On 11/9/2011 9:31 AM, Hermann Himmelbauer wrote:
> Am 09.11.2011 17:08, schrieb Kyle Brantley:
>> On 11/9/2011 8:10 AM, Lonnie Abelbeck wrote:
>>> Hi Hermann,
>>>
>>> I am getting better results than you are.
>>>
>>> I have a the 1MB/1GHz net6501-50, comBIOS 1.40h, and using it as a router...
>>>
>>> [ iperf -s ] --- [eth0/WAN # net6501-50 # eth2/LAN] --- [ iperf -c ... ]
>>>
>>> I get 515 Mbits/sec with the default iperf settings. 'top' is
>>> non-responsive during the test on the console.
>>>
>>> I'm using linux kernel 2.6.35 with the latest e1000e driver from
>>> sourceforge, no-SMP and iptables (IPv4 NAT/IPv6) firewall active.
>>> --
>> A quick test against a device on the same switch as my 6501-70, ran for
>> two minutes:
>>
>> [ ID] Interval Transfer Bandwidth
>> [ 3] 0.0-120.0 sec 11.1 GBytes 797 Mbits/sec
>>
>> Linux localhost.localdomain 2.6.38.6-26.rc1.fc15.x86_64 #1 SMP Mon May 9
>> 20:45:15 UTC 2011 x86_64 x86_64 x86_64 GNU/Linux
>>
>> Not quite full wire speed, but acceptable. It also consumed the 'entire'
>> CPU (as per top, 100% dedicated to software interrupts), leaving the HT
>> core mostly available.
> Thanks a lot for your reply. It seems that there's something wrong with
> my setup, maybe it's the kernel? What bios version do you have? Do you
> have HT enabled?
>
> I would be very much interested in comparing the CPU speed, can you
> perhaps download the following and run "nbench" (already precompiled for
> the 6501)?
>
> http://violin.qwer.tk/~dusty/temp/nbench.tgz
>
> My results are as follows:
>
> ----------------- snip ---------------------
> BYTEmark* Native Mode Benchmark ver. 2 (10/95)
> Index-split by Andrew D. Balsa (11/97)
> Linux/Unix* port by Uwe F. Mayer (12/96,11/97)
>
> TEST : Iterations/sec. : Old Index : New Index
> : : Pentium 90* : AMD K6/233*
> --------------------:------------------:-------------:------------
> NUMERIC SORT : 174.02 : 4.46 : 1.47
> STRING SORT : 23.077 : 10.31 : 1.60
> BITFIELD : 6.8908e+07 : 11.82 : 2.47
> FP EMULATION : 23.429 : 11.24 : 2.59
> FOURIER : 2255.1 : 2.56 : 1.44
> ASSIGNMENT : 3.3704 : 12.82 : 3.33
> IDEA : 646.05 : 9.88 : 2.93
> HUFFMAN : 284.09 : 7.88 : 2.52
> NEURAL NET : 2.1586 : 3.47 : 1.46
> LU DECOMPOSITION : 120.12 : 6.22 : 4.49
> ==========================ORIGINAL BYTEMARK
> RESULTS==========================
> INTEGER INDEX : 9.319
> FLOATING-POINT INDEX: 3.811
> Baseline (MSDOS*) : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
> ==============================LINUX DATA
> BELOW===============================
> CPU : GenuineIntel Genuine Intel(R) CPU @ 1.60GHz
> 1600MHz
> L2 Cache : 512 KB
> OS : Linux 2.6.32-5-686
> C compiler : gcc version 4.4.5 (Debian 4.4.5-8)
> libc : libc-2.11.2.so
> MEMORY INDEX : 2.358
> INTEGER INDEX : 2.302
> FLOATING-POINT INDEX: 2.114
> Baseline (LINUX) : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
> ----------------- snip ---------------------
>
> My /proc/interrupts look a bit different:
>
> -----------------snip ----------------------
> CPU0
> 0: 65 XT-PIC-XT timer
> 2: 0 XT-PIC-XT cascade
> 4: 621 XT-PIC-XT serial
> 7: 13 XT-PIC-XT
> 8: 0 XT-PIC-XT rtc0
> 9: 6 XT-PIC-XT ehci_hcd:usb1, ohci_hcd:usb3,
> ohci_hcd:usb4, ohci_hcd:usb5, serial
> 10: 0 XT-PIC-XT ehci_hcd:usb2, ohci_hcd:usb6,
> ohci_hcd:usb7, ohci_hcd:usb8
> 22: 7585 PCI-MSI-edge eth0-rx-0
> 23: 0 PCI-MSI-edge eth0-tx-0
> 24: 2 PCI-MSI-edge eth0
> 25: 21734 PCI-MSI-edge ahci
> 26: 190244 PCI-MSI-edge eth1-rx-0
> 27: 121159 PCI-MSI-edge eth1-tx-0
> 28: 4 PCI-MSI-edge eth1
> NMI: 2508751 Non-maskable interrupts
> LOC: 3794295 Local timer interrupts
> SPU: 0 Spurious interrupts
> PMI: 2508751 Performance monitoring interrupts
> PND: 2507386 Performance pending work
> RES: 0 Rescheduling interrupts
> CAL: 0 Function call interrupts
> TLB: 0 TLB shootdowns
> TRM: 0 Thermal event interrupts
> THR: 0 Threshold APIC interrupts
> MCE: 0 Machine check exceptions
> MCP: 52 Machine check polls
> ERR: 13
> MIS: 0
> -----------------snip ----------------------
>
> I wonder why I have so much NMI, PMI and PND interrupts?
