Message boards :
Number crunching :
SETI@home Processor Guide - Version 0.4
Message board moderation
Author | Message |
---|---|
Luke Send message Joined: 31 Dec 06 Posts: 2546 Credit: 817,560 RAC: 0 |
This is in its infancy, but I feel a need to collaborate all this information since it is a mess all around the internet, and I plan on adding a few more statistics. Eventually, I'd like to be able to compare RAC to the dollar ratios for all CPU's. SETI@home Processor Guide I may create a GPU guide as well, but that's later down the road. Right now, I need ideas. Give me more columns, more processors, RAC stats, etc. Post here with feedback. Remember, it's still in its infancy, and I will update it tomorrow. - Luke. - Luke. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Another factor to figure in when trying to figure cost to crunch is power consumption. Comparing the i860 & the i930. The i930 uses about 30% more power, but does it give you 30% more performance. I wonder if having average proccess times for the tasks would be good to include. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
j mercer Send message Joined: 3 Jun 99 Posts: 2422 Credit: 12,323,733 RAC: 1 |
Another factor to figure in when trying to figure cost to crunch is power consumption. I feel that power consumption needs to be measured at the wall socket with the many different computer configurations in use today. As comparison my OEM i7 975 XE notebook with GTX 280M & RAID-0 3 X 80GB Intel SATA2 SS consumes about 235 watts. Now compare that to a OC’d i7 920 desktop with HHD & liquid cooling. I have no idea but guessing I would think maybe around 750 watts consumption. I was hoping the “Credit per Watt†thread and spreadsheet would take on a life of its own so we can see some actual costs. Measuring the bang for your buck can be tricky when comparing an OEM to custom owner built rigs. I also understand that many folks are not going to be comfortable with sharing this kind of information. Rightfully so. But if just a few do... ... |
Luke Send message Joined: 31 Dec 06 Posts: 2546 Credit: 817,560 RAC: 0 |
I have just updated the sheet to version 0.1 Changes include: Addition of Intel Core i5 & mobile Core i7 Created "Socket" column Added RAC per dollar data for the i7 975 XE Another factor to figure in when trying to figure cost to crunch is power consumption. Comparing the i860 & the i930. The i930 uses about 30% more power, but does it give you 30% more performance. Hmmm... I'm thinking most of that could be hard to implement. The process times especially since WU's differ in size. - Luke. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Another factor to figure in when trying to figure cost to crunch is power consumption. It would be nice if the cpu makers would state the values for the processors at fulll oad & idle in their general info. Often people use the TDP number, myself included, as the power consumption. When it is really the amount of heat the cooling system needs to handle to keep the processor within it's operating limits. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
archae86 Send message Joined: 31 Aug 99 Posts: 909 Credit: 1,582,816 RAC: 0 |
It would be nice if the cpu makers would state the values for the processors at fulll oad & idle in their general info.They won't, I'd guess. Even back when we did, the guaranteed maximum was the only value with any integrity behind it. We planned in advance to meet it, and tested the units as well. Anything on the data sheet marked "typical" often came straight from marketing. Internally there was not even the pretense that it really represented the center of the product distribution. Because so much of the idle power is leakage, it is especially variable part to part. Since even the active power has a major leakage component, power dissipation is far less predictable sample to sample than in the old days. If you care about power (and I do), you just need to figure on measuring the system yourself at the wall socket. I confess I've not gone the next step and measured at the UPS input. Since the box is not even warm to the touch, it can't be burning much. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
I have just updated the sheet to version 0.1 Yeah it might be hard to do that. Thi will amke it intrestings to see how the CPUs scale for processing S@H. I've noticed since I have 5 systems with old P4's from 3.0GHz to 3.8GHz their processing ability is not as linear as I would have first thought. 2.0GHz RAC ~280 (Northwood) 2.4GHz RAC ~350 (Northwood) 3.0GHz RAC ~600 (Northwood) 3.2GHz RAC ~700 (Northwood) 3.4GHz RAC ~800 (Prescott) 3.6GHz RAC ~830 (Prescott) 3.8GHz RAC ~870 (Prescott) SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Reuben Gathright Send message Joined: 8 Mar 01 Posts: 213 Credit: 14,594,579 RAC: 0 |
