| Author |
Message |
|---|
|
|
|
I thought of this thread being useful for users.
Please post the following
CPU; OS; app; runtime in seconds
AP26 should be run in the following manner (on Linux/Unix)
time ./AP26 366384 366384 0
i start:
Sun UltraSPARC-T1 1000 MHz; Solaris 5.10 (64bit); AP26-1.0; 6329 s
Fujitsu SPARC64 V 1080 MHz; Solaris 5.9 (64bit); AP26-1.0; 2038 s
Intel Xeon E5504 2000 MHz; RHEL 5.3 (64bit); AP26-1.03; 309 s
Intel C2Q Q9550 2833 MHz; RHEL 5.3 (64bit); AP26-1.03; 228 s
The UltraSPARC-T1 is severly hampered by its design using this application, only having one FPU doesn't help him either. I only ported the app to SPARC/Solaris and did no fancy optimization.
I think it is obvious why the SPARC/Solaris-app was not made public. ;)
Another interesting thing is that the single-thread performance of the Nehalem is only on par with the C2Q, maybe Hyperthreading helps a bit here. I do not think the fact that my Nehalem is an engineering sample has any influence on the speed.
I would be pleased if you post your results too, especially the ones from a PS3 or CellBlade. ;)
____________
 |
|
|
|
|
|
Average over 8 AP26 Work Units:
AMD 64 3000+ (2100 MHz); Ubuntu 64bit (Wubi); AP26-BOINC/PG; 887 s
|
|
|
|
|
|
Average over 3 tasks (stopped doing these when I found out it's not efficient on 32-bit OS, but I do have a 64-bit CPU, so I'll probably start again once I have an appropriate OS)
Intel C2D E7200 @ 2.53GHz; Win XP Pro SP2 (32-bit); AP26-BOINC/PG; 1591 s
____________
 |
|
|
 mfl0p Project administrator
Volunteer developer
Send message
Joined: 5 Apr 09
Posts: 212
ID: 38042
Credit: 828,795,534
RAC: 230,901
                     
|
|
The UltraSPARC T1's slowness doesn't come from a weak FPU. It's the same thing that causes the Cell/B.E. to be "slow" at AP26, lack of out-of-order execution processing.
With the current single threaded PG app, PS3 and Cell Blades clocked @ 3.2ghz will complete the test range in 617 seconds. However, the platforms can run 6 and 16 threads, respectively, with no slowdown.
If Rytis decides to release the new apps I have built for Win32/Win64/Linux32 this thread will see some major changes in times. We'll see when he's back in town.
____________
|
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2080
ID: 1178
Credit: 8,000,394,762
RAC: 7,045,892
                                     
|
|
I compiled this list from August to a couple of weeks ago, though I did not include the app number (and since most of the single cores were replaced in August, I cannot add it now--lack of access also means that I cannot do the standard test proposed in the first post). All results are based on an average of at least 10 work units (some up to 20). I also have similar lists for some of the other fixed credit applications (e.g., SGS, 321 Sieve, PSP sieve) if there is interest.
