Another experiment on E. coli
log in

Advanced search

Message boards : News : Another experiment on E. coli

Previous · 1 · 2 · 3 · Next
Author Message
Profile valterc
Project administrator
Project tester
Send message
Joined: 30 Oct 13
Posts: 623
Credit: 34,677,535
RAC: 1
Italy
Message 1188 - Posted: 10 Nov 2017, 10:47:32 UTC
Last modified: 10 Nov 2017, 10:48:04 UTC

There is another experiment on Ec floating around, this one uses *very* computing intensive search parameters. An average workunit should last 3-6 hours. As usual we started just with a few batches, the next Monday, if everything goes well, we will put them all into the system.
One side effect of this is that the work generator should be able, during this experiment, to keep the queue full.

Krümel
Send message
Joined: 31 Oct 16
Posts: 19
Credit: 14,099,551
RAC: 0
Germany
Message 1189 - Posted: 11 Nov 2017, 9:21:44 UTC

2:15 h on my R7 1700. :)

Profile valterc
Project administrator
Project tester
Send message
Joined: 30 Oct 13
Posts: 623
Credit: 34,677,535
RAC: 1
Italy
Message 1190 - Posted: 11 Nov 2017, 10:23:52 UTC - in response to Message 1189.

2:15 h on my R7 1700. :)

better than average!

Profile [VENETO] boboviz
Send message
Joined: 12 Dec 13
Posts: 183
Credit: 4,641,505
RAC: 0
Italy
Message 1191 - Posted: 11 Nov 2017, 19:29:28 UTC - in response to Message 1189.

2:15 h on my R7 1700. :)


Over 6h on my FX6300. :-(
It's time to change my cpu....

Profile valterc
Project administrator
Project tester
Send message
Joined: 30 Oct 13
Posts: 623
Credit: 34,677,535
RAC: 1
Italy
Message 1241 - Posted: 17 Jan 2018, 18:06:21 UTC - in response to Message 1191.
Last modified: 17 Jan 2018, 18:11:57 UTC

I just started with a few batches of a new experiment (Ec-1000) which is the most computationally complex even ran here. A single workunit should run for ~3.4 hours on our reference computer (Intel 4770K running Linux, SSE2 version, hyper-threaded cpu with just one core thread used)

Profile SMTB1963
Avatar
Send message
Joined: 13 Dec 17
Posts: 4
Credit: 37,021
RAC: 0
Message 1242 - Posted: 18 Jan 2018, 15:57:03 UTC - in response to Message 1241.
Last modified: 18 Jan 2018, 16:23:34 UTC

These are running about 5 hours on my Win 10 4770K @ 4GHz...
____________

Profile Buro87 [Lombardia]
Send message
Joined: 23 Nov 16
Posts: 100
Credit: 4,000,541
RAC: 0
Italy
Message 1243 - Posted: 18 Jan 2018, 22:11:24 UTC - in response to Message 1242.

These are running about 5 hours on my Win 10 4770K @ 4GHz...



as explained by Danien at the end of this thread
http://gene.disi.unitn.it/test/forum_thread.php?id=135#1229

with TN-Grid app, ram frequency is more important than cpu speed

Profile SMTB1963
Avatar
Send message
Joined: 13 Dec 17
Posts: 4
Credit: 37,021
RAC: 0
Message 1244 - Posted: 18 Jan 2018, 22:52:38 UTC - in response to Message 1243.
Last modified: 18 Jan 2018, 22:52:52 UTC

Thanks for the link...maybe I'll try OC on my memory then.
____________

Krümel
Send message
Joined: 31 Oct 16
Posts: 19
Credit: 14,099,551
RAC: 0
Germany
Message 1245 - Posted: 19 Jan 2018, 18:57:03 UTC

5h 45min on my R7 1700 @ 2,7 GHz SMT on = 16 threads at a time. RAM @ 3.066 MHz

Profile [VENETO] boboviz
Send message
Joined: 12 Dec 13
Posts: 183
Credit: 4,641,505
RAC: 0
Italy
Message 1246 - Posted: 20 Jan 2018, 10:57:10 UTC - in response to Message 1241.

A single workunit should run for ~3.4 hours on our reference computer


Too much long for my FX6300, over 8 hours.
Is it possible to postpone the deadline for these wus?

Profile Buro87 [Lombardia]
Send message
Joined: 23 Nov 16
Posts: 100
Credit: 4,000,541
RAC: 0
Italy
Message 1247 - Posted: 20 Jan 2018, 14:19:48 UTC

4h 45min on my ryzen 5 1600 stock, SMT on (12WU) ram at 2933mhz

Profile valterc
Project administrator
Project tester
Send message
Joined: 30 Oct 13
Posts: 623
Credit: 34,677,535
RAC: 1
Italy
Message 1248 - Posted: 22 Jan 2018, 10:02:44 UTC - in response to Message 1246.

A single workunit should run for ~3.4 hours on our reference computer


Too much long for my FX6300, over 8 hours.
Is it possible to postpone the deadline for these wus?

The new ones should have a deadline of five days. The previous one was four, if BOINC is working correctly this shouldn't be a problem, just a smaller cache.

