mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Software (https://www.mersenneforum.org/forumdisplay.php?f=10)
-   -   Early Beta of version 24.11 (https://www.mersenneforum.org/showthread.php?t=3934)

Prime95 2005-03-30 23:21

Early Beta of version 24.11
 
Version 24.11 has two major improvements.

1) There is a 64-bit version for 64-bit Windows. It contains all new faster factoring code.

2) An Athlon64 optimization was found for both the 32-bit and 64-bit versions of prime95. You'll get about a 15% performance boost. Still not as fast as a similarly clocked P4, but it is much closer.

Only AMD64 machines should try this version. It is not well QA'ed. Save your work before installing, just in case there is a problem.

You can download from:

Windows: [url]ftp://mersenne.org/gimps/p95v2411.zip[/url]
Windows 64-bit: [url]ftp://mersenne.org/gimps/p64v2411.zip[/url]
Windows NT service: [url]ftp://mersenne.org/gimps/winnt2411.zip[/url]
Linux: [url]ftp://mersenne.org/gimps/mprime2411.tar.gz[/url]
Linux (static link): [url]ftp://mersenne.org/gimps/sprime2411.tar.gz[/url]

Let me know if you find any problems.

Prime95 2005-03-30 23:23

and source is at [url]ftp://mersenne.org/gimps/source24.zip[/url]

If any AMD64 gurus want to try and improve the factoring code in factor64.asm, I'm more than willing to incorporate any improvements.

Peter Nelson 2005-03-31 07:16

Excellent!
 
Excellent to see this and AMD users will be delighted to finally get their hands on it.

Few thoughts occured....

a) will you be building a [B]64 bit version for LINUX [/B] users (eg Suse 9.2 on AMD64 and other distros which are 64-bit)?

b) as the Pentium 6xx series now have Intel EM64T which has almost same instruction set as Athlon64, [B]might the code run on 6xx too?[/B]
I appreciate the architectures are different eg cache, and that there are very few of these in the field yet but it might be forward thinking to support Intel's 64 bit efforts too.

c) will you be able to maintain a single source code tree or have to have different versions for Intel/AMD/32/64 bits? The target architecture (32/64) could be specified at compile time, other differences as existing code by cpu detection. The range of cpu types is likely to further increase ie dualcore where two processors share memory subsystem which will slow mem accesses outside the L2 cache. This may mean different optimisations.

d) please would it be possible to include a short test of trial factoring speed in the benchmark (of the release version) because this would be very useful to know and compare the benefit of your optimisations. Maybe it could just appear when Fullbench option is specified.

sonjohan 2005-03-31 11:17

It is likely to be a silly question, but how do I know wether I have a 64-bit Windows or not??

dsouza123 2005-03-31 13:30

Unless you downloaded Window XP PRO 64 or got it through MSDN
you don't have it. It is/was a beta release from Microsoft, and
hasn't shipped on PCs for retail yet.

Has gone Gold today, shipping to manufacturing, so probably on
shipping PCs later part of April.

Prime95 2005-03-31 14:47

a) I have no immediate plans to try a 64-bit Linux port. I'm not sure if objcopy can convert the MASM object files into ELF64 format.

b) The code should run on Intel 6xx machines too.

c) There is just one source tree. Right now there are 4 different versions of the FFT:

x87 - optimized for Pentium Pro runs on any x87 machine
x87 - optimized for Athlons (also used by P3s and later)
SSE2 - P4 optimized
SSE2 - AMD64 optimized

d) Maybe

penguin22 2005-04-01 17:39

I have been using version 24.6 with the CPUSupportsSSE2=0 option and was wondering if it would be better to delete that line now that this version is out and has support for the features in the A64?

Thanks for your hard work.

Jeff Gilchrist 2005-04-01 17:51

Can anyone with A64/Opteron boxes post some benchmarks with 23.x, 24.6, and 24.11 comparisons please?

Prime95 2005-04-01 20:20

[QUOTE=penguin22]I have been using version 24.6 with the CPUSupportsSSE2=0 option and was wondering if it would be better to delete that line now that this version is out and has support for the features in the A64?[/QUOTE]

Time it both ways. I'm betting the SSE2 code is now faster for the same FFT size.

TheJudger 2005-04-01 20:59

[QUOTE=Prime95]a) I have no immediate plans to try a 64-bit Linux port. I'm not sure if objcopy can convert the MASM object files into ELF64 format.
[/QUOTE]

very sad to hear :(

is it possible for you to make a binary with an other asm compiler for non-windoze os'ses (even if it's a bit slower)?

thejudger

Prime95 2005-04-01 21:21

[QUOTE=TheJudger]is it possible for you to make a binary with an other asm compiler for non-windoze os'ses (even if it's a bit slower)?[/QUOTE]

No. Converting all that assembly code to another assembler format would be a monumental task.

I'm sure the binutils guys will make objcopy work eventually if they haven't done so already. The source is available if someone wants to try a 64-bit linux port.

My next task is more optimizations, especially making use of the extra SSE2 registers in 64-bit mode. Don't expect much - a few percent on AMD64, perhaps a little more on the P4.

TheJudger 2005-04-01 21:27

[QUOTE=Prime95]I'm sure the binutils guys will make objcopy work eventually if they haven't done so already. The source is available if someone wants to try a 64-bit linux port.
[/QUOTE]
So if binutils provite support for MASM -> ELF64 than you would give it a try? :)

[QUOTE=Prime95]My next task is more optimizations, especially making use of the extra SSE2 registers in 64-bit mode. Don't expect much - a few percent on AMD64, perhaps a little more on the P4.[/QUOTE]

Optimizations for ll-tests and/or factoring?

Peter Nelson 2005-04-01 23:24

From what I have read, I believe the latest version of objcopy in binutils DOES have the ability to convert to ELF64 format.

Someone could try to do this from the source provided above.

Obviously this would not include the security module for primenet comms, but if successful, it would not be too much work for George to make an official build using the same process.

Prime95 2005-04-02 01:31

[QUOTE=Peter Nelson]
d) please would it be possible to include a short test of trial factoring speed in the benchmark (of the release version) because this would be very useful to know and compare the benefit of your optimisations.[/QUOTE]

Done. Look for it next time I upload an executable

rainchill 2005-04-02 01:34

Hi, I have an Athlon64 3000+ Hp laptop, with 1Gb of ram. Here are my benchmark results with 23.8
Prime95 version 23.8, RdtscTiming=1
Best time for 384K FFT length: 26.055 ms.
Best time for 448K FFT length: 31.393 ms.
Best time for 512K FFT length: 35.744 ms.
Best time for 640K FFT length: 45.166 ms.
Best time for 768K FFT length: 54.970 ms.
Best time for 896K FFT length: 66.214 ms.
Best time for 1024K FFT length: 74.506 ms.
Best time for 1280K FFT length: 100.525 ms.
Best time for 1536K FFT length: 122.588 ms.
Best time for 1792K FFT length: 147.037 ms.
Best time for 2048K FFT length: 165.960 ms.

And here are my results with 24.11
Prime95 version 24.11, RdtscTiming=1
Best time for 512K FFT length: 32.064 ms.
Best time for 640K FFT length: 38.795 ms.
Best time for 768K FFT length: 46.873 ms.
Best time for 896K FFT length: 56.537 ms.
Best time for 1024K FFT length: 63.192 ms.
Best time for 1280K FFT length: 82.394 ms.
Best time for 1536K FFT length: 102.145 ms.
Best time for 1792K FFT length: 122.718 ms.
Best time for 2048K FFT length: 136.854 ms.

So it seems like a good little speed bump.

Prime95 2005-04-02 19:23

New 24.11 executables are available. From whatsnew.txt:

5) Added timeouts to PrimeNet communications in hopes of avoiding rare hangs
when contacting the PrimeNet server.
6) Fixed rare bug where P-1's GCD could miss a factor.
7) Added trial factoring to the benchmark.
8) Fixed bug in ECM when using zero-padded FFTs.

