Hi,

I have an ultra-spec PC that I built for gaming and for rendering:

CPU: Intel i7-8700K
Motherboard: Asus ROG Maximus X Formula (Intel Z370 chipset), with onboard Intel HD 630 GPU
Graphics: Two (2x) Nvidia GeForce GTX TITAN XP, SLI together
Memory: 64 GB
Windows 10, 64 bit

It came with the mother board's Intel HD 630 GPU disabled (one of the Nvidia Titan XP is my primary display card).

Question: how do I enable the onboard Intel GPU so that Seti@Home sees it as an additional GPU for computing? For example: my MacBook Pro also has two GPUs: Radeon Pro & Intel 630, and both are used by Seti@Home.

I'd like to be able to use the Intel 630 GPU on my PC the same way, in addition to the two GeForce Titan XP's. I've found in the BIOS where to enable the integrated Intel GPU, and, I've installed the drivers in Windows so that Windows sees it (and reports that "the device is working properly"), but Seti@Home doesn't assign any tasks to it.

Any ideas? Has anyone achieved this before? Note that I have two GeForce TITAN XP (not one) and they are in SLI mode. Is that perhaps limiting?

Thank you,

Mark

That's a setting in your BIOS/UEFI, but not all motherboards support running the internal GPU together with an external GPU.
But then, why would you want to? It won't add much to the 2x Titans with their 3840 CUDA cores per GPU. Those two Titans will run rings around the Intel GPU, which will mostly add heat and use up CPU cycles better given to the Titans. .

---

Hi Jord,

Thanks for your response. It is very true that the twin Titan XPs are much faster than the intel 630. However, like on my Mac, it is a GPU, and it can process another thread. So there is a noticeable progress gain on my Mac with it's intel 630 GPU enabled, and as such, I would think there would be a gain (however small, relatively speaking) to using a GPU rather than leaving it dormant. Plus, any openCL app could throw stuff at the intel GPU.

Are you suggesting that the CPU overhead to managing the Intel GPU would be loss in performance overall?

Thank you,

Mark

---

If you're also running work on the CPU, running work on the built-in GPU will slow down work done by the CPU. That's because it's all one item really. But before the if's and however's, you still need to figure if you can enable the internal GPU.

Now, I have just perused your motherboard manual and I can't find any mention of it. However, it may be something has been added at a later BIOS update that then isn't in the manual. (There was an update on the 16th of June)

Edit: ah, it should be in Advanced menu > System Agent (SA) Configuration\Graphics Configuration > iGPU Multi-Monitor setting > Enable
See their FAQ.

---

Hi,

Yes I can enable the Intel GPU, and windows see it in the device manager too. But, setiathome home doesn’t. That was my question (why not?)

Cheers,

Mark

---

Did you enable using the Intel GPU in the project preferences (and for the correct location)?

---

Hi,

Yes, I did.

Thank you,

Mark

---

If that still doesn't work do the following:
Open BOINC Manager->View->Advanced view->Options->Event Log options->check coproc_debug->check work_fetch_debug->Save.
Exit BOINC completely and restart it.
Open BOINC Manager->Projects->Select Seti->click Update.
Press CTRL+SHIFT+E to open Event Log and copy and post all output.

You need to exit &restart BOINC because all coprocessor detection output is only shown at the start of BOINC.

---

Hi, here you go:

