Hardware Benchmark Jest Thread for 100M exponents

[ewmayer]

Originally Posted by retina

There are more than 100 countries in the world. And many don't have presidents.

And over 100 countries have U.S. military bases of one kind or another, surely featuring multiple official portraits of the grinning or glowering, amused or bemused visage [depending on administration and year] of a 4-year-term president from a certain exceptional™ nation. USA! USA! USA! :P

(Now can we please get back onto the topic of world domination not via military hegemony but by way of homebrew cellphone clusters? ;)

[retina]

Originally Posted by kriesel

Hmm, here's a contender for a lowly minion:

[Jan 30 16:48] Worker starting
[Jan 30 16:48] Starting primality test of M332220523 using FFT length 18M, Pass1=2K, Pass2=9K, clm=4
[Jan 30 16:50] Iteration: 10 / 332220523 [0.00%], ms/iter: 5805.210, ETA: 22321d 20:54
[Jan 30 16:52] Iteration: 20 / 332220523 [0.00%], ms/iter: 14515.949, ETA: 55815d 22:12

MPC Transport T3000 nominally 2.1Ghz Pentium M, clocks in at ~800Mhz in prime95 29.4b7 win32 for some reason, estimated 153. years to complete.

Waah

Why the large change in iteration time? Is it throttling with high temps? My previous test was on a 1300Mhz (800MHz less than yours) but the cooling was working perfectly so it never throttled. Theoretically your 2100MHz should still beat me. Maybe time to clean and oil your dusty fan and retry the test.

[kriesel]

Originally Posted by retina

Waah

Why the large change in iteration time? Is it throttling with high temps? My previous test was on a 1300Mhz (800MHz less than yours) but the cooling was working perfectly so it never throttled. Theoretically your 2100MHz should still beat me. Maybe time to clean and oil your dusty fan and retry the test.

Probably the overhead of TightVNC remote desktop service. But there is timing fluctuation in both directions while the system is running undisturbed overnight in ~62F environment. This ancient MPC T3100 is not really a minion, more of a labor camp inmate, being run till it fails with minimal maintenance. I also have a T3000 and it is uh unique too. They each slowed considerably when I added the second GB SODIMM for P-1 memory space. There are timing fluctuations in both directions.