TheJudger 2005-04-02 22:59

9) crashes immediately on CPUs supporting sse2 when "advance/time" exponent is 78000000 :(

Happens with windows-client aswell as the linux-client...
Timing exponents works fine in older versions up to 79.xxxM

With "CPUSupportsSSE2=0" in local.ini timing exponents up to 79.xxxM works fine

[code]
Your choice: 8

Exponent to time (10000000): 78000000
Number of Iterations (10):

Accept the answers above? (Y):
Floating point exception
[/code]

Xyzzy 2005-04-04 00:40

[QUOTE=TheJudger]9) crashes immediately on CPUs supporting sse2 when "advance/time" exponent is 78000000 :([/QUOTE]
[url]http://www.mersenneforum.org/showpost.php?p=52142&postcount=4[/url]

TheJudger 2005-04-04 08:51

xyzzy: I know... but older versions (23.5, 23.9) automaticly switches to x87-code when leaving sse2-range... 24.11 just crashes...

Dresdenboy 2005-04-04 20:15

I posted many prime95 results (incl. 64bit TF numbers vs. 32bit) in the [URL=http://mersenneforum.org/showthread.php?p=52693#post52693]benchmark thread[/URL].

Dresdenboy 2005-04-05 12:01

Trial factoring speedup in 64bit mode:
[code]24.11 24.11
32 bit 64 bit
(ms) (ms) Speedup
6,010 3,786 58,7%
6,032 3,782 59,5%
6,003 4,032 48,9%
6,029 4,140 45,6%
10,961 4,843 126,3%
10,963 5,643 94,3%
13,932 6,816 104,4%
13,836 8,007 72,8%
13,849 7,951 74,2%
13,831 7,934 74,3%[/code]

ET_ 2005-04-05 18:17

[QUOTE=Dresdenboy]Trial factoring speedup in 64bit mode:
[code]24.11 24.11
32 bit 64 bit
(ms) (ms) Speedup
6,010 3,786 58,7%
6,032 3,782 59,5%
6,003 4,032 48,9%
6,029 4,140 45,6%
10,961 4,843 126,3%
10,963 5,643 94,3%
13,932 6,816 104,4%
13,836 8,007 72,8%
13,849 7,951 74,2%
13,831 7,934 74,3%[/code][/QUOTE]

cOOOOl! :showoff:

Luigi

moo 2005-04-05 21:19

d00d

typo_pl 2005-04-06 15:21

I guess I'm the first to download the source code/complain about the source code URL.

Here's a working URL, notice the misplaced 'r':
[URL=ftp://mersenne.org/gimps/source24.zip]ftp://mersenne.org/gimps/srouce24.zip[/URL]

George: Filename now fixed on the server

Washuu 2005-04-06 18:11

Another ProxyServer request...
 
I have troubles connecting to Primenet. So i started reading all threads about it... and investigating.

Now I know: my proxy server (to which I have no access... :wink: ) is configured in that way, so it accepts request

GET [url]http://mersenne.org/cgi-bin/pnHttp.exe?([/url]...)

but DOES NOT

GET [url]http://mersenne.org:80/cgi-bin/pnHttp.exe?([/url]...)

which is used by prime95 communication.

Is there any switch I can use? Or may I put a request for such? :rolleyes:


BTW: there was PrimeNet 4.0 Proxy Server some years ago... Pity that it isn't developed more... :down:

Prime95 2005-04-06 23:42

There is no switch you can use. I'll put in a fix for the next 24.11 update.

sharad 2005-04-07 10:26

Using...

[url]ftp://mersenne.org/gimps/p95v2411.zip:[/url]
AMD Athlon(tm) 64 Processor 3000+
CPU speed: 1809.21 MHz
Windows XP Pro. (32bit)

[code]
[b]Prime95 version 24.6, RdtscTiming=1[/b]
Best time for 512K FFT length: 32.254 ms.
Best time for 640K FFT length: 40.449 ms.
Best time for 768K FFT length: 49.186 ms.
Best time for 896K FFT length: 59.303 ms.
Best time for 1024K FFT length: 66.494 ms.
Best time for 1280K FFT length: 88.572 ms.
Best time for 1536K FFT length: 106.560 ms.
Best time for 1792K FFT length: 128.943 ms.
Best time for 2048K FFT length: 148.450 ms.

[b]Prime95 version 24.11, RdtscTiming=1[/b]
Best time for 512K FFT length: 30.152 ms.
Best time for 640K FFT length: 37.092 ms.
Best time for 768K FFT length: 44.866 ms.
Best time for 896K FFT length: 54.281 ms.
Best time for 1024K FFT length: 60.341 ms.
Best time for 1280K FFT length: 80.006 ms.
Best time for 1536K FFT length: 97.621 ms.
Best time for 1792K FFT length: 117.558 ms.
Best time for 2048K FFT length: 130.864 ms.
[/code]

Nice boost but is it ok to switch to the new version if you are in between a factoring assignment?

SalemTheCat100 2005-04-07 15:11

Athlon 3400+ @2.5ghz timings
 
Running on Windows XP 64bit Build 1433

AMD Athlon(tm) 64 Processor 3400+ overclocked 4%
1gb PC3200 Crucial DDR 3-3-3-8
CPU speed: 2495.91 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 512

Prime95 version 23.8, RdtscTiming=1
Best time for 512K FFT length: 25.650 ms.
Best time for 640K FFT length: 32.008 ms.
Best time for 768K FFT length: 38.921 ms.
Best time for 896K FFT length: 46.940 ms.
Best time for 1024K FFT length: 52.920 ms.
Best time for 1280K FFT length: 70.715 ms.
Best time for 1536K FFT length: 84.584 ms.
Best time for 1792K FFT length: 102.695 ms.
Best time for 2048K FFT length: 116.308 ms.

Prime95 version 24.6 with SSE2, RdtscTiming=1
Best time for 512K FFT length: 25.911 ms.
Best time for 640K FFT length: 32.005 ms.
Best time for 768K FFT length: 39.066 ms.
Best time for 896K FFT length: 47.263 ms.
Best time for 1024K FFT length: 53.246 ms.
Best time for 1280K FFT length: 71.135 ms.
Best time for 1536K FFT length: 84.945 ms.
Best time for 1792K FFT length: 103.066 ms.
Best time for 2048K FFT length: 117.047 ms.

Prime95 version 24.6 no SSE2, RdtscTiming=1
Best time for 512K FFT length: 24.035 ms.
Best time for 640K FFT length: 32.138 ms.
Best time for 768K FFT length: 38.689 ms.
Best time for 896K FFT length: 46.980 ms.
Best time for 1024K FFT length: 52.488 ms.
Best time for 1280K FFT length: 69.806 ms.
Best time for 1536K FFT length: 83.631 ms.
Best time for 1792K FFT length: 100.429 ms.
Best time for 2048K FFT length: 111.799 ms.

Prime95 version 24.11 [B]32bit code[/B], RdtscTiming=1
Best time for 512K FFT length: 23.132 ms.
Best time for 640K FFT length: 27.318 ms.
Best time for 768K FFT length: 33.018 ms.
Best time for 896K FFT length: 40.094 ms.
Best time for 1024K FFT length: 44.604 ms.
Best time for 1280K FFT length: 59.142 ms.
Best time for 1536K FFT length: 72.431 ms.
Best time for 1792K FFT length: 87.165 ms.
Best time for 2048K FFT length: 97.316 ms.

Best time for 58 bit trial factors: 4.852 ms.
Best time for 59 bit trial factors: 4.838 ms.
Best time for 60 bit trial factors: 4.818 ms.
Best time for 61 bit trial factors: 4.840 ms.
Best time for 62 bit trial factors: 8.809 ms.
Best time for 63 bit trial factors: 8.810 ms.
Best time for 64 bit trial factors: 11.193 ms.
Best time for 65 bit trial factors: 11.149 ms.
Best time for 66 bit trial factors: 11.132 ms.
Best time for 67 bit trial factors: 11.088 ms.

Prime95 version 24.11 [B]64bit code[/B], RdtscTiming=1
Best time for 512K FFT length: 23.093 ms.
Best time for 640K FFT length: 27.489 ms.
Best time for 768K FFT length: 33.122 ms.
Best time for 896K FFT length: 40.072 ms.
Best time for 1024K FFT length: 44.526 ms.
Best time for 1280K FFT length: 59.238 ms.
Best time for 1536K FFT length: 72.824 ms.
Best time for 1792K FFT length: 87.254 ms.
Best time for 2048K FFT length: 97.604 ms.

