Does my memory clash with my processor?

  • Thread starter Thread starter RayLopez99
  • Start date Start date
R

RayLopez99

I am going to reinstall Windows 7 on my quad core i5--this time a legal version instead of a pirate version--and as I was going through the specs below I noticed some warning red flags. First, this Intel(R) Core(TM) i5-2400 CPU @ 3.10GHz chip, I bought in early 2011, just as it was newly released, but the memory was from a year earlier, 2010, which is always a potential problem. Then I notice that the Topala Software Solutions freeware program information on the memory states "Warning! Wrong values reported by BIOS".

Then I notice this anomaly of sorts (speed of memory not the same as 'Supported Frequencies', off by a little--is this a mobo problem?):
Speed 667 MHz (DDR3 1333)

Supported Frequencies
457.1 MHz, 533.3 MHz, 609.5 MHz, 685.7 MHz

I am assuming the CPU can adjust to some default slower frequency if the 'supported frequency' is not the same as the actual frequency?

On occasion this system is somewhat unstable it seems, and I'm wondering ifperhaps the BIOS is set up wrong (I had thought it was the pirated Windows). If I do a DBAN clean of the HD and reinstall Windows 7 Pro (legal version) should it detect and do whatever necessary to the memory automatically?Or should I go inside the BIOS and fiddle around (I hate that)? Or perhaps the mobo is all wrong? I will add therefore the motherboard and BIOS information below*

Printout below. Thanks in advance.

RL

http://ark.intel.com/products/52208/Intel-Core-i5-2400S-Processor-6M-Cache-up-to-3_30-GHz?q=i5-2400

As reported by Copyright © 2005-2011 Topala Software Solutions

Number of Logical Processors 4

CPU Name

Intel(R) Core(TM) i5-2400 CPU @ 3.10GHz

CPU Code Name
Sandy Bridge

Vendor
GenuineIntel

Number of Bits 64

Instruction Set
MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, ET64, XD, VMX, SMX, EST

Platform Name
Socket 1155 LGA

Revision
D2

Technology
32 nm

Original Clock
3100 MHz

Original System Clock
100 MHz

Original Multiplier
31.0

CPU Clock
3100 MHz

System Clock
99.8 MHz

Number of Cores
4

Turbo Boost
Enabled

Virtual Technology
Enabled

SLAT
Supported

Hyper Threading
Disabled

///////////
Memory

Top

Property

Value

Memory Summary

Warning!
Wrong values reported by BIOS


Maximum Capacity
16384 MBytes

Maximum Memory Module Size
[unknown]

Memory Slots
2

Error Correction
None


DRAM Frequency
665.2 MHz


Memory Timings
9-9-9-24 (CL-RCD-RP-RAS)


Device Locator
Manufacturer
Kingston

Part Number
99U5471-002.A00LF


Serial Number
711E63ED


Capacity
2048 MBytes


Memory Type
DDR3 (PC3-10700)

Speed
667 MHz (DDR3 1333)

Supported Frequencies
457.1 MHz, 533.3 MHz, 609.5 MHz, 685.7 MHz


Memory Timings
6-6-6-17-23 at 457.1 MHz, at 1.5 volts (CL-RCD-RP-RAS-RC)


Memory Timings
7-7-7-20-27 at 533.3 MHz, at 1.5 volts (CL-RCD-RP-RAS-RC)

Memory Timings
8-8-8-22-30 at 609.5 MHz, at 1.5 volts (CL-RCD-RP-RAS-RC)


Memory Timings
9-9-9-25-34 at 685.7 MHz, at 1.5 volts (CL-RCD-RP-RAS-RC)


Manufacturing Date
2010, Week 37



EPP SPD Support
No


XMP SPD Support
No

Device Locator

Manufacturer
Kingston



Part Number
99U5471-002.A00LF



Serial Number
0F1FD952



Capacity
2048 MBytes



Memory Type
DDR3 (PC3-10700)



Speed
667 MHz (DDR3 1333)



