Re: BIOS F7:Optimized Defaults not being accepted

Benny wrote:


Thanks for agreat explannation Paul.
After posting I did find a post elsewhere that also mentioned the EIST function and how it works.
When I go to System Properties it does show the full "@ 3.16GHz".
The system seems to be running very stable so I won't be flashing the BIOS and the 4 RAM modules are working fine as well.
I think I will loeave well enough alone at the moment.
I opened CPU-Z. The SPD tab showed some interesting info about the 4 modules of RAM.
DIMMs 1 & 3 show the following info:

Memory SPD:

DIMMs 1 & 3
General
Memory type DDR3
Module format UDIMM
Manufacturer (ID) Corsair (7F7F9E0000000000)
Size 1024 MBytes
Max bandwidth PC3-10700H (667 MHz)
Part number CM3X1G1333C9DHX

Attributes
Number of banks 8
Nominal Voltage 1.50 Volts
EPP no
XMP no

Timings table
Frequency (MHz) 444 519 593 667
CAS# 6.0 7.0 8.0 9.0
RAS# to CAS# delay 6 8 9 10
RAS# Precharge 6 8 9 10
TRAS 16 19 22 25
TRC 23 27 31 34

While DIMMS 2 & 4 show the following:

DIMMs 2 & 4
General
Memory type DDR3
Module format UDIMM
Manufacturer (ID) Corsair (7F7F9E0000000000)
Size 1024 MBytes
Max bandwidth PC3-8500F (533 MHz)
Part number CM3X1024-1333C9DHX

Attributes
Number of banks 8
Nominal Voltage 1.50 Volts
EPP no
XMP no

Timings table
Frequency (MHz) 457 533 610 686
CAS# 6.0 7.0 8.0 9.0
RAS# to CAS# delay 6 7 9 10
RAS# Precharge 6 7 9 10
TRAS 18 20 23 26
TRC 24 27 31 35

How can this info be showing different details when they are exactly the same RAM out of exactly the same packages?
One set is showing bandwidth of 667MHz and the other set is showing 533MHz etc etc.

Also the Chipset info shows:

Northbridge Intel P45/P43 rev. A3
Southbridge Intel 82801JR (ICH10R) rev. 00
Graphic Interface PCI-Express
PCI-E Link Width x16
PCI-E Max Link Width x16
Memory Type DDR3
Memory Size 4096 MBytes
Channels Dual (Symmetric)
Memory Frequency 533.3 MHz (5:8)
CAS# 9.0
RAS# to CAS# 9
RAS# Precharge 9
Cycle Time (tRAS) 24
Command Rate 2T


regards
Benny

All I can say is... weird.

I don't think the BIOS has done a particularly bad thing in this case.
At least it selected safe values, even if the values are not as
aggressive as the SPD tables suggest.

On non-enthusiast memory, you expect every column in the SPD to
be valid. The frequencies chosen should preferably be "canonical"
ones, as then it makes it easier to for the BIOS to select the
same column from each DIMM pair, then take the slowest timings
from those two columns.

I suppose the BIOS could sweep for the slowest in both directions.
Like use the 593 column and the 533 column, and end up with
8-9-9-22 at 533.

Why not send your info to Corsair and ask them to explain ?

On an enthusiast memory, you expect the SPD to be largely
bogus. On an enthusiast memory, the customer expectation is
that the board will POST, and then the user can enter
the BIOS and make the "shrink-wrap" settings manually. For
example, if the RAM was supposed to run at 667MHz, an enthusiast
table night stop at 533MHz, and the first time the board starts,
it runs at 533MHz. The user then enters the BIOS and sets 667MHz
and the correct timings at that speed. If the BIOS battery
ever runs flat, the next time the system starts, it is at
533MHz again, and the board successfully POSTs. The user
is never in a position where they're not in control.

There was a time, when SPD problems prevented enthusiast memory
from even POSTing, forcing a user to insert a stick of
cheap RAM from Best Buy, to get the motherboard to start.

The 686 is also a bit weird, as it suggests a set of conditions
above the shrink-wrap rating ? Is the memory tested to those
values, as well as "9-9-9 @ 1333" ? My expectation would be
that any lengthy verification test at the factory, would be
at the shrink-wrap conditions.

As far as the SPD chips themselves are concerned, there are
two approaches.

On my Crucial modules, each module had a custom serial number
burned into the SPD. That implies the SPD is programmed separately
for each chip. Yet the timing tables all match.

I've also seen module pairs, where the SPD obviously come out
of a big barrel of identical chips. They could have been programmed
at the SPD factory for example. All the serial numbers are the
same (meaning the serial number is useless). There was even
one poster, who had a module where the SPD did not match
the module design (meaning they were so sloppy, they used
the wrong SPD chip entirely).

In your Corsair example, I see no logic to speak of :-) In
which case, I'd love to see their tech support explain it.

(Some canonical frequencies might be 400 for DDR3-800,
533 for DDR3-1066, and 667 for DDR3-1333.)

Paul
.

Relevant Pages

  • Re: P4P800se and populating all memory slots?
    ... I've tried using 2x512 basic Crucial modules with 2x512 Samsung ... and take me back to the bios screen). ... it ran the memory ... I'd start by looking at the SPD contents. ...
    (alt.comp.periphs.mainboard.asus)
  • Re: P5WD2 reports incorrect speed of RAM
    ... >> but it should not affect what is displayed in the BIOS. ... >Timing information is stored in the SPD chip on a DIMM. ... >Performance memory operates faster than any JEDEC spec. ... >To see more of what your BIOS has to offer, try settings ...
    (alt.comp.periphs.mainboard.asus)
  • Re: Asus Board / CPU 533 FSB / DDR400 clocked at DDR333 ?
    ... You can check whether the SPD of your memory modules is programmed to ... provide better timings when run at speeds below spec. ... PC100 use that specified faster timings. ...
    (alt.comp.periphs.mainboard.asus)
  • Re: ACTUAL RAM SPEED
    ... The CPU-z memory tab shows '166.7 MHz'. ... Depends on whether you have the BIOS use the SPD settings (what the ... be overclocking your 166MHz memory by running it at 200MHz. ...
    (microsoft.public.windowsxp.hardware)
  • Re: Identify DRAM speed - Alternative to CPUZ?
    ... The idea is, the additional timings are to help auto setup, ... and CPUZ memory tab reports ... if they used those settings in the SPD, ...
    (alt.comp.periphs.mainboard.asus)