.... I am new to this list & have a question about AMD64 hardware. An Intel P3/P4 processor requires 2 clock ticks to complete 1 32-bit floating point add or multiply, more for subtraction & division, more for 64-bit (I think); An SGI R10K/R12K/R14K can do 1 F.P. add & 1 F.P. mult (32 bit or 64 bit) in 1 clock tick, more for subtraction & division. What are the equivalent stats for AMD64 CPUs ? Thanks in advance.
.... I am new to this list & have a question about AMD64 hardware. An Intel P3/P4 processor requires 2 clock ticks to complete 1 32-bit floating point add or multiply, more for subtraction & division, more for 64-bit (I think); An SGI R10K/R12K/R14K can do 1 F.P. add & 1 F.P. mult (32 bit or 64 bit) in 1 clock tick, more for subtraction & division. What are the equivalent stats for AMD64 CPUs ? Thanks in advance. This is an interesting question but the answer is far from simple. The figures you mention are important for actual (math) processing speed, but things like
On Friday 19 March 2004 14:28, William A. Mahaffey III wrote: pipelining, caching and memory accessspeed are probably even more important nowadays. And not to forget the hardware/register based parallelism like MMX/ SSE. If you really want to know every detail, you have to work on a commercially interesting project in order to be allowed to sign an NDA with AMD .... In the meantime have a look at: http://www.cpuid.com/K8/index.php /Henk
Henk Slager wrote:
.... I am new to this list & have a question about AMD64 hardware. An Intel P3/P4 processor requires 2 clock ticks to complete 1 32-bit floating point add or multiply, more for subtraction & division, more for 64-bit (I think); An SGI R10K/R12K/R14K can do 1 F.P. add & 1 F.P. mult (32 bit or 64 bit) in 1 clock tick, more for subtraction & division. What are the equivalent stats for AMD64 CPUs ? Thanks in advance. This is an interesting question but the answer is far from simple. The figures you mention are important for actual (math) processing speed, but things like
On Friday 19 March 2004 14:28, William A. Mahaffey III wrote: pipelining, caching and memory accessspeed are probably even more important nowadays. And not to forget the hardware/register based parallelism like MMX/ SSE. If you really want to know every detail, you have to work on a commercially interesting project in order to be allowed to sign an NDA with AMD .... In the meantime have a look at:
http://www.cpuid.com/K8/index.php
/Henk
-- Check the List-Unsubscribe header to unsubscribe For additional commands, email: suse-amd64-help@suse.com
Interesting. I understand the necessity for good code & compilers to achieve speed-of-light performance, I was just asking what the speed-of-light is under AMD64, i.e. are they just a 64-bit version of P3/P4 (2 ticks / FPop) or more like SGI ?
William A. Mahaffey III wrote:
Interesting. I understand the necessity for good code & compilers to achieve speed-of-light performance, I was just asking what the speed-of-light is under AMD64, i.e. are they just a 64-bit version of P3/P4 (2 ticks / FPop) or more like SGI ?
Have you check AMD's web site? They have a whole section of AMD64 manuals (There is an AMD Athlon64 Tech Docs link is on the front page). There is a 5 volume programmers manual under the Opteron section. I remember flipping through an optimization guide there somewhere but most of it was above my head. I assume that the FPU is based on the Athlon which I remember being faster than the P4 one. -- Shawn D'Alimonte shawnd@mycybernet.net
Shawn D'Alimonte wrote:
William A. Mahaffey III wrote:
Interesting. I understand the necessity for good code & compilers to achieve speed-of-light performance, I was just asking what the speed-of-light is under AMD64, i.e. are they just a 64-bit version of P3/P4 (2 ticks / FPop) or more like SGI ?
Have you check AMD's web site? They have a whole section of AMD64 manuals (There is an AMD Athlon64 Tech Docs link is on the front page). There is a 5 volume programmers manual under the Opteron section.
I remember flipping through an optimization guide there somewhere but most of it was above my head.
I assume that the FPU is based on the Athlon which I remember being faster than the P4 one.
-- Shawn D'Alimonte shawnd@mycybernet.net
-- Check the List-Unsubscribe header to unsubscribe For additional commands, email: suse-amd64-help@suse.com
Yes, that's 1 of the things I don't like about AMD's website, they want you to download 5-volume manuals (usually pdf) & read them offline :-). I was just hoping someone would have an off-the cuff response they would render. I re-read the cpuid site a bit more & I think I may have the answer anyway, thanks.
Hello, I am trying to compile the SuSE 9.0 source on my SLES 8, as the postfix 1.1.12-12 that comes with SLES 8 does not support smtp_tls_server and smtpd_tls_auth_only. I get the following error when I run: rpm -bc /usr/src/packages/SPECS/postfix.spec ... gcc -Wmissing-prototypes -Wformat -DHAS_LDAP -DHAS_PCRE -DUSE_SASL_AUTH -DHAS_SSL -I/usr/include/sasl -O -I. -I../../include -DLINU X2 -c master_wakeup.c gcc -Wmissing-prototypes -Wformat -DHAS_LDAP -DHAS_PCRE -DUSE_SASL_AUTH -DHAS_SSL -I/usr/include/sasl -O -I. -I../../include -DLINU X2 -c master_flow.c gcc -Wmissing-prototypes -Wformat -DHAS_LDAP -DHAS_PCRE -DUSE_SASL_AUTH -DHAS_SSL -I/usr/include/sasl -O -I. -I../../include -DLINU X2 -o master master.o master_conf.o master_ent.o master_sig.o master_avail.o master_spawn.o master_service.o master_status.o master_listen.o master_vars.o master_wakeup.o master_flow.o ../../lib/libglobal.a ../../lib/libutil.a -lldap -llber -lpcre -lssl -lcrypto -lsasl2 -ldb -lnsl -lresolv /usr/lib64/gcc-lib/x86_64-suse-linux/3.2.2/../../../../x86_64-suse-linux/bin/ld: cannot find -lsasl2 collect2: ld returned 1 exit status make: *** [master] Error 1 make: *** [update] Error 1 Bad exit status from /var/tmp/rpm-tmp.23986 (%build) kingfish:/usr/src/packages # What might I do to resolve this. I can the sasl2 libraries: kingfish:/usr/src/packages # ls -ld /usr/lib64/sas* drwxr-xr-x 2 root root 776 Mar 19 17:38 /usr/lib64/sasl drwxr-xr-x 2 root root 1144 Mar 13 08:49 /usr/lib64/sasl2 Thank you - Richard
participants (4)
-
Henk Slager -
Richard Mixon (qwest) -
Shawn D'Alimonte -
William A. Mahaffey III