certainly seems like windows/linux cross validation issues.
all of my invalids from yesterday lost out to a pair of Windows hosts.
all of my valids from yesterday validated against another Linux host.
I am seeing the same thing. It seems ~98% of our valid results are against other Windows systems (both 10 and 11) but basically all of our invalids were invalid against both Linux Mint 20.3 and Ubuntu 20.04.4 LTS among others. Also got about 20 errors while downloading.
Still waiting for a bunch of work units to be validated/invalidated.
Good deal on the .13 app. I saw the same validation pattern on my host.
CUDA-OpenCL looks to be working between Linux hosts on the v0.12 app, and I have a number of validations against opencl-ati, cuda55, and even Petri's special app on the anonymous platform.
Bernd- out of curiosity, I know these are all MEERKAT work units that are currently being sent out. I was reading your original post in this thread and I know that BRP7 is also going to include FAST data. I will admit I am ignorant on the type of data that MEERKAT versus FAST consists of. Is the FAST data going to be so similar to the MEERKAT work units that we will start seeing those mixed in with what we are receiving or is the data so different that it will require additional beta testing because they are completely different? Just curious!
Linux/Windows validation looks better with the new Linux-OpenCL 0.13 app version. I'm trying to build a Linux-CUDA app the same way. May take a bit.
My valid/invalid ratio is rapidly improving with v0.13. I just took a quick look through ~150 v0.13 tasks, and there were zero v0.13 invalids and many tasks which validated against Windows hosts.
FAST data was and still is in the planning and discussion, however I haven't seen a bit of real data yet, not even simulated. I don't know whether we'll get it at all and when, nor what it will look like. We needed to develop and change our pre-processing and application code quite a bit for the MeerKAT data, I really don't know whether we could process FAST data with the same pipeline.
We switched to 0.13 OpenCL for NVidia on Linux. The App is slightly slower than the CUDA version, but validation with Windows CUDA (which is the dominant version) should be better.
I'll be away from my desk for a few weeks and make another attempt with a more modern CUDA version when I'm back.
Apple dropped support for OpenCL and has issued the last box with NVidia GPU what - ten years ago? Unless some enthusiastic volunteer shows up that would want to port the App to Metal and Apple Silicon GPU I don't see that happen at all. As far as E@H is concerned, GPU computing on Apple is dead. We simply don't have the manpower to dive into Apples own universe.
Ian&Steve C.
)
I am seeing the same thing. It seems ~98% of our valid results are against other Windows systems (both 10 and 11) but basically all of our invalids were invalid against both Linux Mint 20.3 and Ubuntu 20.04.4 LTS among others. Also got about 20 errors while downloading.
Still waiting for a bunch of work units to be validated/invalidated.
Linux/Windows validation
)
Linux/Windows validation looks better with the new Linux-OpenCL 0.13 app version. I'm trying to build a Linux-CUDA app the same way. May take a bit.
BM
Good deal on the .13 app. I
)
Good deal on the .13 app. I saw the same validation pattern on my host.
CUDA-OpenCL looks to be working between Linux hosts on the v0.12 app, and I have a number of validations against opencl-ati, cuda55, and even Petri's special app on the anonymous platform.
Bernd- out of curiosity, I
)
Bernd- out of curiosity, I know these are all MEERKAT work units that are currently being sent out. I was reading your original post in this thread and I know that BRP7 is also going to include FAST data. I will admit I am ignorant on the type of data that MEERKAT versus FAST consists of. Is the FAST data going to be so similar to the MEERKAT work units that we will start seeing those mixed in with what we are receiving or is the data so different that it will require additional beta testing because they are completely different? Just curious!
Bernd Machenschalk
)
My valid/invalid ratio is rapidly improving with v0.13. I just took a quick look through ~150 v0.13 tasks, and there were zero v0.13 invalids and many tasks which validated against Windows hosts.
FAST data was and still is in
)
FAST data was and still is in the planning and discussion, however I haven't seen a bit of real data yet, not even simulated. I don't know whether we'll get it at all and when, nor what it will look like. We needed to develop and change our pre-processing and application code quite a bit for the MeerKAT data, I really don't know whether we could process FAST data with the same pipeline.
BM
We switched to 0.13 OpenCL
)
We switched to 0.13 OpenCL for NVidia on Linux. The App is slightly slower than the CUDA version, but validation with Windows CUDA (which is the dominant version) should be better.
I'll be away from my desk for a few weeks and make another attempt with a more modern CUDA version when I'm back.
BM
A Mac app might be ready by
)
A Mac app might be ready by Christmas perhaps?
6dj72cn8 wrote: A Mac app
)
Let them get the real computers sorted first, the Mickey Mouse ones can come later.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
6dj72cn8 wrote:A Mac app
)
Apple dropped support for OpenCL and has issued the last box with NVidia GPU what - ten years ago? Unless some enthusiastic volunteer shows up that would want to port the App to Metal and Apple Silicon GPU I don't see that happen at all. As far as E@H is concerned, GPU computing on Apple is dead. We simply don't have the manpower to dive into Apples own universe.
BM