[Jan 31 05:03] M90552467 stage 1 is 66.28% complete. Time: 7.586 sec.
[Jan 31 05:03] M90552467 stage 1 is 66.28% complete. Time: 7.659 sec.
[Jan 31 05:03] M90552467 stage 1 is 66.28% complete. Time: 7.676 sec.
[Jan 31 05:03] M90552467 stage 1 is 66.28% complete. Time: 7.624 sec.
[Jan 31 05:03] M90552467 stage 1 is 66.28% complete. Time: 10.103 sec.
[Jan 31 05:03] M90552467 stage 1 is 66.28% complete. Time: 3.827 sec.
[Jan 31 05:03] M90552467 stage 1 is 66.28% complete. Time: 7.601 sec.
[Jan 31 05:03] M90552467 stage 1 is 66.28% complete. Time: 7.625 sec.
[Jan 31 05:04] M90552467 stage 1 is 66.28% complete. Time: 7.618 sec.
[Jan 31 05:04] M90552467 stage 1 is 66.28% complete. Time: 7.641 sec.
[Jan 31 05:04] M90552467 stage 1 is 66.29% complete. Time: 7.626 sec.
...
[Jan 31 05:50] M90552467 stage 1 is 66.66% complete. Time: 7.625 sec.
[Jan 31 05:50] M90552467 stage 1 is 66.66% complete. Time: 7.612 sec.
[Jan 31 05:50] M90552467 stage 1 is 66.66% complete. Time: 13.630 sec.
[Jan 31 05:50] M90552467 stage 1 is 66.66% complete. Time: 7.612 sec.
[Jan 31 05:50] M90552467 stage 1 is 66.66% complete. Time: 7.672 sec.
[Jan 31 05:50] M90552467 stage 1 is 66.66% complete. Time: 7.668 sec.
[Jan 31 05:50] M90552467 stage 1 is 66.66% complete. Time: 7.636 sec.
...
[Jan 31 05:56] M90552467 stage 1 is 66.71% complete. Time: 7.631 sec.
[Jan 31 05:56] M90552467 stage 1 is 66.71% complete. Time: 7.654 sec.
[Jan 31 05:56] M90552467 stage 1 is 66.71% complete. Time: 4.881 sec.
[Jan 31 05:56] M90552467 stage 1 is 66.71% complete. Time: 3.610 sec.
[Jan 31 05:56] M90552467 stage 1 is 66.71% complete. Time: 7.176 sec.
[Jan 31 05:56] M90552467 stage 1 is 66.71% complete. Time: 7.591 sec.
[Jan 31 05:56] M90552467 stage 1 is 66.71% complete. Time: 7.655 sec.
[Jan 31 05:56] M90552467 stage 1 is 66.72% complete. Time: 7.583 sec.
[Jan 31 05:57] M90552467 stage 1 is 66.72% complete. Time: 7.619 sec.
...
[Jan 31 06:49] M90552467 stage 1 is 67.17% complete. Time: 7.606 sec.
[Jan 31 06:49] M90552467 stage 1 is 67.17% complete. Time: 7.542 sec.
[Jan 31 06:49] M90552467 stage 1 is 67.17% complete. Time: 7.650 sec.
[Jan 31 06:49] M90552467 stage 1 is 67.17% complete. Time: 9.000 sec.
[Jan 31 06:49] M90552467 stage 1 is 67.17% complete. Time: 7.644 sec.
[Jan 31 06:49] M90552467 stage 1 is 67.18% complete. Time: 7.677 sec.
[Jan 31 06:50] M90552467 stage 1 is 67.18% complete. Time: 7.590 sec.
[Jan 31 06:50] M90552467 stage 1 is 67.18% complete. Time: 13.717 sec.
[Jan 31 06:50] M90552467 stage 1 is 67.18% complete. Time: 7.618 sec.
[Jan 31 06:50] M90552467 stage 1 is 67.18% complete. Time: 7.618 sec.
[Jan 31 06:50] M90552467 stage 1 is 67.18% complete. Time: 7.601 sec.
...
[Jan 31 06:56] M90552467 stage 1 is 67.22% complete. Time: 7.607 sec.
[Jan 31 06:56] M90552467 stage 1 is 67.22% complete. Time: 7.624 sec.
[Jan 31 06:56] M90552467 stage 1 is 67.23% complete. Time: 7.633 sec.
[Jan 31 06:56] M90552467 stage 1 is 67.23% complete. Time: 7.646 sec.
[Jan 31 06:56] M90552467 stage 1 is 67.23% complete. Time: 7.677 sec.
[Jan 31 06:56] M90552467 stage 1 is 67.23% complete. Time: 7.693 sec.
[Jan 31 06:57] M90552467 stage 1 is 67.23% complete. Time: 20.545 sec.
[Jan 31 06:57] M90552467 stage 1 is 67.23% complete. Time: 11.059 sec.
[Jan 31 06:57] M90552467 stage 1 is 67.23% complete. Time: 7.986 sec.
[Jan 31 06:57] M90552467 stage 1 is 67.23% complete. Time: 12.046 sec.
[Jan 31 06:57] M90552467 stage 1 is 67.23% complete. Time: 8.128 sec.
[Jan 31 06:57] M90552467 stage 1 is 67.23% complete. Time: 7.823 sec.
[Jan 31 06:58] M90552467 stage 1 is 67.24% complete. Time: 8.656 sec.
[Jan 31 06:58] M90552467 stage 1 is 67.24% complete. Time: 8.128 sec.
[Jan 31 06:58] M90552467 stage 1 is 67.24% complete. Time: 8.647 sec.

Looking at each iteration, note 15 second or more elapsed time, even while not connected to remote desktop, and increased time when reconnected to remote desktop at the last lines:

[Jan 31 09:01:30] Iteration: 1365 / 332220523 [0.00%], ms/iter: 3030.617, ETA: 11653d 02:41
[Jan 31 09:01:45] Iteration: 1370 / 332220523 [0.00%], ms/iter: 3033.263, ETA: 11663d 06:56
[Jan 31 09:02:00] Iteration: 1375 / 332220523 [0.00%], ms/iter: 3033.212, ETA: 11663d 02:10
[Jan 31 09:02:15] Iteration: 1380 / 332220523 [0.00%], ms/iter: 3036.693, ETA: 11676d 11:27
[Jan 31 09:02:30] Iteration: 1385 / 332220523 [0.00%], ms/iter: 3032.966, ETA: 11662d 03:27
[Jan 31 09:02:46] Iteration: 1390 / 332220523 [0.00%], ms/iter: 3033.936, ETA: 11665d 21:02
[Jan 31 09:03:01] Iteration: 1395 / 332220523 [0.00%], ms/iter: 3032.390, ETA: 11659d 22:18
[Jan 31 09:03:16] Iteration: 1400 / 332220523 [0.00%], ms/iter: 3033.772, ETA: 11665d 05:50
[Jan 31 09:03:31] Iteration: 1405 / 332220523 [0.00%], ms/iter: 3036.067, ETA: 11674d 01:38
[Jan 31 09:03:46] Iteration: 1410 / 332220523 [0.00%], ms/iter: 3049.062, ETA: 11724d 00:52
[Jan 31 09:04:01] Iteration: 1415 / 332220523 [0.00%], ms/iter: 3030.043, ETA: 11650d 21:40
[Jan 31 09:04:17] Iteration: 1420 / 332220523 [0.00%], ms/iter: 3037.430, ETA: 11679d 07:23
[Jan 31 09:04:32] Iteration: 1425 / 332220523 [0.00%], ms/iter: 3037.069, ETA: 11677d 22:07
[Jan 31 09:04:47] Iteration: 1430 / 332220523 [0.00%], ms/iter: 3033.225, ETA: 11663d 03:21
[Jan 31 09:05:02] Iteration: 1435 / 332220523 [0.00%], ms/iter: 3029.420, ETA: 11648d 12:09
[Jan 31 09:05:17] Iteration: 1440 / 332220523 [0.00%], ms/iter: 3035.314, ETA: 11671d 04:08
[Jan 31 09:05:33] Iteration: 1445 / 332220523 [0.00%], ms/iter: 3037.829, ETA: 11680d 20:11
[Jan 31 09:05:48] Iteration: 1450 / 332220523 [0.00%], ms/iter: 3030.651, ETA: 11653d 05:49
[Jan 31 09:06:03] Iteration: 1455 / 332220523 [0.00%], ms/iter: 3032.233, ETA: 11659d 07:48
[Jan 31 09:06:18] Iteration: 1460 / 332220523 [0.00%], ms/iter: 3035.353, ETA: 11671d 07:39
[Jan 31 09:07:02] Iteration: 1465 / 332220523 [0.00%], ms/iter: 8877.448, ETA: 34134d 22:09
[Jan 31 09:07:41] Iteration: 1470 / 332220523 [0.00%], ms/iter: 7785.990, ETA: 29938d 03:06

A recheck of CPU-Z shows it has throttled back from 2100Mhz to 1062 core. The laptop does not seem to be particularly warm. Looks like time to fire up the P3-450 though, going by the minimum ms/iter numbers.

[kriesel]

Originally Posted by retina

Waah

Another contender for slowest. Infinite time per iteration for 100Mdigit, greater than 18.9 years run time for 77M. This is a Pentium-133 running Windows NT4, and prime95 V24.14, which is described in the whatsnew.txt as being capable of exponents up to 596 million. But perhaps that requires SSE2 and does not apply to the plain pentium code path. This has been running self test 1 for an hour (400 iterations).

Of three other antiques I tried, the Pentium-90 and the Pentium3-450 failed to boot, and the AMD Athlon 3200+ with XP declined to run the exponent.

[retina]

Originally Posted by kriesel

Another contender for slowest. Infinite time per iteration for 100Mdigit, greater than 18.9 years run time for 77M. This is a Pentium-133 running Windows NT4, and prime95 V24.14, which is described in the whatsnew.txt as being capable of exponents up to 596 million. But perhaps that requires SSE2 and does not apply to the plain pentium code path. This has been running self test 1 for an hour (400 iterations).

If it can't produce a time for even one iteration then I'd suggest it doesn't count as a result.

[kriesel]

Originally Posted by retina

If it can't produce a time for even one iteration then I'd suggest it doesn't count as a result.

