mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > No Prime Left Behind

Reply
 
Thread Tools
Old 2010-07-21, 20:13   #1
Oddball
 
Oddball's Avatar
 
May 2010

499 Posts
Default LLR benchmarks

I'm thinking of replacing my old Pentium 4 processor, so could you run some or all of these tests and tell me how long it takes to complete?

1.) k value between 100 and 200, n value between 500,000 and 501,000
2.) k value between 100 and 200, n value between 600,000 and 601,000
3.) k value between 300 and 2000, n value between 500,000 and 501,000
4.) k value between 300 and 2000, n value between 600,000 and 601,000
5.) k value above 1M, n value between 500,000 and 501,000

Any chip you have will do, but I'm especially interested in the AMD Phenom II X6 1055T and the Intel core i7 920. If you're not running your comps on stock settings, please say what your overclock is.

Thanks in advance.
Oddball is offline   Reply With Quote
Old 2010-07-21, 23:02   #2
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

186916 Posts
Default

My fastest processor is a Core 2 Quad Q6600, which I don't believe is even sold any more so that won't be of much help, but I know that Gary (a.k.a. gd_barnes on this forum, my co-admin at NPLB and CRUS) has an i7 920 running at stock settings. He's on a camping trip right now until Friday but when he gets back I'll direct him to this thread. (Edit: duh, sorry, I thought this was in the TPS forum. Since it's in the NPLB forum, Gary will surely see it right off the bat when he gets back without needing my direction anyway. )

He also has a number of older AMD Phenom X4's that he's been able to compare not only with his i7 but with a few Q6600's like mine. He found that the Intels were a LOT faster per-core than the AMDs, enough so that he's not buying any more AMDs. It's not so huge a difference that two of his AMDs won't beat one Q6600, but at the time the price difference was not enough to make it worthwhile to go for the AMD. However, it's possible that the difference now of 2 extra cores from the AMD (for ~$100 less, to boot) is enough to make them approximately even in terms of overall crunching ability. Also note that this comparison was with both manufacturers' previous generation of CPUs, the original Phenom vs. the Core 2, so it may not be completely applicable to the comparison of Phenom II vs. Core i7.

One thing you may want to consider: there are now other, faster i7's available for cheaper prices than the 920 that have since come out. When I go here and sort by lowest price, I see that you can get a 2.93 GHz i7 870 (specifically, this) for about $5 less than the i7 920 at 2.66 GHz. I see no other substantial difference between the CPUs, so it seems that the 870 is definitely faster across the board. I would use that as your comparison referential for the Intel side.

On the AMD side, I'm seeing only two options currently for Phenom II X6: the 1055T and 1090T, the latter of which is only a little faster than the former yet costs $100 more. So the 1055T is definitely the way to go on the AMD side.

I believe there are at least a few members of this forum in posession of a Phenom II X6 1055T, and while I'm not sure of the same for the i7 870, even the i7 920 should be a decent comparison point (because the 870 is only going to be faster than that).

Last fiddled with by mdettweiler on 2010-07-21 at 23:09
mdettweiler is offline   Reply With Quote
Old 2010-07-21, 23:17   #3
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

186916 Posts
Default

One other thing you may want to consider is an Intel i5 quad-core instead of an i7. The only difference between the two is that the i5's don't have hyperthreading, and that they're about $100 cheaper--bringing them pretty close to the price of the AMD X6 you're looking at. From the testing Gary's done on his i7 920, he's found that hyperthreading as implemented on the i7 is a sizeable boost for sieving, but not so much for LLRing. So if LLRing is to be the machine's primary task, the i5 provides much better "bang for the buck".

The one I would currently recommend for comparison with the aforementioned i7 870 and Phenom II 1055T is the i5 760. It's clocked a little lower than the i7 870 but shouldn't be much slower. If you want to try overclocking, you can probably actually get them both up to similar levels anyway.
mdettweiler is offline   Reply With Quote
Old 2010-07-22, 04:28   #4
Oddball
 
Oddball's Avatar
 
May 2010

7638 Posts
Default

Quote:
Originally Posted by mdettweiler View Post
One thing you may want to consider: there are now other, faster i7's available for cheaper prices than the 920 that have since come out. When I go here and sort by lowest price, I see that you can get a 2.93 GHz i7 870 (specifically, this) for about $5 less than the i7 920 at 2.66 GHz. I see no other substantial difference between the CPUs, so it seems that the 870 is definitely faster across the board. I would use that as your comparison referential for the Intel side.
I see. I was pretty sure that the i7 870 was in the $500 range when I checked last week, but I guess Intel did a price cut over the weekend. So the 870 will replace the 920 as one of the processors I'm looking at.

