Message boards :
Number crunching :
Francois, Do You Need Help With Bail?
Message board moderation
Author | Message |
---|---|
Fivestar Crashtest Send message Joined: 10 Dec 99 Posts: 226 Credit: 5,377,978 RAC: 0 |
|
W-K 666 Send message Joined: 18 May 99 Posts: 19093 Credit: 40,757,560 RAC: 67 |
Francois' latest skulltrail video is on http://www.tgdaily.com/content/view/36165/135/ he mentions Seti and Folding right at the end. |
AlphaLaser Send message Joined: 6 Jul 03 Posts: 262 Credit: 4,430,487 RAC: 0 |
Nice article! :o) |
NewtonianRefractor Send message Joined: 19 Sep 04 Posts: 495 Credit: 225,412 RAC: 0 |
Francois' latest skulltrail video is on http://www.tgdaily.com/content/view/36165/135/ he mentions Seti and Folding right at the end. So what are the advantages of FB-dimm? This guy claims that it is the best thing in the world, while Tom's Hardware shows that the memory performance stinks... |
W-K 666 Send message Joined: 18 May 99 Posts: 19093 Credit: 40,757,560 RAC: 67 |
Francois' latest skulltrail video is on http://www.tgdaily.com/content/view/36165/135/ he mentions Seti and Folding right at the end. Correct, the chipset, 5400XS, is basically a modified server set, FB ram is used in servers because of its error detection and correction capabilities, this all takes time. |
Brian Silvers Send message Joined: 11 Jun 99 Posts: 1681 Credit: 492,052 RAC: 0 |
Francois' latest skulltrail video is on http://www.tgdaily.com/content/view/36165/135/ he mentions Seti and Folding right at the end. Not only THG, but Anandtech also disagrees with continuing FBD in Skulltrail... Anandtech article (pointed to page 3, which contains the FBD discussion) Again, my opinion on Francois is he's a salesman first and foremost... I'm willing to extend a "benefit of the doubt" that he has some technical skills, else he probably wouldn't have the job, but I still think he's more of a salesman... |
The Gas Giant Send message Joined: 22 Nov 01 Posts: 1904 Credit: 2,646,654 RAC: 0 |
Looks like who? is having a lot of phun with all this and good on him! He is still hopeful of beating the Macs and is enjoying the challenge. Mind you he did come across a little 'sheepish' when discussing the comp between PCs and Macs on SETI@Home. I loved his dig at PS3/Folding@Home as well. |
Francois Piednoel Send message Joined: 14 Jun 00 Posts: 898 Credit: 5,969,361 RAC: 0 |
Francois' latest skulltrail video is on http://www.tgdaily.com/content/view/36165/135/ he mentions Seti and Folding right at the end. hehehhee ... a sales man ... right ... Not at all, Anand and Toms have no clue about the Performance of DDR2 and DDR3 compare to FB-DIMM, especially on 8 cores. I am pretty well know have invented couple of instructions, I probably got them on ebay :) I propose that if somebody has a machine, using DDr2 or 3 that can beat skulltrail to show it to me, otherwise, please drop the arguement, because it is false. Notice the 16 core AMD machine that can not get even close to the MACs on the top 20 computer list??? it is DDR2!!!! The MACs and my machine are all FB-DIMM. In 2 minutes, the sales man proved you wrong ... hahahahha Nothing personal! this is just DATA from the TOP 20 list of SETI ... Who? |
Brian Silvers Send message Joined: 11 Jun 99 Posts: 1681 Credit: 492,052 RAC: 0 |
I think you miss the point. The point is that it is felt that skulltrail would be BETTER with plain DDR2 or DDR3. So, you are stating that Intel tried it, and FB-DIMM was actually better? |
David Send message Joined: 19 May 99 Posts: 411 Credit: 1,426,457 RAC: 0 |
I propose that if somebody has a machine, using DDr2 or 3 that can beat skulltrail to show it to me Not here thats for sure, BUT I'd like to try code thats better than Crunch3r's code for my Quad Core Q6600 using DDR2. I hope that happens soon, or Crunch3r does an update as those Mac's are just gaining too much ground. |
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
I propose that if somebody has a machine, using DDr2 or 3 that can beat skulltrail to show it to me That's all I have ever hoped for from Francois...... I got the hardware thing sussed....... Just hoping for some hardcore core 2 coding from who?...... (No diss on the Crunchmaster). "Freedom is just Chaos, with better lighting." Alan Dean Foster |
JDWhale Send message Joined: 6 Apr 99 Posts: 921 Credit: 21,935,817 RAC: 3 |
Nothing personal! this is just DATA from the TOP 20 list of SETI ... Is the SKT host that you've been running here at maximum memory performance, or have you been holding back, or is the motherboard that you have presented to us one of those manufacture defects. Why does your SKT have bandwith problems that will not allow it to achieve greater than 7GB/sec? That is not enough bandwith to feed 8 cores. Yes, the processors are fast and you can overclock all you want, but the problem is that you have not demonstrated that you can feed the beast and the processors are being starved. My DDR2 PC6400 gets 8 GB/sec and even that is not enough to fully feed my single little old Q6600. How can it be enough bandwith to feed 8 cores that are running faster? Please help me to understand, I beg you. Is it possible that you forgot to flip the switch and open up the second memory channel. |
David Send message Joined: 19 May 99 Posts: 411 Credit: 1,426,457 RAC: 0 |
