[opensuse] Should I go 64-bit?
I believe my hardware supports 64-bit, but what software issues can I expect to run into by installing 11.1 64-bit? Driver problems? Java problems? Thank you, Adam Sailer -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Adam Sailer wrote:
I believe my hardware supports 64-bit, but what software issues can I expect to run into by installing 11.1 64-bit?
Driver problems? Java problems?
Thank you,
Adam Sailer
So far I haven't run into any with 11.1. I haven't used the 11.1 x86_64 box that much though. On 11.0, there are no issues at all with x86_64. I have 3 x86_64 boxes running right now. 2 are 10.3 and one is 11.0/11.1 dual boot. If you have and ATI card -- I would suggest waiting until the next ATI driver is released. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Adam Sailer wrote:
I believe my hardware supports 64-bit, but what software issues can I expect to run into by installing 11.1 64-bit?
Driver problems? Java problems?
The only problems I've had, is some OEMs like Epson who are too ignorant to provide 64-bit drivers for all hardware. SOME of their scanners are an example of this.....one reason I now recommend HP scanners and printers ONLY. Other than that, you probably won't have any problems but SHOULD have some performance improvements. Fred -- A government big enough to give you everything you want, is strong enough to take everything you have. -Thomas Jefferson -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, Jan 22, 2009 at 8:01 AM, Fred A. Miller <fmiller@lightlink.com> wrote:
Adam Sailer wrote:
I believe my hardware supports 64-bit, but what software issues can I expect to run into by installing 11.1 64-bit?
Driver problems? Java problems?
The only problems I've had, is some OEMs like Epson who are too ignorant to provide 64-bit drivers for all hardware. SOME of their scanners are an example of this.....one reason I now recommend HP scanners and printers ONLY. Other than that, you probably won't have any problems but SHOULD have some performance improvements.
Fred
-- A government big enough to give you everything you want, is strong enough to take everything you have. -Thomas Jefferson -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
I tried openSUSE 64 bit for the first time, and so far I havent come across any issues except for choosing between the 32 bit and 64 bit versions of software :) Cheers, Aditya -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Wednesday January 21 2009, Adam Sailer wrote:
I believe my hardware supports 64-bit, but what software issues can I expect to run into by installing 11.1 64-bit?
...
Adam Sailer
The question is, in fact, whether there's any reason to use a 64-bit OS. Do you have even one application that requires the address space afforded by the x86_64 architecture? If not, don't do it. If so, you don't really even have a choice. Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Wed, 21 Jan 2009, Randall R Schulz wrote:- <snip>
The question is, in fact, whether there's any reason to use a 64-bit OS. Do you have even one application that requires the address space afforded by the x86_64 architecture?
If not, don't do it.
I keep seeing this advice given, but have yet to see a real reason for it. Java and Flash work on my 10.3, 11.0 and 11.1 64bit systems, so that's not an issue. I can watch and listen to a variety of media, so it's not that. I haven't noticed any difference in speed between 32bit and 64bit systems on the same hardware. So, given that I don't have a problem yet you're saying not to do it, I'm really curious as to just what issue(s) you found while running a 64bit OS? Regards, David Bolt -- Team Acorn: http://www.distributed.net/ OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s | openSUSE 10.3 32b | openSUSE 11.0 32b | openSUSE 10.2 64b | openSUSE 10.3 64b | openSUSE 11.0 64b | openSUSE 11.1 64b TOS 4.02 | openSUSE 10.3 PPC | RISC OS 3.6 | RISC OS 3.11 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
The question is, in fact, whether there's any reason to use a 64-bit OS. Do you have even one application that requires the address space afforded by the x86_64 architecture?
If not, don't do it.
I keep seeing this advice given, but have yet to see a real reason for it. Java and Flash work on my 10.3, 11.0 and 11.1 64bit systems, so that's not an issue. I can watch and listen to a variety of media, so it's not that. I haven't noticed any difference in speed between 32bit and 64bit systems on the same hardware.
So, given that I don't have a problem yet you're saying not to do it, I'm really curious as to just what issue(s) you found while running a 64bit OS?
