Edited: 4-11-2006. Although mentioned with regard to I875, the first link in the original post wasn't actually very relevant to I875, so the verbiage used to introduce that link was changed. Added some comments relating to the first link. Also added the following:
Start =======
Given that
Originally posted by: Zap
Tested out one stick at a time. One will pass Memtest only at CAS 3, the other will pass at CAS 2.5. Neither will do CAS 2 as advertised. Note that as I'm testing these, NOTHING is overclocked. Completely stock CPU speeds and all RAM latencies are AUTO except for CAS.
The memory is erratic and I vote that you return the modules any way you can. I think the "RamGuy" would be would be your best shot in that regard.
End =======
Originally posted by: Zap
Canterwood = 875P right?
Yep - Canterwood is Intel 875P
With regard to memory and timings, the following might be informative:
http://www.madshrimps.be/?action=getarticle&number=1&artpage=1014&articID=258
Note that the article is dated 2004-11-30 and is S754 oriented - well before the later 939 socket and the latest cores and their improvements to the memory controller. But also note the following statement in that article:
I discovered 2 values which increased performance significantly with these TCCD modules:
DQS Skew Contol: Increase Skew
DRAM Drive Strenght: Level 1
These netted me an increase of ~10Mhz when overclocking at CL2.5 3-3-10. They did not impact OC potential at tight timings though.
If your BIOS has the two parameters noted above, you might try some experiments with those parameters as well.
I've had direct experience with Corsair memory on two occasions.
My first use of Corsair was when I used their 512MB PC133 CL2 modules in 5 PCs for which I provided specifications in 2001. I wanted 1GB of PC133 memory in each box (A7V-133 R1.05 motherboards). At the time, Corsair was the only company providing such modules, which were built using binned Micron chips. We ended up buying 10 of these modules and every one worked perfectly. I was pleased - Corsair had lived up to their good brand name (and high price).
But my second experience with Corsair modules, which co-incided with some reasonably lengthy research I did in determining which memory modules use TCCD chips, considerably tarnished Corsair's reputation in my eyes. I now think of Corsair as "just another memory manufacturer" who seemingly is coasting on their reputation.
What happened? First, I bought a 512MB stick of Corsair PC3200 memory, model CMX512-3200C2PT that was advertised as "new" on eBay. Within one day after that auction ended, I discovered this database:
http://www.techpowerup.com/memdb/
When the Corsair memory arrived, I noted that its revision number was XMS3202 V1.1 (from the database, Winbond BH-6 chips) - if this memory was "new", at best it was "new old stock." I installed it in my VNF3-250 (E6 Sempy) and it didn't work - long beeps. But when I installed it in an ASUS A7V600-X motherboard using the Via KT600 chipset (Athlon XP processor), it worked just fine and passed 6+ hours of continuous memtest86 testing. So I sold it together with the A7V600-X.
I suggest you take a hard look at the revision codes on your memory.
But I've digressed a bit - let's focus on that database. Consider, for example, the entry for "Corsair XMS-PC3200C2 Rev4.2 " Corsair produced those modules using Samsung TCCD and Samsung TCC5 memory chips. Huh? They used
different chips on a given model designation of memory module and didn't even bother to at least change the revision number? That can't exactly please those of us who like to match memory modules right down to matching the memory chips. Thus there is a down side to the current widespread use of heat spreaders on memory modules - those spreaders hide the memory chips. But that may not matter much since the markings on many memory chips nowadays don't reveal their true maker (and that maker's model designation).
To be fair, Corsair's offenses along these lines pale in comparison to similar transgressions by other memory makers, particularly Kingston.
And I haven't even touched on the PCBs used in the various memory modules. See:
http://www.legitreviews.com/article.php?aid=53&pid=1
The article is rather dated - Oct 21, 2004 - but the point it makes probably still applies. I wonder if Corsair changes the Revision number for a given module when they change its PCB.
But I admit - I'm somewhat "picking on" Corsair, probably because their module didn't work in my VNF3-250. In terms of identifying changes to memory modules, they may be better than most with regard to identifying changes to memory modules - they've seemingly got Kingston beaten "like a drum" in this regard.
Facebook
Twitter