14-Jul-18 12:58:20 PM | | cc_config.xml not found - using defaults
14-Jul-18 12:58:20 PM | | Starting BOINC client version 7.10.2 for windows_x86_64
14-Jul-18 12:58:20 PM | | log flags: file_xfer, sched_ops, task
14-Jul-18 12:58:20 PM | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2g zlib/1.2.8
14-Jul-18 12:58:20 PM | | Data directory: C:\ProgramData\BOINC
14-Jul-18 12:58:20 PM | | Running under account Mark
14-Jul-18 12:58:20 PM | | CUDA: NVIDIA GPU 0: TITAN Xp (driver version 398.36, CUDA version 9.2, compute capability 6.1, 4096MB, 3551MB available, 12150 GFLOPS peak)
14-Jul-18 12:58:20 PM | | CUDA: NVIDIA GPU 1: TITAN Xp (driver version 398.36, CUDA version 9.2, compute capability 6.1, 4096MB, 3551MB available, 12150 GFLOPS peak)
14-Jul-18 12:58:20 PM | | OpenCL: NVIDIA GPU 0: TITAN Xp (driver version 398.36, device version OpenCL 1.2 CUDA, 12288MB, 3551MB available, 12150 GFLOPS peak)
14-Jul-18 12:58:20 PM | | OpenCL: NVIDIA GPU 1: TITAN Xp (driver version 398.36, device version OpenCL 1.2 CUDA, 12288MB, 3551MB available, 12150 GFLOPS peak)
14-Jul-18 12:58:20 PM | | OpenCL CPU: Intel(R) Core(TM) i7-8700K CPU @ 3.70GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 7.6.0.611, device version OpenCL 2.1 (Build 611))
14-Jul-18 12:58:20 PM | | Version change (7.8.3 -> 7.10.2)
14-Jul-18 12:58:20 PM | | Host name: SeegerUltimate
14-Jul-18 12:58:20 PM | | Processor: 12 GenuineIntel Intel(R) Core(TM) i7-8700K CPU @ 3.70GHz [Family 6 Model 158 Stepping 10]
14-Jul-18 12:58:20 PM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx smx tm2 pbe fsgsbase bmi1 hle smep bmi2
14-Jul-18 12:58:20 PM | | OS: Microsoft Windows 10: Core x64 Edition, (10.00.17134.00)
14-Jul-18 12:58:20 PM | | Memory: 63.86 GB physical, 73.36 GB virtual
14-Jul-18 12:58:20 PM | | Disk: 1.86 TB total, 1.64 TB free
14-Jul-18 12:58:20 PM | | Local time is UTC -7 hours
14-Jul-18 12:58:20 PM | | VirtualBox version: 5.2.8
14-Jul-18 12:58:20 PM | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 8496994; resource share 100
14-Jul-18 12:58:20 PM | SETI@home | General prefs: from SETI@home (last modified 13-Sep-2014 01:19:17)
14-Jul-18 12:58:20 PM | SETI@home | Computer location: home
14-Jul-18 12:58:20 PM | SETI@home | General prefs: no separate prefs for home; using your defaults
14-Jul-18 12:58:20 PM | | Reading preferences override file
14-Jul-18 12:58:20 PM | | Preferences:
14-Jul-18 12:58:20 PM | | max memory usage when active: 32696.33 MB
14-Jul-18 12:58:20 PM | | max memory usage when idle: 58853.39 MB
14-Jul-18 12:58:20 PM | | max disk usage: 100.00 GB
14-Jul-18 12:58:20 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)
14-Jul-18 12:58:20 PM | | Setting up project and slot directories
14-Jul-18 12:58:20 PM | | Checking active tasks
14-Jul-18 12:58:20 PM | | Setting up GUI RPC socket
14-Jul-18 12:58:20 PM | | Checking presence of 293 project files
14-Jul-18 12:58:20 PM | | Running CPU benchmarks
14-Jul-18 12:58:20 PM | | Suspending computation - CPU benchmarks in progress
14-Jul-18 12:58:51 PM | | Benchmark results:
14-Jul-18 12:58:51 PM | | Number of CPUs: 12
14-Jul-18 12:58:51 PM | | 5524 floating point MIPS (Whetstone) per CPU
14-Jul-18 12:58:51 PM | | 21004 integer MIPS (Dhrystone) per CPU
14-Jul-18 12:59:38 PM | SETI@home | General prefs: from SETI@home (last modified 13-Sep-2014 01:19:17)
14-Jul-18 12:59:38 PM | SETI@home | Computer location: home
14-Jul-18 12:59:38 PM | SETI@home | General prefs: no separate prefs for home; using your defaults
14-Jul-18 12:59:38 PM | | Reading preferences override file