Best time for 58 bit trial factors: 3.037 ms.
Best time for 59 bit trial factors: 3.038 ms.
Best time for 60 bit trial factors: 3.238 ms.
Best time for 61 bit trial factors: 3.320 ms.
Best time for 62 bit trial factors: 3.893 ms.
Best time for 63 bit trial factors: 4.536 ms.
Best time for 64 bit trial factors: 5.480 ms.
Best time for 65 bit trial factors: 6.424 ms.
Best time for 66 bit trial factors: 6.388 ms.
Best time for 67 bit trial factors: 6.372 ms.

Nice work.

Salem

Peter Nelson 2005-04-08 15:55

Benchmark of Trial Factoring (Linux Client)
 
[QUOTE=Prime95]Done. Look for it next time I upload an executable[/QUOTE]

George, thanks for the new factoring benchmark, it's very useful on my Windows XP machines.

For instance comparing Dresdenboy's A64 3000+ with my 3GHz P4, the P4 is still much faster at LL but they are about the same for TF in 32-bit. And in 64-bit, the A64 is significantly faster at TF work.

a)

[B]Users of the TF benchmark stats should note that whilst then benchmark says eg 62 bit this is equivalent to what factoring runs call "to 2^63".[/B] And so on for each bit size. This is correct but could be potentially confusing. I wonder if the factoring could say FROM 2^62 TO 2^63 (for clarity), whereas the benchmark can stay as is.

b)

[B]However I also downloaded the latest Linux (32-bit) version which identifies itself as 24.11.2. Upon execution I found the benchmark output only contains the LL not the factoring tests.[/B]

It's the statically linked one I tried because the dynamic linked one needs libstdc++.so.6 to run. (Suse 9.3 comes out in about a week so I will be reinstalling from it and don't want to mess about with libraries etc before then).

The mprime files on the ftp server have the linux as slightly newer than the Windows version.

Did you intend to recompile and make a newer linux client available from the latest source?
Or does the linux client not support benchmarking of factoring speed?
I assume it's the same source code for both platform so was confused why the linux build doesn't do it (yet). This would be nice for cross-platform comparisons.

Please explain any options I need to set to get the factor benchmark on static mprime, or provide a version which does it from the normal benchmark menu as I expected.

Thanks in anticipation, Peter

c)

P.S. Quite unrelated, but [B]which version of MASM are you using[/B] to turn your assembly files into object code (I'm particularly interested in relation to understanding your 64-bit compilation environment).

Prime95 2005-04-08 20:18

Linux mprime should give you the factoring benchmark too. I probably made a mistake in building or uploading it.

My 64-bit build environment is MS Visual Studio 2005 a.k.a. Whidbey. It cost $10 for a time-limited beta version. I don't know what the final version will cost.

Masm for 32-bit is version 8.00.40607.16
Masm for 64-bit is also version 8.00.40607.16 but is a different executable.

dans 2005-04-11 05:49

SSE2 option under new version
 
[QUOTE=Prime95]Time it both ways. I'm betting the SSE2 code is now faster for the same FFT size.[/QUOTE]

According to my benchmarks, don't use that option anymore!

Benchmark [B]WITH [/B]the "CpuSupportsSSE2=0" option set
AMD Athlon(tm) 64 Processor 3200+
CPU speed: 1994.68 MHz
Prime95 version 24.11, RdtscTiming=1
Best time for 512K FFT length: 30.266 ms.
Best time for 640K FFT length: 40.544 ms.
Best time for 768K FFT length: 48.946 ms.
Best time for 896K FFT length: 59.808 ms.
Best time for 1024K FFT length: 66.461 ms.
Best time for 1280K FFT length: 87.620 ms.
Best time for 1536K FFT length: 107.221 ms.
Best time for 1792K FFT length: 126.338 ms.
Best time for 2048K FFT length: 140.388 ms.
Best time for 58 bit trial factors: 6.119 ms.
Best time for 59 bit trial factors: 6.107 ms.
Best time for 60 bit trial factors: 6.119 ms.
Best time for 61 bit trial factors: 6.110 ms.
Best time for 62 bit trial factors: 11.132 ms.
Best time for 63 bit trial factors: 11.108 ms.
Best time for 64 bit trial factors: 21.321 ms.
Best time for 65 bit trial factors: 21.928 ms.
Best time for 66 bit trial factors: 22.479 ms.
Best time for 67 bit trial factors: 22.696 ms.

Benchmark [B]WITHOUT [/B]"CpuSupportsSSE2=0" option set
AMD Athlon(tm) 64 Processor 3200+
CPU speed: 1994.98 MHz
Prime95 version 24.11, RdtscTiming=1
Best time for 512K FFT length: 28.707 ms.
Best time for 640K FFT length: 67.948 ms.
Best time for 768K FFT length: 77.572 ms.
Best time for 896K FFT length: 50.826 ms.
Best time for 1024K FFT length: 56.583 ms.
Best time for 1280K FFT length: 74.948 ms.
Best time for 1536K FFT length: 91.276 ms.
Best time for 1792K FFT length: 109.248 ms.
Best time for 2048K FFT length: 122.294 ms.
Best time for 58 bit trial factors: 6.096 ms.
Best time for 59 bit trial factors: 6.111 ms.
Best time for 60 bit trial factors: 6.087 ms.
Best time for 61 bit trial factors: 6.109 ms.
Best time for 62 bit trial factors: 11.139 ms.
Best time for 63 bit trial factors: 11.112 ms.
Best time for 64 bit trial factors: 14.203 ms.
Best time for 65 bit trial factors: 14.278 ms.
Best time for 66 bit trial factors: 14.301 ms.
Best time for 67 bit trial factors: 14.138 ms.

TheJudger 2005-04-11 10:43

[QUOTE=Peter Nelson]
[B]However I also downloaded the latest Linux (32-bit) version which identifies itself as 24.11.2. Upon execution I found the benchmark output only contains the LL not the factoring tests.[/B]

It's the statically linked one I tried because the dynamic linked one needs libstdc++.so.6 to run. (Suse 9.3 comes out in about a week so I will be reinstalling from it and don't want to mess about with libraries etc before then).
[/QUOTE]

On my SuSE 9.2 Systems a simple symbolic link helped...
[CODE]
cd /usr/lib/
ln -s libstdc++.so.5 libstdc++.so.6
[/CODE]

Yeah, I know that's quick&dirty ;)

Peter Nelson 2005-04-11 13:14

[QUOTE=TheJudger]On my SuSE 9.2 Systems a simple symbolic link helped...
[CODE]
cd /usr/lib/
ln -s libstdc++.so.5 libstdc++.so.6
[/CODE]

Yeah, I know that's quick&dirty ;)[/QUOTE]

Thanks for the suggestion, that's one way around it, but the statically linked version runs for me so I don't need the dynamic one to find its external library. I will only need the library when I compile it all myself, and I won't be doing that before my 9.3 OS upgrade.

The reason I explained that it was the STATIC build version of mprime was to help George identify the problem (ie which file I had tried executing which exhibited my lack of factoring benchmark problem) and to explain that I was unable to state whether the same problem was present in the dynamic build.

As it turns out, it probably was a simple oversight with his build procedure meaning that the ftp file didn't reflect his latest source.

I'm waiting until he does a re-build and uploads a new static mprime containing the factoring benchmark. This may coincide with the next bugfix, and I was trying to be patient by not asking "Rebuild and upload it now".

ET_ 2005-04-12 11:57

Benchmark timings
 
George, I have a dumb question about v24.11

Looking at the factoring benchmark, I noticed that the best time results are printed in ms.

Assuming that there are about 1,029,407,319 k from 57 to 58 bits (exponent 35000011), and that about 95% of those k are sieved out, we have about 51,470,366 k to check.

Prime95 runs the factoring work between 57 and 58 bits in about 15,450 ms and 58-59 in 31,500 ms (time of Prime95); the benchmark shows a timing of about 7 ms for bit depth comprised between 58 and 61 bits, so I assume the same for 57 bits.