I was also one of the ones who said don´t do it. After following the recent discussions here, I decided to give it a try... and I have had zero problems. YAST is intelligent enough to handle sorting out the libraries as needed, and everything I´ve installed has simply worked. As for performance... I have noticed a few improvements here and there.. one in particular is that I am able to play online videos... eg YouTube full screen now with no lag or stuttering. The sole difference being I have a 64bit install now. I don´t know what the reason is behind the difference... but this one in particular is VERY visible. In my 32bit install any YouTube or similar video would play fine in its little confined window, but if I full screened it, the frame rate would drop to 1 or 2 fps. This happened even after recompiling my kernel for multimedia optimizations. After moving to 64bit, the exact same videos would play smooth at full screen even before recompiling my 64bitkernel. I cant speak for ATI since I avoid that video card as much as possible, but by nVidia drivers work perfectly as well. The nVidia binaries are 64bit aware, and automatically take care of 32bit mode compatibility. Gaming in Cedega or Wine is also fine... no issues at all. So... 64bit is a total success on my setup. I will be continuing to use it on all my computers going forward, and recommending it to the people I support/help out with Linux. C. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
David Bolt wrote:
On Wed, 21 Jan 2009, Randall R Schulz wrote:-
<snip>
The question is, in fact, whether there's any reason to use a 64-bit OS. Do you have even one application that requires the address space afforded by the x86_64 architecture?
If not, don't do it.
I keep seeing this advice given, but have yet to see a real reason for it. Java and Flash work on my 10.3, 11.0 and 11.1 64bit systems, so that's not an issue. I can watch and listen to a variety of media, so it's not that. I haven't noticed any difference in speed between 32bit and 64bit systems on the same hardware.
So, given that I don't have a problem yet you're saying not to do it, I'm really curious as to just what issue(s) you found while running a 64bit OS?
Editing LARGE pics. in Gimp will give you an idea of how much better a 64-bit release is over 32-it. There are other examples, but your logic is sound.....use it! Fred -- "Politicians and diapers need to be changed regularly -- and for the same reason." -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Fred A. Miller wrote:
David Bolt wrote:
So, given that I don't have a problem yet you're saying not to do it, I'm really curious as to just what issue(s) you found while running a 64bit OS?
Editing LARGE pics. in Gimp will give you an idea of how much better a 64-bit release is over 32-it. There are other examples, but your logic is sound.....use it!
Fred
And of course, you can work on really large files with vi. ;-) -- Use OpenOffice.org <http://www.openoffice.org> -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, 22 Jan 2009, James Knott wrote:-
Fred A. Miller wrote:
Editing LARGE pics. in Gimp will give you an idea of how much better a 64-bit release is over 32-it. There are other examples, but your logic is sound.....use it!
Fred
And of course, you can work on really large files with vi. ;-)
There's a thought... <warfare type="religeous" catalyst="preferred_editor"> Does it also mean that Emacs won't be constantly swapping? </warfare> :-) Regards, David Bolt -- Team Acorn: http://www.distributed.net/ OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s | openSUSE 10.3 32b | openSUSE 11.0 32b | openSUSE 10.2 64b | openSUSE 10.3 64b | openSUSE 11.0 64b | openSUSE 11.1 64b TOS 4.02 | openSUSE 10.3 PPC | RISC OS 3.6 | RISC OS 3.11 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thursday January 22 2009, David Bolt wrote:
On Thu, 22 Jan 2009, James Knott wrote:-
Fred A. Miller wrote:
Editing LARGE pics. in Gimp will give you an idea of how much better a 64-bit release is over 32-it. There are other examples, but your logic is sound.....use it!
Fred
And of course, you can work on really large files with vi. ;-)
There's a thought...
<warfare type="religeous" catalyst="preferred_editor"> Does it also mean that Emacs won't be constantly swapping? </warfare>
What it does mean, of course, is that for any given hardware and mix of applications, swapping will be _more_ likely, since all 64-bit code and data are substantially larger than their 32-bit counterparts.
:-)
Regards, David Bolt
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, 22 Jan 2009, Randall R Schulz wrote:-
On Thursday January 22 2009, David Bolt wrote:
On Thu, 22 Jan 2009, James Knott wrote:-
Fred A. Miller wrote:
Editing LARGE pics. in Gimp will give you an idea of how much better a 64-bit release is over 32-it. There are other examples, but your logic is sound.....use it!
Fred
And of course, you can work on really large files with vi. ;-)
There's a thought...
<warfare type="religeous" catalyst="preferred_editor"> Does it also mean that Emacs won't be constantly swapping? </warfare>
What it does mean, of course, is that for any given hardware and mix of applications, swapping will be _more_ likely, since all 64-bit code and data are substantially larger than their 32-bit counterparts.
