OK, after that realization that I had never tried it, and given that the rig that I'm typing on is running BH-5, I decided to try CAS 3 with 2 x 512...would not POST.

Given those results I just had, I would say the CAS 3 is at least part of your problem. The 4 sticks will certainly reduce your overclock.

How about giving a little more info, such as system specs and voltage to the ram. At 225 it is very probable that your sticks will already need more than default voltages.

Illah · Sep 15, 2007

Opty 165 (was previously an Opty 146 until this morning

)
Asrock Dual939 Sata2

4x512 BH-5 @ ~3.1v
-- 2x512 are my original set of BH-5 memtested stable at I think 242 2-2-2-5
-- 2x512 are OCZ PC3500 Plat LE BH-5 tested stable to 225 2-2-2-5 ~3.1v (prob has way more headroom, I was just too lazy to bother going higher)

For testing purposes the CPU was barely overclocked to remove that as a variable. I was running my Opty 146 1:1 with the ram so it varied between 2150 and 2250 for the tests (it's dead stable at 2700).

I'll give cas 2.5 a shot, I didn't know about the cas 3 issue! I was just getting frustrated so I went 3-4-4-8

I knew going 4x512 would cost me a bit but I can live with that. Despite my RAM being stable north of 240 I ran day to day at 225 due to CPU limitations (dividers allowed either 225 or 270, and 270 will obviously never happen). Losing 10MHz isn't the end of the world, I'm not Mr. Super Gamer so the extra GB is worth more to me than a few seconds in a benchmark.

--Illah

RollingThunder · Sep 15, 2007

Illah said:
Opty 165 (was previously an Opty 146 until this morning )
Asrock Dual939 Sata2

4x512 BH-5 @ ~3.1v
-- 2x512 are my original set of BH-5 memtested stable at I think 242 2-2-2-5
-- 2x512 are OCZ PC3500 Plat LE BH-5 tested stable to 225 2-2-2-5 ~3.1v (prob has way more headroom, I was just too lazy to bother going higher)

For testing purposes the CPU was barely overclocked to remove that as a variable. I was running my Opty 146 1:1 with the ram so it varied between 2150 and 2250 for the tests (it's dead stable at 2700).

I'll give cas 2.5 a shot, I didn't know about the cas 3 issue! I was just getting frustrated so I went 3-4-4-8

I knew going 4x512 would cost me a bit but I can live with that. Despite my RAM being stable north of 240 I ran day to day at 225 due to CPU limitations (dividers allowed either 225 or 270, and 270 will obviously never happen). Losing 10MHz isn't the end of the world, I'm not Mr. Super Gamer so the extra GB is worth more to me than a few seconds in a benchmark.

--Illah

Illah,

I'm not going to stick my nose in here too much because Reefa is "Da Man," but your 4 sticks aren't the same brand and possibly have different voltage requirements. Not an ideal situation.

You might want to step up your RAMs in smaller increments from 215 to 225 to see where they fail, then try 2.5 at that point.

That's a LOT of voltage to give those RAMs and you must have had volt mods done to get that much. If you're not using good direct cooling on them, I'd be careful. How much vcore are you using?

Reefa_Madness · Sep 16, 2007

2x512 are OCZ PC3500 Plat LE BH-5

Just a little tidbit that you may or may not know, but those are unique sticks, as they are the only original (not UTT) Winbond BH-5 that I'm aware of that was made using a Brain Power PCB. They have generally been pretty good clockers.

Double check yours to confirm this when you get a chance. On the edge of the module's PCB you should see "BPxxx", just don't recall whether just exactly which version of the BP it will be (xxx = 805 or 815, etc).

Neuromancer · Sep 16, 2007

Something else to consider is OCZ penchant for producing great RAM that just will not run cas 3 at all..
What causes it I do not know but it seems to be a pretty common ailment.
My TCCD would run 200 at 3cas unstably if I gave them 2.9volts. but cas 2 I could get up to around 235 with less.

jonspd · Sep 16, 2007

Yep I think everyone rounded it up it's kinda like putting 91 octane in a jet... it;s just hard for it to run something so slow seeing as it's build to be fast...

Reefa_Madness · Sep 16, 2007

Neur0mancer said:
Something else to consider is OCZ penchant for producing great RAM that just will not run cas 3 at all..
What causes it I do not know but it seems to be a pretty common ailment.
My TCCD would run 200 at 3cas unstably if I gave them 2.9volts. but cas 2 I could get up to around 235 with less.

There was a group of TCCx that was failing at CAS 2.5 after short periods of use, but while I'm not certain about this, I seem to recall that it may have been limited to a few production runs, as opposed to all of their TCCD (or TCC5).

The CAS 3 thing with the Winbond BH-5 is across the board, however, regardless of what brand module was being used. It was an IC level thing that prevented them running CAS 3.

Moto7451 · Sep 16, 2007

More voltage more voltage more voltage

When I was using my UTT they came alive at 3.4v and went from clocking at 210-220MHz to 240MHz. At 3.6v I could to 260MHz. After that more voltage didn't help.

Neuromancer · Sep 16, 2007

Reefa_Madness said:
There was a group of TCCx that was failing at CAS 2.5 after short periods of use, but while I'm not certain about this, I seem to recall that it may have been limited to a few production runs, as opposed to all of their TCCD (or TCC5).

The CAS 3 thing with the Winbond BH-5 is across the board, however, regardless of what brand module was being used. It was an IC level thing that prevented them running CAS 3.

Oh yah I was not implying that ALL TCCx was problematic, just that the problem exists and OCZ has had a problem with sticks not running cas3 and in my case it was TCC ram. (Come to think of it.. it was tcc5 not tccd.. I had the rev 2 v1.1s later i got the tccd but just sold them instead of running them).

Did not know aboiut the BH5 cas 3 issue though

Good to know.

fldrice · Sep 18, 2007

Is that on 2T? 4 dimms ocing on A64 has always been problematic for s939.

215 2-2-2-5 stable, but no post at 225 3-4-4-8?

Illah

Member

Reefa_Madness

DRAM Guru Senior

Illah

Member

RollingThunder

Destroyer of Trolls & Spammers

Reefa_Madness

DRAM Guru Senior

Neuromancer

Member

jonspd

Super Speedy Senior

Reefa_Madness

DRAM Guru Senior

Moto7451

Senior Something

Neuromancer

Member

fldrice

Member

Similar threads