14-Jul-18 12:59:38 PM | | Preferences:
14-Jul-18 12:59:38 PM | | max memory usage when active: 32696.33 MB
14-Jul-18 12:59:38 PM | | max memory usage when idle: 58853.39 MB
14-Jul-18 12:59:38 PM | | max disk usage: 100.00 GB
14-Jul-18 12:59:38 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)
14-Jul-18 12:59:42 PM | SETI@home | Computation for task blc15_2bit_guppi_58227_14378_HIP66840_0038.5135.818.21.44.208.vlar_0 finished
14-Jul-18 12:59:42 PM | SETI@home | Starting task blc15_2bit_guppi_58227_11799_HIP53117_0031.15126.818.21.44.130.vlar_1
14-Jul-18 12:59:44 PM | SETI@home | Started upload of blc15_2bit_guppi_58227_14378_HIP66840_0038.5135.818.21.44.208.vlar_0_r1005850059_0
14-Jul-18 12:59:47 PM | SETI@home | Finished upload of blc15_2bit_guppi_58227_14378_HIP66840_0038.5135.818.21.44.208.vlar_0_r1005850059_0
14-Jul-18 1:00:12 PM | SETI@home | Computation for task blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.97.vlar_1 finished
14-Jul-18 1:00:12 PM | SETI@home | Starting task blc15_2bit_guppi_58227_11799_HIP53117_0031.15126.818.21.44.242.vlar_1
14-Jul-18 1:00:14 PM | SETI@home | Started upload of blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.97.vlar_1_r304653083_0
14-Jul-18 1:00:16 PM | SETI@home | Finished upload of blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.97.vlar_1_r304653083_0
14-Jul-18 1:00:49 PM | SETI@home | Computation for task blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.110.vlar_1 finished
14-Jul-18 1:00:49 PM | SETI@home | Computation for task blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.164.vlar_0 finished
14-Jul-18 1:00:49 PM | SETI@home | Starting task blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.122.vlar_1
14-Jul-18 1:00:49 PM | SETI@home | Starting task blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.117.vlar_1
14-Jul-18 1:00:51 PM | SETI@home | Started upload of blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.164.vlar_0_r1026008233_0
14-Jul-18 1:00:51 PM | SETI@home | Started upload of blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.110.vlar_1_r177998188_0
14-Jul-18 1:00:53 PM | SETI@home | Finished upload of blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.164.vlar_0_r1026008233_0
14-Jul-18 1:00:53 PM | SETI@home | Finished upload of blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.110.vlar_1_r177998188_0
14-Jul-18 1:00:56 PM | SETI@home | Computation for task blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.4.vlar_0 finished
14-Jul-18 1:00:56 PM | SETI@home | Starting task blc15_2bit_guppi_58227_11799_HIP53117_0031.14440.818.21.44.6.vlar_0
14-Jul-18 1:00:58 PM | SETI@home | Started upload of blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.4.vlar_0_r1758012763_0
14-Jul-18 1:01:00 PM | SETI@home | Finished upload of blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.4.vlar_0_r1758012763_0
14-Jul-18 1:02:04 PM | | Re-reading cc_config.xml
14-Jul-18 1:02:04 PM | | log flags: file_xfer, sched_ops, task, coproc_debug, work_fetch_debug
14-Jul-18 1:02:04 PM | | [work_fetch] Request work fetch: Core client configuration
14-Jul-18 1:02:05 PM | SETI@home | [coproc] NVIDIA instance 0; 1.000000 pending for blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.122.vlar_1
14-Jul-18 1:02:05 PM | SETI@home | [coproc] NVIDIA instance 0; 1.000000 pending for blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.117.vlar_1
14-Jul-18 1:02:05 PM | SETI@home | [coproc] NVIDIA instance 0: confirming 1.000000 instance for blc15_2bit_guppi_58227_12160_HIP53954_0032.5710.0.22.45.122.vlar_1