The "one iteration time" in factoring should be 15,450/51,470,366 = 30*10^-4 ms; 7 ms are worth about 23,320 iterations.

As a demonstration, if we divide k by 23,320 and multiply the result by 7, we again have 15,450 ms

May I ask you why did you choose 23,320 iterations for the benchmark? :rolleyes:

Luigi

Prime95 2005-04-12 13:36

You lost me.

The 32-bit factoring assembly code processes a 16KB sieve. This is "one iteration" in the benchmark. When doing trial factoring you control how often prime95 writes to the screen in Options / Preferences.

The 64-bit factoring assembly code processes a 48KB sieve.
This is "one iteration". The benchmarking code divides this time by 3 so you can properly compare it to 32-bit machines.

Frodo42 2005-04-12 22:13

I seeem to be having a problem with sprime when I try to factor a number that requires FFT length of 1024K

I get the following message:
[code]
Starting P-1 factoring with B1=40000, B2=420000
Chance of finding a factor is an estimated 0.92%
P-1 on 55459*2^9011278+1 with B1=40000, B2=420000
Using FFT length 1024K
SUMOUT error occurred.
[/code]

There doesn't seem to be any problems P-1 factoring numbers with lower FFT length (even 1024 zero-padded)
I am testing this on another machine just now ... I haven't had any hardware problems with this computer so far.

Anyhow just thought I would report this ... it may just be a hardware error on my side.

ET_ 2005-04-12 22:26

[QUOTE=Prime95]You lost me.

The 32-bit factoring assembly code processes a 16KB sieve. This is "one iteration" in the benchmark. When doing trial factoring you control how often prime95 writes to the screen in Options / Preferences.

The 64-bit factoring assembly code processes a 48KB sieve.
This is "one iteration". The benchmarking code divides this time by 3 so you can properly compare it to 32-bit machines.[/QUOTE]

I'm lost too :surrender

I'd better look into Prime95 source code before asking dumb questions.
Your answer helped, thank you.

Luigi

Frodo42 2005-04-12 22:40

[QUOTE=Frodo42]I am testing this on another machine just now ... [/QUOTE]

It reproduced the error message, so I guess it's not a hardware failure.

Prime95 2005-04-13 00:43

[QUOTE=Frodo42]It reproduced the error message, so I guess it's not a hardware failure.[/QUOTE]

Is this version 24.11 that is doing this? When I try that number on a P4 I get a zero-padded 1024K FFT and no errors.

Frodo42 2005-04-13 04:36

[QUOTE=Prime95]Is this version 24.11 that is doing this? When I try that number on a P4 I get a zero-padded 1024K FFT and no errors.[/QUOTE]

Yes it's from sprime2411.tar.gz

The line I have in worktodo.ini is:
[code]
Pfactor=55459,2,9011278,1,49,1.5
[/code]
My local.ini:
[code]
OldCpuType=12
OldCpuSpeed=3017
DayMemory=768
NightMemory=768
CPUHours=24
DayStartTime=480
DayEndTime=1320
Pid=0
RollingAverage=1000
RollingStartTime=0
[/code]

Frodo42 2005-04-13 08:07

[QUOTE=Prime95]Is this version 24.11 that is doing this? [/QUOTE]

No I am wrong ... it's not version 2.4.11 even though I thought it was. It was an older version.
Sorry about all this fuss :redface:
It works fine when i actually use ver. 2.4.11

The problem arose because I wrongly assumed that the static binary was called sprime, while it is actually mprime ... I had an sprime binary in the directory from an old package that contained both the sprime and mprime.

Prime95 2005-04-18 23:18

I've just uploaded my third attempt at 24.11.

The only new feature is I've spent time optimizing the SSE2 macros further. P4s will see about a 1% boost and AMD64s will see about a 2.5% boost.

PhilF 2005-04-18 23:40

Earlier in this thread you mentioned only AMD64 machines should run this version. Is it now OK for P4 machines to run this version on exponents currently being tested by version 23.8?

Prime95 2005-04-19 00:03

[QUOTE=PhilF]Earlier in this thread you mentioned only AMD64 machines should run this version. Is it now OK for P4 machines to run this version on exponents currently being tested by version 23.8?[/QUOTE]

It is still beta software. You should be OK, but no guarantees. You must ask yourself if it is worth the small risk for the small speed gain.

Peter Nelson 2005-04-20 23:22

I have tried the latest files on Windows and Linux. As for my earlier post about the static linux build not producing benmarks of factoring, this new version/build has fixed it and benchmarking works fine.

I notice that on Linux the version in About is 24.11.2 whilst I expected the 2 to change to a 3 or something. Additionally when results go in logfile etc I think the version is only reported as the first two numbers and omit the third. Certainly this is the case in Windows Help About.

Just a suggestion but could you treat the third number as a kind of build/release number (ie it changes every time) and arrange for it to be used at all times?

This would help us know which machines have which of the 3 (so far) released versions of 24.11 and provide proper comparison of benchmarks (since efficiency alters with changes) and help bug reports be attributed to the right build.

PhilF 2005-04-20 23:46

As long as we're talking about program improvements, what would be handy for me is a way to specify the path where the 2nd backup file is stored. That way I can keep one on the local drive and one on a network drive, both being constantly updated.

db597 2005-04-21 16:03

[QUOTE=PhilF]As long as we're talking about program improvements, what would be handy for me is a way to specify the path where the 2nd backup file is stored. That way I can keep one on the local drive and one on a network drive, both being constantly updated.[/QUOTE]

That would be a nice feature. Also, if one hardisk crashes, with the backup in the same directory, everything is still lost.

ric 2005-04-22 07:06

Communication issue with the latest beta (Error 1)
 
I'm reporting a small communication issue related to the latest 24.11 beta: when communicating behind a proxy server, I always have Primenet Error 1 as response.

The issue does not arise in the same 24.11 when I communicate without a proxy (i.e. via direct dialup), and is not related to some misconfigured element: the same machine communicates successfully via the same proxy configuration (primenet.ini, and so on) if I only restore version 24.6. I've also tried to regenerate the masked password (just in case) via ProxyMask, to no result.

Did anybody else experience the same behaviour?
ric

PS: environment is WinXP Prof Ed SP1 on a 2Ghz Compaq Evo, non-influent firewall local software (ISS Desktop Protector v7.0, Kerio PF 4)

Prime95 2005-04-22 13:04

[QUOTE=ric]I'm reporting a small communication issue related to the latest 24.11 beta: when communicating behind a proxy server, I always have Primenet Error 1 as response. [/QUOTE]

Please try putting "SendPortNumber=1" in primenet.ini. Then let me know if that helps or not.

ric 2005-04-22 13:21

Yes, now it works nicely via proxy.
 
Did as per your recommendation, restarted, now it communicates nicely via proxy - haven't tried yet the dialup (=no proxy), but it's not urgent.

One last small question: I've browsed rapidly the documentation, but I haven't found any reference to this SendPortNumber parameter. Is this intentional?

George, many thanks again for your recommendation.
ric

eldel 2005-04-22 15:58

Tried it here - still getting Primenet Error 1.

Opteron running W2K3 Enterprise x64 Edition build 3790 with SP1.

Derek

eldel 2005-04-22 17:02

However - a complete stop and restart (reboot the machine) caused it to work. Thanks to ric for the idea.

Prime95 2005-04-22 22:40

SendPortNumber is brand new --- unfortunately.

A user reported that his proxy server did not work because of the ":80" in "http://mersenne.org:80/blahblah" sent by prime95.

So, since the :80 in theory is superfluous, I removed it. But, I added this SendPortNumber ini setting in case some proxy server required the :80.

Alas, we've now proven that some proxy servers require the :80 and some fail with the :80. It's just nuts.

So, I'll make sending the :80 the default (same as all previous versions of prime95) and add a description of SendPortNumber to undoc.txt.

