Prime95 v25.3 - Page 5

drew · 2007-07-22, 20:30

Quote:

Originally Posted by Prime95

Nice catch. I've fixed it (but not uploaded the fix).

Great.

I think I found another problem...actually 2.

I discovered the second problem while trying to report the first. When I grab text from the Edit->Copy menu the EOL characters don't follow the Windows 'CR+LF' standard, so when I paste into notepad it looks like garbage (see attached). It would be great if the windows actually allowed you to highlight text to copy rather than just copying the buffer from a menu.

Both of my tests hit stage 2 P-1 at nearly the same time, since I started them together. It appears to be stuck in a race condition with the two threads continually adjusting their memory configurations. I don't know if it's getting any work done at all in this state. Just to provide more data I have 768 (not 748) MB of RAM allocated in the CPU options dialog.

Attached is a difficult-to-read copy of what I'm seeing from the two threads. This went on continually until I exit Prime95.

Drew

drew · 2007-07-22, 21:41

Quote:

Originally Posted by drew

Both of my tests hit stage 2 P-1 at nearly the same time, since I started them together. It appears to be stuck in a race condition with the two threads continually adjusting their memory configurations. I don't know if it's getting any work done at all in this state. Just to provide more data I have 768 (not 748) MB of RAM allocated in the CPU options dialog.

I should add that I never observed this behavior awhile back when you had us testing P-1 code in this thread. Something must've changed since then to cause this problem. Or perhaps since it doesn't have more P-1 worked queued, it reverts back to the same test.

You also may find this helpful. Look at the repeating pattern in the memory usage. Prime95 is oscillating between around 66 and 400 MB.

Knappo · 2007-07-23, 07:55

I've found another (very) minor issue regarding the systray.
(seen during LL)
If you start Prime it will state "Prime95" in the systray tooltip. It will not give a status unless you open the window of Prime. Then it changes its tooltip to the usual 3,43% of M23165165165165151(and so on...)

chappjc · 2007-07-23, 14:21

Quote:

Originally Posted by drew

Both of my tests hit stage 2 P-1 at nearly the same time, since I started them together. It appears to be stuck in a race condition with the two threads continually adjusting their memory configurations.

I reported this issue above. This was Prime95's response:

Quote:

Originally Posted by Prime95

The way it is supposed to work is one worker grabs the memory and does stage 2. The other three workers having insufficient memory to run stage 2 should start their LL test. They will do stage 2 when the first worker completes stage 2.

BTW drew, I see the same thing in the output when the workers hit stage 2 p-1.

drew · 2007-07-23, 14:39

Quote:

Originally Posted by chappjc

I reported this issue above. This was Prime95's response:

Oh, I see. The impression I had from your post was that thread 1 was still working. Now I realize that none of the threads worked.

I told it to run only 1 test multithreaded in order to get past this hurdle. It's doing a good job keeping both CPUs active this way.

Drew

chappjc · 2007-07-23, 15:35

That was my solution as well:

Quote:

Originally Posted by chappjc

To get around this in the meantime, I changed the number of workers to 1 and the number of threads per worker to 4. It plowed through stage 2 very quickly.

However, when I had the problem, I think one worker was actually working properly, while the others were continually asking for memory and not getting it.

Bundu · 2007-07-25, 00:30

Quote:

Originally Posted by Prime95

2) Scott, has suggested a one window option. Instead of "Main thread", "Comm thread", Worker 1", they would all appear in one window (kinda like v24) with text next to the timestamp that says what thread generated the message. Do you have better or alternative suggestions?

How about using different coloured text to differentiate the "Comm thread", Worker 1"...etc in the same window?

Bundu · 2007-07-25, 01:03

Also, having installed this version over a 24.14 version i copied onto my brother's new dual core laptop i noticed that when I increased the number of worker threads this new version returns the first thread back to the start of its TF.. oops

thread 1

(previously 2^67 87% complete)

Using old style factoring save file
resuming TF of MXXXXX
TF of MXXXX to 2^64

Thread 2

starting
TF of MXXXXXX2 2^64

Anyway, you live you learn

I guess once the "jobs" are done the next ones will be using the 25.3 save files and this won't be a problem?

Thanks

PS - 25.3 detected the hardware correctly Intel Dual core T7100 @1.8Ghz, L1 32kb, L2 2048kb

Denahar · 2007-07-27, 19:10

I have encountered a problem with V25.3 that bothers me already for quite some time.
I run a dual Xeon 5160 workstation (2x2 cores) but whenever I want to let Prime95 run on all four cores it won't let me (and gives me the message that I selected more threads than CPUs detected).
It really irks me since I do know I have four cores, why not giving P95 the full performance (of nearly 13Ghz)?

-D

drew · 2007-07-27, 20:22

Quote:

Originally Posted by Denahar

I have encountered a problem with V25.3 that bothers me already for quite some time.
I run a dual Xeon 5160 workstation (2x2 cores) but whenever I want to let Prime95 run on all four cores it won't let me (and gives me the message that I selected more threads than CPUs detected).
It really irks me since I do know I have four cores, why not giving P95 the full performance (of nearly 13Ghz)?

-D

Hey Denehar,

This is one of those things that people have expressed should be worded more clearly. If you set "Number of worker threads to run=4", then you should set "Number of CPUs to use=1". The CPUs to use are 'per thread'...so if you put '1', then it will use 1 CPU for each of the 4 worker threads.

Alternatively, you can run 2 threads on 2 CPUs apiece, or 1 thread on 4 CPUs. Or any other combination if you want to configure individual threads to run differently.

Prime95 · 2007-07-27, 20:38

Quote:

Originally Posted by Denahar

I run a dual Xeon 5160 workstation (2x2 cores) but whenever I want to let Prime95 run on all four cores it won't let me

In Options/CPU dialog box, has prime95 properly identified your machine as having 4 CPUs?

2007-07-25, 01:03	#52
Bundu Jul 2004 Mid Calder, Scotland 5·37 Posts	Also, having installed this version over a 24.14 version i copied onto my brother's new dual core laptop i noticed that when I increased the number of worker threads this new version returns the first thread back to the start of its TF.. oops thread 1 (previously 2^67 87% complete) Using old style factoring save file resuming TF of MXXXXX TF of MXXXX to 2^64 Thread 2 starting TF of MXXXXXX2 2^64 Anyway, you live you learn I guess once the "jobs" are done the next ones will be using the 25.3 save files and this won't be a problem? Thanks PS - 25.3 detected the hardware correctly Intel Dual core T7100 @1.8Ghz, L1 32kb, L2 2048kb Last fiddled with by Bundu on 2007-07-25 at 01:07

2007-07-27, 19:10	#53
Denahar Mar 2007 Berlin, Germany 2·5 Posts	I have encountered a problem with V25.3 that bothers me already for quite some time. I run a dual Xeon 5160 workstation (2x2 cores) but whenever I want to let Prime95 run on all four cores it won't let me (and gives me the message that I selected more threads than CPUs detected). It really irks me since I do know I have four cores, why not giving P95 the full performance (of nearly 13Ghz)? -D Last fiddled with by Denahar on 2007-07-27 at 19:10

2007-07-23, 07:55	#47
Knappo Aug 2005 69469, Germany 3·5 Posts	I've found another (very) minor issue regarding the systray. (seen during LL) If you start Prime it will state "Prime95" in the systray tooltip. It will not give a status unless you open the window of Prime. Then it changes its tooltip to the usual 3,43% of M23165165165165151(and so on...)