Supported Frequencies
457.1 MHz, 533.3 MHz, 609.5 MHz, 685.7 MHz



Memory Timings
6-6-6-17-23 at 457.1 MHz, at 1.5 volts (CL-RCD-RP-RAS-RC)



Memory Timings
7-7-7-20-27 at 533.3 MHz, at 1.5 volts (CL-RCD-RP-RAS-RC)



Memory Timings
8-8-8-22-30 at 609.5 MHz, at 1.5 volts (CL-RCD-RP-RAS-RC)



Memory Timings
9-9-9-25-34 at 685.7 MHz, at 1.5 volts (CL-RCD-RP-RAS-RC)



Manufacturing Date
2010, Week 37



EPP SPD Support
No



XMP SPD Support
No
//
* MOBO AND BIOS INFO HERE--

Summary



Manufacturer

ASUSTeK Computer INC.



Model

P8H67-M LE



Version

Rev X.0x



Serial Number

MF70B3G05500624



Bridge



North Bridge

Intel Sandy Bridge Revision 09



South Bridge

Intel H67 Revision B3



CPU



Name

Intel(R) Core(TM) i5-2400 CPU @ 3.10GHz



Cpu Socket

Socket 1155 LGA



Max CPU Speed

3800 MHz



Memory



Maximum Capacity

16384 MBytes



Maximum Memory Module Size

[unknown]



Memory Slots

2



System Slots



ISA

0



PCI

4



AGP

0



VL-BUS

0



EISA

0



PCMCIA

0



ExpressCard

0



MCA

0


--------------------------------------------------------------------------------

BIOS



Top



Property

Value



BIOS Properties



BIOS Vendor

American Megatrends Inc.



Serial Number

Syst-em-



BIOS Version

0603



Firmware Version




BIOS Date

02/11/2011



BIOS Size

4096 KB



BIOS Starting Segment

F000h



DMI Version

2.6



Characteristics



supports ISA

No



supports MCA

No



supports EISA

No



supports PCI

Yes



supports PC Card (PCMCIA)

No



supports Plug-and-Play

No



supports APM

No



upgradeable (Flash) BIOS

Yes



allows BIOS shadowing

Yes



supports VL-VESA

No



ESCD support is available

No



supports booting from CD-ROM

Yes



supports selectable boot

Yes



BIOS ROM is socketed

Yes



supports booting from PC Card (PCMCIA)

No



supports Enhanced Disk Drive specification

Yes



supports INT 13 for Japanese NEC 9800 1.2M floppy (3.5-inch, 1024-byte sectors, 360rpm)

No



supports INT 13 for Japanese Toshiba 1.2M floppy (3.5-inch, 360rpm)

No



supports INT 13 5.25-inch/360K floppy services

No



supports INT 13 5.25-inch/1.2M floppy services

Yes



supports INT 13 3.5-inch/720K floppy services

Yes



supports INT 13 3.5-inch/2.88M floppy services

Yes



supports INT 05 print-screen

Yes



supports INT 09 and 8042 keyboard services

Yes



supports INT 14 serial services

Yes



supports INT 17 printer services

Yes



supports INT 10 CGA/Mono video services

No



NEC PC-98

No



supports ACPI

Yes



supports legacy USB

Yes



supports AGP

No



supports booting from I2O device

No



supports booting from LS-120

No



supports booting from ATAPI ZIP drive

No



supports booting from IEEE 1394 device

No



Smart Battery supported

No



BIOS Boot Specification supported

Yes



Function key-initiated Network Service boot supported

No



Enable Targeted Content Distribution

Yes



ACPI



APIC

ALASKA A M I



FACP

ALASKA A M I



HPET

ALASKA A M I



MCFG

ALASKA A M I



SSDT

AMICPU PROC



SLIC

ACRSYS ACRPRDCT


--------------------------------------------------------------------------------
 
On Sunday, December 2, 2012 6:53:06 PM UTC+2, RayLopez99 wrote:

[]

I will add that the system was built by a PC shop overseas in Asia, but that does not mean they knew what they were doing. A lot of times they just want to make a sale, and if the mobo does not support "Sandy Bridge" 100% then I'm screwed. I did not pay a lot for the system however, small consolation. But I think I'm OK, just that on occasion it seems to hang a bit--hopefully not hardware related though.

