February 2005 - openSUSE AMD64 - openSUSE Mailing Lists

Re: Re: Re: [suse-amd64] g++ 3.3.3 -m64 -O3
by Arne 06 Feb '05

06 Feb '05

On 6.2 2005 21:13, Eugene Chu wrote: > >Hi, > >You wrote: > >>Btw it's a big difference: 1.76 vs 9.81 for -O3 and -O0. > >What host are you running on? I have an Opteron 242 (1.6GHz) system, >and the best I could get with that code was 2.74 using -O. > >thanks, >eyc > Hi I am running a AMD Athlon64 2 GHz (3200) with 1 GB RAM (2x512 DDR400 with 2 3 2 5 latency if that matters) Welcome /Arne ------------------------------------------------------------------ Gratis svenskpråkig e-post och förmånliga textmeddelanden http://webmail.surfnet.fi ------------------------------------------------------------------

1 0

Re: [suse-amd64] g++ 3.3.3 -m64 -O3
by Arne 06 Feb '05

06 Feb '05

On 6.2 2005 16:12, insane(a)fmi.uni-sofia.bg wrote: > >I tried to compile the following code with g++ 3.3.3 (g++ -m64 -O3): > >#include <vector> >#include <algorithm> >#include <ctime> >#include <cstdio> > >using namespace std; > >#define BENCH_TYPE double > >int main() >{ > vector<BENCH_TYPE> v; > printf("Generating 10M elements\n"); > for (int i = 0; i < 10000000; i++) > v.push_back((BENCH_TYPE)(rand()*0.5 + rand())); > > printf("generated\nSorting...\n"); > > clock_t start = clock(); > sort(v.begin(), v.end()); > clock_t end = clock(); > > printf("%f\n", (end - start)/((double)CLOCKS_PER_SEC)); > return 0; >} > >The resulting executable doesn't finish in a minute or so and I have to break >its execution. The funny thing is that the same code compiled with g++ -m64 >-O0 finishes in 8 sec or so. What's going on? Any ideas? >I have Suse 9.1 x86_64 with the latest updates of all installed packages. > >I wouldn't bother a lot about this test code, if it wasn't for a more serious >problem: I downloaded and installed the latest x86_64 build of the kernel >2.6.5-7.145-default and the nVidia graphics driver stopped working. I >manually downloaded the latest driver from the nVidia web site: >NVIDIA-Linux-x86_64-1.0-6629-pkg2.run and installed it and when I started X >it was painfully slow and I had to kill it. > >I don't know if the two problems are related. I would appreciate any help on >any of them. Thanks a lot. > >-- >Check the List-Unsubscribe header to unsubscribe >For additional commands, email: suse-amd64-help(a)suse.com > > Maybe not so much help but NVIDIA-Linux-x86_64-1.0-6629-pkg2.run and your test code works fine on SuSE64 9.2, k2.6.8-24.11-default. Can't help you more, good luck! Btw it's a big difference: 1.76 vs 9.81 for -O3 and -O0. ------------------------------------------------------------------ Gratis svenskpråkig e-post och förmånliga textmeddelanden http://webmail.surfnet.fi ------------------------------------------------------------------

1 0

How to turn off disk caching?
by Ed Sherman 06 Feb '05

06 Feb '05

We've recently started running Oracle 10g on SLES9 We were wondering why we are going into swap with 16G of ram and not much running on the machine except for Oracle. Some things we found are: > Subject: Re : Why does system show all 16 GB used and swap also in use > > Paul; > Turn off disk cacheing, make sure that you are using async I/O. > > HTH > Peter > Subject: Re : Why does system show all 16 GB used and swap also in use > > You're probably observing two phenomena here: > > 1. Linux will use as much memory as it can get for "disk > caching" (filesystem buffers, really). You can reduce > this dramatically using the BufferMem parameter, and > by using async I/O. (As already noted...) > 2. In most unixen, shared memory segments get counted multiple > times. The executable code for each process is loaded into > shared (sharable) memory, and the SGA is -- of course -- > shared also. The memory consumed for the "oracle" executable > and for the SGA are commonly reported for *all* processes > that are using them, even though each is present in memory > only once. I search the Internet and find that the buffermem parameter is supposed to be in /proc/sys/vm but I also heard that some extensive work has been done on the virtual memory system in kernel 2.6. I don't have a buffermem parameter to set. Does anyone know how to adjust/turn-off disk caching in SLES9 with the 2.6 kernel? Thanks, Ed

