I'm a little confused with all the optimized downloads.

I have an Intel Core2 Quad which, of course, has SSE3.
I have an NVIDIA GTX750 Ti.
I am running Vista 64-bit.

What exactly should I download and install to get the proper optimized AP7 applications for both my CPU and my GPU?

I'm a little confused with all the optimized downloads.

I have an Intel Core2 Quad which, of course, has SSE3.
I have an NVIDIA GTX750 Ti.
I am running Vista 64-bit.

What exactly should I download and install to get the proper optimized AP7 applications for both my CPU and my GPU?

Did you use the Lunatics installer ?

---

With each crime and every kindness we birth our future.

I have used the Lunatics installer in the past to optimize MBv7 and APv6. I have been running them for several years on both my CPU and GPU.

I have not downloaded or installed anything for APv7 yet due to my confusion over what I should use.

I have used the Lunatics installer in the past to optimize MBv7 and APv6. I have been running them for several years on both my CPU and GPU.

I have not downloaded or installed anything for APv7 yet due to my confusion over what I should use.

Just download the Lunatics installer 0.43.

Since Vista doesn`t support AVX you have to use SSE3 version for AP7.

---

With each crime and every kindness we birth our future.

I will start downloading now.

Thank you, Mike!

I have a GTX 580 & a GTX 570 on a XEON quad-core. I'm running with the Lunatics 0.43 installer, that went in very easily (many thanks to all involved)

My question is can someone please give me tips on the cmdline options to use, currently:-

ap_cmdline_winx86_SSE2_OpenCL_NV has
-use_sleep -unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -hp


ap_cmdline_x64_SSE3_CPU is empty

TIA

---

Happy Crunching,

Graham

We were talking with Mike over in another thread. He recommended the addition of the bolded areas

-tune 1 64 4 1 -tune 2 64 4 1 -oclfft_plan -hp

I don't know if you have used the tune option or not. He said he wanted to post over here, I'm sure he will be around soon.

Zalster

I seem to recall hearing that 64-bit Intel was going to be a requirement for future releases? Looking at the list of apps I see "Mac OS/X 10.3+ 7.00" under AP v7. Is that an app for PPC and/or hardware with OS X.3?

BOINC requires 64-bit Intel Mac and OS X 10.5 beginning from version 7.4.19. This is because NVIDIA dropped support for 32-bit applications from CUDA 6.5/drivers.

I don't what plans the project has for older Macs/CUDAs.

We were talking with Mike over in another thread. He recommended the addition of the bolded areas

-tune 1 64 4 1 -tune 2 64 4 1 -oclfft_plan -hp

I don't know if you have used the tune option or not. He said he wanted to post over here, I'm sure he will be around soon.

Zalster

Again please read the read me before using oclfft_plan.

It needs values as well.
Just start with this example.

-use_sleep -oclFFT_plan 256 16 256.

Next step would be to add tune params.

-use_sleep -unroll 10 -oclFFT_plan 256 16 256 -tune 1 64 4 1 -tune 2 64 4 1.

Make sure unroll needs to be double of your number of CU`s.

---

With each crime and every kindness we birth our future.

I seem to recall hearing that 64-bit Intel was going to be a requirement for future releases? Looking at the list of apps I see "Mac OS/X 10.3+ 7.00" under AP v7. Is that an app for PPC and/or hardware with OS X.3?

BOINC requires 64-bit Intel Mac and OS X 10.5 beginning from version 7.4.19. This is because NVIDIA dropped support for 32-bit applications from CUDA 6.5/drivers.

I don't what plans the project has for older Macs/CUDAs.

Looking at App list AP v6 it included Mac OS/X 10.3+ & Mac OS X/Intel. The decision was probably made to only support 64-bit for Intel MAC for AP v7. Which makes sense given 99% of all MAC machines are 64-bit. That machine will just run AP v6 until they are gone along with MB v7.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

I was wondering if rescheduling would be over with AP7 and I was wrong.
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7374749

I went looking at the host at this URL and after looking at it I wonder if my version of boinc is configured correctly. I am running strictly a CPU with no video cards and boinc indicates my limit is 100 WUs. I always assumed it was because I have one CPU. Have missed a way of configuring boinc so that a single core is considered a logical CPU? That would give me a limit 400 WUs not 100.