RL
 
I notice that the Topala Software Solutions freeware program information
on the memory states
"Warning! Wrong values reported by BIOS".

Several conditions can cause this error report. The easiest
remedied is a dying CMOS button battery.
 
But I think I'm OK, just that on occasion it seems to hang a bit--hopefully not hardware related though. If I do a DBAN clean of the HD and reinstall Windows 7 Pro (legal version) should it detect and do whatever necessary to the memory automatically?

http://bitsum.com/prolasso.php

This is Process Lasso. I can see in the pro version it mentions a
ProBalance act for restraining watchdog rules as various setup
parameters for throttling a CPU. Still, might be worth a look at the
free portable version (although the Pro is also portable), as at least
it would seem likely to have a limited subset of the same parameters,
rather than pull the routine in point of the whole or greater program
intent. Evidently there's some hesitancy issues or that potential
according to program classifications, between arbitration factors and
a deterministic nature to each CPU core when prioritizing
actualization upon a premise of best realized resources.

I do see where it often throttles processes on a couple programs I
use, in a sort of graphical interplay of net responsiveness, I believe
would be a desirable effect, for memory and a couple other factors to
offset. All pretty much a slight of hand, I suspect, in this
hyperthreading scheme of things Intel provides at least on this
processor, generally so identified by most as more than a core. Where
I do have the physical cores, it's not so much resource intensive as
only application by CPU intensiveness, hyperthreading simply wouldn't
cut. A matter of when and how to bring a CPU environment down to its
knees and hard.

Have a look to a residual of indicatives if you can duplicate specific
instances for other than memory dependencies.
 
RayLopez99 said:
I am going to reinstall Windows 7 on my quad core i5--this time a legal version instead of a pirate version--and as I was going through the specs below I noticed some warning red flags. First, this Intel(R) Core(TM) i5-2400 CPU @ 3.10GHz chip, I bought in early 2011, just as it was newly released, but the memory was from a year earlier, 2010, which is always a potential problem. Then I notice that the Topala Software Solutions freeware program information on the memory states "Warning! Wrong values reported by BIOS".

Then I notice this anomaly of sorts (speed of memory not the same as 'Supported Frequencies', off by a little--is this a mobo problem?):
Speed 667 MHz (DDR3 1333)

Supported Frequencies
457.1 MHz, 533.3 MHz, 609.5 MHz, 685.7 MHz

I am assuming the CPU can adjust to some default slower frequency if the 'supported frequency' is not the same as the actual frequency?

On occasion this system is somewhat unstable it seems, and I'm wondering if perhaps the BIOS is set up wrong (I had thought it was the pirated Windows). If I do a DBAN clean of the HD and reinstall Windows 7 Pro (legal version) should it detect and do whatever necessary to the memory automatically? Or should I go inside the BIOS and fiddle around (I hate that)? Or perhaps the mobo is all wrong? I will add therefore the motherboard and BIOS information below*

Printout below. Thanks in advance.

RL

In the interest of saving a few electrons, I've snipped the listing.

The memory chips themselves are flexible. Inside, the chips are "analog"
and not "digital".

Say the clock is 1GHz. The period of one clock cycle is 1 nanosecond.
Say CAS is 13. If you wish to compute the timing, multiply the CAS value
by the clock period. The answer is 13ns.

Any time you then change the clock value, the clock period changes. To
take advantage of the speedy 13ns figure that the memory chip is capable
of supporting inside, you recompute the correct CAS and use it. That's
what the BIOS does, when you use canonical clock values. Those are
the values you can easily select from the menu.

When a motherboard boots for the first time, the settings are
typically all set to "Auto". OK, so what happens in that case ?