Just hoping for some hardcore core 2 coding from who?...... Ditto. For now Crunch3r's code is the best there is for my Quad, and I give him the respect he deserves for excellent code, but there is always something better. I dont care who writes it, and how much better it is, but if any code is better than Crunch3rs code then good - I'll trial and then MAYBE use it and definately give the author the respect he/she deserves. For now Crunch3r's code is the best and I'll run it. Who? I hope your code is better, but if its not then so be it. Every crunch is a S@H crunch here, so as long as the 2 PC's run they will look for ET :) |
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
Just hoping for some hardcore core 2 coding from who?...... My point was.....perhaps poorly expressed...... If Francois would share some of his coding insights........ It might inspire or enable others to continue the quest....... A spark.........just a spark........ "Freedom is just Chaos, with better lighting." Alan Dean Foster |
ML1 Send message Joined: 25 Nov 01 Posts: 20372 Credit: 7,508,002 RAC: 20 |
... It might inspire or enable others to continue the quest....... Nah... Not enough electrons there even for the merest tingle of chip blowing static. All you've got is hope, hype, and vapour. Quite likely it's all just the best efforts of OC-ing, the latest unreleased ES, and keeping the vapour-chill cool enough. I guess this does put an alternate spin on "virtual (-isation)" and "code"... With all the perpetual tease and hype, this gets more like the ACME Da Vinci Code every month. Religion anyone? Good fun! Happy crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
... It might inspire or enable others to continue the quest....... Well, Marty.....I was hopin' for an endpoint...... "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Dr. C.E.T.I. Send message Joined: 29 Feb 00 Posts: 16019 Credit: 794,685 RAC: 0 |
. . . hUrt 'eM Francois - real bad like ;) ps - i've goT tiME SIr! Francois' latest skulltrail video is on http://www.tgdaily.com/content/view/36165/135/ he mentions Seti and Folding right at the end. BOINC Wiki . . . Science Status Page . . . |
ML1 Send message Joined: 25 Nov 01 Posts: 20372 Credit: 7,508,002 RAC: 20 |
Well, Marty.....I was hopin' for an endpoint...... "Closure" (such a horrid bit of jargon) is only what you get for sure when the film ends or if you're following one of those 'people' features in the media. I guess this bit of 'fun' and hype can enjoy a few more teeth (claws) being pulled yet... Let the kitties roar! Meanwhile, you do have a nice fast PC even if it doesn't quite have the promised code to make it scream like no other for s@h. Happy crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
Francois Piednoel Send message Joined: 14 Jun 00 Posts: 898 Credit: 5,969,361 RAC: 0 |
YOu are correct. In simulation, and now, with the hardware in the hands, as soon as you start using more than 4 cores, statistically FB-DIMM is better than DDR2. Most of the DDR2 mem sub systems are design with 8 ways. Most of the algorythm are what we call tri-polar algorythms, they do something like "A # B = C", (# operaration can be complexe, but the idea is that ever core does read 2 streams, and write one) The DDR2 memory subsystems are very well design for 4 cores, and statistically the 8 ways memory controlers does not break the stream too often on 4 cores(2 reads, 1 write per core), on 8 Cores, the statistics are reversed. In the case of 8 cores, you get 2 reads and 1 writes per core times 8, your DDR2 8 ways are totally overwarmed compare to 16 reads, and 8 writes, and on the top of This, DDR memory is a reversal subsystem, you can t read and write at the same time, With so many cores, it is a problem for DDR2. So, I know that a lot of journalist went emotional on the FB-DIMMS, proving how little they know about hardware, to design a memory controler able to handles more than 4 threads, it is a huge efforts, that required a lot of smart minds on the problem, in 2008, intel will be releasing something new about this, I will start explaining more soon, but basically, FB-DIMM todays are the top of art of 8 cores memory controler, it even beats a machine that AMD has with 16 Cores, and beats world records every day! So, yes, if you use only 1 cores out of your 8 cores, may be DDR2 is better, but if you use the full power of the 8 cores, I have outragious data showing that FB-DIMMs does MUCH better. May the Core be with you! Who? |
JDWhale Send message Joined: 6 Apr 99 Posts: 921 Credit: 21,935,817 RAC: 3 |
So, I know that a lot of journalist went emotional on the FB-DIMMS, proving how little they know about hardware, to design a memory controler able to handles more than 4 threads, it is a huge efforts, that required a lot of smart minds on the problem, in 2008, intel will be releasing something new about this, I will start explaining more soon, but basically, FB-DIMM todays are the top of art of 8 cores memory controler, it even beats a machine that AMD has with 16 Cores, and beats world records every day! So what you are saying is that the current SKT memory subsystem is not yet able to fully utilize performance potential of FB-DIMM. Does this explain the 7GB/sec ceiling on memory reads? Maybe we can expect this figure to nearly double, that the true and all powerful SKT is only a few (2-10) months away? |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.