-
### Usage
Does Prefix Caching currently support offloading to the CPU?
If not, is there a plan to support it? Thanks~
-
Platforms: linux, rocm, slow
This test was disabled because it is failing in CI. See [recent examples](https://hud.pytorch.org/flakytest?name=test_remote_cache_load_function_device_cpu_bfloat16_dyn…
-
Platforms: asan, linux, mac, macos
This test was disabled because it is failing in CI. See [recent examples](https://hud.pytorch.org/flakytest?name=test_cache_load_with_guards_static_bounds_device_cp…
-
Platforms: asan, linux, mac, macos, rocm, slow
This test was disabled because it is failing in CI. See [recent examples](https://hud.pytorch.org/flakytest?name=test_cache_load_with_guards_static_boun…
-
Platforms: linux, rocm, slow
This test was disabled because it is failing in CI. See [recent examples](https://hud.pytorch.org/flakytest?name=test_remote_cache_load_function_device_cpu_float32_dynami…
-
**Do you want to request a *feature* or report a *bug*?**
Bug report
**What is the current behavior?**
When I run yarn cache clean, it causes a high cpu node process and does not exit even afte…
-
FBuild fails to generate a valid JSON report (`-report=json`) if no `-cache` variant is used.
I.e.: The following execution:
```
"Overview": {
"cmd line options": " -report=json -summary -clean…
-
SRAM 之所以被称为“静态”存储器,是因为只要处在通电状态,里面的数据就可以保持存在。而一旦断电,里面的数据就会丢失了。在 SRAM 里面,一个比特的数据,需要 6~8 个晶体管。所以 SRAM 的存储密度不高。同样的物理空间下,能够存储的数据有限。不过,因为 SRAM 的电路简单,所以访问速度非常快。
![image](https://user-images.githubusercont…
-
Hi,
Is it possible to increase the cache of each cpu to values larger than 512 KB? Our application doing allocations of 32, 64 size alot and the maximum tcmalloc cache depth for these classes is 204…
-
Hello! In NPKit, we have a thread `NpKit::CpuTimestampUpdateThread()` looping to update the cpu timetamp and the updated value is given to a pointer `cpu_timestamp_`. To synchronize between CPU and GP…