Posts by ralphw

1) Message boards : Number crunching : Best way to get more processing from GTX 1050ti alongside GTX 750ti/ GTX 950 (Message 1873749)
Posted 18 Jun 2017 by Profile ralphw
Post:
I ended up moving an MSI GTX 750 Ti back into this system.

I was expecting to put four GPUs on this motherboard, but I apparently need all of my 750 Ti systems to be
the shorter 5-6" long cards.

Only the first slot of this Gigabyte motherboard really accommodates a full-length card such as MSI's dual-fan GTX 750Ti.
The fan shroud and card length really don't fit well with the other heat sinks and other connectors

I will have to limit myself to the smaller form factor GPUs to mechanically use all of my remaining motherboard slots

I'll see how well the WU averages keep up.
2) Message boards : Number crunching : Best way to get more processing from GTX 1050ti alongside GTX 750ti/ GTX 950 (Message 1872556)
Posted 12 Jun 2017 by Profile ralphw
Post:
Thanks.

That is my primary configuration (GTX 1050 Ti alongside two GTX 750 Ti systems).

The third 750 Ti (from MSI) is currently in an inactive system.
3) Message boards : Number crunching : Best way to get more processing from GTX 1050ti alongside GTX 750ti/ GTX 950 (Message 1872515)
Posted 12 Jun 2017 by Profile ralphw
Post:
Hello,

I have thee models of NVIDIA GPU,

  • NVIDIA 1050 (1) Ti (1)
  • NVIDIA GTX 950 Ti (2)
  • NVIDIA GTX 750 Ti (3)


Spread across two systems

What's the best strategy for having the fast cards (950,1050) process more data?
There are some clients that have loop unrolling options, but is running multiple workunits on a GPU - by setting up an app_info.xml file - really taking advantage of the extra CUDA cores?

4) Questions and Answers : GPU applications : If I were to run a GTX 1060 and a GTX 750 Ti on the same pc how could I max both? (Message 1872118)
Posted 10 Jun 2017 by Profile ralphw
Post:
I have a similar "problem" - GTX 1050 Ti and two GTX 750 Ti systems.
I assume there is an XML file I can tweak.
5) Questions and Answers : GPU applications : Back with linux Mint 18.1 (Message 1871524)
Posted 7 Jun 2017 by Profile ralphw
Post:
Found what was missing:

The use_all_gpus setting in the cc_config.xml (/var/lib/boinc-client/cc_config.xml) was missing.

Now I'm using all of my cude cores:
<use_all_gpus>1</use_all_gpus>


Wed 07 Jun 2017 02:56:35 AM EDT | | CUDA: NVIDIA GPU 0: GeForce GTX 1050 Ti (driver version 375.66, CUDA version 8.0, compute capability 6.1, 4039MB, 3994MB available, 2274 GFLOPS peak)
Wed 07 Jun 2017 02:56:35 AM EDT | | CUDA: NVIDIA GPU 1: GeForce GTX 750 Ti (driver version 375.66, CUDA version 8.0, compute capability 5.0, 2001MB, 1973MB available, 1606 GFLOPS peak)
Wed 07 Jun 2017 02:56:35 AM EDT | | OpenCL: NVIDIA GPU 0: GeForce GTX 1050 Ti (driver version 375.66, device version OpenCL 1.2 CUDA, 4039MB, 3994MB available, 2274 GFLOPS peak)
Wed 07 Jun 2017 02:56:35 AM EDT | | OpenCL: NVIDIA GPU 1: GeForce GTX 750 Ti (driver version 375.66, device version OpenCL 1.2 CUDA, 2001MB, 1973MB available, 1606 GFLOPS peak)
Wed 07 Jun 2017 02:56:35 AM EDT | | [coproc] NVIDIA library reports 2 GPUs
[/code]
6) Questions and Answers : GPU applications : Back with linux Mint 18.1 (Message 1871426)
Posted 6 Jun 2017 by Profile ralphw
Post:
All three VGA devices seem to show appropriately.

$ lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corporation Xeon E3-1200 v3/4th Gen Core Processor Integrated Graphics Controller (rev 06)
01:00.0 VGA compatible controller: NVIDIA Corporation Device 1c82 (rev a1)
02:00.0 VGA compatible controller: NVIDIA Corporation GM107 [GeForce GTX 750 Ti] (rev a2)
7) Questions and Answers : GPU applications : Back with linux Mint 18.1 (Message 1871422)
Posted 6 Jun 2017 by Profile ralphw
Post:
I'm working with Linux Mint 18.1, using the driver settings with Nvidia.