The code itself may be bigger but, looking at the sizes of applications in /bin on both a 32bit and 64bit system, and the libraries in /lib abd /lib64, I wouldn't say the sizes are anything more than about 10% bigger. So, yes, you're likely to want about 10% more memory[0] on a 64bit system than on a 32bit system to be able to run the exact same applications without a need for swap. As for the data, why would it be "substantially" larger on a 64bit system? Pointers are going to be bigger on the 64bit system, as are variables of type "long", since these are 64bit, but those of type "int", and smaller, still occupy the same sizes in both the 32bit and 64bit OSes. Unfortunately, I really can't compare like with like. While I have two 10.3 systems which have the same memory and are running 64bit and 32bit OSes, they have quite different things running on them. The 32bit system is running my web server, has a few minor things that I occasionally connect remotely to use, but is mostly left alone. The 64bit system has quite a heavy load due to the number of desktop applications in use, and could actually do with a substantial amount of memory adding to it[1]. My 11.0 systems are also wildly different, the 64bit system has 2GB, runs my database server and some other stuff, and is presently using 1.5GB with 500MB caches and buffers, and 1.4MB swap use. The 32bit system does virtually nothing but the occasional compilation run, has 256MB of which 100MB is in use, 150MB used for buffers and cache, and is using 38MB of swap. [0] an extra 200MB for a 2GB system, 400MB for a 4GB system, etc. [1] at the present time, it has 2GB of memory and is using about 1.8GB of swap. A very large part of that usage is due to the eleven separate instances of Konqueror running, so that when one decides to segfault and crash, the others remain unaffected. Regards, David Bolt -- Team Acorn: http://www.distributed.net/ OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s | openSUSE 10.3 32b | openSUSE 11.0 32b | openSUSE 10.2 64b | openSUSE 10.3 64b | openSUSE 11.0 64b | openSUSE 11.1 64b TOS 4.02 | openSUSE 10.3 PPC | RISC OS 3.6 | RISC OS 3.11 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday January 23 2009, David Bolt wrote:
On Thu, 22 Jan 2009, Randall R Schulz wrote:-
...
What it does mean, of course, is that for any given hardware and mix of applications, swapping will be _more_ likely, since all 64-bit code and data are substantially larger than their 32-bit counterparts.
The code itself may be bigger but, looking at the sizes of applications in /bin on both a 32bit and 64bit system, and the libraries in /lib abd /lib64, I wouldn't say the sizes are anything more than about 10% bigger.
Use the "size" command, not the file size.
...
As for the data, why would it be "substantially" larger on a 64bit system? Pointers are going to be bigger on the 64bit system, as are variables of type "long", since these are 64bit, but those of type "int", and smaller, still occupy the same sizes in both the 32bit and 64bit OSes.
It's "substantial" because every pointer and every int is twice as big in a 64-bit programming model compared to a 32-bit one.
...
Regards, David Bolt
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Randall R Schulz wrote:
On Friday January 23 2009, David Bolt wrote:
On Thu, 22 Jan 2009, Randall R Schulz wrote:-
...
What it does mean, of course, is that for any given hardware and mix of applications, swapping will be _more_ likely, since all 64-bit code and data are substantially larger than their 32-bit counterparts.
The code itself may be bigger but, looking at the sizes of applications in /bin on both a 32bit and 64bit system, and the libraries in /lib abd /lib64, I wouldn't say the sizes are anything more than about 10% bigger.
Use the "size" command, not the file size.
...
As for the data, why would it be "substantially" larger on a 64bit system? Pointers are going to be bigger on the 64bit system, as are variables of type "long", since these are 64bit, but those of type "int", and smaller, still occupy the same sizes in both the 32bit and 64bit OSes.
It's "substantial" because every pointer and every int is twice as big in a 64-bit programming model compared to a 32-bit one.
...
Regards, David Bolt
Randall Schulz
A 64 bit processor simply has bigger internal registers and is therefore able to handle for instance a 64 bit calculation with one instruction instead of at least 3 for a 32 bit processor. The instruction is capable of being 64 bits in size but I doubt if the instruction set has increased in size to the point where that is necesary. The program counter register's size is determined by the size of the address space and not the size of the data bus. For instance the 64 bit celeron I'm running has address sizes : 36 bits physical, 48 bits virtual in cpuinfo and that would indicate a 36 bit program counter, been out of this for too long, which is expandable to 48 bits. The extent that a 64 bit processor out performs a 32 bit one is dependent on the optimization in the compiler used to compile the program. All this said I haven't ever done assembly programming for anything bigger than a 586. Regards Dave P -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday January 23 2009, Dave Plater wrote:
...
...
No matter what you say or how you wave your hands, you can't fit 10 pounds in a 5-pound bag.
Dave P
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Fri, 23 Jan 2009, Randall R Schulz wrote:-
On Friday January 23 2009, David Bolt wrote:
The code itself may be bigger but, looking at the sizes of applications in /bin on both a 32bit and 64bit system, and the libraries in /lib abd /lib64, I wouldn't say the sizes are anything more than about 10% bigger.
Use the "size" command, not the file size.
