New Multi-Threaded Pi Program - Faster than SuperPi and PiFast

[CRFX]

Been playing with y-cruncher for a bit to stress test. Nice to spot memory errors when messing with timings.
Using an AMD 940 @3.5ghz 14x250mhz, 8gb ram. For the advanced swap runs was using 4 completely different drives, just what I had around. 150 WD VelociRaptor , 34 GB WD Raptor, 1TB WD, 500GB WD. Wound up giving me a bit over 200MB/s read/writing.

Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,499,968,327 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        10,000,000,000
Hexadecimal Digits:    8,304,820,238
Threading Mode:        4 threads
Computation Mode:      Advanced Swap
Swap Disks:            4
Working Memory:        7.03 GB

Start Time:            Sun May 30 03:45:01 2010
End Time:              Sun May 30 09:25:45 2010

Computation Time:      19,213.444 seconds
Total Time:            20,444.781 seconds

CPU Utilization:           212.99 %
Multi-core Efficiency:     53.24 %

Last Digits:
9763261541 1423749758 2083180752 2573977719 9605119144  :  9,999,999,950
9403994581 8580686529 2375008092 3106244131 4758821220  :  10,000,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   afbb8decc6372e5e02c4830fe70bf4be45f24dca0e6d814b1dcb7831b80f25a7



Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,499,971,970 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        5,000,000,000
Hexadecimal Digits:    4,152,410,119
Threading Mode:        4 threads
Computation Mode:      Advanced Swap
Swap Disks:            4
Working Memory:        6.89 GB

Start Time:            Tue Jan 06 00:46:02 2009
End Time:              Tue Jan 06 03:21:46 2009

Computation Time:      8,656.553 seconds
Total Time:            9,344.830 seconds

CPU Utilization:           207.96 %
Multi-core Efficiency:     51.99 %

Last Digits:
4384678622 1397184596 0181195416 0748430457 5386741865  :  4,999,999,950
0914971996 1298184401 9216126684 9425834935 5440797257  :  5,000,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   3e90e3bd4bda69b81b6bba5e8607b4645e3e4d386de0e1e70209677b8adeadb1



Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,499,964,882 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        2,500,000,000
Hexadecimal Digits:    2,076,205,060
Threading Mode:        4 threads
Computation Mode:      Basic Swap
Swap Disks:            1
Working Memory:        5.09 GB

Start Time:            Sun May 30 13:51:33 2010
End Time:              Sun May 30 14:46:07 2010

Computation Time:      3,064.283 seconds
Total Time:            3,274.682 seconds

CPU Utilization:           252.08 %
Multi-core Efficiency:     63.02 %

Last Digits:
0917027898 3554136437 7123165188 3528593128 0032489094  :  2,499,999,950
9228502005 4677489552 2459688725 5242233502 7255998083  :  2,500,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   6d96b2b94cefff2e1c5d4262bea9ea1781508808f286eb8e16b268f48db790d0



Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,500,033,606 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        1,000,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        4 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        4.75 GB

Start Time:            Mon Jan 05 22:48:11 2009
End Time:              Mon Jan 05 23:00:23 2009

Computation Time:      698.078 seconds
Total Time:            732.356 seconds

CPU Utilization:           388.09 %
Multi-core Efficiency:     97.02 %

Last Digits:
6434543524 2766553567 4357021939 6394581990 5483278746  :  999,999,950
7139868209 3196353628 2046127557 1517139511 5275045519  :  1,000,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   8aa91c4c41b343c601786c9d3e1daff73d41adcccde3af4e112263b548cf620a



Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,499,962,659 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        500,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        4 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        2.42 GB

Start Time:            Mon Jan 05 23:03:11 2009
End Time:              Mon Jan 05 23:08:44 2009

Computation Time:      315.981 seconds
Total Time:            333.237 seconds

CPU Utilization:           379.80 %
Multi-core Efficiency:     94.95 %

Last Digits:
3896531789 0364496761 5664275325 5483742003 7847987772  :  499,999,950
5002477883 0364214864 5906800532 7052368734 3293261427  :  500,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   ce7f605a99f54b62064bd9e20452d0334d80ff539a58a59b1623b0a7eeb2b6cd



Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,499,970,806 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        250,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        4 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        1.25 GB

Start Time:            Mon Jan 05 23:11:59 2009
End Time:              Mon Jan 05 23:14:27 2009

Computation Time:      138.842 seconds
Total Time:            147.274 seconds

CPU Utilization:           386.50 %
Multi-core Efficiency:     96.62 %

Last Digits:
3673748634 2742427296 0219667627 3141599893 4569474921  :  249,999,950
9958866734 1705167068 8515785208 0067520395 3452027780  :  250,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   c67582ba838a1805035d66a9aa0012a5b229242cc2f0142706ce4ca19982209a



Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,499,962,238 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        100,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        4 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        536 MB

Start Time:            Mon Jan 05 23:27:29 2009
End Time:              Mon Jan 05 23:28:21 2009

Computation Time:      48.551 seconds
Total Time:            52.071 seconds

CPU Utilization:           379.88 %
Multi-core Efficiency:     94.97 %

Last Digits:
9948682556 3967530560 3352869667 7734610718 4471868529  :  99,999,950
7572203175 2074898161 1683139375 1497058112 0187751592  :  100,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   6aadf6c6c06bcd62113787fa59d8a1a9a24150dd43ca557d1677b4b4ff6a46b1



Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,499,961,443 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        50,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        4 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        317 MB

Start Time:            Mon Jan 05 23:30:54 2009
End Time:              Mon Jan 05 23:31:18 2009

Computation Time:      21.789 seconds
Total Time:            23.675 seconds

CPU Utilization:           378.19 %
Multi-core Efficiency:     94.54 %

Last Digits:
4127897300 0153683630 8346732220 0943329365 1632962502  :  49,999,950
5130045796 0464561703 2424263071 4554183801 7945652654  :  50,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   99b5ff37390549200f27bb917a4efa88987414e8cf0d2545112712f971118986



Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  CRFX


Processor(s):          AMD Phenom(tm) II X4 940 Processor
Logical Cores:         4
Physical Memory:       8,588,873,728 bytes  ( 8.00 GB )
CPU Frequency:         3,499,974,798 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        25,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        4 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        207 MB

Start Time:            Mon Jan 05 23:30:00 2009
End Time:              Mon Jan 05 23:30:11 2009

Computation Time:      9.945 seconds
Total Time:            11.036 seconds

CPU Utilization:           368.58 %
Multi-core Efficiency:     92.14 %

Last Digits:
3803750790 9491563108 2381689226 7224175329 0045253446  :  24,999,950
0786411592 4597806944 2455112852 2554677483 6191884322  :  25,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   9a4bb1bb0b6858fa01b70a9a0ba412b09c8f9011565814aa1db8631060a5e5c7

[poke349]

Updated.

@ Alpha, lol. Gotta love those old dualies... I think there's an FX system somewhere in the list. Don't remember. Been a while.


@ CRFX
Very nice large runs.
Looks like you've got the second entry for 10b for someone from XS (excluding myself) - The only single socket AMD entry.
5.3 hours for 10b: Given that it's XS, not too many people here have the patience for that.

[Boogerlad]

the stress test is even more stressful then linpack? Does it report instability?

[poke349]

the stress test is even more stressful then linpack? Does it report instability?

Only on Core i7 with HT enabled do I see this - which it produces more heat than Linpack and prime95. On most other systems, Linpack and prime95 are still more stressful.

y-cruncher is actually turning out to be one of the better torture tests for memory.
For CPU, Linpack, prime95, and CoreDamage (which is a new one) are still better in most cases.

Yes, it will report errors. (Both in the stress-test and in normal computations.)

Here's an example of an error in the stress-test:
http://www.xtremesystems.org/forums/...&postcount=106
(I grabbed this from the Thuban OC thread.)

[skycrane]

Hey poke, is there a waay to run the large ones in a continus mode?

Where you could start out. At 1b didgets aand have the program run all the tests up to 1t. ?

[poke349]

Hey poke, is there a waay to run the large ones in a continus mode?

Where you could start out. At 1b didgets aand have the program run all the tests up to 1t. ?

