Everest Home Report ?

  • Thread starter Thread starter BF
  • Start date Start date
B

BF

Out of curiosity ran Everest.
CPU report says that the CPU's are running at 1587 MHz. They are nominal 2.4
GHz Xeon's. Is this normal or is something wrong? If wrong, is there a way
to adjust?
I'm not otherwise having any problems, just wondering if I'm missing out on
performance.
Thanks for any info,
BF
 
"BF" said:
Out of curiosity ran Everest.
CPU report says that the CPU's are running at 1587 MHz. They are nominal 2.4
GHz Xeon's. Is this normal or is something wrong? If wrong, is there a way
to adjust?
I'm not otherwise having any problems, just wondering if I'm missing out on
performance.
Thanks for any info,
BF
Click to expand...

Consider some clock speeds: 100MHz, 133MHz, 200MHz

The ratio of 200/133 equals 2.4/1.587 and it could be
your CPU clock is set too low. Without any details about
the processors (their SL code or order code), or the motherboard,
it is pretty hard to speculate further.

Paul
 
Paul said:
Consider some clock speeds: 100MHz, 133MHz, 200MHz

The ratio of 200/133 equals 2.4/1.587 and it could be
your CPU clock is set too low. Without any details about
the processors (their SL code or order code), or the motherboard,
it is pretty hard to speculate further.

Paul
Click to expand...

Paul,
Thanks, is the CPU clock speed user adjustable? If so, how?
What is "SL code" and how does one find out and what does one learn from it?
The motherboard is a Supermicro P4DC6 but I doubt that is of much help. What
other information would be of help and where is to be found?
I appreciate any information, but note that had I not run Everest, I
wouldn't have even known that I had a "problem", so this is not a super
critical situation.
BF
 
"BF" said:
Paul,
Thanks, is the CPU clock speed user adjustable? If so, how?
What is "SL code" and how does one find out and what does one learn from it?
The motherboard is a Supermicro P4DC6 but I doubt that is of much help. What
other information would be of help and where is to be found?
I appreciate any information, but note that had I not run Everest, I
wouldn't have even known that I had a "problem", so this is not a super
critical situation.
BF
Click to expand...

According to this page, the motherboard is FSB400, which means
a 100MHz clock.

http://www.supermicro.com/products/motherboard/Xeon/860/P4DC6.cfm

Either Everest is reporting the speed incorrectly, or you have
taken a processor with a different normal bus clock requirement,
and plugged it into a motherboard that can only support FSB400.

But, based on my math above, I don't see an easy combination of
processor clock choices, to give the 3/2 ratio you are reporting
above. There is no processor that uses a 150MHz clock, so I cannot
make sense of it that way.

What does Everest tell you about the processor ? Do you have the
box that the processor came in ? Are there any numbers printed
on the box ?

You can also download this and double check the results. This
is a popular tool used to report CPU and memory clock info.

http://www.cpuid.com/download/cpu-z-131.zip

If you had a 2.4GHz/FSB533 processor and plugged it into a
motherboard that will only run at FSB400, the resulting speed
is 1.8GHz. Yet, you report 1.587, so that doesn't work out.

On the Intel processorfinder.intel.com site, if I select
"Xeon", then filter on 2.4GHz and socket 603, I get six
processors returned. They all have FSB400 (clock=100MHz)
This is an example of one: SSPEC=SL6YV order=BX80532KC2400D

http://processorfinder.intel.com/scripts/details.asp?sSpec=SL6YV

OK. A possible explanation, is if the CPU clock was set to 66Mhz,
when it should be 100MHz. That would give the right ratio of
results to expected value.

If I look in the manual, I see nothing to set the clock!

http://www.supermicro.com/manuals/motherboard/860/MNL-0636.pdf

OK, that leaves one other possibility. I could be mistaken,
but I think the multiplier in the processor, on a Xeon,
can be turned down. Your processor should be 100 x 24 = 2400MHz.
Maybe the multiplier in the BIOS is set to 16 instead of 24 ?
See PDF page 72 of the above manual, and the setting labelled
"CPU Clock Ratio", on how to set the multiplier.

Either the processor input clock (which is one quarter of the
FSB) or the multiplier, is not what it is supposed to be. On
many desktop motherboards, the only adjustment is the clock.
On ye olde Xeon, the multiplier could be an option as well.

