Re: 2 servers, identical hardware, different speed

Hi Mikael,

Mikael Pettersson wrote:
In article <1157944792.652006.252050@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
OtisUsenet <otis_usenet@xxxxxxxxx> wrote:
Hi,

I have 2 _nearly_ identical servers that, after benchmarking (and
kernel compiles) turned out to perform VERY differently (one is 4 or
more times slower than the other one).
Both machines are 3GHz P4s, single CPU, but with Hyperthreading, 2 SATA
disks in RAID 1.
The slower machine actually has more RAM (4GB) than the faster one
(1GB).

I can't think of anything that would cause 4x+ difference in
performance of two nearly identical systems. Does anyone have any
ideas where to look for clues? Anything in the BIOS perhaps?

Check that all usable memory listed in the BIOS-e820 map
[run dmesg] is covered by write-back /proc/mtrr entries.

If I understand you correctly, then I think this is not the cause of
the performance issues in this case. This is a box with 4GB of RAM,
and its /proc/mtrr looks like this:

reg00: base=0x00000000 ( 0MB), size=2048MB: write-back, count=1
reg01: base=0x80000000 (2048MB), size=1024MB: write-back, count=1
reg02: base=0xc0000000 (3072MB), size= 512MB: write-back, count=1
reg03: base=0xe0000000 (3584MB), size= 256MB: write-back, count=1
reg04: base=0xf0000000 (3840MB), size= 128MB: write-back, count=1
reg05: base=0xf8000000 (3968MB), size= 32MB: write-back, count=1

That looks ok, right?

It's not uncommon for BIOSen to make a mess of this and
report some memory as being usable but failing to also
make it cacheable, leading to serious performance losses.

The benchmarks I run didn't really test memory and were not memory
intensive. From what I can tell, it looks like they test CPU
performance (various computation-intensive benchmarks like sqrts,
recursions, etc.), system call overhead, process creation, file-system
throughput... but nothing very memory intensive.

Thanks.

Is this is what's happening, then possible remedies include:
a) reduce the amount of physical RAM
b) boot with mem= to prevent the kernel from using the
uncached areas
c) a BIOS upgrade
--
Mikael Pettersson (mikpe@xxxxxxxx)
Computing Science Department, Uppsala University

.

Relevant Pages

  • Re: forth and virtual memory
    ... too, maybe even the same order, so ordering the blocks by allocation ... on systems with too little memory ... What Java is known for, and what it actually does, are two distinct ... My measurements indicate that some of the benchmarks (from SpecJVM98, ...
    (comp.lang.forth)
  • Re: [PATCH 0/11] Avoiding fragmentation with subzone groupings v26
    ... a standard kernel was getting < 1% of memory as large ... I don't have a list of real workloads that break anti-frag yet so so I want to get anti-frag out there and see does it help people who really care about hugepages or not. ... Small differences in aim9 seem to make very little difference to other benchmarks like kbuild. ... echo Attempting load of hugetlbfs module ...
    (Linux-Kernel)
  • Re: MM B&V Benchmark weights
    ... Then some benchmarks need to be fixed to bring out the intended scoring bias ... For an application that never uses more than for example 64MB, memory usage ...
    (borland.public.delphi.language.basm)
  • Re: MM managers and win98 compatibility
    ... The following are my test results under Windows 98 SE. ... should give a good starting point for the authors of the different memory ... The following memory managers didn't pass all validations: ... Now, regarding benchmarks run under Windows 98 SE, the following memory ...
    (borland.public.delphi.language.basm)
  • Re: sig 11 - repeatedly - but only in 1 program.
    ... >> It has also been known to be caused by memory faults, ... >> victim of that is kernel compiles, which you say work ok for you. ... > I've managed to make substandard machines freeze in half a minute by ... > starting a GCC bootstrap. ...
    (comp.os.linux.development.system)
Loading