Quote:
My fastest processor is a Core 2 Quad Q6600
Just out of curiosity, how's the multicore efficiency for your Q6600? Lets say that if three cores are idle and only one core is doing the work, a test takes 100 seconds to complete. But if all 4 cores are running one test each, how long would it take for those 4 tests to complete? Will it be in the 100-105 second range, the 150-160 second range, or something else?
Oddball is offline   Reply With Quote
Old 2010-07-22, 07:54   #5
henryzz
Just call me Henry
 
henryzz's Avatar
 
"David"
Sep 2007
Cambridge (GMT/BST)

2·2,861 Posts
Default

Quote:
Originally Posted by Oddball View Post
Just out of curiosity, how's the multicore efficiency for your Q6600? Lets say that if three cores are idle and only one core is doing the work, a test takes 100 seconds to complete. But if all 4 cores are running one test each, how long would it take for those 4 tests to complete? Will it be in the 100-105 second range, the 150-160 second range, or something else?
If you get good memory(at least 1066Mhz DDR2 which is now gettting more and more pricey) it isn't too bad although intel cpus post core 2 are much better because of a built in memory controller. I really wouldn't suggest getting one though as you can get a Athlon II 620 x4 cheaper(plus with now cheaper DDR3 memory). My dad's Athlon II 620 x4 @ stock 2.5 Ghz is slightly faster at llring and amds are usually better at sieving. Plus it has a built in memory controller so running all four cores shouldn't be as much of a slowdown as a Q6600. If you are interested in overclocking then my 620 overclocks further than the Q6600 although the Q6600 has a rubbish motherboard and the 620 a good one. These really are at the cheaper end of the market now though. Most people here get i5/7s these days I think.

edit: Could you create a file with all the tests you want run in it? I will then run it on both pcs. I would suggest working on a sieving benchmark as well.

Last fiddled with by henryzz on 2010-07-22 at 08:04
henryzz is online now   Reply With Quote
Old 2010-07-22, 15:36   #6
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

624910 Posts
Default

Quote:
Originally Posted by henryzz View Post
If you get good memory(at least 1066Mhz DDR2 which is now gettting more and more pricey) it isn't too bad although intel cpus post core 2 are much better because of a built in memory controller. I really wouldn't suggest getting one though as you can get a Athlon II 620 x4 cheaper(plus with now cheaper DDR3 memory). My dad's Athlon II 620 x4 @ stock 2.5 Ghz is slightly faster at llring and amds are usually better at sieving. Plus it has a built in memory controller so running all four cores shouldn't be as much of a slowdown as a Q6600. If you are interested in overclocking then my 620 overclocks further than the Q6600 although the Q6600 has a rubbish motherboard and the 620 a good one. These really are at the cheaper end of the market now though. Most people here get i5/7s these days I think.

edit: Could you create a file with all the tests you want run in it? I will then run it on both pcs. I would suggest working on a sieving benchmark as well.
Hmm, that's interesting to hear. I didn't know an Athlon II is actually slightly faster than a Q6600 now. In that case, the speed difference between the Phenom II X6 and the i7 870 may not be as big as I was thinking. It's looking more and more like the Phenom II X6, due to its two more cores for less money, may be the way to go.

Oddball, in the absence of a Phenom II X6 benchmark, henryzz's Athlon II X4 should provide a roughly similar benchmark. They're both of the same underlying architecture (AMD K10.5), so it should at least give you a rough idea of where it is relative to Intel i-architecture cores.
mdettweiler is offline   Reply With Quote
Old 2010-07-22, 15:53   #7
henryzz
Just call me Henry
 
henryzz's Avatar
 
"David"
Sep 2007
Cambridge (GMT/BST)

572210 Posts
Default

Quote:
Originally Posted by mdettweiler View Post
Hmm, that's interesting to hear. I didn't know an Athlon II is actually slightly faster than a Q6600 now. In that case, the speed difference between the Phenom II X6 and the i7 870 may not be as big as I was thinking. It's looking more and more like the Phenom II X6, due to its two more cores for less money, may be the way to go.