2 1

g++ 3.3.3 -m64 -O3
by insane＠fmi.uni-sofia.bg 06 Feb '05

06 Feb '05

I tried to compile the following code with g++ 3.3.3 (g++ -m64 -O3): #include <vector> #include <algorithm> #include <ctime> #include <cstdio> using namespace std; #define BENCH_TYPE double int main() { vector<BENCH_TYPE> v; printf("Generating 10M elements\n"); for (int i = 0; i < 10000000; i++) v.push_back((BENCH_TYPE)(rand()*0.5 + rand())); printf("generated\nSorting...\n"); clock_t start = clock(); sort(v.begin(), v.end()); clock_t end = clock(); printf("%f\n", (end - start)/((double)CLOCKS_PER_SEC)); return 0; } The resulting executable doesn't finish in a minute or so and I have to break its execution. The funny thing is that the same code compiled with g++ -m64 -O0 finishes in 8 sec or so. What's going on? Any ideas? I have Suse 9.1 x86_64 with the latest updates of all installed packages. I wouldn't bother a lot about this test code, if it wasn't for a more serious problem: I downloaded and installed the latest x86_64 build of the kernel 2.6.5-7.145-default and the nVidia graphics driver stopped working. I manually downloaded the latest driver from the nVidia web site: NVIDIA-Linux-x86_64-1.0-6629-pkg2.run and installed it and when I started X it was painfully slow and I had to kill it. I don't know if the two problems are related. I would appreciate any help on any of them. Thanks a lot.

1 0

Problem after kernel upgrade
by Stacey Creamer 05 Feb '05

05 Feb '05

I did a kernel upgrade last night, using YOU and just accepting all of the suggested updates. Because the kernel fixes required a reboot, I did so, but now it does not come up to a login screen. I tried rebooting again and this time using fail-safe. The last line processed says ReiserFS: hde2: replayed 22 transactions in 7 seconds. Then it hangs with a rapidly blinking cursor. I have very little Linux experience. Can someone help me fix this? Thanks!

2 2

Need help finding a supported SCSI card.
by Black, Alain 05 Feb '05

05 Feb '05

Greetings all, I've got an Dual Opteron SLES 8 SP3 installation to which I'm trying to add a tape library. The tape library (2 drive SDLT and one robot) is LVD VHDCI. I found the Adaptec AHA39160 which has 2 VHDCI but is listed as "problematic" in the SuSE HardWare Database. Adaptec makes an AHA29160-LP which has an external VHDCI connector, but it is not listed in the HWDB, however the AHA29160 is listed and supported, but it has different connectors, are those close enough to believe that the AHA29160-LP is also supported or would that be pushing my luck? What I'm looking for is a PCI/PCI-X SCSI card preferably with 2 external LVD VHDCI connectors, I have room for 2 cards if need be, that is fully supported. I also found an LSI Logic card which would meet my needs, but am either using the wrong vendor name or they are not listed in the HWDB. -Alain.

6 5

Xchm compile help
by Wayne Brennan 05 Feb '05

05 Feb '05

I am having trouble compiling 'chmlib', ./configure doesn't return errors but a make returns many errors, including: src/chm_lib.c:174:2: #error "Please define the sized types for your platform in chm_lib.c" src/chm_lib.c:229: error: parse error before "Int16" src/chm_lib.c: In function `_unmarshal_int16': src/chm_lib.c:231: error: `pLenRemain' undeclared (first use in this function) ... I have seen a "compile with -m32" for 32 bit apps mentioned before, but I'm not sure how to fit this into any of my configure or make instructions.... I would also like to know how I can install a 32bit version of Firefox (to run flash plugin)... thanks for any help, Wayne

2 3

RE: [suse-amd64] Need help finding a supported SCSI card.
by Black, Alain 04 Feb '05

