Let's buy GIMPS an Opteron! - Page 9

Dresdenboy · 2003-09-30, 06:05

Maybe a reason for this error is, that I used the link as description and URL, which lengthens the whole construction.

The best solution for the link on this board, which doesn't like long URL's, is to go directly to Chip-Architect and go to the desired articles.

Regards,
DB

guido72 · 2003-09-30, 06:22

Quote:

Originally posted by GP2
Here's the good link, the other one just had some space and "<br/>" mixed in...

http://www.chip-architect.com/news/2...4bit_Core.html

Maybe the mods for this forum could just edit Dresdenboy's original post to fix this.

[Edit: well it's not Dresdenboy's fault: it's the board! It introduced the same garbage characters for me as it did for him. ]

Try:

http://

www.chip-architect.com/news/

2003_09_21_Detailed_Architecture_of_AMDs_64bit_Core.html

Many thanks GP2. I saw the unusual chars in the link immediately after having post...

Xyzzy · 2003-10-14, 00:07

After 24 days uptime I finally have taken the box down for an hour or two so I can install the CD-ROM we bought...

The box will be back up ASAP...

Xyzzy · 2003-10-14, 01:46

And we're back in business...

BTW, the box turned in its first LL test recently...

[Fri Sep 5 06:48:11 2003]
UID: TeamOpteron/Opteron-140, M19474769 completed P-1, B1=235000, B2=5463750, WZ2: F7A40580
[Sat Sep 27 18:36:07 2003]
UID: TeamOpteron/Opteron-140, M19474769 is not prime. Res64: xxxxxxxxxxxxxxxx. WZ2: 71F814E1,3076795,00000000

Dresdenboy · 2003-10-15, 10:44

AMD CodeAnalyst for Linux available!

http://www.amd.com/us-en/Processors/...2_3604,00.html

That should solve our problem of accessing the performance counters.

Xyzzy · 2003-10-16, 16:28

http://www.developwithamd.com/appPar...orms/index.cfm

Can we use that?

Which Code Analyst do we want? The 32-bit one or the 64-bit one?

Xyzzy · 2003-10-16, 16:38

Okay, I think the Code Analyst is installed... Look for it in /usr/local/bin

There are man pages too...

I hope the 32-bit version is what we want...

If you crash the box please email me...

Dresdenboy · 2003-10-19, 15:04

Thanks, the 32bit version should be fine for most programmers here.

They mention to use oprofile (or something similar) for profiling together with CodeAnalyst. But I'm not sure if oprofile supports all possible events. They list only a few on their website.

bayanne · 2003-10-22, 08:42

Latest news from bovine

Code:

We're close to having an optimized AMD64 (x86_64) client with a
hand-optimized natively 64-bit RC5-72 core in it.  If you're
interested in the progress of this, you can follow this bug for
status.  http://n0cgi.distributed.net/bugs/show_bug.cgi?id=3387
Currently we're trying to tweak a few more cycles out of this core
before releasing something.

Prime95 · 2003-10-22, 18:02

I tried to run tracegen and it just hangs. Ideas anyone?

I'm tracing an mprime built on my local Linux system.

Joe O · 2003-11-10, 19:56

I was just wondering how this project was doing.

Prime95 · 2003-11-10, 23:44

Quote:

Originally posted by Joe O
I was just wondering how this project was doing.

Badly stalled. The bottleneck has been identified - we can't get data into and out of the FPU fast enough. No workaround has been found (Opteron articles suggest we should be able to do more loads and stores per clock cycle than we are seeing).

Further analysis via AMD's CodeAnalyst is difficult - it crashes my P4 and hangs when running mprime on the Opteron. I'm entering a very busy period, but it seems the next best step is to recode the assembly snipets I'm trying to benchmark in gas and build a 64-bit executable to run CodeAnalyst on.

optim · 2003-11-12, 20:39

because somebody in another message reffered to "734pin" athlon64 etc:

AMD 64bit models:

Athlon 64 3200+: 754pin, 2GHz, single channel memory, about $500
Athlon64 FX: 940pin, 2.2GHz, dual channel memory, about $700-900, identical to Opteron 14x, No SMP support.
Opteron 140/142/144/146: 940pin, 1.4ghz/1.6/1.8/2ghz, dual channel memory, about $240 for model 140 or 240 at 1.4ghz
Opteron models 240 are for dual processor systems and have SMP support.

prices are not right, they just indicate the "level" of the expenses needed to buy the cpu.

pakaran · 2003-11-18, 00:37

Quote:

Originally posted by Dresdenboy
I wish you good luck!

And you are right, SuSe distributions are freely downloadable. Only when you buy a SuSe distribution, you usually pay for the media, books and support.

IIRC you have to download files - they don't distribute ISOs. Depending on your Linux/unix experience this may/may not be a problem. For me, even being comfortable with some of the odder distros, I'd prefer to download something I could install from an ISO.

SalemTheCat100 · 2003-12-18, 01:43

After having donations for funding an Opteron system and identifing the bottle neck - coding ceases.