I have a GTX 1050 Ti (Pascal) and GTX 750 Ti (Maxwell)

The 375 drivers are supposed to recognize both, but don't seem to.


Eventually, I'll want to put the third Nvidia card back in.
nvidia-smi
Tue Jun  6 00:37:27 2017       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 375.66                 Driver Version: 375.66                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 105...  Off  | 0000:01:00.0     Off |                  N/A |
| 35%   33C    P0    35W /  75W |      0MiB /  4038MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX 750 Ti  Off  | 0000:02:00.0     Off |                  N/A |
| 40%   26C    P0     1W /  38W |      0MiB /  2000MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID  Type  Process name                               Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

When SETI starts, this is in the event log:
Tue 06 Jun 2017 12:30:42 AM EDT |  | CUDA: NVIDIA GPU 0: GeForce GTX 1050 Ti (driver version 375.66, CUDA version 8.0, compute capability 6.1, 4039MB, 3996MB available, 2274 GFLOPS peak)
Tue 06 Jun 2017 12:30:42 AM EDT |  | CUDA: NVIDIA GPU 1 (not used): GeForce GTX 750 Ti (driver version 375.66, CUDA version 8.0, compute capability 5.0, 2001MB, 1973MB available, 1606 GFLOPS peak)
Tue 06 Jun 2017 12:30:42 AM EDT |  | OpenCL: NVIDIA GPU 0: GeForce GTX 1050 Ti (driver version 375.66, device version OpenCL 1.2 CUDA, 4039MB, 3996MB available, 2274 GFLOPS peak)
Tue 06 Jun 2017 12:30:42 AM EDT |  | OpenCL: NVIDIA GPU 1 (ignored by config): GeForce GTX 750 Ti (driver version 375.66, device version OpenCL 1.2 CUDA, 2001MB, 1973MB available, 1606 GFLOPS peak)



Suggests for what to try next are welcome. Mint 18.1 seems a bit better behaved than 17.2 or 18.0,

But I'm not sure what is missing.
8) Questions and Answers : GPU applications : NVIDIA fan controls (Linux MINT 18, Driver Version 367.57, two NVidia GTX 950s) (Message 1831601)
Posted 20 Nov 2016 by Profile ralphw
Post:
The problem seems to be that manual fan control is enabled by the "Coolbits" option in xorg.conf

Problem 1 (solved): xorg.conf gets overwritten with new values upon reboot. Enabling nogpumanager in GRUB doesn't fix this in Linux MINT 18, so I had to use restore to using chattr +i on the /etc/X11/xorg.conf file.

Problem 2 (working on): even though the line
"Coolbits" "12" is now preserved for both device entries, only one NVidia device allows for manual control of fan speed. Working on a solution using this Folding@home page as a guide:

https://foldingforum.org/viewtopic.php?p=267165
It requires setting up multiple X Servers, one per GPU.

I'd prefer to easier configuration options for multiple GPUs and multiple monitors, along with a more aggressive adaptive fan control system. I'm not overclocking, but SETI@home pushes GPUs hard.
9) Questions and Answers : GPU applications : NVIDIA fan controls (Linux MINT 18, Driver Version 367.57, two NVidia GTX 950s) (Message 1831561)
Posted 20 Nov 2016 by Profile ralphw
Post:
So I'm thinking that there is a driver compatibility issue with Linux Mint 18, that might be preventing things from working. This guide shows use of the PPA install method with Linux MINT 18.
https://johners.tech/2016/07/installing-the-latest-nvidia-graphics-drivers-on-linux-mint-18/

I was using the PPA for graphics drivers, but am going to try switching to the .run installation method (downloading the driver directly from NVidia).
(This post talks about CUDA support, but recommends NOT using the PPA method - https://devtalk.nvidia.com/default/topic/955464/cuda-setup-and-installation/gtx-1080-cuda-8-0-on-linux-mint-18-problems-setting-up-/#reply)

Not sure if this will work any better, but I need to try something to combat the frustration of (works with 14.04 / doesn't work with MINT 18) syndrome.

Fan controls in NVIDIA seem important when I'm running GPU apps on SETI.
90 degree GPU temps seem too high.
10) Questions and Answers : GPU applications : NVIDIA fan controls (Linux MINT 18, Driver Version 367.57, two NVidia GTX 950s) (Message 1831536)
Posted 20 Nov 2016 by Profile ralphw
Post:
Those look interesting, but the NVIDIA driver controls are supposed to be able to regulate this. Aside from googling (which I'm trying), how can I be sure the driver version is "proper" and supports the fan speed setting capability?
11) Questions and Answers : GPU applications : NVIDIA fan controls (Linux MINT 18, Driver Version 367.57, two NVidia GTX 950s) (Message 1831446)
Posted 19 Nov 2016 by Profile ralphw
Post:
I'm running SETI@home 8.0 applications on Linux Mint 18 (based on Ubuntu 16.04).
GPU temps are extemely high, and I feel I need better speed controls for the GPU fans.