The BSEL pins tell the clock chip something about the default
clock to use. On more modern processors, where Intel has locked
things a bit more, this is even simpler, in that the base clock
is stuck at 100MHz. The multiplier information, is stored inside
the processor some how, with a "low" and "high" value. So those
things can be set by the BIOS pretty easily. The "low" value
gets used, when the desktop is idle, and a lower power P-state
is desired.

The processor has a clock generator for the memory. The clock value
can be modified by the ratio of simple integers. That's how
the processor can make 533MHz, 667MHz, and the like, clock
sources for memory. (It depends on the generation of processor,
whether all the clocks are made by an external generator, or
some of them are made by PLLs running inside the processor.
The method used, has an effect on timing margins, which is
why the method may change from one processor generation to another.)

If you artificially push the BCLK off frequency, the memory
clocks will have similarly scaled values. But if everything
is nominal or auto, the clocks should be pretty close to
the canonical ratios.

The processor might have the peak nominal value, set up for
DDR3-1333 memory. If you install DDR3-1600 memory, perhaps
the Auto setting still results in the configuration running
at DDR3-1333. If the memory is enthusiast type, supports XMP,
the motherboard supports XMP, the memory subsystem might
automatically be set up to whatever the correct value should be.

Now, say I'm a cheap skate, and I buy DDR3-1066 memory. The
BIOS knows it can't run that at DDR3-1333, so it automatically
selects the correct ratio, and runs the RAM at DDR3-1066. The
SPD chip on the DIMM, has a couple of values for the timing
parameters, such that the CAS is also adjusted properly.
So maybe it's CAS13 at DDR3-1333 and CAS10 at DDR3-1066,
both yielding the same analog time constant inside the
memory chip (so the memory stays happy). [Note - those
values are to indicate that different values get used, without
me worrying about whether the math is exactly right in this
case.]

So without really going into any details, the answer lies in
the BIOS. Observing the current values in something like CPU-Z,
then visiting the BIOS and seeing how it is set up. A Dell or
HP or Acer motherboard, might not have any BIOS memory settings
at all. Then, the BIOS gets to choose whatever it feels like,
based on the SPD table from the DIMM. A home-builder motherboard
from Asus or Gigabyte, would have an extensive BIOS setting page
for the memory.

You do not need to adjust all the settings. When you don't
know what a setting does, set it to "Auto". You can use the
canonical memory speed menu, to select 533 or 667 or whatever
is appropriate for your DDR3-1066 or DDR3-1333 memory. For
those values, perhaps the BIOS computed CAS value will be
correct. When you do various overclocking tricks to a
motherboard, then the arithmetic becomes a lot more complicated,
and the user needs to visit an enthusiast web site, to learn
more about properly setting it up. The reason that is
necessary, is no BIOS really works "right". They all
have their quirks, where they're not displaying the
"real" frequencies involved. If you use CPUZ in Windows,
you may learn about these quirks. So even without
enthusiast info to go on, you can take baby steps
yourself to discover the multiplier chain for
yourself. The secret is to work in tiny steps, and
then observe which parameters are changing, and
which other parameters did *not* get adjusted
properly. (For example, change BCLK from 100MHz to
101MHz, and see which other frequencies are affected.)
But for simple canonical changes, like leaving all settings
at Auto, except setting the memory clock to 533 or 667
for DDR3-1066 or DDR3-1333 memory, that should work
pretty well without a lot of fuss.

I don't really think the memory is clashing with
the processor. It really depends on how crazy your
shop techs are, as to how messed up it is. If they're
the "super-overclocker" types, it could be a real mess.
No shop really wants the customer coming back for
free maintenance, when they screw it up. So there
isn't a lot of incentive to twiddle every possible
setting in there.

Don't forget to test with memtest86+. The download
is half way down the web page (scroll down). If you
detect errors this way, then the memory settings
would be badly screwed up. This test isn't that
strenuous. Running a single complete pass (pass count
equals one), is sufficient. Might take 20 minutes
or so. So even if you're super-afraid of the BIOS,
you can still help yourself knowledge-wise, without
entering the BIOS.

http://www.memtest.org/

Paul
 
Have a look to a residual of indicatives if you can duplicate specific

instances for other than memory dependencies.