I bet if this was an Intel processor we would have had continuous work until a fix was found.

This type of non-support for AMD is why I have turned off 12 of my 14 prime95 clients.

SALEM

Xyzzy · 2003-12-18, 04:29

Actually, there are people logged into the Opteron quite a bit working on stuff... Just because it isn't posted here doesn't mean work isn't being done...

BTW, if you want an account to help let me know!

GP2 · 2003-12-18, 05:39

Salem,

Your attitude is unhelpful.

The machines you turned off are no doubt non-64 Athlons which wouldn't benefit from the Opteron optimization anyway. So what's the point?

Instead of conspiracy theories, let's try to come up with constructive "thinking-outside-the-box" solutions. I started a thread for this:

http://www.mersenneforum.org/showthr...&threadid=1742

gbvalor · 2003-12-18, 06:24

Salem,

Since I've been working in this opteron on Glucas, their speed has rised about a 15%. And other people is working in mprime/prime95 trying to do the things better and faster

Guillermo

Xyzzy · 2003-12-18, 07:15

Code:

Linux opteron 2.6.0 #2 Thu Dec 18 01:22:39 EST 2003 x86_64 unknown unknown GNU/Linux

Now the two Broadcom Gig-E ethernet ports work! We were using a crappy SMC (rtl8139) card up until now...

SalemTheCat100 · 2003-12-18, 17:38

Quote:

Originally posted by gbvalor
Salem,

Since I've been working in this opteron on Glucas, their speed has rised about a 15%. And other people is working in mprime/prime95 trying to do the things better and faster

Guillermo

Well it's good

to see that work is continuing on the AMD64 client.

After the hollidays I will be configuring my dual Opteron 240 system as my main work station on Windows 2000. I am willing to help in testing.

Is there a site to download these alpha/beta AMD64 clients?

Thanks

SALEM

gbvalor · 2003-12-18, 21:39

Quote:

After the hollidays I will be configuring my dual Opteron 240 system as my main work station on Windows 2000. I am willing to help in testing.

Is there a site to download these alpha/beta AMD64 clients?

Glucas is designed mainly for Unix/Linux/Mac Oses. Some people have compiled and tried it on windows with cgywin, I didn't

Prime95 is the proper client for this OS, but I don't know how is the state of development/optimization of it.

Guillermo

gbvalor · 2003-12-18, 21:45

Quote:

Linux opteron 2.6.0 #2 Thu Dec 18 01:22:39 EST 2003 x86_64 unknown unknown GNU/Linux

I don't know whether is because the compiler or the kernel, but my timings have dropped about a 4% with this new kernel.

Anyone have also detected the same on mprime?

Guillermo.

Xyzzy · 2003-12-19, 05:18

Quote:

Originally posted by gbvalor
I don't know whether is because the compiler or the kernel, but my timings have dropped about a 4% with this new kernel.

Anyone have also detected the same on mprime?

Yes, I saw a slight speed loss as well... Maybe I compiled the kernel wrong? I used the default menuconfig settings, minus SMP...

aaronl · 2003-12-19, 06:18

Quote:

Originally posted by Xyzzy
Yes, I saw a slight speed loss as well... Maybe I compiled the kernel wrong? I used the default menuconfig settings, minus SMP...

Did you enable kernel preemption (CONFIG_PREEMPT)? In my slightly outdated kernel source it looks like it's enabled by default on x86 but not x86-64.

Linux 2.6 runs the scheduler at a much higher frequency by default. This adds some overhead and could be part of the cause. I'd be personally interested in what changing HZ in include/asm-x86_64/param.h from 1000 to 100 does to the timings.

I'm tempted to raise this issue in the Linux forum so that people who run crunching farms can adjust their kernel settings optimally. I'm confident that lower HZ would improve mprime timings slightly (HZ=25 would probably be even better on dedicated GIMPS boxen). I also think that turning off kernel preemption would be a win, but this is harder to predict. Someone who has a spare machine to test with (not me) should run some benchmarks so we could verify this. It would also be interesting to see how 2.6 compares to 2.4 after factors like HZ and preemption are taken into account.

PrimeCruncher · 2003-12-20, 21:07

From this CNet news report:

Quote:

The new kernel also monitors for new events more frequently--1,000 times per second instead of 100--a fact that slows down the system about 1 percent, Morton said in an October presentation about the kernel.

In addition, 2.6 requires somewhat more memory to run and shows worse performance when it has to use hard drives as extra memory under heavy loads, Morton said.

Don't know much about Linux but that sounds like the culprit right there to me...