Paul
 
Out of curiosity ran Everest.
CPU report says that the CPU's are running at 1587 MHz. They are nominal 2.4
GHz Xeon's. Is this normal or is something wrong? If wrong, is there a way
to adjust?
I'm not otherwise having any problems, just wondering if I'm missing out on
performance.
Thanks for any info,
Click to expand...


I suspect your FSB is set to 100MHz when it should be set to
200MHz (800 pseudo-MHz in Intelspeak).

Run another util. for confirmation of reported speed like
"CPU-Z"
 
kony said:
I suspect your FSB is set to 100MHz when it should be set to
200MHz (800 pseudo-MHz in Intelspeak).

Run another util. for confirmation of reported speed like
"CPU-Z"
Click to expand...

Thanks Kony and Paul.
Way more information than I can absorb tonight.
I'll revisit it tomorrow when I'm a bit more alert.
BF
 
Thanks Kony and Paul.
Way more information than I can absorb tonight.
I'll revisit it tomorrow when I'm a bit more alert.
Click to expand...


Well, I seem to be numerically challenged today, what I
/meant/ to write was that it seemed your FSB was set to
133MHz when it should be set to 200MHz.
 
kony said:
Well, I seem to be numerically challenged today, what I
/meant/ to write was that it seemed your FSB was set to
133MHz when it should be set to 200MHz.
Click to expand...

Kony,
Again from Everest, seems you may be correct, see below.
Can the clock speed be changed, maybe a jumper? Of course I could go to
Supermicro web site and look around, but asking you is much easier.

Front Side Bus Properties:
Bus Type Intel NetBurst
Bus Width 64-bit
Real Clock 99 MHz (QDR)
Effective Clock 397 MHz
Bandwidth 3174 MB/s

Thanks,
BF
 
"BF" said:
Kony,
Again from Everest, seems you may be correct, see below.
Can the clock speed be changed, maybe a jumper? Of course I could go to
Supermicro web site and look around, but asking you is much easier.

Front Side Bus Properties:
Bus Type Intel NetBurst
Bus Width 64-bit
Real Clock 99 MHz (QDR)
Effective Clock 397 MHz
Bandwidth 3174 MB/s

Thanks,
BF
Click to expand...

Your clock should be 100MHz. The QDR (quad data rate) means the
effective clock would be FSB400, the rating of all the 2.4GHz
socket 603 Xeons. In other words, within measurement error, your
clock is perfect and does not need adjustment.

You need to verify the multiplier of the Xeon. I suspect it is set
at x16 and not the x24 it should be. Use your utilities to see the
value of the multiplier, and if it is incorrect, fix it in
the BIOS.

Paul
 
Your clock should be 100MHz. The QDR (quad data rate) means the
effective clock would be FSB400, the rating of all the 2.4GHz
socket 603 Xeons. In other words, within measurement error, your
clock is perfect and does not need adjustment.

You need to verify the multiplier of the Xeon. I suspect it is set
at x16 and not the x24 it should be. Use your utilities to see the
value of the multiplier, and if it is incorrect, fix it in
the BIOS.

Paul
Click to expand...

Thanks Paul, that did the trick. Now the real question, will it make any
real difference to performance? The multiplier had been set at 16 with 16
min / 24 max. It's now set at 24 with 16 min / 24 max.
Most of my performance needs are 3-D solid modeling and a bit of FEA.
BF
 
"BF" said:
Thanks Paul, that did the trick. Now the real question, will it make any
real difference to performance? The multiplier had been set at 16 with 16
min / 24 max. It's now set at 24 with 16 min / 24 max.
Most of my performance needs are 3-D solid modeling and a bit of FEA.
BF
Click to expand...

For problems which are compute bound, the machine will now
be 50% faster. If there is a significant memory component
(a cache buster - all data coming from main memory), then
it will be no faster. It will just "wait" faster, if the problems
you work on are memory bound.

I expect you'll get the most meaningful results, when you
benchmark the applications you regularly use. Sometimes
what a faster processor buys you, is faster execution of
a batch processing job, but the desktop might not feel
much different than it did before. If you have a dual
Xeon, your desktop experience was probably already
"buttery smooth", so the improvement at the desktop level
might be inperceptable.

HTH,
Paul
 
Back
Top