Discussion:
Sysbench 0.9.5d overflow?
(too old to reply)
Dariusz Piatkowski
2011-01-23 05:01:37 UTC
Permalink
Raw Message
Hey guys!

I recently upgraded to some new hardware and managed to get a 4 core Phenom II
X2 555 based system up and running.

Still tuning it and I am using Sysbench as a measuring tool

I ran into what I believe may be an 'overflow' error. The "CPU Integer" test
appears to report max values of 17624.800. I get this value for both the
Heapsort & Sieve tests...Hanoi seems to be OK, clocking in at 17946.000, but
Dhrystone appears to have rolled-over and is currently posting 5251.200.

Here are some details:

1) Phenom2 X2 555 - 4 core unlocked, 3.2 Gig (Baseline configuration)
CPU integer
==========
Dhrystone - 25160.300
Hanoi - 14837.00
Heapsort - 17624.800
Sieve - 17624.800
TOTAL = 20262.283

CPU float
========
Linpack - 1834.240
Flops - 2293.100
FFT - 1783.400
TOTAL = 999.467

2) Phenom2 X2 555 - 4 core unlocked, overclocked to 4.0 Gig
CPU integer
==========
Dhrystone - 5251.200
Hanoi - 17946.00
Heapsort - 17624.800
Sieve - 17624.800
TOTAL = 12560.069

CPU float
========
Linpack - 2290.670
Flops - 2863.900
FFT - 1998.850
TOTAL = 1195.491

So..as you can see the Integer stuff is going funny...the float keeps on moving
up as I expected it.

Is anyone maintaining this code at all? Is the source available?
Trevor Hemsley
2011-01-23 20:55:26 UTC
Permalink
Raw Message
On Sun, 23 Jan 2011 05:01:37 UTC in comp.os.os2.bugs, "Dariusz Piatkowski"
Post by Dariusz Piatkowski
Hey guys!
I recently upgraded to some new hardware and managed to get a 4 core Phenom II
X2 555 based system up and running.
Still tuning it and I am using Sysbench as a measuring tool
I ran into what I believe may be an 'overflow' error. The "CPU Integer" test
appears to report max values of 17624.800. I get this value for both the
Heapsort & Sieve tests...Hanoi seems to be OK, clocking in at 17946.000, but
Dhrystone appears to have rolled-over and is currently posting 5251.200.
Is anyone maintaining this code at all? Is the source available?
I've sent debug versions of the CPU int code to Darius and he's sent me back the
results. I'll take a look at them and try to work out what is going on here.
--
Trevor Hemsley, Brighton, UK
Trevor dot Hemsley at ntlworld dot com
Dave Yeo
2011-01-23 21:16:30 UTC
Permalink
Raw Message
Post by Trevor Hemsley
I've sent debug versions of the CPU int code to Darius and he's sent me back the
results. I'll take a look at them and try to work out what is going on here.
While on the subject, benchmarking a 500 GB drive consistently produces
this here,

12-22-2010 20:15:25 SYS3175 PID af78 TID 0003 Slot 00f5
F:\UTILS\SYSBENCH\SYSBENCH.EXE
c0000005
00013b88
P1=00000001 P2=0000fc10 P3=XXXXXXXX P4=XXXXXXXX
EAX=0000fc10 EBX=0000fc10 ECX=005bfc73 EDX=00000000
ESI=00000000 EDI=ffffffff
DS=0053 DSACC=f0f3 DSLIM=ffffffff
ES=0053 ESACC=f0f3 ESLIM=ffffffff
FS=150b FSACC=00f3 FSLIM=00000030
GS=0000 GSACC=**** GSLIM=********
CS:EIP=005b:00013b88 CSACC=f0df CSLIM=ffffffff
SS:ESP=0053:005bfa5c SSACC=f0f3 SSLIM=ffffffff
EBP=005bfae0 FLG=00012286

SYSBENCH.EXE 0001:00003b88

Dave
Trevor Hemsley
2011-01-24 01:21:24 UTC
Permalink
Raw Message
On Sun, 23 Jan 2011 21:16:30 UTC in comp.os.os2.bugs, Dave Yeo
Post by Dave Yeo
While on the subject, benchmarking a 500 GB drive consistently produces
this here,
Which test exactly is that in? Unfortunately I don't seem to have a map file
from the old release so can't translate it back.
--
Trevor Hemsley, Brighton, UK
Trevor dot Hemsley at ntlworld dot com
Dave Yeo
2011-01-24 02:45:13 UTC
Permalink
Raw Message
Post by Trevor Hemsley
On Sun, 23 Jan 2011 21:16:30 UTC in comp.os.os2.bugs, Dave Yeo
Post by Dave Yeo
While on the subject, benchmarking a 500 GB drive consistently produces
this here,
Which test exactly is that in? Unfortunately I don't seem to have a map file
from the old release so can't translate it back.
Average Transfer Rate, Last track bwds.
Is there a newer release?
Dave

Loading...