In task manager you only see system memory not GPU memory consumption.

That´s exactly why this issue is wierd, why the main memory usage rises to that kind of level on a determinate WU while all others remains low?

When the system memory usage rises to that large number the system runs out of memory very fast.

yep, exactly system (OS, CPU.., not GPU) memory increase expected with issue I described earlier.

I honestly dont think so.
I had 3 30/30 in a row 2 days ago and on my 4GB sys mem host it would have crashed.
Two were running at the same time with those sharp timings.
I never had this happen.

Like i said earlier i watched closely while benching Juans task i had 104K constantly during this bench and it is overflow task.
I think this is host dependend.

---

With each crime and every kindness we birth our future.

In task manager you only see system memory not GPU memory consumption.

That´s exactly why this issue is wierd, why the main memory usage rises to that kind of level on a determinate WU while all others remains low?

When the system memory usage rises to that large number the system runs out of memory very fast.

yep, exactly system (OS, CPU.., not GPU) memory increase expected with issue I described earlier.

I honestly dont think so.
I had 3 30/30 in a row 2 days ago and on my 4GB sys mem host it would have crashed.
Two were running at the same time with those sharp timings.
I never had this happen.

Like i said earlier i watched closely while benching Juans task i had 104K constantly during this bench and it is overflow task.
I think this is host dependend.

Talking about timings, do you have any of these extra "-oclfft_plan ", and "-tune " things which would fit my HD7870? -use_sleep, is that something for me too?

For now, I only use: -unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -instances_per_device 2

Oh, and only if they do not lead to the release of "Magic Smoke" :-)

use_sleep is necessary only on nvidia cards to reduce CPU usage.
Your card is easy to handle.

-unroll 16 -oclfft_plan 256 16 256 -ffa_block 12288 -ffa_block_fetch 6144 -tune 1 64 4 1 -tune 2 64 4 1

---

With each crime and every kindness we birth our future.

Like i said earlier i watched closely while benching Juans task i had 104K constantly during this bench and it is overflow task.
I think this is host dependend.

With exactly same command line as he provided?
I configured memory working set logger via perfmon (and learnt smth new , LoL) so now log into file current working set each 10 seconds (configurable).

Will run his workunit later with same command line (but with ATi app cause NV card can't accept same command line).

Like i said earlier i watched closely while benching Juans task i had 104K constantly during this bench and it is overflow task.
I think this is host dependend.

With exactly same command line as he provided?
I configured memory working set logger via perfmon (and learnt smth new , LoL) so now log into file current working set each 10 seconds (configurable).

Will run his workunit later with same command line (but with ATi app cause NV card can't accept same command line).

Yes, sure.

---

With each crime and every kindness we birth our future.

I think this is host dependend.

I belive is not, why? it´s happening on a totaly diferent OS & GPU, the same happening on a 2x690 who runs Win Server. The only thing they have in common is they use the same old I5 CPU.

After few test i made here remotely, i could confirm the problem is aparently related to the size of the -ffa_block, a 4096 block makes some WU ussing 150MB, larger sizes huge increase the memory usage, a 8192 makes the WU uses about 250MB and so on, the 1GB apears when you use a 16k block size.

For some reason who i cant understand until now, my hosts simply apears to ignore the -oclfft_plan 256 16 256 switch.

---

-oclfft_plan 256 16 256 switch.

yep, as I said earlier there is no confirmation from app in log (stderr) that switch was recognized.

I think this is host dependend.

I belive is not, why? it´s happening on a totaly diferent OS & GPU, the same happening on a 2x690 who runs Win Server. The only thing they have in common is they use the same old I5 CPU.

After few test i made here remotely, i could confirm the problem is aparently related to the size of the -ffa_block, a 4096 block makes some WU ussing 150MB, larger sizes huge increase the memory usage, a 8192 makes the WU uses about 250MB and so on, the 1GB apears when you use a 16k block size.

For some reason who i cant understand until now, my hosts simply apears to ignore the -oclfft_plan 256 16 256 switch.

-oclFFT_plan is case sensitive.
Make sure FFT is upper case or just snip it from the read me.
-use_sleep -unroll 16 -oclFFT_plan 256 16 256 -ffa_fetch 8192 -ffa_fetch_block 4096-tune 1 64 4 1 -tune 2 64 4 1

Make sure you get this message in stderr.

oclFFT plan class overrides requested: global radix 256; local radix 16; max workgroup size 256

---

With each crime and every kindness we birth our future.

-oclFFT_plan is case sensitive.

Maybe is a good ideia change that, all the other switches uses lower case letters.