Okay, here's the output for the first 20 files in /bin for a 64bit and 32bit 10.3: davjam@adder:~> find /bin -type f | sort | xargs size 2>/dev/null | head -20 text data bss dec hex filename 1478 552 8 2038 7f6 /bin/arch 129290 1832 20704 151826 25112 /bin/ash 688532 4072 33584 726188 b14ac /bin/ash.static 14174 1024 392 15590 3ce6 /bin/basename 629025 22776 16880 668681 a3409 /bin/bash 33679 3440 17352 54471 d4c7 /bin/blkparse 23164 2944 248 26356 66f4 /bin/blktrace 17867 1032 392 19291 4b5b /bin/cat 38433 1128 424 39985 9c31 /bin/chgrp 34962 1088 456 36506 8e9a /bin/chmod 40631 1168 424 42223 a4ef /bin/chown 3342 640 24 4006 fa6 /bin/chvt 61757 1544 568 63869 f97d /bin/cp 116523 4888 1936 123347 1e1d3 /bin/cpio 45135 3408 488 49031 bf87 /bin/date 6049 720 16 6785 1a81 /bin/dbus-cleanup-sockets 315436 4004 1448 320888 4e578 /bin/dbus-daemon 9545 848 8 10401 28a1 /bin/dbus-monitor 12490 936 16 13442 3482 /bin/dbus-send davjam@cobra-mk3:~> find /bin -type f | sort | xargs size 2>/dev/null | head -20 text data bss dec hex filename 1154 280 4 1438 59e /bin/arch 108311 1008 19196 128515 1f603 /bin/ash 612700 2356 26752 641808 9cb10 /bin/ash.static 12238 532 388 13158 3366 /bin/basename 584686 12016 14992 611694 9556e /bin/bash 34193 2572 13156 49921 c301 /bin/blkparse 20044 1924 176 22144 5680 /bin/blktrace 16151 516 388 17055 429f /bin/cat 35580 568 420 36568 8ed8 /bin/chgrp 32041 544 420 33005 80ed /bin/chmod 37658 588 420 38666 970a /bin/chown 2598 324 12 2934 b76 /bin/chvt 57727 788 524 59039 e69f /bin/cp 106815 2756 1768 111339 1b2eb /bin/cpio 44019 2328 420 46767 b6af /bin/date 5597 364 16 5977 1759 /bin/dbus-cleanup-sockets 289830 2160 1024 293014 47896 /bin/dbus-daemon 8653 428 8 9089 2381 /bin/dbus-monitor 12070 472 8 12550 3106 /bin/dbus-send Yes, they are bigger, taking up more space on disc and more memory, but nowhere near double the space. So, while the 64bit binaries are close to 10% bigger and you'd probably need at least 10% more memory to avoid using swap[0], that isn't really "substantial" especially since you seem to be assuming that they are going to require double the memory.
...
As for the data, why would it be "substantially" larger on a 64bit system? Pointers are going to be bigger on the 64bit system, as are variables of type "long", since these are 64bit, but those of type "int", and smaller, still occupy the same sizes in both the 32bit and 64bit OSes.
It's "substantial" because every pointer and every int is twice as big in a 64-bit programming model compared to a 32-bit one.
Well, you're partially right. On a 64bit system, the pointers are indeed 64bit, longs are 64bits, but ints are still 32bit: davjam@adder:~> uname -m ; cat hello.c ; gcc -O3 -Wall -o hello.64 hello.c ; ./hello.64 ; ls -l hello.64 x86_64 #include <stdio.h> #include <stdlib.h> int main(int argc,char **argv) { printf("Hello world\n\n"); printf("sizeof(int)\t%lu\n", (unsigned long)sizeof(int)); printf("sizeof(long)\t%lu\n", (unsigned long)sizeof(long)); printf("sizeof(char *)\t%lu\n", (unsigned long)sizeof(char *)); exit(0); } Hello world sizeof(int) 4 sizeof(long) 8 sizeof(char *) 8 -rwxr-xr-x 1 davjam users 10866 2009-01-23 14:57 hello.64 And just as a comparison: davjam@cobra-mk3:~> uname -m ; cat hello.c ; gcc -O3 -Wall -o hello.32 hello.c ; ./hello.32 ; ls -l hello.32 i686 #include <stdio.h> #include <stdlib.h> int main(int argc,char **argv) { printf("Hello world\n\n"); printf("sizeof(int)\t%lu\n", (unsigned long)sizeof(int)); printf("sizeof(long)\t%lu\n", (unsigned long)sizeof(long)); printf("sizeof(char *)\t%lu\n", (unsigned long)sizeof(char *)); exit(0); } Hello world sizeof(int) 4 sizeof(long) 4 sizeof(char *) 4 -rwxr-xr-x 1 davjam users 9100 2009-01-23 14:59 hello.32 As you can see from the above, cobra-mk3 is running the 32bit version of 10.3, while adder is running the 64bit version. [0] Of course, the only real way to avoid using swap is to have sufficient memory available and to disable swap altogether. Regards, David Bolt -- Team Acorn: http://www.distributed.net/ OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s | openSUSE 10.3 32b | openSUSE 11.0 32b | openSUSE 10.2 64b | openSUSE 10.3 64b | openSUSE 11.0 64b | openSUSE 11.1 64b TOS 4.02 | openSUSE 10.3 PPC | RISC OS 3.6 | RISC OS 3.11 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday January 23 2009, David Bolt wrote:
On Fri, 23 Jan 2009, Randall R Schulz wrote:-
On Friday January 23 2009, David Bolt wrote:
The code itself may be bigger but, looking at the sizes of applications in /bin on both a 32bit and 64bit system, and the libraries in /lib abd /lib64, I wouldn't say the sizes are anything more than about 10% bigger.
