Heavy Metal! - high cpu count

Message boards : Number crunching : Heavy Metal! - high cpu count
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · Next

AuthorMessage
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 3160
Credit: 1,282,604,591
RAC: 6,640
United States
Message 1984601 - Posted: 11 Mar 2019, 19:58:27 UTC - in response to Message 1984591.  

More "Heavy Metal" :)

https://setiathome.berkeley.edu/show_host_detail.php?hostid=8431394

About 80 cores worth!

Tom


This is the same system you posted back in Message ID 1960507

This system is on the 2nd page of the LeaderBoard and moving up. It has "only" 80 'cpus' and [4] NVIDIA Quadro P4000 (4095MB) driver: 382.59 OpenCL: 1.2's

https://setiathome.berkeley.edu/show_host_detail.php?hostid=8431394

Tom

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 1984601 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 4996
Credit: 276,046,078
RAC: 462
Message 1984669 - Posted: 12 Mar 2019, 4:03:02 UTC - in response to Message 1984601.  

More "Heavy Metal" :)

https://setiathome.berkeley.edu/show_host_detail.php?hostid=8431394

About 80 cores worth!

Tom


This is the same system you posted back in Message ID 1960507

This system is on the 2nd page of the LeaderBoard and moving up. It has "only" 80 'cpus' and [4] NVIDIA Quadro P4000 (4095MB) driver: 382.59 OpenCL: 1.2's

https://setiathome.berkeley.edu/show_host_detail.php?hostid=8431394

Tom


Now its on the 3rd page. That must be why I didn't recognize it :)

Tom
A proud member of the OFA (Old Farts Association).
ID: 1984669 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 4996
Credit: 276,046,078
RAC: 462
Message 1994784 - Posted: 22 May 2019, 19:58:30 UTC

https://setiathome.berkeley.edu/show_host_detail.php?hostid=8703499

If I am reading it right, this system is nearly brand new. Just started here on April 15, 2019

88 cores.
A proud member of the OFA (Old Farts Association).
ID: 1994784 · Report as offensive
Profile TimeLord04
Volunteer tester
Avatar

Send message
Joined: 9 Mar 06
Posts: 20374
Credit: 33,933,039
RAC: 23
United States
Message 1994786 - Posted: 22 May 2019, 20:32:49 UTC - in response to Message 1994784.  
Last modified: 22 May 2019, 20:40:34 UTC

https://setiathome.berkeley.edu/show_host_detail.php?hostid=8703499

If I am reading it right, this system is nearly brand new. Just started here on April 15, 2019

88 cores.

Probably 44 Physical Cores/88 Threads. Most likely Dual Xeon CPUs on the MOBO to do this.

If you look at my 4 Core i7 7700K @ 4.2GHz, "iMac 18,3" Hackintosh, System ID: 7952666 you'll see that BOINC is reporting 8 Cores. It's really 4 Cores/8 Threads.

In fact, my curiosity made me search the Processor.

E5-2699 v4 Xeon. 22 Cores/44Threads. (e-Bay: $976.89 - USED)

From Dell, this thing, new, goes for $8K+ NEW, and $10.5K+ NEW Non-Dell. So, NOT for the faint-hearted! I wonder what the MOBO goes for to support two of these MONSTROSITIES!!!???


TL

[EDIT:]

Just looked at one of his Completed and Validated Units:

Task 7700486492

Name blc25_2bit_guppi_58340_51498_HIP25486_0067.16842.409.20.29.76.vlar_1

Workunit 3485197161

Created 22 May 2019, 10:29:32 UTC
Sent 22 May 2019, 16:43:41 UTC
Report deadline 14 Jul 2019, 21:43:23 UTC
Received 22 May 2019, 19:59:07 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x00000000)
Computer ID 8703499

Run time 2 hours 50 min 2 sec
CPU time 2 hours 41 min 33 sec

Validate state Valid
Credit 83.47
Device peak FLOPS 3.44 GFLOPS
Application version SETI@home v8 v8.08 (alt)
windows_x86_64

Peak working set size 87.77 MB
Peak swap size 88.95 MB
Peak disk usage 0.06 MB
Stderr output

<core_client_version>7.14.2</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_v8 8.00 Visual Studio/Microsoft Visual C++
libboinc: 7.6.33