>
> Many thanks for help!
>
> Best Regards,
> Hermann
>
>
TEST : Iterations/sec. : Old Index : New Index
: : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT : 168.16 : 4.31 : 1.42
STRING SORT : 17.53 : 7.83 : 1.21
BITFIELD : 5.5322e+07 : 9.49 : 1.98
FP EMULATION : 24.93 : 11.96 : 2.76
FOURIER : 2437.4 : 2.77 : 1.56
ASSIGNMENT : 3.2209 : 12.26 : 3.18
IDEA : 618.52 : 9.46 : 2.81
HUFFMAN : 283.77 : 7.87 : 2.51
NEURAL NET : 2.3631 : 3.80 : 1.60
LU DECOMPOSITION : 135.23 : 7.01 : 5.06
==========================ORIGINAL BYTEMARK
RESULTS==========================
INTEGER INDEX : 8.607
FLOATING-POINT INDEX: 4.193
Baseline (MSDOS*) : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA
BELOW===============================
CPU : Dual GenuineIntel Genuine Intel(R) CPU @
1.60GHz 1600MHz
L2 Cache : 512 KB
OS : Linux 2.6.38.6-26.rc1.fc15.x86_64
C compiler : gcc version 4.4.5 (Debian 4.4.5-8)
libc : libc-2.11.2.so
MEMORY INDEX : 1.969
INTEGER INDEX : 2.292
FLOATING-POINT INDEX: 2.326
Baseline (LINUX) : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.
Numbers are more or less the same. Some are higher, some are lower, they
average out okay I think. I'm sure that some of the oddities resulted
from me using the static version you provided (from Debian) on my
soekris (running Fedora), but those oddities are nearly non-existant.
My /proc/interrupts:
CPU0 CPU1
0: 954868031 0 IO-APIC-edge timer
2: 0 0 XT-PIC-XT-PIC cascade
4: 2780 0 IO-APIC-edge serial
8: 20 0 IO-APIC-edge rtc0
16: 0 0 IO-APIC-fasteoi PCIe PME, ehci_hcd:usb2,
ohci_hcd:usb6, ohci_hcd:usb7, ohci_hcd:usb8
17: 0 0 IO-APIC-fasteoi PCIe PME
18: 0 0 IO-APIC-fasteoi PCIe PME, mmc0, mmc1
19: 36 0 IO-APIC-fasteoi PCIe PME, ehci_hcd:usb1,
ohci_hcd:usb3, ohci_hcd:usb4, ohci_hcd:usb5, ath9k
40: 26231 0 PCI-MSI-edge ahci
44: 2648790 0 PCI-MSI-edge p1p2-rx-0
45: 645679 0 PCI-MSI-edge p1p2-tx-0
46: 251779 0 PCI-MSI-edge p1p2
NMI: 268106 267987 Non-maskable interrupts
LOC: 954877380 954869765 Local timer interrupts
SPU: 0 0 Spurious interrupts
PMI: 268106 267987 Performance monitoring interrupts
IWI: 0 0 IRQ work interrupts
RES: 8345 3662 Rescheduling interrupts
CAL: 44 36 Function call interrupts
TLB: 2689 2117 TLB shootdowns
TRM: 0 0 Thermal event interrupts
THR: 0 0 Threshold APIC interrupts
MCE: 0 0 Machine check exceptions
MCP: 3183 3184 Machine check polls
ERR: 1
MIS: 0
I'm running a stock Fedora kernel, and the 1.40h BIOS with HT enabled. I
should note that I ran a few iperf tests alone, and then re-ran them
with top (refresh rate @ 0.5s) running alongside in another ssh session.
This resulted in a notable drop in performance -- from 821Mbit to
569Mbit just now.
Additionally, and I suspect that this is the real source of your iperf
issue, is that iperf seems to perform notably better the longer you test
it for.
Time Speed (/s)
------------------
5s 335
10s 339
20s 395
30s 468
40s 531
50s 672
60s 527
70s 605
80s 581
90s 558
100s 569
110s 765
120s 821
--Kyle
_______________________________________________
Soekris-tech mailing list
[email protected]
http://lists.soekris.com/mailman/listinfo/soekris-tech
