Message boards : Number crunching : who will be the next king of the hill?
Author | Message |
---|---|
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
take a look at this monster: 351446 compare to 341221 (Number 2, soon number 1) The daily average is almost double of the "Top20" number 2 Looks like Dual socket XEON quad core will be the king of the hill for quit some time ... You obvisouly don't need hypertransport to be fast, that was a nice hype, but confirmed to be untrue. Opteron is again ... I will be surprise if GrandFather 4x4 can make it in the top 20 list. Core 2 Quad core is the Godfather! on the other side, look at this 268485, a new prototype from AMD, Jul 2006 ... what can it be? K8L or 65nm K8???? It gets a 0.004167 credit per seconds while the Number 2 gets 0.00760 341221 credit per seconds for each processor core. To compare the top 8 CPU Opteron gets 0.003367 credit per second per core, making this new "prototype machine" faster than the K8 based core, but way behind Core 2 The Proto Owner Mats Pxxxxxxx Created 4 Jul 2006 17:20:25 UTC Total Credit 155,053.79 Recent average credit 1,366.09 CPU type AuthenticAMD AMD Engineering Sample <====== HO HO! never saw this one before! Number of CPUs 4 Operating System Linux 2.6.17-1.2139_FC4smp Memory 6072.48 MB Cache 1024 KB Measured floating point speed 1346.79 million ops/sec <== Measured integer speed 2297.07 million ops/sec <== Average upload rate 16.95 KB/sec Average download rate Unknown Average turnaround time 1.82 days Maximum daily WU quota per CPU 100/day Of course, you should not take into account the 2 scored with arrows, because we all know those a "quick and dirty" estimate for the unit ordering mecanism. Now, the real question ... what frequency it is running at???? Who? |
zombie67 [MM] Send message Joined: 11 Feb 06 Posts: 316 Credit: 6,621,003 RAC: 1 |
|
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
take a look at this monster: Two XEONs Quad Core ... that's what we all need :) who? |
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
take a look at this monster: I forgot to say, and to be sure to be transparent, I did merge it with its secret identity ID it was using before the "launch" it got 300 credits from there. who? |
FluffyChicken![]() Send message Joined: 1 Nov 05 Posts: 1260 Credit: 369,635 RAC: 0 |
take a look at this monster: What Hz is the Xeon running, defualt 2.66 or at 4 like the other. Team mauisun.org |
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
take a look at this monster: At the stock 2.66GHZ, insane, isn't it? who? |
FluffyChicken![]() Send message Joined: 1 Nov 05 Posts: 1260 Credit: 369,635 RAC: 0 |
take a look at this monster: Yes. (how come you've not joined team intel over here like you have at seti ?) Team mauisun.org |
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
one week of running time, and 2,092 RAC :) vrommmmm Still running without hypertransport :) hehehehe who? |
Mats Petersson Send message Joined: 29 Sep 05 Posts: 225 Credit: 951,788 RAC: 0 |
And why would hypertransport make any sort of difference to Rosetta? Anyone claiming that would be as clever as saying that a Ferrari is better than a normal car in London City rush-hour traffic - it won't, because there's no room to benefit from the speed. As we've discussed before, a good machine (with large enough caches) will have 97%+ cache-hit-rate with Rosetta... You are the one who said you wanted to see a stop to performance gain claims that are based on "other" improvements. You've got a brand new core in your processor, it's not doing much memory access, I presume, to so even if you had a shared bus at 100MHz, it shouldn't make much, if any, difference at all. -- Mats |
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
And why would hypertransport make any sort of difference to Rosetta? Anyone claiming that would be as clever as saying that a Ferrari is better than a normal car in London City rush-hour traffic - it won't, because there's no room to benefit from the speed. As we've discussed before, a good machine (with large enough caches) will have 97%+ cache-hit-rate with Rosetta... You are the one who said you wanted to see a stop to performance gain claims that are based on "other" improvements. You've got a brand new core in your processor, it's not doing much memory access, I presume, to so even if you had a shared bus at 100MHz, it shouldn't make much, if any, difference at all. I agree with you. Hypertransport is useless on Rosetta :), so does it on all the crunching workloads: A well programmed crunching algorythm will always fit in the case, or close to it. I actually see performance improvement on Rosetta when i increase the front side bus speed, it is due to the fact that 3% can be compressed to 1% :) In the racing world, 2% is great :) who? |