Oddball, in the absence of a Phenom II X6 benchmark, henryzz's Athlon II X4 should provide a roughly similar benchmark. They're both of the same underlying architecture (AMD K10.5), so it should at least give you a rough idea of where it is relative to Intel i-architecture cores.
Please bear in mind that that is 2.5 Ghz v 2.4 Ghz but yes the Athlon II 620 x4 is slightly faster than the Q6600. Apparently the Athlon II suffers because of its lack of L3 cache but that doesn't affect our size of tests much just GIMPS.
BTW I think the Phenom only had the SSE2 slowness only partly fixed Phenom II was the first to be completely fixed by amd so thats why Gary's phenoms are so slow. He also might consider amd again now if he wants even more pcs in the future. Its £83(127$) for a 2.9 Ghz Athlon II 635 at my local shop right now(they have got so cheap they have stopped selling the 620). Thats a very reasonable price for a quad.

Last fiddled with by henryzz on 2010-07-22 at 15:54
henryzz is online now   Reply With Quote
Old 2010-07-22, 16:02   #8
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3×2,083 Posts
Default

Quote:
Originally Posted by henryzz View Post
Please bear in mind that that is 2.5 Ghz v 2.4 Ghz but yes the Athlon II 620 x4 is slightly faster than the Q6600. Apparently the Athlon II suffers because of its lack of L3 cache but that doesn't affect our size of tests much just GIMPS.
BTW I think the Phenom only had the SSE2 slowness only partly fixed Phenom II was the first to be completely fixed by amd so thats why Gary's phenoms are so slow. He also might consider amd again now if he wants even more pcs in the future. Its £83(127$) for a 2.9 Ghz Athlon II 635 at my local shop right now(they have got so cheap they have stopped selling the 620). Thats a very reasonable price for a quad.
Fascinating--I didn't know that about the Phenoms and SSE2, but that would go a long way towards explaining why Gary's AMDs are so much slower than his Intels. It would also jive with what I've heard elsewhere on the forum about Phenom IIs being a little slower, but not hugely so, than i5/i7's.
mdettweiler is offline   Reply With Quote
Old 2010-07-27, 06:07   #9
Oddball
 
Oddball's Avatar
 
May 2010

1F316 Posts
Default

Quote:
Originally Posted by henryzz View Post
edit: Could you create a file with all the tests you want run in it? I will then run it on both pcs.
Sorry for the delay in responding, but here are the tests I'd like you to run:

11819883*2^390000-1
20193*2^480000-1
309*2^1000043-1
Oddball is offline   Reply With Quote
Old 2010-07-27, 06:23   #10
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

52×173 Posts
Default

Oddball-
I have a core2 quad 2.66 (8400? i forget model number) overclocked to 3.2Ghz with DDR2-800 running at 800. 4 cores testing 128k FFT or smaller is less than 1% slower per test than a single core running.

For small FFTs (I would guess 256k or smaller), there is zero hit to performance running 4 copies of LLR on core2.
-Curtis

Last fiddled with by VBCurtis on 2010-07-27 at 06:23
VBCurtis is offline   Reply With Quote
Old 2010-08-06, 14:00   #11
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

186916 Posts
Default

Oddball, I saw over in the TPS variable-n sieve thread that you mentioned your new computer--just curious, which one did you end up getting?
mdettweiler is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Benchmarks MurrayInfoSys Information & Answers 3 2011-04-14 17:10
benchmarks Unregistered Information & Answers 15 2009-08-18 16:44
Benchmarks for i7 965 lavalamp Hardware 21 2009-01-06 04:32
Benchmarks for 24.12 Prime95 Software 60 2005-06-11 07:35
Benchmarks Vandy Hardware 6 2002-10-28 13:45

All times are UTC. The time now is 13:35.

Thu Oct 1 13:35:25 UTC 2020 up 21 days, 10:46, 2 users, load averages: 2.23, 2.06, 2.03

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, 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.