>nvidia-smi
>Sat Nov 19 12:00:51 2016
>+-----------------------------------------------------------------------------+
>| NVIDIA-SMI 367.57 Driver Version: 367.57 |
>|-------------------------------+----------------------+----------------------+
>| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
>| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
>|===============================+======================+======================|
>| 0 GeForce GTX 950 Off | 0000:01:00.0 On | N/A |
>| 0% 40C P0 23W / 125W | 214MiB / 1996MiB | 0% Default |
>+-------------------------------+----------------------+----------------------+
>| 1 GeForce GTX 950 Off | 0000:02:00.0 Off | N/A |
>| 0% 25C P8 6W / 125W | 1MiB / 1996MiB | 0% Default |
>+-------------------------------+----------------------+----------------------+

....


Despite playing with the "coolbits" settings in the Xorg.conf file, no controls to set the fans are ever activated.
It also seems like the settings should be under the "device" section, but the nvidia-settings CLI option puts them in the "screens" section.

I have two NVidia GTX 950s in this system.

When runnning SETI@home, GPU temps would peak briefly at 90 degrees C, which is about 40 degrees too hot.
So I'm looking for a way to turn the fans on more or continuously when running GPU work units.
12) Questions and Answers : Wish list : Request: Linux client use DMI info (base board manuf/product) to populate "Model" field for computer (Message 1758007)
Posted 22 Jan 2016 by Profile ralphw
Post:
When you look at computer details, the "Product Name" field shows the same information as what's in the "Model" field in the summary list of computers:

The detail page for this mac shows "MacBookPro11,3" in the Product Name field.

On the summary page (listing all computers), the same field is visible under the column "Model". I don't know why the field name changes between the summary and detail page...

So Boinc must support the transmission of this information somehow.
13) Questions and Answers : Wish list : Request: Linux client use DMI info (base board manuf/product) to populate "Model" field for computer (Message 1757649)
Posted 20 Jan 2016 by Profile ralphw
Post:
When I look at "My Computers", I see only one system that has populated the computer "Model field".

I'd love for this to populate for an PC motherboard as well, what would I have to do to manually populate this?

Can it be done at a workunit level?

(Here is what "dmidecode -t baseboard" shows on the system I'm using right now)

Handle 0x0002, DMI type 2, 15 bytes
Base Board Information
Manufacturer: Gigabyte Technology Co., Ltd.
Product Name: Z97X-SOC Force
14) Message boards : Number crunching : Intel / OpenCL binaries for Linux (2015) (Message 1756480)
Posted 15 Jan 2016 by Profile ralphw
Post:
My experiment with Beignet doesn't appear to be successful with Astropulse 7.0.8.
My machine has three GPUs (Nvidia GTX 750Ti), plus the Intel HD 4600 graphics.

I think it's getting confused, thinking it sees two Nvidias, but trying to use the Intel. Would trying an older NVidia driver help?



https://setiathome.berkeley.edu/result.php?resultid=4662135832
....

<core_client_version>7.2.42</core_client_version>
<![CDATA[
<message>
process exited with code 193 (0xc1, -63)
</message>
<stderr_txt>
Running on device number: 0
OpenCL platform detected: NVIDIA Corporation
Number of OpenCL devices found : 2
BOINC assigns slot on device #0.
Info: BOINC provided OpenCL device ID used
Used GPU device parameters are:
Number of compute units: 5
Single buffer allocation size: 256MB
Total device global memory: 2047MB
max WG size: 1024
local mem type: Real
FERMI path used: yes
-unroll default value used: 5
-ffa_block default value used: 1280
-ffa_block_fetch default value used: 640
AstroPulse v7.08
Linux 64 bit, Rev 2751, OpenCL version by Raistmer, GPU mode
V7, by Raistmer ported to Linux by Lunatics.kwsn.net team.
by Urs Echternacht
ffa threshold mods by Joe Segur
SSE3 dechirping by JDWhale using SSE3 emulation