I see where you're going. The batch mode only does ram only computations.
I guess I could add a mode selection too... It'd be something I can easily add to v0.5.4. (which I'm planning a late summer release - been really busy right now since I'm graduating in a couple of weeks... Then I have a hella crazy summer coming up... lolz)

The "Standard", and "SuperPi" sizes auto-check the digits. It does it by checking the last 10 digits with known results that are hard-coded into the program.

At the time I wrote the benchmark feature, I only had access to digits up to 100b and 128G. Which is where the benchmark options stop.
I guess I could extend them now. I have all the digits up to 250b, and I also have the 100 digits leading up to 500b, 1t, and 2.5t.
(I still haven't deleted the digits from the 250b run that I did on my workstation back in the beginning of April. )

*I suppose I could ask Shigeru Kondo or Fabrice Bellard for the results for 256G, 512G, 1T, and 2T...



But in all seriousness, unless you're going on vacation or something, is a batch mode even necessary for anything really large?
At some point, I should probably add cmdline options to allow for scripting.

[skycrane]

ive got some good and bad news about the new version, for the first few runs, it was faster than the original, but in the middle of the pack it was slower, and running the large numbers, it is definatly faster... for the 10b its just over 10,000 sec. it droped almost 3000 sec of the time.

im doing a 25b run right now, ill get the new numbers to ya when this ones done

[poke349]

ive got some good and bad news about the new version, for the first few runs, it was faster than the original, but in the middle of the pack it was slower, and running the large numbers, it is definatly faster... for the 10b its just over 10,000 sec. it droped almost 3000 sec of the time.

im doing a 25b run right now, ill get the new numbers to ya when this ones done

Wait, so it's slower for "middle"? Where's "middle"?
I do recall you having 32GB of ram and running 5b all in ram. I dunno what you're running right now though.

There really shouldn't any cases where v0.5.3 is slower than v0.5.2.

[skycrane]

hey poke, here are my new runs

[poke349]

hey poke, here are my new runs

Woah! Nice!!!
That's the first 25b from XS. (barring myself)

As for why 2.5b and 5b are slower, I have a feeling you're running some slightly different settings than last time. Maybe it has something to do with the memory?

And on the other end:
v0.5.3 is never more than 15% faster than v0.5.2, yet some of the smaller runs are LOT faster than that.

The reason why v0.5.3 should never be slower than v0.5.2, is that it simply has fewer operations. Nothing else is different.
Same memory access patterns, same function calls - everything is the same. Just less of it. So it should be architecture independent.

[Alpha]

i may have to run some crazy runs over the next while. my file server at home has 7x 1tb drives (maybe upping to 8 this weekend)....

2gb ram + Q8400 = slow....

[poke349]

i may have to run some crazy runs over the next while. my file server at home has 7x 1tb drives (maybe upping to 8 this weekend)....

2gb ram + Q8400 = slow....

Nice! Anything more than 4 HDs is enough to get some serious speed.
Going from 1 HD to 4 HDs raises CPU utilization from 10% to 60% on my Core i7 rig.


I can't wait till I get home for the summer. I've got 6 x 2 TB next to me right now. I wanna see what they can do when I throw them in my workstation.
(two of them are filled with Anime, but that won't be a problem since I have a couple of empty 2TB WD green drives waiting for me at home)


Just wondering...

Are there any other programs that reward having a lot of memory?
As far as I can tell, all benchmarks are CPU, GPU, memory speed, disk speed. But there's nothing on memory or disk quantity.

[skycrane]

wish me luck on my 3rd try for 50b, im not sure whats wrong, it just wont finish

not sure if its the program, my os, or the hdd. ill let you know in a few hours, its at 24% now

[poke349]

wish me luck on my 3rd try for 50b, im not sure whats wrong, it just wont finish

not sure if its the program, my os, or the hdd. ill let you know in a few hours, its at 24% now

3 times?!?!

What errors are you getting? If I can see a screenshot or something, I can usually tell right away whether it's software or hardware.

Good luck.

[skycrane]

not sure what errors im getting, the comp just locks up and i get no video. so ive had to reset the comp. but it looks good im at 48% now im keeping a track on it every hour or so....

