| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Beta Testing
Forum: Beta Test Support Forum
Thread: Beta Test for Help Conquer Cancer - GPU v6.56 |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 89
|
|
| Author |
|
|
Sabrina Tarson
Advanced Cruncher United States Joined: Jun 27, 2012 Post Count: 149 Status: Offline Project Badges: 
|
I too am running POEM@home at resource 1, and World Community Grid at Resource 100.
----------------------------------------If when the HCC GPU app comes out, and I stick with this approach, will POEM finish all its work units and not request anymore until HCC runs out of work? ---------------------------------------- [Edit 1 times, last edit by Chase Tarson at Oct 5, 2012 7:17:55 PM] |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges: 
|
I am assuming that POEM will request 1 unit of work for every 100 units of HCC, but there hasn't been that much HCC work yet to actually check it out. But at least it got 5 HCC work units for me, and then went back to POEM after they were completed, so it has the right idea. We just need to check the proportions when the full program is implemented. There have been all sorts of BOINC GPU scheduling bugs, and that may be the limiting factor for a while yet.
The real problem I see is that the HCC project will now be over so quickly that we will need something else to work on, though POEM is perfectly acceptable as an interim project for me. |
||
|
|
Sabrina Tarson
Advanced Cruncher United States Joined: Jun 27, 2012 Post Count: 149 Status: Offline Project Badges:
|
Jim, I have to ask, is there a way to disable POEM from using the CPU, and just use the GPU. I have it set not to use CPU, and only my AMD Card, but the GPU task requires 1 CPU + 1 GPU. Anyway to make it only use the GPU?
----------------------------------------I know this is for discussion of the beta, but there is no private messaging on this forum. |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
Jim, I have to ask, is there a way to disable POEM from using the CPU, and just use the GPU. I have it set not to use CPU, and only my AMD Card, but the GPU task requires 1 CPU + 1 GPU. Anyway to make it only use the GPU? I know this is for discussion of the beta, but there is no private messaging on this forum. To answer your question, no. POEM requires CPU cycles to run the GPU tasks because part of the operation can't be ported to the GPU. Same rule applies here to the HCC GPU tasks. It requires CPU cycles to finish the task for the same reasons.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
  |
||
|
|
Sabrina Tarson
Advanced Cruncher United States Joined: Jun 27, 2012 Post Count: 149 Status: Offline Project Badges:
|
Darn. Looks like I have to stick to Milkway@home for a GPU backup, seeing as it only uses (.05CPU + 1GPU).
----------------------------------------Its different if a World Community Grid task takes away 1 core just for a GPU, because it is all working for the same project, but I don't want the GPU backup project to take a core away from my main CPU project. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Its different if a World Community Grid task takes away 1 core just for a GPU, because it is all working for the same project, but I don't want the GPU backup project to take a core away from my main CPU project. Don't worry, the concern doesn't seem it would last for a long time.We are in a transition. We now have CPUs with integrated GPU-functionality, and the next race for the big three (AMD, nVidia, and Intel): GPUs with integrated CPU-functionality. A hardware-based integration of CPU-functionality in a GPU-chip, has the best shot at performance and efficiency at both the compute aspect and at the energy-use aspect than what the current model (separate CPU and separate GPU) for GPU-computing allows. ; |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Darn. Looks like I have to stick to Milkway@home for a GPU backup, seeing as it only uses (.05CPU + 1GPU). Its different if a World Community Grid task takes away 1 core just for a GPU, because it is all working for the same project, but I don't want the GPU backup project to take a core away from my main CPU project. Not the house expert, but as it is, the situation could differ depending on ATI/AMD or NVidia and their handling of OpenCL translating into a higher or lower CPU load from little to a lot. You may not want to, but the overall efficiency is generally served with this CPU core reservation. The application is a one stop shop, 1 compile for AMD/ATI, 1 compile for NVidia, no matter the card capabilities. System tweaks, if there are any applicable that do not compromise the task result, have to be set by the user. It's impossible to maintain compiles for every single family of GPU cards and driver versions... it's way worse than dealing with the CPU varieties and OSses. [Edit 1 times, last edit by Former Member at Oct 6, 2012 9:19:55 AM] |
||
|
|
Bearcat
Master Cruncher USA Joined: Jan 6, 2007 Post Count: 2803 Status: Offline Project Badges: 
|
Its different if a World Community Grid task takes away 1 core just for a GPU, because it is all working for the same project, but I don't want the GPU backup project to take a core away from my main CPU project. Don't worry, the concern doesn't seem it would last for a long time.We are in a transition. We now have CPUs with integrated GPU-functionality, and the next race for the big three (AMD, nVidia, and Intel): GPUs with integrated CPU-functionality. A hardware-based integration of CPU-functionality in a GPU-chip, has the best shot at performance and efficiency at both the compute aspect and at the energy-use aspect than what the current model (separate CPU and separate GPU) for GPU-computing allows. ; It may be efficient but would have to worry about burning it up faster than normal whereas a dedicated GPU can be swapped out pretty fast if it burns up. Would like to see some crunching numbers from folks crunching with a AMD APU or intel 3 & 4000 series to see how it compares to dedicated GPU's. Anyone know? (On the A-10 5700, it takes a little under 4 minutes to get to 99.415%, and then another 45 secs or so to finish up and start uploading.) Found one. One of my teammates is running a AMD A10-5700. Not bad considering my ATI 5670 is close to the same time. Might have to look into this.
Crunching for humanity since 2007!
---------------------------------------- [Edit 2 times, last edit by Bearcat at Oct 6, 2012 2:03:31 PM] |
||
|
|
Simplex0
Advanced Cruncher Sweden Joined: Aug 14, 2008 Post Count: 83 Status: Offline Project Badges:
|
I second that Bearcat. integrate a gpu with the same power as an high en AMD or Nvidia card in to a CPU would create a tremendous heat on the small surface of the chip, I would be very surprised it that happened any soon. If they taking aim on that the HCC beta will be a never ending story, I'm glad IF the ongoing beta test ever will be finished.
|
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
These aren't Intel CPUs on the GPU we are talking about. The next-generation Nvidia GPU (Maxwell) is going to include an ARM processor, to handle various OpenCL chores that I don't understand, but should be good news for reducing the usage of the system CPU. With the continued advance in semiconductor processing (reducing lithography size), I expect that it will be at least as power efficient as the present-generation GPUs, and probably more so.
----------------------------------------[Edit 1 times, last edit by Jim1348 at Oct 6, 2012 6:49:35 PM] |
||
|
|
|