Use the "size" command, not the file size.
Okay, here's the output for the first 20 files in /bin for a 64bit and 32bit 10.3:
...
Yes, they are bigger, taking up more space on disc and more memory, but nowhere near double the space. So, while the 64bit binaries are close to 10% bigger and you'd probably need at least 10% more memory to avoid using swap[0], that isn't really "substantial" especially since you seem to be assuming that they are going to require double the memory.
When I improve a metric of my software by 10%, that _is_ substantial, and conversely if I make some measure 10% worse. You're also ignoring dynamically allocated memory, which is very common in contemporary appliations, often far exceeding the statically allocated space requirements that are shown by the "size" command. Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Randall R Schulz wrote:
On Friday January 23 2009, David Bolt wrote:
On Fri, 23 Jan 2009, Randall R Schulz wrote:-
On Friday January 23 2009, David Bolt wrote:
The code itself may be bigger but, looking at the sizes of applications in /bin on both a 32bit and 64bit system, and the libraries in /lib abd /lib64, I wouldn't say the sizes are anything more than about 10% bigger.
Use the "size" command, not the file size.
Okay, here's the output for the first 20 files in /bin for a 64bit and 32bit 10.3:
...
Yes, they are bigger, taking up more space on disc and more memory, but nowhere near double the space. So, while the 64bit binaries are close to 10% bigger and you'd probably need at least 10% more memory to avoid using swap[0], that isn't really "substantial" especially since you seem to be assuming that they are going to require double the memory.
When I improve a metric of my software by 10%, that _is_ substantial, and conversely if I make some measure 10% worse.
You're also ignoring dynamically allocated memory, which is very common in contemporary appliations, often far exceeding the statically allocated space requirements that are shown by the "size" command.
Randall Schulz
i've been testing and noting bugs since 10.3 and at 11.1 64 bit is more stable than 32 bit. Regards Dave P -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Dave Plater wrote:
i've been testing and noting bugs since 10.3 and at 11.1 64 bit is more stable than 32 bit. Regards Dave P
Which might be only indicative of more people reporting more problems on the 32-bit versions. I know I would as I only run the 32-bit version now. Randall Schulz said it very eloquently: "The question is in fact whether there's any reason to use a 64-bit OS." I migrated a number of servers to 64bit about 12 months ago, and then migrated them back to 32bit some 9-10 months later. These servers were all the same - AMD 64bit, 2Gb RAM. My thinking was "64 has to be better than 32", if nothing else then because it's 64-bit OS/software on a 64-bit CPU. Well, I was wrong. Because of the increase in memory-usage with 64-bit openSUSE, each server ended upn being able to handle less load running 64bit than 32bit. /Per -- /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday 23 January 2009 16:07:01 Per Jessen wrote:
Randall Schulz said it very eloquently:
"The question is in fact whether there's any reason to use a 64-bit OS."
He /can/ be eloquent, can't he! (Hi All! I'm back from Pluto... he he)
I migrated a number of servers to 64bit about 12 months ago, and then migrated them back to 32bit some 9-10 months later. These servers were all the same - AMD 64bit, 2Gb RAM. My thinking was "64 has to be better than 32", if nothing else then because it's 64-bit OS/software on a 64-bit CPU. Well, I was wrong. Because of the increase in memory-usage with 64-bit openSUSE, each server ended upn being able to handle less load running 64bit than 32bit.