Xyzzy 2005-04-23 02:29

Windows Server 2003 Enterprise x64 Edition

[code]AMD Athlon(tm) 64 Processor 3400+
CPU speed: 2402.97 MHz
CPU features: RDTSC, CMOV, Prefetch, 3DNow!, MMX, SSE, SSE2
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 512
Prime95 version 24.11, RdtscTiming=1
Best time for 512K FFT length: 23.648 ms.
Best time for 640K FFT length: 27.952 ms.
Best time for 768K FFT length: 33.717 ms.
Best time for 896K FFT length: 40.733 ms.
Best time for 1024K FFT length: 45.252 ms.
Best time for 1280K FFT length: 60.393 ms.
Best time for 1536K FFT length: 73.928 ms.
Best time for 1792K FFT length: 89.212 ms.
Best time for 2048K FFT length: 99.531 ms.
Best time for 58 bit trial factors: 3.154 ms.
Best time for 59 bit trial factors: 3.160 ms.
Best time for 60 bit trial factors: 3.368 ms.
Best time for 61 bit trial factors: 3.451 ms.
Best time for 62 bit trial factors: 4.038 ms.
Best time for 63 bit trial factors: 4.706 ms.
Best time for 64 bit trial factors: 5.677 ms.
Best time for 65 bit trial factors: 6.666 ms.
Best time for 66 bit trial factors: 6.631 ms.
Best time for 67 bit trial factors: 6.627 ms.[/code]

Xyzzy 2005-04-23 13:14

1 Attachment(s)
If I add this to the top of my worktodo.ini I get an immediate crash...

[code]ECM=791,11000000,0,736,0,0,1,0[/code]Right now it is at the end of my worktodo.ini and it hasn't crashed... The status button shows everything okay...

Xyzzy 2005-04-23 13:34

When I tell Prime95 to do ECM via the menu, it adds this entry to worktodo.ini:

[code]ECM2=1,2,791,1,11000000,0,736,0,0[/code]But this crashes Prime95 if I move it to the top as well...

Prime95 2005-04-23 21:15

[QUOTE=Xyzzy]But this crashes Prime95 if I move it to the top as well...[/QUOTE]

CPU type? 64 or 32-bit prime95? any old 2^791+1 ecm save files hanging around?

Xyzzy 2005-04-23 21:52

[QUOTE=Prime95]CPU type? 64 or 32-bit prime95? any old 2^791+1 ecm save files hanging around?[/QUOTE]3400+ AMD64... 64-bit Prime95... No save files... It happens with and without lowm.txt and lowp.txt in the folder...

Prime95 2005-04-24 03:13

Is "CpuSupportsSSE2=0" set in local.ini?

Xyzzy 2005-04-24 08:47

[QUOTE=Prime95]Is "CpuSupportsSSE2=0" set in local.ini?[/QUOTE]No... If I add that line to local.ini Prime95 crashes no matter what I do...

I can (I think!) provide remote desktop access to this box if you want...

Which form of the ECM line is the correct one for me to use?

Mystwalker 2005-04-24 10:16

Both should work.
ECM is the old syntax, when only 2^x +/-1 was supported.
ECM2 is the new syntax, for numbers of the form k*b^n +/-1.

2^791+1 works for both, and AFAIK, there is no difference.

kierkegaard 2005-04-24 14:05

Hi, I was trying the beta version 24.11 of Prime95 and noticed that it has some troubles using a proxy server. On the same machine with the same configuration version 23.8 works ok. But using 24.11 I get an "error 1". I tried setting "Debug=1" and noticed that the first request issued by the program to the proxy is successful, while the second is invalid. And I mean it is an invalid URL (unforunately I can't report it because at the moment I don't have direct access to the machine), as part of the address is missing (if I remember right, it's something of the form "http://mersenne.org[xxx]", where the "[xxx]" is something related to the version and to the previous request; so at least one "/" is missing). I hope this information is enough, sorry if I can't post the exact log messages.

Prime95 2005-04-25 01:10

[QUOTE=kierkegaard]Hi, I was trying the beta version 24.11 of Prime95 and noticed that it has some troubles using a proxy server. [/QUOTE]

For now, add "SendPortNumber=1" to primenet.ini

Prime95 2005-04-25 01:11

xyzzy, I'm stumped. Can you email the *.ini files?

Xyzzy 2005-04-25 13:49

YGM...

Nebob 2005-04-26 23:38

AMD Athlon(tm) 64 Processor 2800+
CPU speed: 1799.98 MHz
CPU features: RDTSC, CMOV, Prefetch, 3DNow!, MMX, SSE, SSE2
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 512
Prime95 version 24.11, RdtscTiming=1
Best time for 512K FFT length: 31.383 ms.
Best time for 640K FFT length: 37.562 ms.
Best time for 768K FFT length: 45.233 ms.
Best time for 896K FFT length: 54.496 ms.
Best time for 1024K FFT length: 60.727 ms.
Best time for 1280K FFT length: 80.939 ms.
Best time for 1536K FFT length: 99.224 ms.
Best time for 1792K FFT length: 119.936 ms.
Best time for 2048K FFT length: 133.851 ms.
Best time for 58 bit trial factors: 4.216 ms.
Best time for 59 bit trial factors: 4.212 ms.
Best time for 60 bit trial factors: 4.489 ms.
Best time for 61 bit trial factors: 4.606 ms.
Best time for 62 bit trial factors: 5.397 ms.
Best time for 63 bit trial factors: 6.280 ms.
Best time for 64 bit trial factors: 7.583 ms.
Best time for 65 bit trial factors: 8.902 ms.
Best time for 66 bit trial factors: 8.848 ms.
Best time for 67 bit trial factors: 8.839 ms.

Prime95 2005-04-27 00:47

XYZZY has found a bug in the 64-bit GCD code. Do not use the 64-bit executable to do P-1 or ECM. I don't have a fix yet.

Prime95 2005-04-28 01:47

Xyzzy, I've uploaded a fix for the GCD bug in the 64-bit executable.

Xyzzy 2005-04-28 01:50

[QUOTE=Prime95]Xyzzy, I've uploaded a fix for the GCD bug in the 64-bit executable.[/QUOTE]Cool, I'll test it ASAP...

Xyzzy 2005-04-28 13:33

1 Attachment(s)
It still crashes immediately...

I tried "ECM=791,11000000,0,736,0,0,1,0" first... Since I have not run Prime95 for a few days, it contacted PrimeNet and then I saw the ECM line appear before it crashed... Usually it crashes so fast I can't even get the window open... Any time I tried to run it after, since the PrimeNet deal was done, it crashed immediately... I also tried "ECM2=1,2,791,1,11000000,0,736,0,0" with the same results...

I tried running a high priority task in the background, to try to slow down the crashing, and I was able to capture this screen shot... (I sure wish Prime95 had an option to log everything to a text file!)

Prime95 2005-04-28 22:59

Xyzzy, try again. The debug version worked OK, so I built the release version and uploaded it without testing. Unfortunately, the C++ optimizer was assuming certain registers were saved across assembly calls. I've fixed the assembly code to save those registers.

Interestingly, Windows-64 and Linux assume different registers are saved. Makes writing portable assembly code a little tricky.

Xyzzy 2005-04-28 23:13

It doesn't crash...

:smile:

I'm curious though... How do we know it gets the correct results? Do we have a list of known numbers to run through it that cover all the possible situations? An example is the script they distribute with GMP-ECM that has a pile of test numbers...

Peter Nelson 2005-04-29 16:54

Client hangs on network
 
Hi I am using 24.11. (third release) as the static binary on Suse Linux.

After a while now I noticed a pattern of a recurring problem not previously experienced with older versions.

That is my software will hang when doing some network comms.

I am NOT using a proxy server, firewall or NAT addressing, and my network is fine.

What happens is the normally high cpu usage goes right down, and CTRL-C will not exit the app. If I close the console window and re-invoke the software (reboot not required) processing continues until the next hang.

Now maybe I'm noticing this more than others because I set my client to check in with the server automatically quite often to keep my individual report page up to date with expected completions etc.

I know there have been probs with server down recently but I think its more than that cos not seen this on previous versions.

Have seen the "ERROR 1" like others but not every time.

As it takes me a while to notice the processing stalled, I keep losing maybe a day of time when work could be getting done GRRR.

I believe there was some attempted improvement to the functionality of client when server was down? Could this maybe need further refinement to ensure it's not a backward step?