Wow. Impressive. But apparently this is not a static 'one time only' setup which is what I want, but a 'dynamic, on the fly, different changes for different programs run' setup, more for overclockers and gamers than me.

RL



In addition to ProBalance, there are countless features allowing the user to take full automated control of the processes on their PC. You can have a wide range operations performed, or settings applied, each time a process is run. Core optimization technologies allow you to choose on what CPUs/cores a process should run, as well as what their CPU priority class and I/O priority class should be. You can also disallow certain processes from running, log all processes run, and even set various other process rules. These rules, along with many others, include automatically restarting or terminating a process after it reaches some CPU or memory threshold. For license enforcement, you can limit the number of instances of a process that can be running. A gaming mode allows for easy process priority optimization for avidgamers.
 

Thanks Paul. For noob me, I got two takeaways from reading your post: usememtest.org's test (which as I recall, I did, and I set it up for overnight, not 15 minutes, because the test took that long and it passed), and, useAUTO in BIOS for settings and let the uP figure it out on bootup. Sounds good to me! I'm not into performance, but stability. And as I say the slight crashes or freezes (no BSOD) I've had I think are due to stuff like a bad pirate program, or maybe like Don P says a bad CMOS battery on the mobo,and I don't think my memory is seated improperly but I'll double check since I'll be adding a new graphics card to replace the on-chip Intel built-inGPU.

RL
 
RayLopez99 said:
Thanks Paul. For noob me, I got two takeaways from reading your post: use memtest.org's test (which as I recall, I did, and I set it up for overnight, not 15 minutes, because the test took that long and it passed), and, use AUTO in BIOS for settings and let the uP figure it out on bootup. Sounds good to me! I'm not into performance, but stability. And as I say the slight crashes or freezes (no BSOD) I've had I think are due to stuff like a bad pirate program, or maybe like Don P says a bad CMOS battery on the mobo, and I don't think my memory is seated improperly but I'll double check since I'll be adding a new graphics card to replace the on-chip Intel built-in GPU.

RL

The first part of any technical endeavor is "observation".
Comparing BIOS settings to CPUZ viewed settings while
in Windows.

On some of my motherboards here, I change the main
Auto setting to Manual. Only to get the settings to
all be displayed in the BIOS. Not to adjust anything.
I then make very few adjustments to the memory. Like
you, I might wish to correct the memory clock, since
the BIOS might default to the wrong canonical value.
For example, the old DDR400 low-CAS memory, wouldn't
use the guaranteed aggressive settings by default,
and the BIOS does that to ensure it starts the first
time (i.e. make a mistake, reset the CMOS, and try again).
So they may not have set up things, to "go aggressive"
immediately.

There are many numeric values in the BIOS memory settings
page, that I don't know how to set up. One setting,
for example, is a function of round trip flight time,
between the processor and memory slot, using the copper
traces as the path. The manufacturer knows the correct value.
I've seen hobbyists guess at it (kinda shocking to me, since
it's so easy to get wrong). But for a first visit, the plan
isn't to twiddle everything. Just make sure the clock and
maybe tCAS, look reasonable.

I had one motherboard, where overclocking the processor core,
I cranked the memory clock back one notch. And... the BIOS
ignored my memory setting, and ran the memory much faster
than I'd planned. The funny part is, the computer booted
just fine. Since I was immediately going to use CPUZ, I
found the bug right away. (My DDR2-800 CAS5 memory was
run at DDR2-1066 CAS5.) So observation is the key,
as much as anything. Visiting the enthusiast sites, to learn
more, is a second step, and only if you're interested.

On boards like Gigabyte, the memory page may be hidden.
There can be a function key you press, to get the advanced
memory page to appear. Whereas, on my Asus, the memory page
is always there, but you may need to switch the main setting
from Auto to Manual, just to get it to display the numbers
it is using.

Paul
 
Back
Top