Are you saying I should now downgrade my 11.1 installation to the 32-bit version? I waited *forever* to upgrade my increasingly cranky previous system and when I took the plunge, I installed the 'native' 64-bit flavor! regards, Carl -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday January 23 2009, Carl Hartung wrote:
On Friday 23 January 2009 16:07:01 Per Jessen wrote:
Randall Schulz said it very eloquently:
"The question is in fact whether there's any reason to use a 64-bit OS."
He /can/ be eloquent, can't he!
On occasion, perhaps. But he really does place his commas deliberately, even if somewhat Clemensesquely... (Twain was known for his over-the-top use of commas.)
(Hi All! I'm back from Pluto... he he)
What's it like? Has word reached its denizens that they no longer live on a planet?
...
Are you saying I should now downgrade my 11.1 installation to the 32-bit version? I waited *forever* to upgrade my increasingly cranky previous system and when I took the plunge, I installed the 'native' 64-bit flavor!
Well, I know someone who has made it a habit of re-installing every time something fails to suit his tastes and exceeds his ability to rectify the situation, but it's not something I can recommend. I counted it as a good transition when it took only a week to get my 11.1 installation the way I like it. If past is prolog, the next time I install a new OS on this system, it'll be openSUSE 13.0.
regards,
Carl
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Carl Hartung wrote:
I migrated a number of servers to 64bit about 12 months ago, and then migrated them back to 32bit some 9-10 months later. These servers were all the same - AMD 64bit, 2Gb RAM. My thinking was "64 has to be better than 32", if nothing else then because it's 64-bit OS/software on a 64-bit CPU. Well, I was wrong. Because of the increase in memory-usage with 64-bit openSUSE, each server ended upn being able to handle less load running 64bit than 32bit.
Are you saying I should now downgrade my 11.1 installation to the 32-bit version? I waited *forever* to upgrade my increasingly cranky previous system and when I took the plunge, I installed the 'native' 64-bit flavor!
No, whether you're better off with 32bit or 64bit depends entirely on your system and your software. A heavily loaded database server with 32Gb of memory will no doubt be better off with a 64bit OS and apps. My servers only have 2Gb memory each, and because I'm running several hundred processes on each server, I ended up having room for less when everything was 64bit. -- /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Per Jessen wrote:
Carl Hartung wrote:
I migrated a number of servers to 64bit about 12 months ago, and then migrated them back to 32bit some 9-10 months later. These servers were all the same - AMD 64bit, 2Gb RAM. My thinking was "64 has to be better than 32", if nothing else then because it's 64-bit OS/software on a 64-bit CPU. Well, I was wrong. Because of the increase in memory-usage with 64-bit openSUSE, each server ended upn being able to handle less load running 64bit than 32bit. Are you saying I should now downgrade my 11.1 installation to the 32-bit version? I waited *forever* to upgrade my increasingly cranky previous system and when I took the plunge, I installed the 'native' 64-bit flavor!
No, whether you're better off with 32bit or 64bit depends entirely on your system and your software. A heavily loaded database server with 32Gb of memory will no doubt be better off with a 64bit OS and apps. My servers only have 2Gb memory each, and because I'm running several hundred processes on each server, I ended up having room for less when everything was 64bit.
I have boxes of all flavors 32bit - 64bit. I have 1 32-bit laptop and 1 64-laptop. 4 32-bit servers and 2 64-bit servers and several other desktops. I guess it has been a little over a year since building my first 64-bit server (opteron based) on 10.3. My primary server at work is still 32-bit while at home I have the 64-bit box. There isn't enough of a difference to compel me to lug the 64-bit box to work and replace the 32-bit box. The improvements are not dramatic, but good, on server throughput, etc.. Large compiles seem to go faster, though I haven't timed the difference. By far, my server at work is the busiest, but even in that regard, it is hardly even stressed at the busiest of times with the 13 clients it has plus the mail and web load on it. As mentioned already in this thread, gimp on my 64-bit laptop seems faster, but gimp on my old p4 3.3GHz 32-bit laptop is plenty fast as well. I think the thrust of the thread started as a "should I", or "shouldn't I" update from 32-bit on my 64-bit box. To me the answer is that there are no drawbacks to running the 64-bit release on your 64-bit hardware. openSuSE supports both equally and so do 3rd party drivers, etc. (In a majority of cases there isn't any difference) On the plus side, you will see good performance increases in the instances that 64-bit addressing makes a difference. If all my hardware was 64-bit, I'd be running the 64-bit release on all of it. With that said, I am equally happy with my 32-bit boxes and see no compelling reason to run out and change hardware. The bottom line for me is that I have seen (zero) drawbacks from the 64-bit systems and I have seen good benefits where it makes a difference. I don't consider that some packages are 5% or so larger to support the 64-bit addressing -- that's just the reality of the additional code needed to make use of the additional address space. (OK, I did (past tense) see a difference with java for web browsing not having a 64-bit plugin, but sun has since rectified that problem) Match the OS to your hardware. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
David C. Rankin wrote:
I think the thrust of the thread started as a "should I", or "shouldn't I" update from 32-bit on my 64-bit box. To me the answer is that there are no drawbacks to running the 64-bit release on your 64-bit hardware.