[poke349]

not sure what errors im getting, the comp just locks up and i get no video. so ive had to reset the comp. but it looks good im at 48% now im keeping a track on it every hour or so....

The only case I think of that may cause that (aside from instability or software bug), is having the pagefile on one of the swap drives. But I doubt this is the case because I've never encountered it after I fixed a related bug between v0.1.0 and v0.2.1. (even when i tried cause it intentionally)

Windows sometimes does some really stupid things when a memory hog eats up all the ram (such as with y-cruncher) And if it decides to thrash the pagefile on one of the swapdrives (after the power-saving turns off the screen), the system will hang for a VERY LONG TIME. High and real-time priority may also cause this.

I really have no idea, but if you hear the HDs going crazy during a hang, then this might be the case.
I have my swap drives on separate physical drives from the boot or pagefile.

At least these are my test scenarios.
I haven't really tested it much with swapping on the boot drive since I have dedicated crunch drives on both my test-machines. (I also use ram-drives for smaller tests...)

[poke349]

So I'm finally home after graduation... Now I get to play with my workstation again.


Now about those 2TB drives that I had mentioned earlier...
Well... I took them home... ALL of them...
And getting through airport security took an embarrassingly long time - they had check every single one of my HDs (10 total) and then bomb-swipe everything...


I took these pictures right after I finished moving all my Anime to that 2TB WD Green drive that's on the floor.
That way, I can clear up all the fast 7200 RPM drives that I have in that machine...

To get this to work, I had to evict that old 750GB (the other HD that's on the floor). Then... A bit of case modding.
I still need to get an air filter for the front HDs...

(click to enlarge)

[skycrane]

looks nice there poke.... i hope you get some nice runs in.

looks like ive figured out the problem, with all the mem and extra drives and fans i had running on my rig, it looks like the psu was not up to the task, looks like it took a crap on me, ill know more this weekend when i get home to work on it.

[poke349]

looks nice there poke.... i hope you get some nice runs in.

looks like ive figured out the problem, with all the mem and extra drives and fans i had running on my rig, it looks like the psu was not up to the task, looks like it took a crap on me, ill know more this weekend when i get home to work on it.

Ah, what PSU was it?

Speaking of PSU issues, I'm not entirely sure that mine will hold up with the extra load of HDs.
It's a 1000W, but I don't know how much the extra 6 Hitachi drives will pull.

Right now, I'm running 100b on it as a quick benchmark. When it finishes, I need to do a maximum power-draw test to see if my PSU will hold up.
Like, I'm gonna run all the following at the same time:

• Stress test with 40GB ram.
• Advanced Swap run using 20GB of ram and all 11 HDs.
• Crysis at Max Settings.
• Run the DVD drive.
• All fans to max speed.

Also, here's my new HD-setup:
Every single SATA port is occupied...
Every single HD drive bay is also occupied... (including the 5-in-3 box in the front)



On Motherboard:
SATA 0: 1.5 TB Seagate Barracuda (Boot)
SATA 1: 2.0 TB Hitachi Deskstar - C0
SATA 2: 1.0 TB Seagate Barracuda - A0
SATA 3: 1.0 TB Seagate Barracuda - A1
SATA 4: 1.0 TB Seagate Barracuda - A2
SATA 5: 1.0 TB Seagate Barracuda - A3

Controller: SuperMicro AOC-SAT2-MV8
SATA 0: 2.0 TB Hitachi Deskstar - C1
SATA 1: Front eSATA port (no HD plugged in)
SATA 2: 2.0 TB Hitachi Deskstar - B0
SATA 3: 2.0 TB Hitachi Deskstar - B1
SATA 4: 2.0 TB Hitachi Deskstar - B2
SATA 5: 2.0 TB Hitachi Deskstar - B3
SATA 6: Back eSATA port (no HD plugged in)
SATA 7: Back eSATA port (no HD plugged in)

*The BD-ROM drive is a DAEMON Tools virtual drive. The other one is the "real" DVD drive.
*The external 1.5 TB is just a USB drive and is not part of the setup. This is the one that's sitting on top of the rig in those pictures.

