![]() |
|
|
#1 |
|
Oct 2004
112 Posts |
I have two identical machines running Zerahstar Linux (live CD)
Well, almost identical because one runs about 20% slower than the other. ps -a -F comfirms that all the CPU cycles goes to mprime. The CPU option shows the correct CPU speed for both machines - 3ghz P4. If I run option 18 (The benchmark) it confirms the slower machine. Any ideas how to figure out what's different? |
|
|
|
|
|
#2 |
|
"Mike"
Aug 2002
25×257 Posts |
Maybe run "dmesg" on both of them and then diff the results to see if there are any differences... Maybe run "lsmod" and verify the same modules are loaded...
|
|
|
|
|
|
#3 |
|
Bamboozled!
"πΊππ·π·π"
May 2003
Down not across
10,753 Posts |
Check the BIOS is identical, both version and configuration data. Memory timings can have a big effect if they are set particularly badly and mprime is very memory intensive.
Check the memory is indeed identical! Paul |
|
|
|
|
|
#4 |
|
"Nancy"
Aug 2002
Alexandria
2,467 Posts |
Do both cpus have the same core? I think the Prescott is slower than the Northwood at the same clock speed, and both cores are sold in 3GHz cpus. Try /proc/cpuinfo .
Alex |
|
|
|
|
|
#5 |
|
Mar 2003
New Zealand
13×89 Posts |
Another possibility is that one of them is overheating and the thermal throttle is engaging. 2.6 kernels can detect this and will report it in syslog if the compile time options was selected.
Do you have onboard video or a video card that shares main memory? They can slow down mprime quite a lot depending on bios settings and video mode used. |
|
|
|
|
|
#6 |
|
Oct 2004
3 Posts |
Thanks for the awesome feedback so far. Update:
First, I checked the CPUs using /proc/cpuinfo. Identical. Then I checked the BIOS settings. Speed settings identical. Then I checked CPU temp after running a while. OK on both. Then I checked memory - PC3200 on both. Then I ran dmesg, as Xyzzy suggested. Identical. But wait! There's more! I ran lsmod... Aha, a small difference! Below is the output from "diff 1 2" < snd_via82xx 24612 0 < snd_ac97_codec 62596 1 snd_via82xx --- > snd_intel8x0 31148 0 > snd_ac97_codec 62596 1 snd_intel8x0 Whoa! One mobo is Intel, one is VIA!!!!!! More detective work on the two mobos revealed this: One is an ECS 848P-A using the Intel 848P chipset. One is an ECS PT800CE-A using the ViA PT800 chipset. Guess which one's faster? Why? |
|
|
|
|
|
#7 | |
|
Mar 2003
New Zealand
13×89 Posts |
Quote:
|
|
|
|
|
|
|
#8 |
|
Oct 2004
3 Posts |
Yup, the Intel 848 board was faster than the PT800.
Situation Update: I returned the ECS PT800CE-A board to the store. I got the ABIT VT7 board, which ALSO uses the VIA PT800 chipset. The ABIT board runs just fine - around 30% faster than the ECS (VIA) and on par with the ECS (Intel) board. Sheesh - this took waaaayyyy too many hours to solve... Still not 100% sure why though... |
|
|
|
![]() |
| Thread Tools | |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Two identical tests? | CuriousKit | PrimeNet | 5 | 2015-04-10 01:10 |
| Identical PenD Dual-Core CPUs, one is 2.25x faster | NBtarheel_33 | Hardware | 5 | 2008-11-12 03:24 |
| Exponents with identical digits | GP2 | Marin's Mersenne-aries | 2 | 2007-11-19 19:03 |
| P-1: two stages, two machines? | James Heinrich | Marin's Mersenne-aries | 23 | 2005-04-05 18:31 |
| Machines | R.D. Silverman | NFSNET Discussion | 12 | 2004-07-02 12:12 |