Build features: Non-graphics OpenCL USE_OPENCL_NV OCL_ZERO_COPY OPENCL_WRITE COMBINED_DECHIRP_KERNEL SMALL_CHIRP_TABLE TWIN_FFA FFTW BLANKIT USE_INCREASED_PRECISION SSE2 64bit
System: Linux x86_64 Kernel: 3.16.0-38-generic
CPU : Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
8 core(s), Speed : 4300.000 MHz
L1 : 64 KB, Cache : 8192 KB

Number of OpenCL platforms: 2


OpenCL Platform Name: NVIDIA CUDA
Number of devices: 2
Max compute units: 5
Max work group size: 1024
Max clock frequency: 1254Mhz
Max memory allocation: 536821760
Cache type: Read/Write
Cache line size: 128
Cache size: 81920
Global memory size: 2147287040
Constant buffer size: 65536
Max number of constant args: 9
Local memory type: Scratchpad
Local memory size: 49152
Queue properties:
Out-of-Order: Yes
Name: GeForce GTX 750 Ti
Vendor: NVIDIA Corporation
Driver version: 352.63
Version: OpenCL 1.2 CUDA
Extensions: cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_copy_opts cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64
Max compute units: 5
Max work group size: 1024
Max clock frequency: 1254Mhz
Max memory allocation: 536821760
Cache type: Read/Write
Cache line size: 128
Cache size: 81920
Global memory size: 2147287040
Constant buffer size: 65536
Max number of constant args: 9
Local memory type: Scratchpad
Local memory size: 49152
Queue properties:
Out-of-Order: Yes
Name: GeForce GTX 750 Ti
Vendor: NVIDIA Corporation
Driver version: 352.63
Version: OpenCL 1.2 CUDA
Extensions: cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_copy_opts cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64


OpenCL Platform Name: Experiment Intel Gen OCL Driver
DRM_IOCTL_I915_GEM_APERTURE failed: Invalid argument
Assuming 131072kB available aperture size.
May lead to reduced performance or incorrect rendering.
get chip id failed: -1 [22]
param: 4, val: 0
SIGSEGV: segmentation violation
Stack trace (15 frames):
../../projects/setiathome.berkeley.edu/astropulse_7.08_x86_64-pc-linux-gnu__opencl_nvidia_100(boinc_catch_signal+0x4d)[0x4c6fdd]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x10340)[0x7fc8c8194340]
/usr/lib/x86_64-linux-gnu/libdrm_intel.so.1(drm_intel_bufmgr_gem_enable_reuse+0x0)[0x7fc8c2f4d0d0]
/usr/lib/beignet/libcl.so(+0x13b58)[0x7fc8c4a5eb58]
/usr/lib/beignet/libcl.so(+0x13d07)[0x7fc8c4a5ed07]
/usr/lib/beignet/libcl.so(+0x13e21)[0x7fc8c4a5ee21]
/usr/lib/beignet/libcl.so(+0x13f08)[0x7fc8c4a5ef08]
/usr/lib/beignet/libcl.so(+0xf45d)[0x7fc8c4a5a45d]
/usr/lib/beignet/libcl.so(+0xf521)[0x7fc8c4a5a521]
../../projects/setiathome.berkeley.edu/astropulse_7.08_x86_64-pc-linux-gnu__opencl_nvidia_100[0x4885df]
../../projects/setiathome.berkeley.edu/astropulse_7.08_x86_64-pc-linux-gnu__opencl_nvidia_100[0x488dfc]
../../projects/setiathome.berkeley.edu/astropulse_7.08_x86_64-pc-linux-gnu__opencl_nvidia_100[0x461896]
../../projects/setiathome.berkeley.edu/astropulse_7.08_x86_64-pc-linux-gnu__opencl_nvidia_100[0x46a205]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5)[0x7fc8c719dec5]
../../projects/setiathome.berkeley.edu/astropulse_7.08_x86_64-pc-linux-gnu__opencl_nvidia_100[0x40bd89]

Exiting...

</stderr_txt>
]]>
15) Questions and Answers : Unix/Linux : One of my computers is "lost" (Message 1756478)
Posted 15 Jan 2016 by Profile ralphw
Post:
Sometimes BOINC assigns separate identities to the same computer by mistake. You can correct this by merging old identities with the newest one.

This list below seems pretty strange.

It looks like I should ALWAYS do an "add project" by signing up with email/password, instead of using the authenticator in the account_info.xml file.

>> Here's the output from when I went into the "merge computers" section.