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
And why would hypertransport make any sort of difference to Rosetta? Anyone claiming that would be as clever as saying that a Ferrari is better than a normal car in London City rush-hour traffic - it won't, because there's no room to benefit from the speed. As we've discussed before, a good machine (with large enough caches) will have 97%+ cache-hit-rate with Rosetta... You are the one who said you wanted to see a stop to performance gain claims that are based on "other" improvements. You've got a brand new core in your processor, it's not doing much memory access, I presume, to so even if you had a shared bus at 100MHz, it shouldn't make much, if any, difference at all. I am upgrading the top machine to Vista, so, it will not be running over night. who? |
Michael G.R. Send message Joined: 11 Nov 05 Posts: 264 Credit: 11,247,510 RAC: 0 |
How's the CPU overhead on Vista? Much of a difference, or are the requirements mostly for GPU and RAM? |
zombie67 [MM] Send message Joined: 11 Feb 06 Posts: 316 Credit: 6,621,003 RAC: 1 |
I am upgrading the top machine to Vista, so, it will not be running over night. Be sure to tell us the new machine id#, so we can continue to track it. Thanks! Reno, NV Team: SETI.USA |
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
I am upgrading the top machine to Vista, so, it will not be running over night. You can t miss them .. both my machines are in the top 2 :-P just enjoying it, sorry ;) Who? |
![]() ![]() Send message Joined: 15 Dec 05 Posts: 761 Credit: 285,578 RAC: 0 |
then did you mistyoe the thread title - should it have been who? will be the next king of the hill. |
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
then did you mistyoe the thread title - should it have been The real king is Core 2, not me :) The Dual Xeon 5355 is about to get a RAC of 3000, i think it will stop at 3600. who? |
![]() ![]() Send message Joined: 2 Jan 06 Posts: 111 Credit: 1,389,340 RAC: 0 |
|
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
I was thinking awhile back about how much a dual quad core would cost to build for a crunching box. What would it cost to build a serious cruncher? The dual quad core xeon machine can be put together for 4500$ .. I know it is expensive, but it is a killer who? |
MikeMarsUK Send message Joined: 15 Jan 06 Posts: 121 Credit: 2,637,872 RAC: 0 |
If your constraint is 'most crunch for $" rather than "most crunch per box", you'd be better off with several overclocked quad core2s boxes instead for the same $4,500 ... i.e., like Who?'s quad system rather than his octo system. PS we're experimenting with an SSE2 version of the climate model using Intel Fortran 9.1, will be a month or two until we know whether it gives the same climate physics as the non-SSE2. The previous time SSE2 was tried it gave different results, but looking better this time round. ![]() ![]() |
![]() Send message Joined: 2 Apr 06 Posts: 213 Credit: 1,366,981 RAC: 0 |
If your constraint is 'most crunch for $" rather than "most crunch per box", you'd be better off with several overclocked quad core2s boxes instead for the same $4,500 ... i.e., like Who?'s quad system rather than his octo system. you can buy several nissan and drive them in parallel, or you can buy a lambo ... bragging right gets different with several nissan... and if you use more machine, your density of computing will be terrible .. a lot of space, and your power usage will be almost double ... your idea sound like a good idea, but if you considere that people doing processing only pay attention to processing power density and electric power, that is a very bad idea. Based on this, the 8 Cores machine is the best choice.
well, SSE2 did not try to do anything, the programmer who did try to generate SSE2 didn t properly configure the SSE2 setting for x87 compatibility ... that is reality, SSE2 does to try by its own to change the calculation mode. the computers don t do errors, the errors comes from the operators! who? |
Message boards :
Number crunching :
who will be the next king of the hill?
©2024 University of Washington
https://www.bakerlab.org