Work Unit Info:
...............
WU true angle range is : 0.011972
Optimal function choices:
--------------------------------------------------------
name timing error
--------------------------------------------------------
v_BaseLineSmooth (no other)
v_avxGetPowerSpectrum 0.000080 0.00000
avx_ChirpData_b 0.003074 0.00000
v_avxTranspose4x16ntw 0.001128 0.00000
AK SSE folding 0.001396 0.00000
Triplet: peak=11.38574, time=52.55, period=37.48, d_freq=7893452242.13, chirp=2.2707, fft_len=256
Pulse: peak=7.691181, time=45.99, period=19.57, d_freq=7893445943.49, score=1.041, chirp=-4.2927, fft_len=4k
Pulse: peak=8.018477, time=45.99, period=19.57, d_freq=7893445942.68, score=1.085, chirp=-4.3105, fft_len=4k
Spike: peak=26.73931, time=85.9, d_freq=7893450051.83, chirp=-15.121, fft_len=128k
Spike: peak=29.0706, time=85.9, d_freq=7893450051.84, chirp=-15.126, fft_len=128k
Spike: peak=24.87879, time=85.9, d_freq=7893450051.81, chirp=-15.127, fft_len=128k
Spike: peak=30.1932, time=85.9, d_freq=7893450051.84, chirp=-15.131, fft_len=128k
Spike: peak=26.23547, time=85.9, d_freq=7893450051.81, chirp=-15.132, fft_len=128k
Spike: peak=29.95384, time=85.9, d_freq=7893450051.84, chirp=-15.136, fft_len=128k
Spike: peak=26.31975, time=85.9, d_freq=7893450051.81, chirp=-15.137, fft_len=128k
Spike: peak=28.32604, time=85.9, d_freq=7893450051.84, chirp=-15.141, fft_len=128k
Spike: peak=25.10128, time=85.9, d_freq=7893450051.81, chirp=-15.142, fft_len=128k
Spike: peak=25.45118, time=85.9, d_freq=7893450051.84, chirp=-15.146, fft_len=128k
Pulse: peak=2.195631, time=45.82, period=3.255, d_freq=7893442342.54, score=1.029, chirp=-19.301, fft_len=64
Pulse: peak=7.642792, time=45.9, period=22.31, d_freq=7893442649.67, score=1.003, chirp=-33.633, fft_len=2k
Pulse: peak=9.708794, time=45.99, period=24.7, d_freq=7893450401.9, score=1.032, chirp=-33.651, fft_len=4k
Pulse: peak=1.306258, time=45.84, period=1.556, d_freq=7893443961.74, score=1.043, chirp=35.052, fft_len=512
Pulse: peak=1.615052, time=45.82, period=2.116, d_freq=7893445940.08, score=1.115, chirp=37.748, fft_len=256
Pulse: peak=3.266259, time=45.86, period=6.957, d_freq=7893452347.65, score=1.003, chirp=-39.452, fft_len=1024
Pulse: peak=2.717923, time=45.86, period=5.54, d_freq=7893452672.16, score=1.031, chirp=-44.561, fft_len=1024
Pulse: peak=4.509082, time=45.86, period=9.992, d_freq=7893450871.11, score=1.016, chirp=52.153, fft_len=1024
Pulse: peak=2.715545, time=45.84, period=5.029, d_freq=7893451338.67, score=1.01, chirp=62.867, fft_len=512
Pulse: peak=4.210422, time=45.82, period=8.495, d_freq=7893451288.07, score=1.054, chirp=64.713, fft_len=128
Pulse: peak=6.498947, time=45.9, period=17.67, d_freq=7893452112.4, score=1.026, chirp=-66.451, fft_len=2k
Pulse: peak=1.090146, time=45.84, period=1.325, d_freq=7893447170.98, score=1.045, chirp=81.174, fft_len=512
Triplet: peak=10.70693, time=45.54, period=16.64, d_freq=7893447424.23, chirp=-82.061, fft_len=2k
Pulse: peak=6.728341, time=45.9, period=17.45, d_freq=7893452912.7, score=1.063, chirp=-94.301, fft_len=2k
Pulse: peak=7.646245, time=45.9, period=18.67, d_freq=7893451475.28, score=1.01, chirp=98.133, fft_len=2k

Best spike: peak=30.1932, time=85.9, d_freq=7893450051.84, chirp=-15.131, fft_len=128k
Best autocorr: peak=17.12929, time=85.9, delay=1.683, d_freq=7893445500.97, chirp=-27.649, fft_len=128k
Best gaussian: peak=0, mean=0, ChiSq=0, time=-2.124e+11, d_freq=0,
score=-12, null_hyp=0, chirp=0, fft_len=0
Best pulse: peak=1.615052, time=45.82, period=2.116, d_freq=7893445940.08, score=1.115, chirp=37.748, fft_len=256
Best triplet: peak=11.38574, time=52.55, period=37.48, d_freq=7893452242.13, chirp=2.2707, fft_len=256

Flopcounter: 38654324279242.281250

Spike count: 10
Autocorr count: 0
Pulse count: 16
Triplet count: 2
Gaussian count: 0
12:58:30 (65224): called boinc_finish(0)

</stderr_txt>
]]>
TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos
ID: 1994786 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 13971
Credit: 40,757,560
RAC: 67
United Kingdom
Message 1994787 - Posted: 22 May 2019, 20:37:31 UTC - in response to Message 1994786.  