Bit of an old thread you've picked here, David, but perhaps the key bit here is "to me". To me, there were distinct drawbacks in going 64bit. To others there will (hopefully) be distinct advantages. /Per -- Per Jessen, Zürich (-2.63°C) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday 23 January 2009 04:20:33 pm Carl Hartung wrote:
On Friday 23 January 2009 16:07:01 Per Jessen wrote: ...
My thinking was "64 has to be better than 32", if nothing else then because it's 64-bit OS/software on a 64-bit CPU. Well, I was wrong. Because of the increase in memory-usage with 64-bit openSUSE, each server ended upn being able to handle less load running 64bit than 32bit.
Are you saying I should now downgrade my 11.1 installation to the 32-bit version? I waited *forever* to upgrade my increasingly cranky previous system and when I took the plunge, I installed the 'native' 64-bit flavor!
If you use programs that work with large data sets, which on desktop are video or images, or web pages with a lot of images that has to be decompressed, or flash videos from YouTube, you can see substantial improvement. I use GIMP and images are not really huge (about 2048x1536 from 3M pixel camera), and any complex effect is applied much faster then on 32 bit. If it would be possible to put whole image in CPU cache it would 2x faster. The other example mentioned by Larry, is full screen YouTube videos that run smoother. There are other conditions that apply, like size of RAM, hard disk speed, to final effective speed. For instance, Per example where he needed a bit more RAM, otherwise system will start to swap and performance went down. The decisive factor for or against 64 bit is your usage of the computer, and your hardware configuration. -- Regards, Rajko -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Dave Plater escribió: 64 bit is more
stable than 32 bit.
May have something to do with the fact more developers use it. -- "We have art in order not to die of the truth" - Friedrich Nietzsche Cristian Rodríguez R. Software Developer Platform/OpenSUSE - Core Services SUSE LINUX Products GmbH Research & Development http://www.opensuse.org/
Randall R Schulz escribió:
When I improve a metric of my software by 10%, that _is_ substantial, and conversely if I make some measure 10% worse.
Ok, the compiler produces bigger code and so what ? it is just an small part of the story.. -- "We have art in order not to die of the truth" - Friedrich Nietzsche Cristian Rodríguez R. Software Developer Platform/OpenSUSE - Core Services SUSE LINUX Products GmbH Research & Development http://www.opensuse.org/
On Saturday January 24 2009, Cristian Rodríguez wrote:
Randall R Schulz escribió:
When I improve a metric of my software by 10%, that _is_ substantial, and conversely if I make some measure 10% worse.
Ok, the compiler produces bigger code and so what ? it is just an small part of the story..
My point was simply to respond to the point about swapping by saying that bigger programs will have an increased probability of swapping on any given hardware configuration (amount of RAM) and every single program will be bigger when compiled for and run on a 64-bit architecture. And it's not just the code, it's also the data that are inflated on a 64-bit model. Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Saturday 24 January 2009 08:06:10 Randall R Schulz wrote:
My point was simply to respond to the point about swapping by saying that bigger programs will have an increased probability of swapping on any given hardware configuration (amount of RAM) and every single program will be bigger when compiled for and run on a 64-bit architecture. And it's not just the code, it's also the data that are inflated on a 64-bit model.
I've been around long enough to clearly remember the same arguments that were made in favor of 8-bit systems when 16-bit systems were new, and 16-bit systems when 32-bit systems were new. There are undoubtedly specific applications today that will run better on a 32-bit system. However, few computers specifically target a single such application. We expect ours to be able to run whatever we decide to put on it, today or next year, or even three or four years from now. I think that makes a very compelling argument for exploiting the 64-bit capability that's been available in most of our processors for a good many years now. So long that Microsoft won't be releasing any new 32-bit operating systems beyond Server 2008. RAM and disk storage are so relatively cheap these days, that it hardly matters if an operating system needs a little more of either one in order to run smoothly. However, on obsolete hardware, a 32-bit operating system - such as it was designed for - might be the only viable option. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
James Knott wrote:
Fred A. Miller wrote:
David Bolt wrote:
So, given that I don't have a problem yet you're saying not to do it, I'm really curious as to just what issue(s) you found while running a 64bit OS?