2003-11-12, 20:39	#213
optim Nov 2003 European Union 1101000₂ Posts	because somebody in another message reffered to "734pin" athlon64 etc: AMD 64bit models: Athlon 64 3200+: 754pin, 2GHz, single channel memory, about $500 Athlon64 FX: 940pin, 2.2GHz, dual channel memory, about $700-900, identical to Opteron 14x, No SMP support. Opteron 140/142/144/146: 940pin, 1.4ghz/1.6/1.8/2ghz, dual channel memory, about $240 for model 140 or 240 at 1.4ghz Opteron models 240 are for dual processor systems and have SMP support. prices are not right, they just indicate the "level" of the expenses needed to buy the cpu. Last fiddled with by optim on 2003-11-12 at 20:48

2003-12-18, 01:43	#215
SalemTheCat100 Oct 2002 2·13 Posts	Typical AMD support - NONE After having donations for funding an Opteron system and identifing the bottle neck - coding ceases. I bet if this was an Intel processor we would have had continuous work until a fix was found. This type of non-support for AMD is why I have turned off 12 of my 14 prime95 clients. SALEM

2003-12-18, 07:15	#219
Xyzzy "Mike" Aug 2002 20065₈ Posts	Code: Linux opteron 2.6.0 #2 Thu Dec 18 01:22:39 EST 2003 x86_64 unknown unknown GNU/Linux Now the two Broadcom Gig-E ethernet ports work! We were using a crappy SMC (rtl8139) card up until now...

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Opteron is Hyperthreaded ?	bgbeuning	Information & Answers	3	2016-01-10 08:26
Opteron web server...	Xyzzy	Lounge	14	2003-11-05 23:07
Opteron Bottleneck??	Prime95	Hardware	31	2003-09-17 06:54
AMD Opteron	naclosagc	Software	27	2003-08-10 19:14
What will an AMD Opteron be classified as ?	dsouza123	Software	4	2003-08-02 14:29

2003-09-30, 06:05	#201
Dresdenboy Apr 2003 Berlin, Germany 19² Posts	Maybe a reason for this error is, that I used the link as description and URL, which lengthens the whole construction. The best solution for the link on this board, which doesn't like long URL's, is to go directly to Chip-Architect and go to the desired articles. Regards, DB

2003-10-14, 00:07	#203
Xyzzy "Mike" Aug 2002 5·17·97 Posts	After 24 days uptime I finally have taken the box down for an hour or two so I can install the CD-ROM we bought... The box will be back up ASAP...

2003-10-14, 01:46	#204
Xyzzy "Mike" Aug 2002 5·17·97 Posts	And we're back in business... BTW, the box turned in its first LL test recently... [Fri Sep 5 06:48:11 2003] UID: TeamOpteron/Opteron-140, M19474769 completed P-1, B1=235000, B2=5463750, WZ2: F7A40580 [Sat Sep 27 18:36:07 2003] UID: TeamOpteron/Opteron-140, M19474769 is not prime. Res64: xxxxxxxxxxxxxxxx. WZ2: 71F814E1,3076795,00000000

2003-10-15, 10:44	#205
Dresdenboy Apr 2003 Berlin, Germany 169₁₆ Posts	AMD CodeAnalyst for Linux available! http://www.amd.com/us-en/Processors/...2_3604,00.html That should solve our problem of accessing the performance counters.

2003-10-16, 16:28	#206
Xyzzy "Mike" Aug 2002 5·17·97 Posts	http://www.developwithamd.com/appPar...orms/index.cfm Can we use that? Which Code Analyst do we want? The 32-bit one or the 64-bit one?

2003-10-16, 16:38	#207
Xyzzy "Mike" Aug 2002 5×17×97 Posts	Okay, I think the Code Analyst is installed... Look for it in /usr/local/bin There are man pages too... I hope the 32-bit version is what we want... If you crash the box please email me...

2003-10-19, 15:04	#208
Dresdenboy Apr 2003 Berlin, Germany 19² Posts	Thanks, the 32bit version should be fine for most programmers here. They mention to use oprofile (or something similar) for profiling together with CodeAnalyst. But I'm not sure if oprofile supports all possible events. They list only a few on their website.

2003-10-22, 18:02	#210
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 1110101110111₂ Posts	I tried to run tracegen and it just hangs. Ideas anyone? I'm tracing an mprime built on my local Linux system.

2003-11-10, 19:56	#211
Joe O Aug 2002 525₁₀ Posts	I was just wondering how this project was doing.

2003-12-18, 04:29	#216
Xyzzy "Mike" Aug 2002 5·17·97 Posts	Actually, there are people logged into the Opteron quite a bit working on stuff... Just because it isn't posted here doesn't mean work isn't being done... BTW, if you want an account to help let me know!

2003-12-18, 05:39	#217
GP2 Sep 2003 5031₈ Posts	Salem, Your attitude is unhelpful. The machines you turned off are no doubt non-64 Athlons which wouldn't benefit from the Opteron optimization anyway. So what's the point? Instead of conspiracy theories, let's try to come up with constructive "thinking-outside-the-box" solutions. I started a thread for this: http://www.mersenneforum.org/showthr...&threadid=1742

2003-12-18, 06:24	#218
gbvalor Aug 2002 3·37 Posts	Salem, Since I've been working in this opteron on Glucas, their speed has rised about a 15%. And other people is working in mprime/prime95 trying to do the things better and faster Guillermo