04 Feb '05

Thank you Kees, I was able to find the LSI Logic LSI22320-R which states in it's documentation that it supports SLES 8. I will go a head and order this card. -Alain -----Original Message----- From: Kees Hoekzema [mailto:kees@tweakers.net] Sent: Friday, February 04, 2005 11:32 AM To: suse-amd64(a)suse.com Cc: Black, Alain Subject: Re: [suse-amd64] Need help finding a supported SCSI card. On Friday 04 February 2005 19:02, Black, Alain wrote: > I also found an LSI Logic card which would meet my needs, but am either > using the wrong vendor name or they are not listed in the HWDB. The LSI Logic cards I am using with suse all work fine. They are probably listed under 'megaraid' or something similar. -kees

1 0

numbercruncher system configuration
by jens molle 03 Feb '05

03 Feb '05

Hi List, i want to build a new numbercruncher for use with ansys software ( non-linear finite elements computation), amd 64 preferred. For the best price/performance ratio i think of a single processor machine ( ansys does scale poorly with nonlinear problems anyway) with 4 GB RAM. What systems can you recommend , are there any experiences with the new ASUS Board A8N-SLI ? thanks in advance --------------------------------- Gesendet von Yahoo! Mail - Jetzt mit 250MB kostenlosem Speicher

1 0

SATA, K8V Deluxe MB and SUSE 9.2
by Mitch Martinez 03 Feb '05

03 Feb '05

Has any one had any luck in getting a ASUS K8V Deluxe MB and SUSE 9.2 to see an IDE drive on the built in SATA bus? I've been told that it's possible but I haven't been able to see one of my drives on the bus, I've looked for a scsi device (no luck), Here's my Sysinfo: MB: ASUS K8V Deluxe CPU: AMD Athlon(tm) 64 Processor 3200+ MEM: 1540984 kB VID: G-Force FX5900 128 MB MON: ViewSonic P810 OS: SUSE 9.2 Pro CPU:

3 5

Re: [suse-amd64] SUSE 9.2 FTP distribution and ISOs
by Andreas Jaeger 02 Feb '05

02 Feb '05

"Mikhail Kuzminsky" <kus(a)free.net> writes: > In message from Andreas Jaeger <aj(a)suse.de> (Tue, 25 Jan 2005 09:01:14 > +0100): >> >>I'm not sure whether you have read the following, so I want to >>announce it now: >>http://www.novell.com/products/linuxprofessional/downloads/suse_linux/index… >> >>We have released our FTP installation for both x86-64 (AMD64, EM64T) >>and x86 - and also an 3.5 GB installable ISO image that supports both >>x86-64 and x86. > Sorry, does this iso image correspond to special evaluation version > or personal version of SuSE Linux? More exactly, the question is about > differences w/SuSE 9.2 Professional. Is the difference (between this > iso image and Professional version) only in a set of applications > binary rpms (sendmail, HPC applications etc) ? The 3.5 GB ISO contains less than half the packages that the 9.2 Professional DVD contains (which is 8.5 GB). It supports both 32-bit x86 and 64-bit x86-64 platforms. A major difference to the 9.2 DVD is that development packages are completely missing. The ftp installation contains all the packages that are also on the 9.2 Professional DVD, Andreas -- Andreas Jaeger, aj(a)suse.de, http://www.suse.de/~aj SUSE Linux Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany GPG fingerprint = 93A3 365E CE47 B889 DF7F FED1 389A 563C C272 A126

