-
Hi,
i have built CNTK GPU successfully, and i can run the FeedForwardNet.py example.
However when i try to run the same example using GPU (try_set_default_device(cntk.device.gpu(0)))
I get the foll…
-
So many shaders developers want a proper voxelization support, and the basic idea is to load the scene data into a texture3d.
Well there are a lot of different people want different voxelize method…
-
Correct structure of CMP1 tag (terminology and structure from internals of CRX decoder). In terms of CRX decoder this CMP1 is essentially image header for encoded image (exists for each image track). …
-
I want to implement information that automatically reads some addresses from the knx bus at regular intervals.
respect
-
For partyB, there is no security icon relative to the address.
We should display the security icon, which will lower the chances the user input a wrong address with a bad Copy and Paste.
-
The MIST core has timing issues and does not get synthesized correctly. Some implementations do seem to work nicely but others fail from the very beginning.
Constraints and operating conditions need…
-
There is a 12x performance deterioration running our models between RC1 - > RC2.
Looks MPI related given drastic increase in gradient aggregation (Windows/MSMPI/RDMA)
On a positive note, GPU memor…
ghost updated
5 years ago
-
my tf(SDHC) card test on pc have 20MB/s read。
and I use the SDMMC_Text.ino to test，only 0.7MB/s。
I test three 4G tf card，one 2g,read slow as 0.5~0.7MB/S。the 2G card has 1MB write。
how to change…
-
**Is your feature request related to a problem? Please describe.**
Today OIDs in pg_catalog are based on the hash of database ID and table ID.
This makes it hard / impossible to join between vtabl…
-
It was told, that PWM and tone don't work at the same time. It was told also, that tone can only make sound for one pin. But I implemented an example for PWM and tone at the same time and tone with se…