14-Jul-18 1:02:05 PM | SETI@home | [coproc] NVIDIA instance 1: confirming 1.000000 instance for blc15_2bit_guppi_58227_12507_HIP53337_0033.5157.818.22.45.117.vlar_1
14-Jul-18 1:02:09 PM | | [work_fetch] ------- start work fetch state -------
14-Jul-18 1:02:09 PM | | [work_fetch] target work buffer: 43200.00 + 43200.00 sec
14-Jul-18 1:02:09 PM | | [work_fetch] --- project states ---
14-Jul-18 1:02:09 PM | SETI@home | [work_fetch] REC 41252.540 prio -1.022 can request work
14-Jul-18 1:02:09 PM | | [work_fetch] --- state for CPU ---
14-Jul-18 1:02:09 PM | | [work_fetch] shortfall 0.00 nidle 0.00 saturated 102539.19 busy 0.00
14-Jul-18 1:02:09 PM | SETI@home | [work_fetch] share 1.000
14-Jul-18 1:02:09 PM | | [work_fetch] --- state for NVIDIA GPU ---
14-Jul-18 1:02:09 PM | | [work_fetch] shortfall 0.00 nidle 0.00 saturated 106645.83 busy 0.00
14-Jul-18 1:02:09 PM | SETI@home | [work_fetch] share 1.000
14-Jul-18 1:02:09 PM | | [work_fetch] ------- end work fetch state -------
14-Jul-18 1:02:09 PM | | [work_fetch] No project chosen for work fetch
14-Jul-18 1:02:24 PM | SETI@home | update requested by user
14-Jul-18 1:02:24 PM | | [work_fetch] Request work fetch: project updated by user
14-Jul-18 1:02:24 PM | | [work_fetch] ------- start work fetch state -------
14-Jul-18 1:02:24 PM | | [work_fetch] target work buffer: 43200.00 + 43200.00 sec
14-Jul-18 1:02:24 PM | | [work_fetch] --- project states ---
14-Jul-18 1:02:24 PM | SETI@home | [work_fetch] REC 41252.540 prio -1.022 can request work
14-Jul-18 1:02:24 PM | | [work_fetch] --- state for CPU ---
14-Jul-18 1:02:24 PM | | [work_fetch] shortfall 0.00 nidle 0.00 saturated 101931.40 busy 0.00
14-Jul-18 1:02:24 PM | SETI@home | [work_fetch] share 1.000
14-Jul-18 1:02:24 PM | | [work_fetch] --- state for NVIDIA GPU ---
14-Jul-18 1:02:24 PM | | [work_fetch] shortfall 0.00 nidle 0.00 saturated 106335.32 busy 0.00
14-Jul-18 1:02:24 PM | SETI@home | [work_fetch] share 1.000
14-Jul-18 1:02:24 PM | | [work_fetch] ------- end work fetch state -------
14-Jul-18 1:02:24 PM | SETI@home | [work_fetch] request: CPU (0.00 sec, 0.00 inst) NVIDIA GPU (0.00 sec, 0.00 inst)
14-Jul-18 1:02:24 PM | SETI@home | Sending scheduler request: Requested by user.
14-Jul-18 1:02:24 PM | SETI@home | Reporting 5 completed tasks
14-Jul-18 1:02:24 PM | SETI@home | Not requesting tasks: don't need (CPU: job cache full; NVIDIA GPU: job cache full)
14-Jul-18 1:02:25 PM | SETI@home | Scheduler request completed
14-Jul-18 1:02:25 PM | | [work_fetch] Request work fetch: RPC complete
14-Jul-18 1:02:30 PM | | [work_fetch] ------- start work fetch state -------
14-Jul-18 1:02:30 PM | | [work_fetch] target work buffer: 43200.00 + 43200.00 sec
14-Jul-18 1:02:30 PM | | [work_fetch] --- project states ---
14-Jul-18 1:02:30 PM | SETI@home | [work_fetch] REC 41252.540 prio -0.022 can't request work: scheduler RPC backoff (297.87 sec)
14-Jul-18 1:02:30 PM | | [work_fetch] --- state for CPU ---
14-Jul-18 1:02:30 PM | | [work_fetch] shortfall 0.00 nidle 0.00 saturated 101891.10 busy 0.00
14-Jul-18 1:02:30 PM | SETI@home | [work_fetch] share 0.000
14-Jul-18 1:02:30 PM | | [work_fetch] --- state for NVIDIA GPU ---
14-Jul-18 1:02:30 PM | | [work_fetch] shortfall 0.00 nidle 0.00 saturated 106206.06 busy 0.00
14-Jul-18 1:02:30 PM | SETI@home | [work_fetch] share 0.000
14-Jul-18 1:02:30 PM | | [work_fetch] ------- end work fetch state -------
14-Jul-18 1:02:30 PM | | [work_fetch] No project chosen for work fetch