---

-oclFFT_plan is case sensitive.

Maybe is a good ideia change that, all the other switches uses lower case letters.

Like Raistmer said its for advanced users.
Everybody can snip it out of the read me.

Sorry for the typo.

My mistake.

---

With each crime and every kindness we birth our future.

Try to use a smaller -ffa_block and you will see, the memory hogging WU will use a lot less memory (i like the name)

---

Try to use a smaller -ffa_block and you will see, the memory hogging WU will use a lot less memory (i like the name)

I'll drop back to my original (since v6) ffa_block and -ffa_block_fetch, as well as my unroll.

Unroll will be bigger then.
Watch out.

---

With each crime and every kindness we birth our future.

Interesting, Sten-Arne uses ATI, not NV like me, so aparently the problem is in the code itself and sure not host related.

---

Interesting, Sten-Arne uses ATI, not NV like me, so aparently the problem is in the code itself and sure not host related.

It is since i dont get this high memory usage even with biggest unroll and ffa_fetch values.
Same on my sons 7850.

---

With each crime and every kindness we birth our future.

Try to use a smaller -ffa_block and you will see, the memory hogging WU will use a lot less memory (i like the name)

I'll drop back to my original (since v6) ffa_block and -ffa_block_fetch, as well as my unroll.

Unroll will be bigger then.
Watch out.

Hmm, is unroll 12 bigger than 16? Am I missing something here? Is it backwards?

I meant if you remove unroll in comand line.

---

With each crime and every kindness we birth our future.

Interesting, Sten-Arne uses ATI, not NV like me, so aparently the problem is in the code itself and sure not host related.

It is since i dont get this high memory usage even with biggest unroll and ffa_fetch values.
Same on my sons 7850.

Could this be memory related? This old computer only have slow DDR2 memory, single channel.

Yes, possible.

---

With each crime and every kindness we birth our future.

Interesting, Sten-Arne uses ATI, not NV like me, so aparently the problem is in the code itself and sure not host related.

It is since i dont get this high memory usage even with biggest unroll and ffa_fetch values.
Same on my sons 7850.

Did you try a large ffa_block? On my tests, unroll, ffa_fetch does not produce the problem, only large ffa_block on the range of 16K, smaller number produces only small memory usage.

---

Interesting, Sten-Arne uses ATI, not NV like me, so aparently the problem is in the code itself and sure not host related.

It is since i dont get this high memory usage even with biggest unroll and ffa_fetch values.
Same on my sons 7850.

Did you try a large ffa_block? On my tests, unroll, ffa_fetch does not produce the problem, only large ffa_block on the range of 16K, smaller number produces only small memory usage.

I did run your units with the settings you did.
No problem on 2 different hosts.

---

With each crime and every kindness we birth our future.

Hm... Looks like no matter if I write in English or in Russian... nobody cares to read anyway :/

http://setiathome.berkeley.edu/forum_thread.php?id=75863&postid=1590653

No need to invent new entities w/o need.

Interesting, Sten-Arne uses ATI, not NV like me, so aparently the problem is in the code itself and sure not host related.

It is since i dont get this high memory usage even with biggest unroll and ffa_fetch values.
Same on my sons 7850.

Could this be memory related? This old computer only have slow DDR2 memory, single channel.

Yes, possible.

I have 3 ATI Host running the -oclFFT_plan with DDR2, only 1 card May be affected with higher Pulse counts. The Other 5 cards seem to be doing well. But, only 1 Host is running Windows...

Interesting, Sten-Arne uses ATI, not NV like me, so aparently the problem is in the code itself and sure not host related.

It is since i dont get this high memory usage even with biggest unroll and ffa_fetch values.
Same on my sons 7850.

Did you try a large ffa_block? On my tests, unroll, ffa_fetch does not produce the problem, only large ffa_block on the range of 16K, smaller number produces only small memory usage.

I did run your units with the settings you did.
No problem on 2 different hosts.

Wierd... what i could tell you for sure, the same problem happening on few of my hosts with diferent CPU/GPU´s combinations, and not in all of the WU´s about one in 10 WU only.

Could be a CPU related problem? You use AMD CPU, I and Sten-Arne uses low end Intel CPU´s, who knows?

Let´s wait to see if Raistmer found something. For now with small ffa_block the memory hugging is relatively small (less than 150MB) so is not a headache since all my hosts has 8GB at least.

<edit> yes i read

I'll check if there is some additional memory leak or not

that´s exacly why i sugest wait to see if some aditional memory leak could be happening.

---