i7 (w/ HT)
Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz [Intel64 Family 6 Model 26 Stepping 5]; Vista HP (64-bit); 1439s
Core2
Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz [x86 Family 6 Model 15 Stepping 11]; Vista HP (32-bit); 1605s
Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz [x86 Family 6 Model 23 Stepping 10]; Win7 Ultimate (64-bit); 881s
Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz [x86 Family 6 Model 23 Stepping 10]; XP Pro (32-bit); 1325s
Intel(R) Core(TM)2 Duo CPU E6750 @ 2.66GHz [x86 Family 6 Model 15 Stepping 11]; XP Pro (32-bit); 1455s
Intel(R) Core(TM)2 Duo CPU T8100 @ 2.10GHz [x86 Family 6 Model 23 Stepping 6]; Vista HP (32-bit); 1873s
Intel(R) Core(TM)2 CPU T5600 @ 1.83GHz [x86 Family 6 Model 15 Stepping 6]; XP Home (32-bit); 2262s
Athlon 64
AMD Athlon(tm) 64 X2 Dual Core Processor 5600+ [x86 Family 15 Model 67 Stepping 3]; Vista HP (32-bit); 1657s
AMD Athlon(tm) 64 X2 Dual Core Processor 4200+ [x86 Family 15 Model 43 Stepping 1]; XP Media Ctr (32-bit); 2154s
AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ [x86 Family 15 Model 43 Stepping 1]; ; XP Home (32-bit); 2309s
Core Duo
Intel(R) Core(TM) Duo CPU T2450 @ 2.00GHz [x86 Family 6 Model 14 Stepping 12]; Vista HP (32-bit); 2313s
Pentium M
Intel(R) Pentium(R) M processor 1.73GHz [x86 Family 6 Model 13 Stepping 8]; XP Pro (32-bit); 2759s
Intel Atom (w/ HT)
Intel(R) Atom(TM) CPU N270 @ 1.60GHz [x86 Family 6 Model 28 Stepping 2]; XP Home (32-bit); 6959s
Pentium D (HT on 3.73.EE edition)
Intel(R) Pentium(R) D CPU 3.73GHz [x86 Family 15 Model 6 Stepping 4]; XP Pro (32-bit); 3101s
Intel(R) Pentium(R) D CPU 3.00GHz [x86 Family 15 Model 4 Stepping 4] ; XP Pro (32-bit); 2590s
Pentium 4 (Prescott w/ HT)
Intel(R) Pentium(R) 4 CPU 3.80GHz [x86 Family 15 Model 4 Stepping 10]; XP Pro (32-bit); 3098s
Intel(R) Pentium(R) 4 CPU 3.60GHz [x86 Family 15 Model 4 Stepping 3]; XP Pro (32-bit); 3250s
Intel(R) Pentium(R) 4 CPU 3.60GHz [x86 Family 15 Model 4 Stepping 1] ; XP Pro (32-bit); 3288s
Intel(R) Pentium(R) 4 CPU 3.40GHz [x86 Family 15 Model 4 Stepping 1] ; XP Pro (32-bit); 3456s
Intel(R) Pentium(R) 4 CPU 3.20GHz [Family 15 Model 3 Stepping 4]; Ubuntu 9.04 (32-bit); 3507s
Pentium 4 (Cedar Mill w/ HT)
Intel(R) Pentium(R) 4 CPU 3.00GHz [x86 Family 15 Model 6 Stepping 2]; XP Pro (32-bit); 3875s
Pentium 4 (Northwood w/ HT)
Intel(R) Pentium(R) 4 CPU 3.00GHz [x86 Family 15 Model 2 Stepping 9]; XP Pro (32-bit); 4836s
Intel(R) Pentium(R) 4 CPU 3.06GHz [x86 Family 15 Model 2 Stepping 7] ; XP Pro (32-bit); 4976s
Intel(R) Pentium(R) 4 CPU 2.60GHz [x86 Family 15 Model 2 Stepping 9] ; XP Pro (32-bit); 5664s
Pentium 4 (Northwood)
Intel(R) Pentium(R) 4 CPU 2.80GHz [x86 Family 15 Model 2 Stepping 7] ; XP Pro (32-bit); 3470s
Intel(R) Pentium(R) 4 CPU 2.53GHz [x86 Family 15 Model 2 Stepping 7] ; XP Pro (32-bit); 4050s
Intel(R) Pentium(R) 4 CPU 2.40GHz [x86 Family 15 Model 2 Stepping 7]; XP Pro (32-bit); 4159s
Intel(R) Pentium(R) 4 CPU 2.00GHz [x86 Family 15 Model 2 Stepping 4] ; XP Pro (32-bit); 4944s
____________
141941*2^4299438-1 is prime!
 |
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2080
ID: 1178
Credit: 8,000,394,762
RAC: 7,045,892
|
If Rytis decides to release the new apps I have built for Win32/Win64/Linux32 this thread will see some major changes in times. We'll see when he's back in town.