---

I dont see any detection of the intel 630 GPU. However in the Device Managert it doest list it, see attached image

[/img]

---

sorry I cant figure out how to post an image here... can you help?[/img]

---

That Windows sees the GPU doesn't mean BOINC should too. To be able to use the GPU in BOINC it needs drivers, not drivers supplied by Windows, but drivers supplied by Intel. Drivers that contain OpenCL. That your CPU is detected as OpenCL capable is nice, but it shows that the OpenCL drivers on the system (most probably the Nvidia OpenCL component) aren't updated for your Intel GPU.

So go to https://downloadcenter.intel.com/ and get the Graphics drivers (probably these) and install them.

Then exit BOINC & restart it. (A thing I asked you to do before after you added coproc_debug, but that you forgot) and post your log.

These forums don't have an option to store images. If you want to show images you have to store them externally on a third party image site. To use the img tags, it requires that the image name ends on .bmp, .jpg, .png or .gif

---

Dear Jord,

It worked! I was able to enable the integrated Intel HD 630 GPU with the latest OpenCL drivers (per your link), and its doing precisely what I expected, namely: Seti@Home sees it and sends it a separate compute thread.

This is fascinating: I am now able to compute 14 separate Seti@Home threads with the Intel 630 enabled, where as before I was only able to do 12.

BEFORE Intel HD 630 enabled:
12 simultaneous compute threads: 10x running on my CPU (Intel i7-8700K, 6 cores), and 1x running on each of my NVidia Titan Xp (I have two of them, SLI connected)

AFTER Intel HD 630 enabled:

14 compute threads: 11x running on my CPU, 1x running on each of the NVidia Titan Xp (two Titans), and 1x running on the Intel HD 630 GPU.

Interestingly as well that the Intel HD 630 GPU takes less time per Seti@Home compute task than a the CPU. It takes approx 56 to 59 minutes to complete one Seti@Home task for the intel GPU, where as it takes slightly over an hour on the CPU. This Intel GPU compute performance is consistent with my MacBook which has two GPUs: ATI Radeon GPU and the same Intel 630 GPU.

So, mission accomplished (greater compute performance), and a surprise that my CPU can now run 11x threads simultaneously as well. I do not know why, do you?

Later I'll run some full system benchmark software and see if there is indeed an overall system performance boost, and I'll share the results from before & after here. The way the OpenCL drivers work (to my understanding) is that there should be an across the system performance boost as the OpenCL drivers send various compute tasks to all of the GPUs in the system (different things for different applications).

Yes, it is true that enabling the Intel HD 630 GPU will use additional system memory, but I have 64 GB (enough to spare). I've also been told that the overall system performance may drop because of the added bandwidth requirements of moving data to/from the Intel GPU. However, the GPU can work quite fast on certain tasks (such as Seti@Home). I'll play with this and share my overall system performance findings.

Thank you again for your help Jord, and I hope this is useful to anyone wishing to tweak their system for max capable performance.

Cheers,

Mark

---

An update:

Average processing time per Seti@Home work unit on the Intel 630 GPU: 35 minutes

Average processing time per work unit on CPU: 1h5m to 1h25m

Average processing time per work unit on NVidia Titan Xp: 3 min to 4 min

Cheers,

Mark

---

An update:

Average processing time per Seti@Home work unit on the Intel 630 GPU: 35 minutes

Average processing time per work unit on CPU: 1h5m to 1h25m

Average processing time per work unit on NVidia Titan Xp: 3 min to 4 min

Cheers,

Mark

Hurray!

---

A proud member of the OFA (Old Farts Association).

1) I have been told you don't want an SLI bridge in place for Seti gpu processing. You should be able to run upto 3 tasks on each gpu (w/o Intel gpu).

2) I just got done testing with and without an Intel HD 4000 gpu. The total speed of the cpu processing goes up significantly without the Intel internal gpu being used in Seti processing. So task time went down more than 15 minutes per task in my case. Yours will probably be different.

HTH,
Tom

---

A proud member of the OFA (Old Farts Association).

Hi Tom,

I wonder what is different about our setup, because my RAC goes up by ~4% WITH intel GPU enabled (at the system level), and down ~4% with the intel DISABLED... thoughts?

Thank you
Mark

---

Hi Tom,

I wonder what is different about our setup, because my RAC goes up by ~4% WITH intel GPU enabled (at the system level), and down ~4% with the intel DISABLED... thoughts?

Thank you
Mark

If your RAC does that over a 2-4 week testing period then the difference is probably real.

Why? I have read that Intel has been making its internal gpu's better and better due to the pressure of products like AMD's "All in One" which where including standard Radeon gpu's as part of the motherboard.
It is possible that your Intel gpu is significantly better than the HD 4000 I was testing on. It certainly is a different model.

You might download GPU-Z and examine all your gpus. If they aren't running above 90% most of the time then you probably want to increase the # of tasks running on them (in parallel). Right now, I can't think of a way to do it to the Titans and not do it to the Intel. That might slow your whole computer down.

HTH,
Tom

---

A proud member of the OFA (Old Farts Association).

Third generation Intel HD Graphics versus eight generation Intel HD Graphics.

Model number 	  Execution units 	Shading units 	Boost Clock (MHz)    max GFLOPS 
HD Graphics 4000 	16 	             128 	      1300 	        332.8 

HD Graphics 630 	24 	             192       	    1000–1150        384 − 441.6 

---

Tom, you bring up an interesting question: how do I force more parallel work unit tasks on a GPU? at the moment my entire SaH experience only has one work task per GPU, i've never seen more than 1 (as opposed to CPU where you have one work unit task per thread, and typically 2 threads per core (on Intels))

Thanks!

Mark

---