I think when it hangs it is right after the "Contacting server with expected completion date _____"

Maybe there is some loop where client waits for a server's response but goes round infinitely waiting for it. Needs to ignore and get on with some more iterations.

It seems exponents can complete and report results and get new exponents fairly ok, I think it's the checking in that falls over.

I would appreciate a fix.

Peter Nelson 2005-04-29 17:37

More network 24.11 client probs
 
I was trying to further clarify what goes wrong. Can confirm the machine has good inet connectivity ie google loads fine in a browser.

At the moment (?) when I started mprime it said contacting server with expected completion dates (since TF there are a few) but it waits about 4 sec between each then at the last one freezes. Restarted, same again.

Then I thought from menu mprime -m could force manual update see if that worked. Then I thought for same reason in menu mprime -m option 21 About primenet server.

Tried it contacted server reported server version ok.

Tried it again - attempts to contact server. Client hangs and can't CTRL-C.

So, this appears to be intermittent (maybe on server). From another machine I can get my individual report status page (so database is up no error 3) but if I click reload, stops halfway and is slow to load. Same again. Same again. So server appears to be heavily loaded with something intermittent perhaps.

[B]So at times when server is busy, client can't even reliably query the About server version without hanging. Clearly the same problem is affecting other client-server comms intermittently. The client should recover gracefully but it doesn't.[/B]

FYI my round trip ping time to the server is about 170 ms. Most of that is on the London-Washington transatlantic hop.

Trying some more 21. About Server.....
Works works slowly works slowly hangs!

So at the moment the server responds straight away, or takes a while 3, 4, 8 secs maybe or waits longer and it crashes the app.

The problem is easiest to see here because you can make it communicate on demand. Hope you can replicate this problem and debug the comms routines.

Prime95 2005-04-29 21:00

Peter, I'm trying it on Linux and Windows right now and, of course, it works like a champ.

Can you add "Debug=1" to primenet.ini and tell me the last lines output before it hangs?

24.11 is "improved". I make two new calls. One says don't send me a SIG_EPIPE signal. The other sets a timeout on recv requests.

db597 2005-04-29 23:32

Is 24.11 updated?
 
I was wondering if the recent updates to 24.11 in this thread are available at [url]http://www.mersenne.org/gimps/[/url]. All the copies of 24.11 there are all dated 18th of April - which is 2 weeks ago. But the bug fixes discussed in this thread have been made just in the last few days. Where do we get this latest alpha?

Prime95 2005-04-29 23:45

[QUOTE=db597]I was wondering if the recent updates to 24.11 in this thread are available at [url]ftp://www.mersenne.org/gimps/[/url]. ?[/QUOTE]

Yes. The bugs fixed since April 18th only affected the 64-bit Windows port.

Peter Nelson 2005-04-30 06:23

[QUOTE=Prime95]Peter, I'm trying it on Linux and Windows right now and, of course, it works like a champ.

Can you add "Debug=1" to primenet.ini and tell me the last lines output before it hangs?

24.11 is "improved". I make two new calls. One says don't send me a SIG_EPIPE signal. The other sets a timeout on recv requests.[/QUOTE]

Thanks for trying.

My problem is only seen on the linux version. I haven't seen windows version hang on About primenet, possibly when checking in (though I don't remember clearly). So to have a chance to see what I see use the linux build.

Of course, mine now is behaving itself too. I have done About primenet server (menu 21) again and of course it reports and doesn't hang. This is probably just as you saw.

Nothing on my client is changed from earlier.

I have added the debug setting as requested and will try to catch it hanging again, and post what it produces.

There cannot be much going on in the code to request the server version, so thought that this manifestation of the prob might be easier to debug and figure out what goes wrong.

My problem is that it is unpredictable when it will hang (to demonstrate to you). Some server comms work fine, then another time (probably when the server is slow/busy) they stall, which grinds all testing to a standstill. I do not know in advance that a hang will occur and don't watch my console 24x7 to restart it, so unfortunately I lose a bit of time.

How long is your new timeout on read requests set to? If the time limit is exceeded without reply, does the code simply continue working or do something else? For me things seem ok when the server responds in time but when it times out things do not go back to normal work.

Prime95 2005-04-30 14:22

Timeout is set to 90 seconds. If a timeout occurs an error should be displayed and computation should continue.

Peter Nelson 2005-04-30 15:43

I tried some menu 21 again today (about 15:30 GMT/UTC).

Times as follows:

About 20-25 secs, 4 secs, 1 sec, 8 sec, hang!

I let it sit hung for maybe five minutes (certainly longer than your 90 sec timeout) but it did not return to the menu, or display server info.

Neither did it display any error message. What message would it give on timeout and where would it report this message (on console screen, logfile etc)?

I checked the prime.log file and there is no record of an error.

akruppa 2005-04-30 16:11

Maybe George can make an mprime build with debugging symbols, then you could attach a debugger to the process when it hangs and see where it is stuck. Peter, do you have gdb installed?

Alex

Peter Nelson 2005-04-30 16:24

Using menu item 13 Manual comms, I found this....

sometimes it works and sends expected completion dates of all (maybe 4) TF work.

Other times it displays Contacting primenet server and hangs.
Other times it displays:

Sending expected completion date for M......
Sending expected completion date for M......
then hangs before sending the last two updates.

This appears on screen and is also logged to prime.log
However no messages appear about a timeout.

I conclude from the above that the stage when it hangs is not always the same.

My next plan was to log ethernet packets of the session using Ethereal and analyse what happens at the TCP/IP level.

THEN I re-read your post and realised you said put Debug=1 in PRIMENET.INI rather than PRIME.INI (note this is not documented in undoc.txt maybe it ought to be)

The next post contains the (surprising results of doing this).....

Peter Nelson 2005-04-30 16:35

FIRST HALF OF LOG (bold is my addition)
 
(menus shrunk for brevity)

pnelson@82-70-139-74:/usr/pnelson1/mprime/d> ./mprime -m
Main Menu

1. Test/Primenet
Your choice: 21

Contacting PrimeNet Server.
host = mersenne.org, port = 80
IP-addr = 64.66.6.250
GET [url]http://mersenne.org/cgi-bin/pnHttp.exe?ps&4&.&[/url]. HTTP/1.0

[B]Error in send timeout call: 22
Error in receive timeout call: 22[/B]
RECV: HTTP/1.1 200 OK
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:14:09 GMT
Content-type: text/html

pnResult=0
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]

Version: 4.0.031
Server Name: PrimeNet Server 1997-2003 Entropia, Inc.
Email Addr: [email]primenet@mersenne.org[/email], [url]http://mersenne.org/primenet[/url]

Hit enter to continue:
Main Menu

1. Test/Primenet
Your choice: 21

Contacting PrimeNet Server.
host = mersenne.org, port = 80
IP-addr = 64.66.6.250
GET [url]http://mersenne.orgHTTP/1.1[/url] 200 OK
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:14:09 GMT
Content-type: text/html

pnResult=0
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEma5333�/cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

[B]Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 400 Bad Request[/B]
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:14:35 GMT
Content-Length: 407
Content-Type: text/html

[B]<html><head><title>Error 400</title>[/B]

<meta name="robots" content="noindex">
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1"></head>

<body>

[B]<h2>HTTP Error 400</h2>

<p><strong>400 Bad Request</strong></p>

<p>Due to malformed syntax, the request could not be understood by the server. The client should not repeat the request without modifications.</p>[/B]

</body></html>
Trying relative URL
GET /cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 200 OK
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:14:36 GMT
Content-type: text/html


RECV: pnResult=0
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]

Version: 4.0.031
Server Name: PrimeNet Server 1997-2003 Entropia, Inc.
Email Addr: [email]primenet@mersenne.org[/email], [url]http://mersenne.org/primenet[/url]

Hit enter to continue:
Main Menu

1. Test/Primenet
Your choice: 21

Contacting PrimeNet Server.
host = mersenne.org, port = 80
IP-addr = 64.66.6.250
GET [url]http://mersenne.orgpnResult=0[/url]
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]
/cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 400 Bad Request
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:14:55 GMT
Content-Type: text/html
Content-Length: 87