Alessio Susi
Avatar
Send message
Joined: 7 Jun 15
Posts: 15
Credit: 661,981
RAC: 0
Italy
Message 1250 - Posted: 31 Jan 2018, 22:13:14 UTC - in response to Message 1247.
Last modified: 31 Jan 2018, 22:19:14 UTC

Why don't you try to overclock your Ryzen CPUs? With the stock cooler you can do it to 3.6-3.7 GHz without overheat it. I have a Ryzen 7 1700 at 3.9 GHz and it works perfectly.
____________
ASUS X570 E-Gaming
AMD Ryzen 9 3950X, 16 core / 32 thread 4.4 GHz
AMD Radeon Sapphire RX 480 4GB Nitro+
Nvidia GTX 1080 Ti Gaming X Trio
4x16 GB Corsair Vengeance RGB 3466 MHz

Alessio Susi
Avatar
Send message
Joined: 7 Jun 15
Posts: 15
Credit: 661,981
RAC: 0
Italy
Message 1251 - Posted: 31 Jan 2018, 22:20:45 UTC
Last modified: 31 Jan 2018, 22:21:20 UTC

However, 0:40 ~ 4:00 hours for every WU.
AMD Ryzen 7 1700 at 3.9 GHz
RAM 2x8GB 3200 MHz CL16

Profile Buro87 [Lombardia]
Send message
Joined: 23 Nov 16
Posts: 100
Credit: 4,000,541
RAC: 0
Italy
Message 1252 - Posted: 31 Jan 2018, 23:00:45 UTC - in response to Message 1251.

However, 0:40 ~ 4:00 hours for every WU.
AMD Ryzen 7 1700 at 3.9 GHz
RAM 2x8GB 3200 MHz CL16


My bottleneck is ram bandwidth because i have only 1 stick

as explained by Daniel at the end of this thread
http://gene.disi.unitn.it/test/forum_thread.php?id=135#1229
in my case, overclock on CPU is not so good

m
Send message
Joined: 27 May 17
Posts: 4
Credit: 8,488,193
RAC: 0
United Kingdom
Message 1253 - Posted: 1 Feb 2018, 11:00:54 UTC - in response to Message 1248.


The new ones should have a deadline of five days. The previous one was four, if BOINC is working correctly this shouldn't be a problem, just a smaller cache.


I still see deadlines of four days, not five, what am I missing?. My hosts only run a few hours per day and the slow ones really struggle to complete work on time.

Profile valterc
Project administrator
Project tester
Send message
Joined: 30 Oct 13
Posts: 623
Credit: 34,677,535
RAC: 1
Italy
Message 1254 - Posted: 1 Feb 2018, 14:41:35 UTC - in response to Message 1253.


The new ones should have a deadline of five days. The previous one was four, if BOINC is working correctly this shouldn't be a problem, just a smaller cache.


I still see deadlines of four days, not five, what am I missing?. My hosts only run a few hours per day and the slow ones really struggle to complete work on time.

My fault... I thought I did that but actually I didn't.... Anyway, the batch of slow ones is almost completed.

Krümel
Send message
Joined: 31 Oct 16
Posts: 19
Credit: 14,099,551
RAC: 0
Germany
Message 1255 - Posted: 2 Feb 2018, 21:22:55 UTC - in response to Message 1250.

Why don't you try to overclock your Ryzen CPUs? With the stock cooler you can do it to 3.6-3.7 GHz without overheat it. I have a Ryzen 7 1700 at 3.9 GHz and it works perfectly.


My R7 1700 is runnig @ 2,7 GHz to save Energie.
Electricity is quiet expansive in Germany.
At this speed the R7 uses less than 50 watts.
@ 3.6 GHz it would be around 90 watts or more. :)

Profile Buro87 [Lombardia]
Send message
Joined: 23 Nov 16
Posts: 100
Credit: 4,000,541
RAC: 0
Italy
Message 1256 - Posted: 3 Feb 2018, 17:56:57 UTC - in response to Message 1255.

Why don't you try to overclock your Ryzen CPUs? With the stock cooler you can do it to 3.6-3.7 GHz without overheat it. I have a Ryzen 7 1700 at 3.9 GHz and it works perfectly.


My R7 1700 is runnig @ 2,7 GHz to save Energie.
Electricity is quiet expansive in Germany.
At this speed the R7 uses less than 50 watts.
@ 3.6 GHz it would be around 90 watts or more. :)



You just set cpu multiplier to x27? Or you have add an undervolt?

I will try to set my r5-1600 from 3.4ghz (1,07v vcore) to 3-3.1ghz
I also test if CPU stay less time waiting for memory

Krümel
Send message
Joined: 31 Oct 16
Posts: 19
Credit: 14,099,551
RAC: 0
Germany
Message 1257 - Posted: 3 Feb 2018, 18:54:06 UTC - in response to Message 1256.
Last modified: 3 Feb 2018, 18:55:00 UTC

I undervolted a little bit. Offset is -0,05.
2,7 GHz @ 0,85 Volt

Previous · 1 · 2 · 3 · Next
Post to thread

Message boards : News : Another experiment on E. coli


Main page · Your account · Message boards


Copyright © 2024 CNR-TN & UniTN