Editing LARGE pics. in Gimp will give you an idea of how much better a 64-bit release is over 32-it. There are other examples, but your logic is sound.....use it!
Fred
And of course, you can work on really large files with vi. ;-)
Hehehehehehehe......if anyone was going to say that, it'd be you or maybe that whacked out Aussie, Basil! ;) Fred -- "Politicians and diapers need to be changed regularly -- and for the same reason." -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
I've just switched from 11.1 32 bit to 64 bit and these are my initial observations :- the nvidia run package visibly compiles at what seems to be twice the speed so I should imagine compiling a kernel would also be considerably faster and K3b completes an md5 sum calculation of a dvd iso about 30% faster. My cpu's (a celeron dual core 64 bit) cores are running cooler. I should imagine that video processing would be considerably faster too but I have yet to try. Regards Dave P -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, 22 Jan 2009, Fred A. Miller wrote:- <snip>
Editing LARGE pics. in Gimp will give you an idea of how much better a 64-bit release is over 32-it. There are other examples, but your logic is sound.....use it!
I don't have any problems using a 32bit OS to edit the pictures I take. Then again, I only have a 10MP camera, so they aren't really that big. Regards, David Bolt -- Team Acorn: http://www.distributed.net/ OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s | openSUSE 10.3 32b | openSUSE 11.0 32b | openSUSE 10.2 64b | openSUSE 10.3 64b | openSUSE 11.0 64b | openSUSE 11.1 64b TOS 4.02 | openSUSE 10.3 PPC | RISC OS 3.6 | RISC OS 3.11 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Randall R Schulz wrote:
On Wednesday January 21 2009, Adam Sailer wrote:
I believe my hardware supports 64-bit, but what software issues can I expect to run into by installing 11.1 64-bit?
...
Adam Sailer
The question is, in fact, whether there's any reason to use a 64-bit OS. Do you have even one application that requires the address space afforded by the x86_64 architecture?
Very well said. -- /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
I went 64-bit. Everything is working *as well* as the 32 bit ver of 11.1 - That said, I'm noticing some of the same problems that I was having with the 32-bit version: Compiz screen refresh issue - from some of the other forums, I think this is an Nvidia driver issue. Sometimes *frequently* window contents don't update/refresh until I move the window/resize, etc. Any ideas on this? Hard lock after about an hour. This *may* be related to the issue above. After about an hour of use, the gui locks hard - can't ctrl-alt-backspace or ctrl-alt-f1. Can only reboot. Next time it happens I'll try to ssh into it and see if I can see anything weird. Thank you, Adam -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thursday January 22 2009, Adam Sailer wrote:
...
Hard lock after about an hour. This *may* be related to the issue above. After about an hour of use, the gui locks hard - can't ctrl-alt-backspace or ctrl-alt-f1. Can only reboot. Next time it happens I'll try to ssh into it and see if I can see anything weird.
Is it possible you're falling victim to the ReiserFS + Beagle -> Lockup phenomenon?
Thank you,
Adam
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Is it possible you're falling victim to the ReiserFS + Beagle -> Lockup phenomenon? Randall Schulz I'm not using ReiserFS though.... Thanx -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, Jan 22, 2009 at 4:22 PM, Adam Sailer <ASailer@uwb.edu> wrote:
I went 64-bit.
Everything is working *as well* as the 32 bit ver of 11.1 - That said, I'm noticing some of the same problems that I was having with the 32-bit version:
Compiz screen refresh issue - from some of the other forums, I think this is an Nvidia driver issue. Sometimes *frequently* window contents don't update/refresh until I move the window/resize, etc. Any ideas on this?
Hard lock after about an hour. This *may* be related to the issue above. After about an hour of use, the gui locks hard - can't ctrl-alt-backspace or ctrl-alt-f1. Can only reboot. Next time it happens I'll try to ssh into it and see if I can see anything weird.
I tried the KDE factory repository with compiz. About an hour is how long I could go with out it locking up. Fortunately cntrl-alt-f1 was working. I did init 3; init 5 several times. Then I threw in the towel and regressed back to what came on the DVD + the online security patches. It has been far more stable for me since I did that. Greg -- Greg Freemyer Litigation Triage Solutions Specialist http://www.linkedin.com/in/gregfreemyer First 99 Days Litigation White Paper - http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf The Norcross Group The Intersection of Evidence & Technology http://www.norcrossgroup.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (15)
-
Adam Sailer -
Aditya Bhiday -
Carl Hartung -
Clayton -
Cristian Rodríguez -
Dave Plater -
David Bolt -
David C. Rankin -
Fred A. Miller -
Greg Freemyer -
James Knott -
Jerry Houston -
Per Jessen -
Rajko M. -
Randall R Schulz