MONSTROSITIES would be the Intel-Xeon E5-4669 v4 @ $7000 ea
ID: 1994787 · Report as offensive
Profile TimeLord04
Volunteer tester
Avatar

Send message
Joined: 9 Mar 06
Posts: 20374
Credit: 33,933,039
RAC: 23
United States
Message 1994788 - Posted: 22 May 2019, 20:47:38 UTC - in response to Message 1994787.  

MONSTROSITIES would be the Intel-Xeon E5-4669 v4 @ $7000 ea


NEW FROM DELL: E5-4669 v4 $28,123.94

INSANE!!!!!


TL
TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos
ID: 1994788 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 4996
Credit: 276,046,078
RAC: 462
Message 1995771 - Posted: 29 May 2019, 13:16:10 UTC

Here is the first AMD (server) EYPC seti cruncher I have noticed on the LeaderBoard. Looks like he is running a Dual MB with SMT (96 cores).
And a stock Linux version of Boinc.

https://setiathome.berkeley.edu/show_host_detail.php?hostid=7186677

He is running along with a RAC of 58k+ on shear "muscle" :)

Located here: https://setiathome.berkeley.edu/top_hosts.php?sort_by=expavg_credit&offset=80

Tom
A proud member of the OFA (Old Farts Association).
ID: 1995771 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 4996
Credit: 276,046,078
RAC: 462
Message 2013329 - Posted: 25 Sep 2019, 23:10:13 UTC
Last modified: 25 Sep 2019, 23:15:28 UTC

Thank you Moderator.

Just ran across a new to me "Heavy Metal" system: https://setiathome.berkeley.edu/show_host_detail.php?hostid=8819281

According to Intel https://ark.intel.com/content/www/us/en/ark/products/192437/intel-xeon-gold-6230-processor-27-5m-cache-2-10-ghz.html this is a 20c/40t cpu chip which makes it likely they are running a dual cpu motherboard (its reporting 80 cpus).

Since the operator is also running
[4] NVIDIA GeForce RTX 2080 Ti (4095MB)
gpus, with Tbar's gpu app under linux, the system is cranking out a lot of RAC's.

After looking at some of the processing times on the gpu tasks I have to wonder if the "-nobs" command line is being used? The wall clock time is significantly slower than the cpu time.

According to "details" the system is crunching cpu tasks. I am not sure how many cpu threads are being used though. I finally fell over some CPU tasks and while one was probably a noise bomb the other two ran in less than a hour & 45 minutes with is more than brisk for a "2.10Ghz" server CPU.

So welcome, welcome :)

Tom
A proud member of the OFA (Old Farts Association).
ID: 2013329 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 11807
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2013339 - Posted: 26 Sep 2019, 1:04:56 UTC

He's also leaving performance on the table by not switching over to the CUDA101 app.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2013339 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 4996
Credit: 276,046,078
RAC: 462
Message 2015114 - Posted: 11 Oct 2019, 23:26:24 UTC - in response to Message 2013329.  
Last modified: 11 Oct 2019, 23:36:28 UTC


Just ran across a new to me "Heavy Metal" system: https://setiathome.berkeley.edu/show_host_detail.php?hostid=8819281


He hasn't connected since Oct 10th.... so he has either re-deployed the system, something broke or he is experiencing the California wind driven power outage first hand. Ummm.... his profile is in German. I don't think he is in California :)

Tom
A proud member of the OFA (Old Farts Association).
ID: 2015114 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 11807
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2015177 - Posted: 12 Oct 2019, 16:16:39 UTC - in response to Message 2015114.  

Matthias said he got real busy at work and crunching had to take a backseat for a while.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2015177 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9772
Credit: 572,710,851
RAC: 3,799
Panama
Message 2015188 - Posted: 12 Oct 2019, 17:54:56 UTC - in response to Message 2015177.  

Matthias said he got real busy at work and crunching had to take a backseat for a while.