Host 1161925 has an incompatible OS: (Linux, Darwin)
Host 1284323 has an incompatible OS: (Linux, Microsoft Windows 2000)
Host 2175038 has an incompatible CPU: (GenuineIntel Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz [Family 6 Model 60 Stepping 3], GenuineIntel Intel(R) Pentium(R) 4 CPU 3.00GHz)
Host 5070334 has an incompatible OS: (Linux, Microsoft Windows XP)
Host 5080967 has an incompatible OS: (Linux, Darwin)
Host 6437752 has overlapping lifetime: (13 Jan 2016 - 15 Jan 2016), (25 Feb 2012 - 14 Jan 2016)
Host 7582484 has an incompatible OS: (Linux, Darwin)
Host 7834880 has an incompatible CPU: (GenuineIntel Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz [Family 6 Model 60 Stepping 3], GenuineIntel Intel(R) Pentium(R) CPU J2900 @ 2.41GHz [Family 6 Model 55 Stepping 8])
Host 7837118 has overlapping lifetime: (13 Jan 2016 - 15 Jan 2016), (29 Nov 2015 - 15 Jan 2016)
Host 7874920 has overlapping lifetime: (13 Jan 2016 - 15 Jan 2016), (5 Jan 2016 - 15 Jan 2016)
No hosts are eligible for merging with this one.
16) Questions and Answers : Unix/Linux : One of my computers is "lost" (Message 1755914)
Posted 13 Jan 2016 by Profile ralphw
Post:
Yes, both are visible now.

I'm still wondering what happened, since it was missing from the earlier list of "my computers".

I did put the USB->Wireless adapter back in its original place, and turned that machine back on.
17) Questions and Answers : Unix/Linux : One of my computers is "lost" (Message 1755896)
Posted 13 Jan 2016 by Profile ralphw
Post:
A processor with a XEON CPU, Locutus, is no longer listed under my 'computers'.

How does BOINC/SETI determine the Computer ID?
My stats no longer seem to list 'locutus', a XEON 3550 on a HP Z400 board.

I did two things within an hour of each other.

First, I moved the USB-> Wireless adapter from 'locutus' to a different computer, named 'Asimov'. Did the computer ID move with the Mac address?
(They run the same version of Linux mint, if that matters.)

I also moved my SETI account info.xml file for SETI from a third computer to Asimov.

But I still see 'Sagan' (oldest computer - core i7-920) and 'Asimov' (newest computer, Intel 4790k) as independent systems.

I'm stumped as to how this happened.

- Ralph
18) Message boards : Number crunching : Intel / OpenCL binaries for Linux (2015) (Message 1754032)
Posted 5 Jan 2016 by Profile ralphw
Post:
Is there direct support for the Intel Graphics (HD 4600) chip?

I'm now using a motherboard with an Intel i7 4790k, plus a couple of NVIDIA GEFORCE GTX 950s. Having all three doing GPU tasks would be wonderful.

The OS X Science app seems to be able to do this, though it seems both chips are using OpenCL support:


  • NVIDIA GeForce GT 750M (2048MB) OpenCL: 1.2,
  • INTEL Iris Pro (1536MB) OpenCL: 1.2



If I need to install Darwin (Unix) on this Linux (Unix) box, I'll give that a shot if it means more RAC.

19) Message boards : News : SETI@home Version 8 has been released. (Message 1753770)
Posted 4 Jan 2016 by Profile ralphw
Post:
Thanks for the update. I had noticed my RAC had dropped off, more due to networking issues than anything else.

I see that I'm running a combination of SETI@home v7 and SETI@home v8.

Do I need to do anything to the BOINC software at this point?

Computing for science on Unix,
20) Message boards : Number crunching : Different Setting for more than 1 GPU (Message 1745663)
Posted 29 Nov 2015 by Profile ralphw
Post:
If I could chose the time-space tradeoffs for each card, that would be helpful.
While Astropulse is more constrained by memory bandwidth, it would be great if there were explicit controls available to maximize GPU utilization for the specific architecture, for *all* of the SETI@home science clients.

<device0> (for example, NVidia 950 w/ 4G RAM)
<unroll>32</unroll>
<ffa_block>12288</ffa_block>
<ffa_block_fetch>16384</ffa_block_fetch>
</device0>

<device1> (for example, NVidia 750 with 2G RAM)
<unroll>16</unroll>
<ffa_block>6144</ffa_block>
<ffa_block_fetch>8192</ffa_block_fetch>
</device1>

I'm making assumptions here:
- that more loop unrolling is better
- more unrolling needs more RAM on the host and the GPU, but it gives more memory bandwidth.
- it's sad that compilers can't implement this in 2015.


Is there a utility that can be run to calculate the optimal SETI settings for a specific GPU, from benchmarking runs? (NVidia, AMD, or Intel)


Next 20


 
©2017 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.