If/When he does, I'll reproduce the above table (as much as I can) for direct comparisons.
____________
141941*2^4299438-1 is prime!
|
|
|
RytisVolunteer moderator
Project administrator
Send message
Joined: 22 Jun 05
Posts: 2648
ID: 1
Credit: 21,592,315
RAC: 88
               
|
|
The apps are live now.
____________
|
|
|
|
|
|
Thx, cool!
____________
|
|
|
|
|
|
Here's results so far with AP26 1.04:
Intel C2D E7200 @ 2.53GHz; Win XP Pro SP2 (32-bit); 978 s ( -613s from previous test, 39% faster )
Pentium 4 3.06GHz; WinXP Pro SP2 (32-bit); 2584.6 s
One thing to note; these are running in BOINC, so it's running one WU per core (or thread, in the case of the P4 with HTT)
____________
|
|
|
|
|
|
Got around to installing a 64-bit linux, so here's the new AP26 runtimes
Intel C2D E7200 @ 2.53GHz; Ubuntu 9.04 (64-bit); 530 s
____________
|
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2080
ID: 1178
Credit: 8,000,394,762
RAC: 7,045,892
|
|
Okay...here are some early results for the new version 1.04 application on several of the machines. For most machines, the new AP26 app is about 35%-45% faster. The two exceptions are the Pent M/Core Duo machines (23%-29% faster) and the Athlon x2 on 64-bit (7% faster).
i7 (w/ HT)
Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz [Intel64 Family 6 Model 26 Stepping 5]; Vista HP (64-bit); 919s
Core2
Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz [x86 Family 6 Model 15 Stepping 11]; Vista HP (32-bit); 1010s
Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz [x86 Family 6 Model 23 Stepping 10]; XP Pro (32-bit); 793s
Intel(R) Core(TM)2 Duo CPU E6750 @ 2.66GHz [x86 Family 6 Model 15 Stepping 11]; XP Pro (32-bit); 893s
Intel(R) Core(TM)2 Duo CPU T8100 @ 2.10GHz [x86 Family 6 Model 23 Stepping 6]; Vista HP (32-bit); 1146s
Athlon 64
AMD Athlon(tm) 64 X2 Dual Core Processor 5600+ [x86 Family 15 Model 67 Stepping 3]; Win7 Enterprise (64-bit); 755s
Core Duo
Intel(R) Core(TM) Duo CPU T2450 @ 2.00GHz [x86 Family 6 Model 14 Stepping 12]; Vista HP (32-bit); 1643s
Pentium M
Intel(R) Pentium(R) M processor 1.73GHz [x86 Family 6 Model 13 Stepping 8]; XP Pro (32-bit); 2126s
Pentium D (HT on 3.73.EE edition)
Intel(R) Pentium(R) D CPU 3.73GHz [x86 Family 15 Model 6 Stepping 4]; XP Pro (32-bit); 2054s
Pentium 4 (Prescott w/ HT)
Intel(R) Pentium(R) 4 CPU 3.80GHz [x86 Family 15 Model 4 Stepping 10]; XP Pro (32-bit); 1903s
Intel(R) Pentium(R) 4 CPU 3.60GHz [x86 Family 15 Model 4 Stepping 3]; XP Pro (32-bit); 2122s
Intel(R) Pentium(R) 4 CPU 3.60GHz [x86 Family 15 Model 4 Stepping 1] ; XP Pro (32-bit); 2091s
Intel(R) Pentium(R) 4 CPU 3.40GHz [x86 Family 15 Model 4 Stepping 1] ; XP Pro (32-bit); 2194s
Pentium 4 (Cedar Mill w/ HT)
Intel(R) Pentium(R) 4 CPU 3.00GHz [x86 Family 15 Model 6 Stepping 2]; XP Pro (32-bit); 2434s
Pentium 4 (Northwood w/ HT)
Intel(R) Pentium(R) 4 CPU 3.00GHz [x86 Family 15 Model 2 Stepping 9]; XP Pro (32-bit); 2662s
Intel(R) Pentium(R) 4 CPU 2.40GHz [x86 Family 15 Model 2 Stepping 9] ; XP Pro (32-bit); 3313s
Pentium 4 (Northwood)
Intel(R) Pentium(R) 4 CPU 2.40GHz [x86 Family 15 Model 2 Stepping 7]; XP Pro (32-bit); 2660s
Pentium III (Coppermine)
Intel Pentium III 996MHz [x86 Family 6 Model 8 Stepping 10]; Win2k Pro (32-bit); 3660s
Great Job, Bryan!
