| 1 |
Hello, |
| 2 |
|
| 3 |
On Mon, Nov 13, 2017 at 9:46 AM, Peter Humphrey <peter@××××××××××××.uk> wrote: |
| 4 |
> On Monday, 13 November 2017 15:12:56 GMT Daniel Frey wrote: |
| 5 |
>> On 11/13/17 02:59, Peter Humphrey wrote: |
| 6 |
>> > Hello list, |
| 7 |
>> > |
| 8 |
>> > I'm hunting a problem with cooling in this box, and I've got as far as |
| 9 |
>> > suspecting my new AMD WX 5100 GPU. |
| 10 |
>> > |
| 11 |
>> > One of my BOINC projects causes the GPU temperature, as shown by |
| 12 |
>> > gkrellm, to shoot up to 75C or more and cause intolerable system |
| 13 |
>> > cooling noise. If I suspend that project but leave the other seven |
| 14 |
>> > running, the temperature returns to what I hope is a normal 55C. Those |
| 15 |
>> > seven projects are supposed to use the GPU, but I'm not sure whether |
| 16 |
>> > they do in fact. |
| 17 |
>> > |
| 18 |
>> > Is there any way I can monitor what is using the GPU, to find out? |
| 19 |
>> |
| 20 |
>> I don't know if there's a utility for consumer level cards that can do |
| 21 |
>> this. I do remember for Nvidia there's nvidia-smi but I don't think it |
| 22 |
>> will list processes for desktop cards. |
| 23 |
> |
| 24 |
> This isn't consumer grade (look it up in your local shops ;-) ): |
| 25 |
> |
| 26 |
> # lspci -v -s 01:00.0 |
| 27 |
> 01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] |
| 28 |
> Ellesmere [Radeon Pro WX 5100] (prog-if 00 [VGA controller]) |
| 29 |
> Subsystem: Advanced Micro Devices, Inc. [AMD/ATI] Ellesmere [Radeon |
| 30 |
> Pro WX 5100] |
| 31 |
> Flags: bus master, fast devsel, latency 0, IRQ 34, NUMA node 0 |
| 32 |
> Memory at c0000000 (64-bit, prefetchable) [size=256M] |
| 33 |
> Memory at d0000000 (64-bit, prefetchable) [size=2M] |
| 34 |
> I/O ports at e000 [size=256] |
| 35 |
> Memory at fbe00000 (32-bit, non-prefetchable) [size=256K] |
| 36 |
> Expansion ROM at 000c0000 [disabled] [size=128K] |
| 37 |
> Capabilities: [48] Vendor Specific Information: Len=08 <?> |
| 38 |
> Capabilities: [50] Power Management version 3 |
| 39 |
> Capabilities: [58] Express Legacy Endpoint, MSI 00 |
| 40 |
> Capabilities: [a0] MSI: Enable+ Count=1/1 Maskable- 64bit+ |
| 41 |
> Capabilities: [100] Vendor Specific Information: ID=0001 Rev=1 |
| 42 |
> Len=010 <?> |
| 43 |
> Capabilities: [150] Advanced Error Reporting |
| 44 |
> Capabilities: [200] #15 |
| 45 |
> Capabilities: [270] #19 |
| 46 |
> Capabilities: [2b0] Address Translation Service (ATS) |
| 47 |
> Capabilities: [2c0] Page Request Interface (PRI) |
| 48 |
> Capabilities: [2d0] Process Address Space ID (PASID) |
| 49 |
> Capabilities: [320] Latency Tolerance Reporting |
| 50 |
> Capabilities: [328] Alternative Routing-ID Interpretation (ARI) |
| 51 |
> Capabilities: [370] L1 PM Substates |
| 52 |
> Kernel driver in use: amdgpu |
| 53 |
> |
| 54 |
>> The only other generic ones I can think of are cuda-z and gputop. Have |
| 55 |
>> you tried one of those? Although I don't think it'll give you the |
| 56 |
>> information you need either. |
| 57 |
> |
| 58 |
> As it's AMD, not nVidia, nvidia-smi and cuda aren't suitable. I hadn't heard |
| 59 |
> of GPU Top - thanks. I'll have a look at it. |
| 60 |
> |
| 61 |
> I forgot to add that I'm using the proprietary dev-libs/amdgpu-pro-opencl |
| 62 |
> because mesa hasn't caught up yet. |
| 63 |
> |
| 64 |
|
| 65 |
The level of detail you want will likely necessitate the use of a GPU |
| 66 |
debugger. AMD provides CodeXL, located at |
| 67 |
https://gpuopen.com/compute-product/codexl/. I suggest looking at the |
| 68 |
profiling features. |
| 69 |
|
| 70 |
You may want to communicate your findings to the relevant BOINC projects. |
| 71 |
|
| 72 |
Cheers, |
| 73 |
R0b0t1 |
Archives