Humor me...I talk to myself

RONNIE

Well-Known Member
USA team member
Ronnie, I am doing well and I am running only Rosetta@home. I de-tuned the mining rig by disconnecting one power supply and the video cards. It runs the Intel i5-7600 @4.0GHz.

I have had a very productive spring as shown in these Facebook photos: Sunroom, Sunroom2, and Eight gallons of stain.

I am extremely disappointed with the WCG GPU client and only wish to run projects that are looking for cures for diseases. I like Rosetta, and I am a University of Washington alumnus (Oceanography, '69). It was nice to reach number 3 with the mining rig at Milkyway@home, but the reward was only for my ego.
The house looks very beautiful! Congratulations!
 

Vester

Well-Known Member
USA team member
I reconfigured the miner to run WCG OPNG, cloned the SSD to a SSHD, and promptly ran out of GPU work units. I'll let it stew a while and see if I get more GPU work. It ran six GPU of 8 before it ran out of work. I have it setup to run 24 at a time.
 

RONNIE

Well-Known Member
USA team member
I reconfigured the miner to run WCG OPNG, cloned the SSD to a SSHD, and promptly ran out of GPU work units. I'll let it stew a while and see if I get more GPU work. It ran six GPU of 8 before it ran out of work. I have it setup to run 24 at a time.
Ok! Have you also increased the Boinc cache so that it is prepared to download more tasks as soon as they are available to be sent by the server? I've set my cache to 1 full day, but @Nick Name told us he is getting enough tasks using a lower value.
 

Vester

Well-Known Member
USA team member
I had 0.5 days, but changed it to 1.0 days cache. I am running boinc 7.15.0, so the problem is the lack of work from the work server.
Thirty-three valid so far this afternoon.
 

RONNIE

Well-Known Member
USA team member
I had 0.5 days, but changed it to 1.0 days cache. I am running boinc 7.15.0, so the problem is the lack of work from the work server.
Thirty-three valid so far this afternoon.
I don't know exactly how the Boinc program work but if I'm not wrong after some hours it will download more tasks of the various types.
 

Vester

Well-Known Member
USA team member
I could get a fabulous RAC if work were available all of the time. The miner has not had any WUs in the past three or four hours.
 

Vester

Well-Known Member
USA team member
There is a topic at World Community Grid about Graphics Card Performance. The topic is already over ten pages long.

Graphics card performance is meaningless on a project that cannot provide enough work to keep a graphics busy more than 1% of the time. My mining rig went about 24 hours without one AMD GPU work unit, and I am running it at 50 capacity to save power. If it doesn't get better, I may reduce it to one graphics card.
 

Vester

Well-Known Member
USA team member
Nick Name, did you see how well my HD 7990s do? Your Radeon VII can do even better (about 1.7 times better than one Radeon HD 7990 with two GPUs).
 

Vester

Well-Known Member
USA team member
There's nothing like a good rant to change things. The mining rig, 4HD7990, has done about 75 work units since I posted above.
 

RONNIE

Well-Known Member
USA team member
There is a topic at World Community Grid about Graphics Card Performance. The topic is already over ten pages long.

Graphics card performance is meaningless on a project that cannot provide enough work to keep a graphics busy more than 1% of the time. My mining rig went about 24 hours without one AMD GPU work unit, and I am running it at 50 capacity to save power. If it doesn't get better, I may reduce it to one graphics card.
Things again are not working well here: today I've received lots of Intel GPU tasks, but nothing for Nvidia. Perhaps I will move back to F@H+WCG/OPN. I will see...
 

Nick Name

Administrator
USA team member
Nick Name, did you see how well my HD 7990s do? Your Radeon VII can do even better (about 1.7 times better than one Radeon HD 7990 with two GPUs).
I may give it a shot but I could never get F@H working so I don't have much hope for WCG either. Results outside of MilkyWay have been mixed. Most of the other projects I've tried just flat out won't work. PrimeGrid runs, at least some it, but not that well compared to my 2080 Ti. I'm sure part of the problem is that I'm running Windoze 8 on this machine and there's probably some driver compatibility problem.
 

Vester

Well-Known Member
USA team member
I gave up on WCG today. My mining rig got one work unit in 24 hours. They aren't serious about GPUs.

The mining rig, 4HD7990, is running flat out at Milkyway@home. Just for my ego.
 

RONNIE

Well-Known Member
USA team member
I gave up on WCG today. My mining rig got one work unit in 24 hours. They aren't serious about GPUs.

The mining rig, 4HD7990, is running flat out at Milkyway@home. Just for my ego.
Understandably so... Perhaps in the future they will get it right. But I don't think Milkyway is only for your ego (and for our team)! And what about F@H?
 

Vester

Well-Known Member
USA team member
re: Folding@home, my legacy Radeon HD 7990 graphics cards, eight years old, run OpenCL 1.2, not 2.1, with the drivers that I have tried.
 

Vester

Well-Known Member
USA team member
After using MSI Afterburner, one of the graphics cards (two GPUs) acquired a minimum of 750 MHz VRAM speed in Radeon Settings instead of 150 MHz. I had had one error prior to removal of Afterburner. I also replaced one of the twelve fans on the graphics cards (stopped running). I installed the Blockchain Beta driver because the Radeon Settings with a recent driver was sketchy, slow to open, to respond, and frequently crashed on the Windows 10 miner. All is good now.
 
Last edited:

Vester

Well-Known Member
USA team member
Were you having some kind of problem?
The rig hasn't managed to run 24 hours continuously for the last three or four days without a problem of some sort. I got up this morning and one of the video cards was not doing any work (fans running). I restarted and BIOS indicated that I had a problem with that card. I shutdown and disconnected/reconnected the PCIe x1 riser connection, the two USB connections, and the PCIe x16 riser. It runs well again after a restart.

That video card was the one that ran a WCG OPNG task for over an hour last night before completion. EVERY time I run WCG tasks, I have poor computer response. A 1250 Watt power supply and one Radeon HD 7990 failed during WCG Beta testing, not necessarily fully due to WCG warping the heck out of voltage regulation circuits. I have detached from WCG for the last time.

An aside: Have you Windows 10 users monitored your video cards in Task Manager? It is good. Click on the Performance tab, then click on a GPU and select "Compute 1" from a Dropdown menu to the right. You can see the GPU's computing utilization as well as the GPU temperature.
 

Vester

Well-Known Member
USA team member
BOINC Manager crashes are well-documented and until yesterday my computers had been unaffected. I am keeping my hands off of the mining rig to see if it happens again as it has three times in the past two days. If BOINC Manager crashes again, I will try a clean reinstallation.

Edit: It crashed again after a dirty reinstallation without any human input. I will let it finish tasks in queue, uninstall, remove any remaining files, and remove any remaining BOINC registry entries before reinstalling.
 
Last edited:
Top