I was wondering if rescheduling would be over with AP7 and I was wrong.
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7374749

I went looking at the host at this URL and after looking at it I wonder if my version of boinc is configured correctly. I am running strictly a CPU with no video cards and boinc indicates my limit is 100 WUs. I always assumed it was because I have one CPU. Have missed a way of configuring boinc so that a single core is considered a logical CPU? That would give me a limit 400 WUs not 100.

The limits are
CPU = 100 (No matter how many cores, or CPU sockets) I would love for my 2 socket 24 core server to have a limit of 200 CPU tasks. :/
GPU = 100 Per GPU per vendor. 4 NV cards = limit of 400 GPU tasks. 1 NV & 2 ATI = limit of 300 GPU (200 ATI & 100 NV)
The host that is being pointed to has 4 GPU's. So should have a total limit of 500 in progress. Out of their 1432 tasks In Progress they have 100 CPU and 1332 GPU.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Many thanks for the pointers, Mike and Zalster.

I'll try some changes over this weekend & see how it all goes.

---

Happy Crunching,

Graham

I have set up an offline bench with some optimzed params.

Knabench NV

Knabench ATI

Just unpack to a empty folder and double click APbench_213.cmd.

The result will be stored in Testdatas folder.

---

With each crime and every kindness we birth our future.

Well, after having done 189 non-outlier tasks, (Consecutive valid tasks 306), the credits for AP v7, haven't even come close to that of AP v6.

Ditto. My total RAC is down more than 20 percent since the introduction of V7, and I'm pretty sure I've completed enough work on both machines at this point. Ugh. Good news I guess is that work is completing without errors, and I guess I did my app_config.xml correctly.

---

Hit first 500 credit mark late last night UTC. My AP V7 now averaging around 495 give or take. I remember when AP V7 first deployed at Beta, every wu was awarded 1400 credits for several days, dropped down to 200 and then slowly climbed to current levels... 625-750

Keifer

---

Well, after having done 189 non-outlier tasks, (Consecutive valid tasks 306), the credits for AP v7, haven't even come close to that of AP v6. It's rising slowly though, but is still not consistently at even 500 Credits.

Patience young man, patience....:-)

Edit, added: We've been running the new AP v7 now since October 9th, and still no official announcement of this new app on the board. Who stole the keyboards of the SETI project admins?

I have just seen the average graph in my BOINC Manager start to move upward. With the 19th the lowest point. Also the lowest credit I am seeing granted for a non blanked AP task is around 515.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

APV6 versus APV7 seems to be a lot closer in my case. The credit total per APV7 WU does indeed still seem to be less when compared to APV6 WUs but they also seem to be completing sooner to make up for it. A very unscientific look at my results with a sample size of 20 WUs per group produces the following for my CPU.


APV6 4-Oct-14 28.1 seconds/credit

APV7 12-Oct-14 55.0 seconds/credit
APV7 14-Oct-14 43.9 seconds/credit
APV7 20-Oct-14 29.4 seconds/credit

I have no idea where it will eventually settle out at but even if a future sample with more WUs takes a downward tick, I don't think it will be as bad as I first thought. It might even be better once the Linux 32 bit SSE3 app shows up.

CreditScrew seems to decide now that the credit/AP WU, should drop again. Up until yesterday, the credit was rising day by day, and yesterday, many of my task was given 500 Cr or more.

Today though, the Cr/Wu is dropping badly, and most task gets well below 500 Cr.

Strange creature that CreditScrew.

It just levels out.
Jason mentioned he think it will be ~400 so thats not to shabby.

---

With each crime and every kindness we birth our future.

CreditScrew seems to decide now that the credit/AP WU, should drop again. Up until yesterday, the credit was rising day by day, and yesterday, many of my task was given 500 Cr or more.

Today though, the Cr/Wu is dropping badly, and most task gets well below 500 Cr.

Strange creature that CreditScrew.

It just levels out.
Jason mentioned he think it will be ~400 so thats not to shabby.

Today, so far, my computer has 26 gpu AP tasks that have gone full distance and validated, and except for two, they have all got over 490 credits. The other two got 434 and 454 credits.

Looking at the V6 tasks still visible, the RAC for the GPU would have been between about 19,000. For the V7 tasks, mainly due to the lack of penalty for high blanking, the RAC for the GPU will probably be >22,000.