____________
141941*2^4299438-1 is prime!
|
|
|
mfl0pProject administrator
Volunteer developer
Send message
Joined: 5 Apr 09
Posts: 212
ID: 38042
Credit: 828,795,534
RAC: 230,901
|
|
Nice test results, thanks for creating the list.
Looking at that list, I have to laugh at the mistake that was Netburst :) |
|
|
|
|
|
Times look much better.
Is there going to be a 1.04 for OSX 10.4+ Intel?
____________
 |
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2080
ID: 1178
Credit: 8,000,394,762
RAC: 7,045,892
|
Looking at that list, I have to laugh at the mistake that was Netburst :)
Yeah, the AP26 app shines a nice blinding light on that directional error.
Differences on other PrimeGrid apps aren't quite as bad, though. On the sieves, the distance between the P4(HT) and Core 2 is somewhat less than the 3 to 1 or so ratio seen on the AP26 app. The Athlon x2's are also only marginally better than the P4 on the sieves. On the SGS LLR, the Core 2 advantage drops to around 2 to 1, and the Athlon x2 advantage completely disappears (e.g., the Prescott 3.4 HT is slightly faster than the Athlon x 4200+ [939-pin variety]).
____________
141941*2^4299438-1 is prime!
|
|
|
|
|
|
AMD Phenom(tm) II X4 955BE Processor [AMD64 Family 16 Model 4 Stepping 2] @ 3.4ghz; Windows 7 Ultimate 64Bit v7600; 541s |
|
|
|
|
|
I had a very curious question that I figured you guys might be able to answer, would there be a way my Ipod Touch 8 gig could Crunch AP26 work units? At all? I mean that would be so cool if there would be a way, I have the docking station and well it works wireless as well too so that would work as well. Any ideas or suggestions?? Thanks so much guys I really appreciate it.
Another question would be if it can't do AP26 because of Boinc, then could it do PRPNET projects?? or like Primorial Sieving?? That would be really cool if it could some how do those. I mean I barely use the thing and it would put it to good use for how much I paid for it if it could do some sort of prime crunching.
____________
John M. Johnson "Novex" |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
 
|
I had a very curious question that I figured you guys might be able to answer, would there be a way my Ipod Touch 8 gig could Crunch AP26 work units? At all? I mean that would be so cool if there would be a way, I have the docking station and well it works wireless as well too so that would work as well. Any ideas or suggestions?? Thanks so much guys I really appreciate it.
Another question would be if it can't do AP26 because of Boinc, then could it do PRPNET projects?? or like Primorial Sieving?? That would be really cool if it could some how do those. I mean I barely use the thing and it would put it to good use for how much I paid for it if it could do some sort of prime crunching.
IIRC, it uses an ARM CPU. You might be able to build genefer, phrot, PRPNet, and some sievers on it, but not pfgw or llr. |
|
|
|
|
|
I haven't noticed any increase in performance for the AP26 work units since the release of version 1.04.
Computer:
Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz [Intel64 Family 6 Model 26 Stepping 4] with Vista 64-bit OS
Average run time is ~26 minutes.