[B]<html><head><title>Error</title></head><body>The parameter is incorrect. </body></html>
Trying relative URL[/B]
GET /cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 200 OK
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:14:55 GMT
Content-type: text/html


RECV: pnResult=0
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]

Version: 4.0.031
Server Name: PrimeNet Server 1997-2003 Entropia, Inc.
Email Addr: [email]primenet@mersenne.org[/email], [url]http://mersenne.org/primenet[/url]

Hit enter to continue:

Peter Nelson 2005-04-30 16:38

LOG CONTINUATION
 
Main Menu

1. Test/Primenet
Your choice: 21

Contacting PrimeNet Server.
host = mersenne.org, port = 80
IP-addr = 64.66.6.250
GET [url]http://mersenne.orgpnResult=0[/url]
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]
/cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

[B]Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 400 Bad Request[/B]
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:15:05 GMT
Content-Type: text/html
Content-Length: 87

[B]<html><head><title>Error</title></head><body>The parameter is incorrect. </body></html>
Trying relative URL[/B]
GET /cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

[B]Error in send timeout call: 22
Error in receive timeout call: 22[/B]
RECV: HTTP/1.1 200 OK
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:15:06 GMT
Content-type: text/html


RECV: pnResult=0
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]

Version: 4.0.031
Server Name: PrimeNet Server 1997-2003 Entropia, Inc.
Email Addr: [email]primenet@mersenne.org[/email], [url]http://mersenne.org/primenet[/url]

Hit enter to continue:
Main Menu

1. Test/Primenet
Your choice: 21

Contacting PrimeNet Server.
host = mersenne.org, port = 80
IP-addr = 64.66.6.250
GET [url]http://mersenne.orgpnResult=0[/url]
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]
/cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

[B]Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 400 Bad Request[/B]
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:15:15 GMT
Content-Type: text/html
Content-Length: 87

[B]<html><head><title>Error</title></head><body>The parameter is incorrect. </body></html>
Trying relative URL[/B]
GET /cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 200 OK
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:15:16 GMT
Content-type: text/html


RECV: pnResult=0
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]

Version: 4.0.031
Server Name: PrimeNet Server 1997-2003 Entropia, Inc.
Email Addr: [email]primenet@mersenne.org[/email], [url]http://mersenne.org/primenet[/url]

Hit enter to continue:
Main Menu

1. Test/Primenet
Your choice: 21

Contacting PrimeNet Server.
host = mersenne.org, port = 80
IP-addr = 64.66.6.250
GET [url]http://mersenne.orgpnResult=0[/url]
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]
/cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 400 Bad Request
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:15:20 GMT
Content-Type: text/html
Content-Length: 87

[B]<html><head><title>Error</title></head><body>The parameter is incorrect. </body></html>
Trying relative URL[/B]
GET /cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 200 OK
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:15:20 GMT
Content-type: text/html


RECV: pnResult=0
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]

Version: 4.0.031
Server Name: PrimeNet Server 1997-2003 Entropia, Inc.
Email Addr: [email]primenet@mersenne.org[/email], [url]http://mersenne.org/primenet[/url]

Hit enter to continue:
Main Menu

1. Test/Primenet
Your choice: 21

Contacting PrimeNet Server.
host = mersenne.org, port = 80
IP-addr = 64.66.6.250
GET [url]http://mersenne.orgpnResult=0[/url]
versionNumber=4
buildID=4.0.031
primenetServerName=PrimeNet Server 1997-2003 Entropia, Inc.
adminEmailAddr=primenet@mersenne.org, [url]http://mersenne.org/primenet[/url]
/cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

[B]Error in send timeout call: 22
Error in receive timeout call: 22
RECV: HTTP/1.1 400 Bad Request[/B]
Server: Microsoft-IIS/4.0
Date: Sat, 30 Apr 2005 16:15:24 GMT
Content-Type: text/html
Content-Length: 87

[B]<html><head><title>Error</title></head><body>The parameter is incorrect. </body></html>[/B]

[B][THIS IS WHERE IT HUNG! - PETER][/B]

Peter Nelson 2005-04-30 16:52

Comment on log
 
Well, I'm sure some of these errors aren't healthy!

And note the fourth "AdminEmail" response is corrupted. Perhaps other bits of the comms get similarly affected.

Note that although "parameter is incorrect" is usually followed by

Trying relative URL
GET /cgi-bin/pnHttp.exe?ps&4&.&. HTTP/1.0

This did not happen when the client hung.

Well perhaps there is some network congestion or introduction of errors somewhere between my PC and primenet server but generally checksums should detect and discard such packets for TCP/IP retransmission.

In any case this does not seem to affect any other apps at my end. Even if such errors occur, the client should be able to detect and deal with problems without locking up!

Over to you, George

Prime95 2005-04-30 18:28

Thanks Peter - that is a wealth of valuable information. I'll keep you posted on developments

Prime95 2005-04-30 19:43

Peter, try the newly uploaded Linux executables.

There was a bug in the implementation of the new SendPortNumber option. I'll rebuild Windows executables soon.

For some reason, your system does not support setting send and receive timeouts. My Fedora release does.

I cannot explain the hang. It looks like recv returned the data. This should be followed by another call to recv which should return zero (no more bytes to receive). I can only explain the hang by assuming this recv call did not return. This should not happen.

Peter Nelson 2005-05-01 00:03

George, that was quick!

[B]I have downloaded and installed the latest STATIC executable.

This appears to have cured the problem.[/B]

I only get code 200 http and not 400 at all.

Have tried lots of Help About Primenet but cannot obtain errors or hang.

Certainly it takes the primenet server a variable amount of time (maybe 1-15 seconds) to come back but it does (certainly for now).

I also tested the menu 13 Manual Communication option repeatedly and also could not get a hang.

I will let you know if it hangs again when checking in exponents but am quite hopeful it won't now.

So, this appears to have fixed the problem. Thankyou.

The distro this machine is running is Suse 9.1 Professional.
I can look up specific files if you want to verify support for particular network features. The NIC is a Marvell integrated one on a Gigabyte 865G based motherboard.

I can try the client on Suse 9.3 if you wish to see whether the timeout feature is supported on that more recent distro. Suse have a point release approx every six months.

Fedora Core is very popular in USA, and to a lesser extent in Europe although it's widespread deployment means it is well supported.
Over on this side of the Atlantic we like the Novell/Suse distro which originated in Germany and I think it has the lead in popularity over redhat/fc over here. I do have a FC3 disk around. Other people in our Linux user group use Gentoo, Mandrake, Debian etc. Perhaps there is a way to query whether a certain feature is supported before making use of it, or else use config options in prime.ini to enable/disable their use.

Kind regards, Peter

Mike 2005-05-03 01:43

Intel(R) Pentium(R) 4 CPU 2.60GHz
CPU speed: 2612.57 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 8 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 version 23.8, RdtscTiming=1
Best time for 384K FFT length: 13.807 ms.
Best time for 448K FFT length: 16.536 ms.
Best time for 512K FFT length: 18.868 ms.
Best time for 640K FFT length: 22.671 ms.
Best time for 768K FFT length: 27.598 ms.
Best time for 896K FFT length: 32.743 ms.
Best time for 1024K FFT length: 36.528 ms.
Best time for 1280K FFT length: 47.860 ms.
Best time for 1536K FFT length: 59.169 ms.
Best time for 1792K FFT length: 70.219 ms.
Best time for 2048K FFT length: 79.702 ms.



Intel(R) Pentium(R) 4 CPU 2.60GHz
CPU speed: 2612.51 MHz
CPU features: RDTSC, CMOV, Prefetch, MMX, SSE, SSE2
L1 cache size: 8 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 version 24.11, RdtscTiming=1
Best time for 512K FFT length: 18.406 ms.
Best time for 640K FFT length: 22.227 ms.
Best time for 768K FFT length: 27.372 ms.
Best time for 896K FFT length: 32.520 ms.
Best time for 1024K FFT length: 36.297 ms.
Best time for 1280K FFT length: 47.394 ms.
Best time for 1536K FFT length: 59.044 ms.
Best time for 1792K FFT length: 69.496 ms.
Best time for 2048K FFT length: 78.504 ms.
Best time for 58 bit trial factors: 11.232 ms.
Best time for 59 bit trial factors: 11.281 ms.
Best time for 60 bit trial factors: 11.224 ms.
Best time for 61 bit trial factors: 11.291 ms.
Best time for 62 bit trial factors: 12.231 ms.
Best time for 63 bit trial factors: 12.195 ms.
Best time for 64 bit trial factors: 14.845 ms.
Best time for 65 bit trial factors: 14.821 ms.
Best time for 66 bit trial factors: 14.797 ms.
Best time for 67 bit trial factors: 14.754 ms.