15 seconds on Mp50*

[retina]

Originally Posted by kriesel

15 seconds on Mp50*

Mp50 < 100M. Disqualified

[kriesel]

Originally Posted by retina

Hehe, still no one can beat my time.

Richard Crandall's lucas.c, recompiled for Windows X64 via gcc, run on a Xeon E5645, 13.4 seconds/iteration.

Lucas.c for Win64, (c) 1995 R. E. Crandall
(fair use small mods by Woltman, Kriesel)
Memory allocated for x
CN Memory successfully allocated
SN Memory successfully allocated
Permute Memory successfully allocated
Scrambled Memory successfully allocated
Two_to_phi Memory successfully allocated
Two_to_minusphi Memory successfully allocated
2019/02/03 12:32:16  start of 10 iterations
1 maxerr: 0.000000
iter: 1
2 maxerr: 0.000000
iter: 2
3 maxerr: 0.000000
iter: 3
4 maxerr: 0.000000
iter: 4
5 maxerr: 0.000000
iter: 5
6 maxerr: 0.000015
iter: 6
7 maxerr: 0.000031
iter: 7
8 maxerr: 0.000122
iter: 8
9 maxerr: 0.000244
iter: 9
10 maxerr: 0.000488
iter: 10
332220523  1001101111011011010010010001110111110100110000000000000000000010
 (interim)
2019/02/03 12:34:30  end.

[kriesel]

Originally Posted by ewmayer

(Now can we please get back onto the topic of world domination not via military hegemony but by way of homebrew cellphone clusters? ;)

Sounds like building out SKYNET.

[retina]

Originally Posted by kriesel

Richard Crandall's lucas.c, recompiled for Windows X64 via gcc, run on a Xeon E5645, 13.4 seconds/iteration.

Sure, I could take the same software and run it on the Banias CPU (which it still in working order BTW). Hah, maybe I could write a naive BASIC program and get iteration times of months. But that it kind of out of the scope here I think. Aren't we looking to run the test as fast as it can go on the given system?

What is the iteration time for P95 on the E5645 system? Because that CPU should run rings around the Banias.

[kriesel]

Originally Posted by retina

Sure, I could take the same software and run it on the Banias CPU (which it still in working order BTW). Hah, maybe I could write a naive BASIC program and get iteration times of months. But that it kind of out of the scope here I think. Aren't we looking to run the test as fast as it can go on the given system?

What is the iteration time for P95 on the E5645 system? Because that CPU should run rings around the Banias.

dual E5645 system, Win 7, 6 cores, 140 ms/iter, all 12 cores 104 ms/iter (399 days est to completion) in prime95 29.5b6

This will have to do until I get it recompiled for the Pentium 133 I have running occasionally with NT 4. (Its per iteration times for M32220523 or higher would be glacial since it has only 80MB ram, so there's likely to be a lot of paging to disk.) Here, p=2³¹-1, which is not supported in prime95, on the E5645, gave 166 seconds/iteration. Projection is 11,330 years to complete. See https://www.mersenneforum.org/showpo...12&postcount=3 for E5645 single-core Lucas.c timings and working sets at various exponents.

Memory requirements will prevent running this test on my 4K SRAM 1Mhz 6502 Synertek SYM-1 board. (Unless there's a very compact program that could use cassette or open reel tape for large storage...)

Lucas.c for Win64, (c) 1995 R. E. Crandall
(fair use small mods by Woltman, Kriesel)
Memory allocated for x
CN Memory successfully allocated
SN Memory successfully allocated
Permute Memory successfully allocated
Scrambled Memory successfully allocated
Two_to_phi Memory successfully allocated
Two_to_minusphi Memory successfully allocated
2019/02/03 12:47:00  start of 10 iterations
1 maxerr: 0.000000
iter: 1
2 maxerr: 0.000000
iter: 2
3 maxerr: 0.000000
iter: 3
4 maxerr: 0.000000
iter: 4
5 maxerr: 0.000000
iter: 5
6 maxerr: 0.000000
iter: 6
7 maxerr: 0.000000
iter: 7
8 maxerr: 0.000000
iter: 8
9 maxerr: 0.000001
iter: 9
10 maxerr: 0.000001
iter: 10
2147483647  1001101111011011010010010001110111110100110000000000000000000010
 (interim)
2019/02/03 13:14:44  end.