It looks like these 11 drives can sustain 1.0 - 1.1 GB/s of sequential IO. (as measured while running y-cruncher)
That's a bit more than I expected since I thought the syncing overhead would lead to diminishing return... I guess 11 drives isn't enough to reach that point yet.


Either way... 64GB of ram + 11 hard drives = programmer's dream
It doesn't get much better than this for someone with a college student budget...

Though I won't be running too many benchmarks on it. It's my primary dev-machine for the summer. (I left my i7 rig in Chicago.)
So I'll be writing and testing a lot of new code on it.

[poke349]

Here it is: 100b with 11 hard drives

(click to enlarge)


5 HDs:
31 hours
74% CPU utilization

11 HDs:
28.3 hours
83% CPU utilization

Overall, the scaling is crap because it's almost entirely limited by CPU/memory.
With 64GB of ram and 5 HDs, most of the time is already spent on computation rather than disk access. So adding more HDs is at the point of diminishing return.
But for those with less ram, it will be much more limited by disk access.

The disk bandwidth itself does seem to scale very well:
5 HDs ~ 500 MB/s
11 HDs ~ 1.0 - 1.1 GB/s

[mAJORD]

Bump!!

Poor thread had died a little..

Some first time Thuban Benchmark runs.

X6 1055T @ 4.18Ghz 1.47v Air Cooling (Noctua)

DDR3 1600 6-7-7-18

25M: 6.432s
50M: 13.513s
100M: 29.053s

Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  None Specified - You can edit this in "Username.txt".


Processor(s):          AMD Phenom(tm) II X6 1055T Processor
Logical Cores:         6
Physical Memory:       4,293,386,240 bytes  ( 4.00 GB )
CPU Frequency:         4,186,106,687 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        25,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        8 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        208 MB

Start Time:            Sat Jul 10 21:55:04 2010
End Time:              Sat Jul 10 21:55:12 2010

Computation Time:      6.432 seconds
Total Time:            7.519 seconds

CPU Utilization:           465.62 %
Multi-core Efficiency:     77.60 %

Last Digits:
3803750790 9491563108 2381689226 7224175329 0045253446  :  24,999,950
0786411592 4597806944 2455112852 2554677483 6191884322  :  25,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   7399e762628c843c17516c127bb1d9c247a600b48d916d0305a6ccd64f40b5cb

Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  None Specified - You can edit this in "Username.txt".


Processor(s):          AMD Phenom(tm) II X6 1055T Processor
Logical Cores:         6
Physical Memory:       4,293,386,240 bytes  ( 4.00 GB )
CPU Frequency:         4,186,102,591 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        50,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        8 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        318 MB

Start Time:            Sat Jul 10 21:56:45 2010
End Time:              Sat Jul 10 21:57:01 2010

Computation Time:      13.513 seconds
Total Time:            15.350 seconds

CPU Utilization:           495.00 %
Multi-core Efficiency:     82.50 %

Last Digits:
4127897300 0153683630 8346732220 0943329365 1632962502  :  49,999,950
5130045796 0464561703 2424263071 4554183801 7945652654  :  50,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   9460f9a2ad510bd802005fc6310ae79eeba4e362cc8b28af10724205cb851b55

Validation Version:    1.1

Program:               y-cruncher - Gamma to the eXtReMe!!!     ( www.numberworld.org )
                       Copyright 2008-2010 Alexander J. Yee    ( a-yee@northwestern.edu )


User:                  None Specified - You can edit this in "Username.txt".


