mersenneforum.org  

Go Back   mersenneforum.org > Search Forums

Showing results 1 to 25 of 89
Search took 0.02 seconds.
Search: Posts Made By: rgiltrap
Forum: PrimeNet 2009-11-28, 09:20
Replies: 19
Views: 1,408
Posted By rgiltrap
Not sure why you want RAID6 when you only have 4...

Not sure why you want RAID6 when you only have 4 disks in the system. You would be better off performance wise with RAID 10 or even RAID5 2+1 with 1 spare.

There's a big performance boost going...
Forum: PrimeNet 2009-11-28, 00:36
Replies: 19
Views: 1,408
Posted By rgiltrap
As an architect for Sun Microsystems this is my...

As an architect for Sun Microsystems this is my bread and butter. Unfortunately the original config appears to have expired so I can't review it. Can you post the proposed config again?

Also, how...
Forum: News 2009-06-20, 07:24
Replies: 631
Views: 62,061
Posted By rgiltrap
The bigger the cache the more latency it will...

The bigger the cache the more latency it will incur as the distance to the last piece of cache increases. Hence we have very small L1 cache sizes, moderate size L2 and large L3 each with an...
Forum: News 2009-06-19, 20:34
Replies: 631
Views: 62,061
Posted By rgiltrap
Hi Tony, I checked the specs and that's...

Hi Tony,

I checked the specs and that's using Nehalem EP processors which scale to 2 sockets x 4 cores x 2 threads. Is this what you were using for your verification?

The B505 blade is running...
Forum: News 2009-06-16, 10:31
Replies: 631
Views: 62,061
Posted By rgiltrap
You are assuming one CPU... Bull makes very BIG...

You are assuming one CPU... Bull makes very BIG systems. I believe Nehalem EX can scale to 64 sockets making a potential 512 cores or 1K threads!
Forum: News 2009-06-16, 06:14
Replies: 631
Views: 62,061
Posted By rgiltrap
It is unlikely that hyper threading will provide...

It is unlikely that hyper threading will provide any benefit. Hyperthreading is beneficial when there are cycles lost waiting for main memory being brought into the cache.

With highly parallel...
Forum: News 2009-06-15, 23:30
Replies: 631
Views: 62,061
Posted By rgiltrap
In a futile attempt to get this thread back on...

In a futile attempt to get this thread back on track...

The Mlucas (http://www.hogranch.com/mayer/README.html) 3.x (prerelease) verification run has completed.

The verification was run on a Sun...
Forum: News 2009-06-15, 11:02
Replies: 631
Views: 62,061
Posted By rgiltrap
Mlucas / Sparc64 verification is 95% complete. ...

Mlucas / Sparc64 verification is 95% complete.

Residues match at 40,000,000.

Will be finished in about 12 hours from now (just after 4pm PST).
Forum: News 2009-06-14, 05:51
Replies: 631
Views: 62,061
Posted By rgiltrap
Tony's 2nd verification result is in... ...

Tony's 2nd verification result is in...

M42643801 is prime! G29: 00000000,0
NEW MERSENNE PRIME DISCOVERED !!!!

This is the completion of the original Glucas on Bull Novascale 16 way Itanium 2...
Forum: News 2009-06-13, 10:15
Replies: 631
Views: 62,061
Posted By rgiltrap
I'm 71% done (10pm NZT) with Mlucas / Sparc /...

I'm 71% done (10pm NZT) with Mlucas / Sparc / Solaris and can confirm matching residues with Tony's run at 30 Million Iterations.

M42643801 Iter# = 30000000 [0.0182 sec/iter] Res64:...
Forum: News 2009-06-12, 21:20
Replies: 631
Views: 62,061
Posted By rgiltrap
Congratulations Tony on reclaiming your...

Congratulations Tony on reclaiming your verification crown. All hail le Dauphin.

Also my apologies for my little misdirection. It's amazing how many posts were generated from one little sentence...
Forum: News 2009-06-12, 08:52
Replies: 631
Views: 62,061
Posted By rgiltrap
Mea culpa, mea culpa, mea maxima culpa. You are...

Mea culpa, mea culpa, mea maxima culpa. You are correct I meant 1 billion digits. :redface:
Forum: News 2009-06-12, 07:47
Replies: 631
Views: 62,061
Posted By rgiltrap
Just a word of warning... as Kevin's result has...