Seem like he started a new host today. A single 2080Ti one.
ID: 2015188 · Report as offensive
Profile RueiKe Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 14 Feb 16
Posts: 490
Credit: 378,512,430
RAC: 785
Taiwan
Message 2015852 - Posted: 18 Oct 2019, 14:24:47 UTC

I have a new build using Rome 7702P with 64 cores up and running, Nexon . With no GPU, it is producing about 120K per day. Averages more than 1 WU per minute. I plan to add a Vega20 card over the weekend. This will not be a dedicated cruncher, as I have built it for an analytics project I am working on, but I will turn it over to SETI whenever it is not being used.

It runs out of work very easily. I hope that we will soon see more flexibility in max CPU work to download as core counts increase.
ID: 2015852 · Report as offensive
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 3160
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2015856 - Posted: 18 Oct 2019, 14:52:19 UTC - in response to Message 2015852.  

I have a new build using Rome 7702P with 64 cores up and running, Nexon . With no GPU, it is producing about 120K per day. Averages more than 1 WU per minute. I plan to add a Vega20 card over the weekend. This will not be a dedicated cruncher, as I have built it for an analytics project I am working on, but I will turn it over to SETI whenever it is not being used.

It runs out of work very easily. I hope that we will soon see more flexibility in max CPU work to download as core counts increase.


I guess you have disabled SMT?
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2015856 · Report as offensive
Profile RueiKe Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 14 Feb 16
Posts: 490
Credit: 378,512,430
RAC: 785
Taiwan
Message 2015858 - Posted: 18 Oct 2019, 15:02:46 UTC - in response to Message 2015856.  

I have a new build using Rome 7702P with 64 cores up and running, Nexon . With no GPU, it is producing about 120K per day. Averages more than 1 WU per minute. I plan to add a Vega20 card over the weekend. This will not be a dedicated cruncher, as I have built it for an analytics project I am working on, but I will turn it over to SETI whenever it is not being used.

It runs out of work very easily. I hope that we will soon see more flexibility in max CPU work to download as core counts increase.


I guess you have disabled SMT?

Yes, I prefer it off for the development work I’m doing.
ID: 2015858 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 11807
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2015859 - Posted: 18 Oct 2019, 15:13:03 UTC

That's a good question. Does BOINC simply ignore the HT cores since the count moves the host beyond 100 cores. Or do you have to disable SMT for BOINC to even work?
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2015859 · Report as offensive
Profile RueiKe Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 14 Feb 16
Posts: 490
Credit: 378,512,430
RAC: 785
Taiwan
Message 2015861 - Posted: 18 Oct 2019, 15:46:32 UTC - in response to Message 2015859.  

That's a good question. Does BOINC simply ignore the HT cores since the count moves the host beyond 100 cores. Or do you have to disable SMT for BOINC to even work?

I ran it for a while with SMT enabled. It just ran all 100 WUs across the 128 available threads, leaving 28 unused. Doesn’t make sense to use SMT since even 36 extra WUs isn’t enough buffer. If I get a bunch of shorties, cores will be idle until work download catches up.
ID: 2015861 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 4996
Credit: 276,046,078
RAC: 462
Message 2015928 - Posted: 19 Oct 2019, 3:56:52 UTC - in response to Message 2015861.  

That's a good question. Does BOINC simply ignore the HT cores since the count moves the host beyond 100 cores. Or do you have to disable SMT for BOINC to even work?

I ran it for a while with SMT enabled. It just ran all 100 WUs across the 128 available threads, leaving 28 unused. Doesn’t make sense to use SMT since even 36 extra WUs isn’t enough buffer. If I get a bunch of shorties, cores will be idle until work download catches up.


Does upping the number of downloads/uploads you run in parallel help that at all?

I am assuming the 100 cpu tasks / system is a Boinc limit which means running other projects won't really help. I suppose you could also run non-Boinc projects...

Tom
A proud member of the OFA (Old Farts Association).
ID: 2015928 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 4996
Credit: 276,046,078
RAC: 462
Message 2015929 - Posted: 19 Oct 2019, 3:58:16 UTC

Maybe I should ask to rename this thread "Heavy Mettle" and see if it generates more conversation.....

:)

Tom
A proud member of the OFA (Old Farts Association).
ID: 2015929 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 18876
Credit: 416,307,556
RAC: 380
United Kingdom
Message 2015933 - Posted: 19 Oct 2019, 6:56:24 UTC

The one hundred tasks per CPU is a SETI limit, other projects will almost certainly have other limits, or no limit at all.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 2015933 · Report as offensive
Previous · 1 · 2 · 3 · Next

Message boards : Number crunching : Heavy Metal! - high cpu count


 
©2020 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.