Processor(s):          AMD Phenom(tm) II X6 1055T Processor
Logical Cores:         6
Physical Memory:       4,293,386,240 bytes  ( 4.00 GB )
CPU Frequency:         4,186,095,904 Hz

Program Version:       0.5.3 Build 9134b (fix 2) (x64 SSE3 - Windows ~ Kasumi)
Constant:              Pi
Algorithm:             Chudnovsky Formula
Decimal Digits:        100,000,000
Hexadecimal Digits:    Disabled
Threading Mode:        8 threads
Computation Mode:      Ram Only
Swap Disks:            0
Working Memory:        537 MB

Start Time:            Sat Jul 10 21:59:04 2010
End Time:              Sat Jul 10 21:59:37 2010

Computation Time:      29.053 seconds
Total Time:            32.408 seconds

CPU Utilization:           508.38 %
Multi-core Efficiency:     84.73 %

Last Digits:
9948682556 3967530560 3352869667 7734610718 4471868529  :  99,999,950
7572203175 2074898161 1683139375 1497058112 0187751592  :  100,000,000

Timer Sanity Check:        Passed
Frequency Sanity Check:    Passed
ECC Recovered Errors:      0

----

Checksum:   73ba1f3018563cda1aa47c6081c1138cc54bf882f4aca3ef1f40a9236efdaab2

[poke349]

Thanks for the bump! Apparently multi-threaded Pi isn't very interesting in the SuperPi-addicted world of benchmarking, lol...

Those thubans numbers are right on with the quad-core i7s. And cheaper too . Looks like AMD is making a nice come back...
There still slower than the Gulftowns, but that doesn't count since they're in completely different price brackets. lol

I can't wait for bulldozer... If it's gonna beat out Sandy Bridge by that much (especially with all the new instructions sets it will have), I might consider switching to the AMD compiler. (There's no way the Intel compiler will support AMD-only instructions. )

EDIT: Looks like the AMD x86 Open64 Compiler is only for Linux... meh...

[mAJORD]

should interest ppl! , it makes a great bench, and responds to mem performance enough to make it fun.

Ive got some performance scaling from ver 0.44 to 0.53 ill have to post when i get home.. (if iever get out of this damn que im standing in!)

[Calmatory]

Why aren't you just using GCC? I'd consider it way more vendor-neutral than Intel's compiler, being competitive speed-wise and for sure more portable solution.

Edit: The only problem I see with GCC compilers is the poor performing standard library, glibc. Agner Fog has shown this(by comparing the speed to Intel's, Microsoft's, Apple's and some other implementations) and preached about this for years without any real effect.

Actually, I'd see that there could be some potential to even test this out by people.

[poke349]

Why aren't you just using GCC? I'd consider it way more vendor-neutral than Intel's compiler, being competitive speed-wise and for sure more portable solution.

Edit: The only problem I see with GCC compilers is the poor performing standard library, glibc. Agner Fog has shown this(by comparing the speed to Intel's, Microsoft's, Apple's and some other implementations) and preached about this for years without any real effect.

Actually, I'd see that there could be some potential to even test this out by people.

The Intel Compiler seems to beat GCC in optimizing my code on all architectures including both Intel and AMD. (Sometimes by as much as 40%.)*
For that matter, even Visual Studio beats GCC for the vast majority of my code.
I use intrinsics instead of assembly. So the compiler matters since it needs to schedule the instructions properly to minimize stalls...

(A lot of high-performance programs are hand-optimized in assembly, so the compiler doesn't matter that much for those.)


The greatest thing I've noticed about the Intel Compiler is its ability to optimize non-branching code.
I do a lot of very aggressive loop unrolling and forced function inlining. Although they significantly speed up the code on all compilers, ICC gets the most benefit since all that unrolling and inlining creates huge sections of code (200 - 5000 instructions) that are completely branch-free and consist only of data moves and arithmetic.

So in the end ICC almost always ends up being the fastest - even for AMD processors.
Without any loop-unrolling, Visual Studio comes out on top. GCC is somewhere in between...

But if GCC is the only way for me to access AMD's FMA4 and XOP then I'll try it. It might even end up being faster than using the Intel Compiler with just AVX.


*Also, I don't use any of the libraries, nor do I use any C++ classes. Most of the code will actually compile in C.

The only "libraries" I use are:
memcpy() + memset()
printf() + scanf() - and their unicode versions.
exit() + system("pause")
The FileIO functions fread() + fwrite() + getpos() + etc...
WinAPI threads or OpenMP
WinAPI raw fileIO functions
sin() and cos(), but that is only in the "Constructing FFT lookup tables..." step - which takes a negligible amount of time to being with...
log(), but it is used very rarely, none in anything performance critical