1 0

Kernel BUG at direct_io:544
by Ahmon Dancy 02 Feb '05

02 Feb '05

Has anyone seem this problem before? Kernel BUG at direct_io:544 invalid operand: 0000 [1] SMP CPU 1 Modules linked in: usbserial parport_pc lp parport edd freq_table thermal processor fan button battery ac autofs4 raw ipv6 evdev joydev sg st sr_mod ext3 jbd e100 mii tg3 ohci_hcd hw_random usbcore dm_snapshot dm_mod reiserfs 3w_9xxx sd_mod scsi_mod Pid: 6359, comm: oracle Tainted: MG (2.6.8-24.10-smp SL92_BRANCH-200412221154270000) RIP: 0010:[<ffffffff8019fd55>] <ffffffff8019fd55>{dio_new_bio+197} RSP: 0018:00000100f1daf9a8 EFLAGS: 00010246 RAX: 0000000000000020 RBX: 0000000000000000 RCX: 000001007f6bab30 RDX: 0000000000000048 RSI: 0000000000000000 RDI: 00000100f8833380 RBP: 00000100f9f96400 R08: 0000000000000000 R09: 00000000000000f9 R10: 00000000094789d5 R11: 0000000000000001 R12: 0000000000045889 R13: 0000000000000000 R14: 0000000000045889 R15: 0000000000000009 FS: 0000002a97336e40(0000) GS:ffffffff804e2e80(0000) knlGS:000000005568e460 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 00000000007ebe50 CR3: 00000000f9f82000 CR4: 00000000000006e0 Process oracle (pid: 6359, threadinfo 00000100f1dae000, task 00000100eac2c030) Stack: 00000100f9f96400 0000000000000000 0000000000000001 0000000000000000 0000000000000000 ffffffff8019ff82 0000000000000007 0000000000000000 00000100f9f96400 ffffffff801a0cb5 Call Trace:<ffffffff8019ff82>{dio_send_cur_page+114} <ffffffff801a0cb5>{__blockdev_direct_IO+2165} <ffffffffa003aa8c>{:reiserfs:reiserfs_direct_IO+76} <ffffffffa003ec40>{:reiserfs:reiserfs_get_blocks_direct_io+0} <ffffffff8015ac34>{generic_file_direct_IO+100} <ffffffff8015acfb>{generic_file_direct_write+123} <ffffffff8015b959>{generic_file_aio_write_nolock+713} <ffffffff8015ba9e>{generic_file_write_nolock+158} <ffffffffa005955f>{:reiserfs:do_journal_end+3199} <ffffffff80137a80>{autoremove_wake_function+0} <ffffffff8015bcf5>{generic_file_write+101} <ffffffffa0042548>{:reiserfs:reiserfs_file_write+488} <ffffffff8015a0d5>{filemap_nopage+405} <ffffffff801681fc>{do_no_page+1532} <ffffffff80169990>{handle_mm_fault+368} <ffffffff80362141>{__down_read+49} <ffffffff8021206d>{__up_read+29} <ffffffff801222e8>{do_page_fault+504} <ffffffff8017c455>{file_move+37} <ffffffff80179abe>{filp_open+62} <ffffffff8017bb93>{vfs_write+243} <ffffffff8017bcd3>{sys_write+83} <ffffffff8011074e>{system_call+126} Code: 0f 0b 13 d2 38 80 ff ff ff ff 20 02 bf d0 00 00 00 48 8b 9d RIP <ffffffff8019fd55>{dio_new_bio+197} RSP <00000100f1daf9a8>

1 0

relocation truncated to fit: R_X86_64_PC32
by orecchio＠unime.it 02 Feb '05

02 Feb '05

I have a AMD OPTERON 64bit with SUSE LINUX SERVER 9.1. I tried to compile a software (tomoDD) in fortran with some subroutine also in C and I obtained this error message tomoDD.o(.text+0x65e): In function `MAIN__': : relocation truncated to fit: R_X86_64_PC32 contrl_ tomoDD.o(.text+0x5083): In function `MAIN__': : additional relocation overflows omitted from the output collect2: ld returned 1 exit status make: *** [tomoDD] Error 1 Any help will be greatfully received. thanks, Barbara

1 0

Max Number of Sata Drives
by Joel Wiramu Pauling 01 Feb '05

01 Feb '05

Hi not strictly a suse amd64 question, but i'm hoping someone on this list will be able to answer. Is the number of sata devices (and cards) stackable. I.e I have an onboard controller which supports 2 channel, can I then go and add another 4 channel sata controller card and use them all at the same time (i.e the as a striped 0 array onboard, + raid array on the 4 port card). Is there a limit to this setup, can I have say 2 4 port cards with a total of 10 drives (raided or not). Kind regards Joel W

1 0