My P4 loves 24.11! :bounce:

moo 2005-05-03 03:17

give benchmarks for other versions.... mike

dave_0273 2005-05-03 11:27

[QUOTE=moo]give benchmarks for other versions.... mike[/QUOTE]

Mike has already posteded benchmarks for 2 different versions of prime95. However, he isn't really getting a speed boost because (i am assuming) that his P4 is 32bit, and 24.11 only gave a speed boost to 64 bit chips.

Can someone please correct me if I am wrong?

[QUOTE=Mike]My P4 loves 24.11! [/QUOTE]

I am a little confused. I don't see a significant speed boost... is there another reason why you like it over 23.8 or am I just reading the numbers wrong?

Mike 2005-05-03 12:44

The difference isn't significant, but version 24 is a TAD faster. :lol:

Best time for 512K FFT length: 18.868 ms. vs. 18.406 ms. (0.462 ms better.)
Best time for 1024K FFT length: 36.528 ms. vs. 36.297 ms. (0.231 ms better.)
Best time for 2048K FFT length: 79.702 ms. vs. 78.504 ms. (1.198 ms better.)

I'll take whatever advantage, however slight it is.

Joe O 2005-05-03 12:51

Starting with 512K FFT length and finishing with 2048K FFT length he has shown the following improvements:

2.51%
2.00%
0.83%
0.69%
0.64%
0.98%
0.21%
1.04%
1.53%

Peter Nelson 2005-05-03 14:20

1. The last FIVE posts would have been better placed under the so-called "perpetual benchmark thread".

2. You will see for example, I ran the new 24.11 client on a 630 machine and posted there not here.

3. If you are running/benchmarking this beta client please say EXACTLY what version of 24.11 you have (ie by date downloaded from the ftp server or subversion number) since it has been through at least FOUR updates already.

4. Yes, the improvement on P4 is only slight but it is a little faster which I welcome.

5. I like 24.11 because of the ability to benchmark trial factoring.

6. I also like it because of the massive performance jump if you are running in 64 bit mode (eg Athlon64 with 64 bit windows beta).

7. I hope that not too long in the future, we can get this for 64-bit Linux eg on Athlon64 or Intel 6xx / 8xx with EM64T.

8. My own opinion is that this thread is best left for feedback of experiences with the beta like bug reports, testing queries etc.

Dresdenboy 2005-05-03 15:11

EM64T results
 
I asked on the Aceshardware forum, if someone could test Prime95 21.11 on an EM64T CPU with 64 bit OS. And got results :smile:
[quote=Kevin Sheppard]3.0 Ghz Nocona's running CTP 2 (1433).

64 bit.

[code]
[Tue May 03 15:48:51 2005]
Compare your results to other computers at http://www.mersenne.org/bench.htm
That web page also contains instructions on how your results can be included.

Intel(R) Xeon(TM) CPU 3.00GHz
CPU speed: 2992.96 MHz
CPU features: RDTSC, CMOV, Prefetch, MMX, SSE, SSE2
L1 cache size: 16 KB
L2 cache size: 1024 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 version 24.11, RdtscTiming=1
Best time for 512K FFT length: 18.243 ms.
Best time for 640K FFT length: 21.581 ms.
Best time for 768K FFT length: 26.084 ms.
Best time for 896K FFT length: 31.616 ms.
Best time for 1024K FFT length: 35.122 ms.
Best time for 1280K FFT length: 45.999 ms.
Best time for 1536K FFT length: 55.608 ms.
Best time for 1792K FFT length: 67.340 ms.
Best time for 2048K FFT length: 75.183 ms.
Best time for 58 bit trial factors: 8.183 ms.
Best time for 59 bit trial factors: 8.164 ms.
Best time for 60 bit trial factors: 8.472 ms.
Best time for 61 bit trial factors: 8.775 ms.
Best time for 62 bit trial factors: 11.129 ms.
Best time for 63 bit trial factors: 13.693 ms.
Best time for 64 bit trial factors: 15.448 ms.
Best time for 65 bit trial factors: 17.129 ms.
Best time for 66 bit trial factors: 16.990 ms.
Best time for 67 bit trial factors: 16.923 ms.
[/code]

32bit.

[code]
[Tue May 03 15:49:40 2005]
Compare your results to other computers at http://www.mersenne.org/bench.htm
That web page also contains instructions on how your results can be included.

Intel(R) Xeon(TM) CPU 3.00GHz
CPU speed: 2992.76 MHz
CPU features: RDTSC, CMOV, Prefetch, MMX, SSE, SSE2
L1 cache size: 16 KB
L2 cache size: 1024 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 version 24.11, RdtscTiming=1
Best time for 512K FFT length: 18.239 ms.
Best time for 640K FFT length: 21.563 ms.
Best time for 768K FFT length: 26.156 ms.
Best time for 896K FFT length: 31.685 ms.
Best time for 1024K FFT length: 35.210 ms.
Best time for 1280K FFT length: 46.194 ms.
Best time for 1536K FFT length: 55.848 ms.
Best time for 1792K FFT length: 67.308 ms.
Best time for 2048K FFT length: 75.251 ms.
Best time for 58 bit trial factors: 9.176 ms.
Best time for 59 bit trial factors: 9.214 ms.
Best time for 60 bit trial factors: 9.108 ms.
Best time for 61 bit trial factors: 9.139 ms.
Best time for 62 bit trial factors: 12.831 ms.
Best time for 63 bit trial factors: 12.847 ms.
Best time for 64 bit trial factors: 14.657 ms.
Best time for 65 bit trial factors: 14.765 ms.
Best time for 66 bit trial factors: 14.673 ms.
Best time for 67 bit trial factors: 14.828 ms.[/code][/quote]

Link: [URL=http://aceshardware.com/forums/read_post.jsp?id=115129205&forumid=1]Original posting[/URL]

I made a table comparing the TF results:
[code] 32 bit 64 bit
Best time for 58 bit trial factors: 9.176 ms. 8.183 ms
Best time for 59 bit trial factors: 9.214 ms. 8.164 ms
Best time for 60 bit trial factors: 9.108 ms. 8.472 ms
Best time for 61 bit trial factors: 9.139 ms. 8.775 ms
Best time for 62 bit trial factors: 12.831 ms. 11.129 ms
Best time for 63 bit trial factors: 12.847 ms. 13.693 ms
Best time for 64 bit trial factors: 14.657 ms. 15.448 ms
Best time for 65 bit trial factors: 14.765 ms. 17.129 ms
Best time for 66 bit trial factors: 14.673 ms. 16.990 ms
Best time for 67 bit trial factors: 14.828 ms. 16.923 ms[/code]

Peter Nelson 2005-05-05 01:45

Please display 32 or 64 bit for benchmarking
 
Just an idea, but now that a 64 bit version exists, from the number of benchmarks getting posted it is not always immediately obvious whether the machine is running in 32 or 64 bit mode.

When the benchmark prints out version number, processor speed etc could it possibly also display 32bit or 64bit which should be readily available to the client as they are distinct executables.

This will standardise where to find this info in a benchmark and save people having to say or ask which it is.

penguin22 2005-05-10 20:21

Dual Booting with both new 24.11's
 
Hello,

I'm dual booting now with Windows XP Pro and XP Pro x64 Edition and would like to use the same data files to work on the same numbers between boots. Is this supported with this new version of Prime95 24.11? And if so what would be the best practice.

I have a common drive M that stays the same on both boots, but the C & D drive swap depending what OS I'm in. This shouldn't be a problem though I wouldn't think as long as I can direct the Prime95 clients to work from the data located on the M drive (if I can do that).

Any help would be greatly appreciated. Thanks!


All times are UTC. The time now is 09:15.

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