Any ideas as to why this is?
Cheers!
Alan
____________
  |
|
|
mfl0pProject administrator
Volunteer developer
Send message
Joined: 5 Apr 09
Posts: 212
ID: 38042
Credit: 828,795,534
RAC: 230,901
|
|
the workunit length was doubled to 6 K per workunit from 3 K about the time the 1.04 apps were released.
____________
|
|
|
|
|
|
Ah, yes, that makes sense...that is quite sneaky:)
____________
|
|
|
|
|
|
Just to add to the list since there are very few AMD processors listed.
AMD Athlon(tm) 64 Processor 3000+ (2GHz)
[Family 15 Model 44 Stepping 2]
Linux 2.6.31-15-generic (Ubuntu 9.10)
~1775s (~29.5m) per WU based on the average of the last 20 reported WU's
AMD Turion(tm) 64 X2 Mobile Technology TL-58 (1.9GHz)
[x86 Family 15 Model 104 Stepping 2]
Microsoft Windows Vista Home Premium x86 Edition, Service Pack 2 (32Bit)
~3388s (~56.5m) per WU based on the average of the last 20 reported WU's
Since I just did one of those "milestone" birthdays and the holidays are right around the corner (and I've been relatively good this year; really I have!!) I'll be adding one more AMD to this list next week -
And just in time for the Winter Solstice Challenge
A shiny new (oh I just can't wait) -
Phenom II X4 965 BE
____________
There's someone in our head but it's not us. |
|
|
|
|
|
AMD X2 5600 on Mandriva 2009.1
run at around 32 minutes on app 1.04 |
|
|
|
|
|
Core 2 Duo T7700 @ 2.4GHz
Mac OSX 10.5
1665.16 second average over the last 20 completed tasks.
PPC G4 @ 1.42 GHz
Mac OSX 10.4
9804.15 second average over the last 20 completed tasks. |
|
|
|
|
|
checked my amd x64 on windows XP PhenomII 940 OCed to 3.5Ghz
results average around 19.5 minutes |
|
|
|
|
|
So here are the AP26 stats for my new toy.
AMD Phenom II X4 965 Processor (3.4GHz)
[Family 16 Model 4 Stepping 2]
Linux 2.6.31-16-generic (Ubuntu 9.10)
~839s (~14m) per WU based on the average of the last 20 reported WU's
AP26 Ver. 1.04 (as I neglected to note on my previous post)
____________
There's someone in our head but it's not us. |
|
|
|
|
|
Intel Core2 Quad Q9300 @ 3.226 GHz (O/C from 2.5); Win7 7100 (64bit); AP26-1.04; 1395s (avg of last 6 WUs)
AMD Phenom II X2 550 BE @ 3.7 GHz (O/C from 3.1); Win7 7600 (64bit); AP26-1.04; 1470s (avg of last 5 WUs)
Nvidia 8600GT @ 608 Mhz Core, 1350 Mhz Shader, 450 Mhz Memory; Win7 7100 (64bit); AP26-1.0 CUDA; 3259s (avg of last 3 WUs)
These are all the 35.41 credit WUs.
____________
 |
|
|
|
|
|
Intel C2Q 9400, OCed from 2.66 to 3.2 GHz; Win7 Build 7100; AP26 Search v1.04; 1402 Secs Average
& with this baby also running
NVidia 9800GT, 600 MHz Core, 900 MHz Memory, 1500 MHz Shader; ForceWare 195.62; AP26 Search v1.00 (cuda23); 1196 Secs Average |
|
|
|
|
|
Intel Atom Z530 @1.6 GHz - XP32-embedded:
141763432 97329849 19 Dec 2009 10:17:59 UTC 19 Dec 2009 16:12:27 UTC Completed and validated 18,996.56
141763404 97329821 19 Dec 2009 10:17:59 UTC 19 Dec 2009 16:12:27 UTC Completed and validated 19,222.97 |
|
|