You must also take into account the memory bus on the motherboard that you are testing folks. If you think SETI is slow on a dual channel DDR 400Mhz board, try single channel! Overclock with the MSI G31M3-L and Intel E8600 3.33Ghz Intel D865GLC Socket 478 Motherboard ~How To Overclock The Eee ASUS 1005HA Netbook To 1.9Ghz~ |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
My i7 with a GTX 275 and GTX 260 and triple channel memory, H50 with extra fan and 3 case fans draws 4 amps and my voltage measured is at 120V so my machine is using 480 watts from the wall. I slit the power cable and put my true RMS amp probe on it and then checked the voltage where it plugs in. I use a corsair 750 PS. Official Abuser of Boinc Buttons... And no good credit hound! |
Luke Send message Joined: 31 Dec 06 Posts: 2546 Credit: 817,560 RAC: 0 |
I've just had the chance in between building my new desk to update and release version 0.2. Changes Include:
All desktop Nehalem products now listed, including unreleased processors. Creation of a "Die Size" column Started work on "Core / Penryn" & "Sandy Bridge / Ivy Bridge" - change between sheets as needed. Addition of RAC per dollar for all Core i7 products (except those unreleased & the 870) Help Needed:
Find Die Size for all mobile Core i7 processors Find SETI@home RAC for Core i7 870 All other suggestions and comments are welcome... - Luke. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
I've just had the chance in between building my new desk to update and release version 0.2. Are you using http://processorfinder.intel.com for information? SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Luke Send message Joined: 31 Dec 06 Posts: 2546 Credit: 817,560 RAC: 0 |
I've just had the chance in between building my new desk to update and release version 0.2. Nope. I'm using a variety of reliable sources. Processor finder doesn't seem to list up and coming chips. - Luke. |
gizbar Send message Joined: 7 Jan 01 Posts: 586 Credit: 21,087,774 RAC: 0 |
Hey Luke! Don't forget that there's Team Green too! Not as prolific at the moment as Team Blue, but it is there, and I wouldn't be able to type this message to you if it wasn't! regards, Gizbar. A proud GPU User Server Donor! |
S@NL - eFMer - efmer.com/boinc Send message Joined: 7 Jun 99 Posts: 512 Credit: 148,746,305 RAC: 0 |
The Dutch Seti did a real test. http://www.seti.nl/cputijden_boinc.php The best way to go is to use the same WU on all computers. To get the same meaningful results. In other words a benchmark WU. On problem is newer CPU's go faster when there is only one WU running. So running more WU at the exact same time is needed. |
Luke Send message Joined: 31 Dec 06 Posts: 2546 Credit: 817,560 RAC: 0 |
Hey Luke! Ah... AMD. The younger child. Personally, I side with Intel. I no nothing about AMD except for some basic names, Opteron, Athlon, Phenom. I will look into adding them somehow. Thanks Gizbar. - Luke. |
Luke Send message Joined: 31 Dec 06 Posts: 2546 Credit: 817,560 RAC: 0 |
|
Luke Send message Joined: 31 Dec 06 Posts: 2546 Credit: 817,560 RAC: 0 |
I've had some time to 'release' another update. No new processors however. Changes in 0.2.2: Added release date, life cycle stage, release price and current price to all Core 2 Duo Allendale processors Addition of performance data for Core i5 750 & Core i7 870. Created column "Performance Test v7 CPU Mark" If you have any of the processors listed... could you be so kind to suspend S@H and download PassMark Performance Test v7. Then go to Tests > CPU > Run "All" and post back with your average CPU Mark (displayed after the tests and listed as the first item). Thanks folks. All other suggestions are welcome. - Luke. - Luke. |
Smariga Send message Joined: 13 Jun 99 Posts: 49 Credit: 30,454,070 RAC: 0 |
Core i7 860 oc'd to 3360mhz - CPU Mark 6798.5 |
Niko Send message Joined: 14 Dec 09 Posts: 123 Credit: 70,041 RAC: 0 |
Intel i5 750 @ 2.67GHz CPU Mark: 3,736.0 |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
i7 920 @ 3.2 CPUmark 5713 With HT off if that matters. EDIT: Default settings HT on only CPUmark 4802, I see I need a hard drive desperately mine barely shows in the comparison, my CDRom is pretty weak also. Official Abuser of Boinc Buttons... And no good credit hound! |
©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.