Just a word of warning... as Kevin's result has possibly shown, x86 CPUs are susceptible to silent errors.

Processors like Itanium, Sparc, Power have significant portions of silicon dedicated to...
Forum: News 2009-06-11, 07:04
Replies: 631
Views: 62,061
Posted By rgiltrap
After a suggestion from my colleague Tom I have...

After a suggestion from my colleague Tom I have managed to get a minor performance improvement.

The speed has improved from a consistent 0.0183 secs/iteration to 0.0182 secs/iteration. :lol:

My...
Forum: News 2009-06-09, 19:50
Replies: 631
Views: 62,061
Posted By rgiltrap
What would be very useful for comparison is if...

What would be very useful for comparison is if (once you have finished this run) you would try your four cores against an exponent that uses a FFT length of 4096K (something around 77mil).

I get...
Forum: Hardware 2009-06-08, 12:01
Replies: 22
Views: 1,702
Posted By rgiltrap
It would cost very little but the cost of the...

It would cost very little but the cost of the services to get it running would be extortionate! It's the IBM way :wink:

The CPU speed is very misleading. An Itanium2 at 5Ghz would do many times...
Forum: News 2009-06-08, 06:34
Replies: 631
Views: 62,061
Posted By rgiltrap
Ahhh... I see now. ATH is checking George's fake...

Ahhh... I see now. ATH is checking George's fake number.

That explains the difference.
Forum: News 2009-06-08, 03:37
Replies: 631
Views: 62,061
Posted By rgiltrap
Mine are quite different?!? 1Mil Res64:...

Mine are quite different?!?
1Mil Res64: B888A4E085915350
2Mil Res64: E58A6CA45863341A
3Mil Res64: 48267344B9976A4D
4Mil Res64: 105007A65A192DF9
5Mil Res64: 07558BC6A5CA6316

I know that the...
Forum: News 2009-06-07, 19:55
Replies: 631
Views: 62,061
Posted By rgiltrap
Tony, Kevin, I'm running freshly compiled...

Tony, Kevin,

I'm running freshly compiled code, as such I get a little paranoid :drama:.

Would you please confirm for me that we have matching residues?

Iter# = 400000 Res64:...
Forum: News 2009-06-07, 05:09
Replies: 631
Views: 62,061
Posted By rgiltrap
Due to my late start I'm expecting sometime on...

Due to my late start I'm expecting sometime on June 16th EST. I'm currently processing at 18.3ms per iteration so I don't expect to catch up with Kevin or T.Rex as they appear to be processing at a...
Forum: News 2009-06-06, 23:00
Replies: 631
Views: 62,061
Posted By rgiltrap
I have finally managed to get access to a server...

I have finally managed to get access to a server and currently are running at a speed of 10.95% completion per day.

This is only about 60% of the speed we should be getting as I'm being forced to...
Forum: News 2009-06-05, 20:53
Replies: 631
Views: 62,061
Posted By rgiltrap
The Mlucas / Sun team (Ernst, Tom & I) are a...

The Mlucas / Sun team (Ernst, Tom & I) are a little slow off the mark this time. We're still trying to find an available server. Also, this exponent will force us to use a much larger FFT than is...
Forum: News 2009-06-05, 05:56
Replies: 631
Views: 62,061
Posted By rgiltrap
George has probably gone to bed now but I'll send...

George has probably gone to bed now but I'll send him an email and should be able to get a 16 way run started in about 12 hours from the time of this post.
Forum: News 2009-06-04, 18:01
Replies: 299
Views: 49,817
Posted By rgiltrap
It's also the time those of us (from Sun, Bull...

It's also the time those of us (from Sun, Bull etc) with access to big tin, wheel out Mlucas & Glucas and do our best to squeeze every last bit of performance out of 16 cores to bring calm again to...
Forum: Mlucas 2009-05-04, 21:45
Replies: 14
Views: 7,369
Posted By rgiltrap
First things first have you done the self test? ...

First things first have you done the self test?

You need to run "Mlucas -s m" and see if that works first.

If that is successful send us your worktodo.ini

Cheers, Rob.
Showing results 1 to 25 of 89

 
All times are UTC. The time now is 13:27.

Thu May 6 13:27:34 UTC 2021 up 28 days, 8:08, 0 users, load averages: 2.19, 1.72, 1.77

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.