is there a windows equivalent of the linux command "perf stat"? For example to see frontend stalls, cache misses and other performance counter data?
perf is Linux-only profiler capable to access hardware event counters (cache miss, cpu stalls, etc). This profiler supports many CPUs, but can't be used in MS Windows.
For Windows you may try profilers from your CPU vendor:
VTune from/for Intel ($$$)
CodeAnalyst/CodeXL from/for AMD (free)
Intel PCM from/for Intel (free) - https://software.intel.com/en-us/articles/intel-performance-counter-monitor (needs some compilation with DDK kit to get msr.dll driver and pcm.exe tool)
Visual Studio Profiler from Microsoft (part of Studio, some counters for several CPUs) - https://msdn.microsoft.com/en-us/library/bb385772.aspx "CPU Counters", "Portable Events", VSPerfCmd/VSPerfMon command line tool (it may be wrong with some events on recent CPUs)
Some people who find this question might be actually looking for xperf, or the new ETW stuff which is replacing it which could be downloaded here.
Related
I have a Laptop with Intel(R) HD Graphics 520 GPU in it. I added OpenCL developer package to Cygwin. I have found a small Mandelbrot-set calculator program for OpenCL in C on GitHub. It is for Apple, so I modified the Makefile to use the proper headers and settings for gcc. Now the code compiles and executes nicely (bmp file created):
$ ./mandelbrot.exe
Device 0: GenuineIntel pthread-Intel(R) Core(TM) i5-6300U CPU # 2.40GHz
I have two questions:
#1. How can I add (if it is possible) the Intel GPU to the /etc/OpenCL/vendors list? I tried to install from Intel site the Intel CPU runtime for OpenCL Applications for Windows OS and Intel Graphics Technology driver package, but I do not know where can I find the proper OpenCL dll I can point in the intel.idc file.
#2. In /etc/OpenCL/vendors I have found a pocl.icd file pointing to cygpocl-2.dll. I assume this is the pthread library. But it seems to me it is running only a single thread, although I have 4 cores. Should I do any modification to run it in multiple threads? I debugged the code and it seems that as there are only one device found, so it runs only on one thread. In the initialization function it sets the device_work_size property for processing a stirp per device of the final bmp. But as there is only one device, the whole bmp is processed by one run (one clEnqueueNDRangeKernel and one clEnqueueReadBuffer is called).
UPDATE
I have installed Intel(R) Graphics – Windows* DCH Drivers. It installed graphics drivers. I have found the intelocl64.dll (as /cygdrive/c/Program Files (x86)/Common Files/Intel/Shared Libraries/intel64/intelocl64.dll). I put this whole path into /etc/OpenCL/vendors/intel.icd file. So far, so good. Now it cannot even find pthread device... Bah...
I don't think it is possible to use the GPU from cygwin. I would recommend to either build a native Windows binary (e.g. with Visual Studio or Intel DPC++) or use WSL. See https://github.com/intel/compute-runtime/blob/master/WSL.md for requirements.
I am kinda new to Android Studio & stuff. So today, I was installing the Android Studio with the SDK Manager. All was going smooth until an error came up which says:
Unable to install Intel HAXM
Your CPU does not support required features (VT-x or SVM).
Unfortunately, your computer does not support hardware accelerated virtualization.
Here are some of your options:
Use a physical device for testing
Develop on a Windows/OSX computer with an Intel processor that
supports VT-x and NX
Develop on a Linux computer that supports VT-x or SVM
Use an Android Virtual Device based on an ARM system image (This
is 10x slower than hardware accelerated virtualization)
I've attached a pic of my system specs. Can someone please throw some light on this issue?
Thanks
It is because you had not intialize virtual technology in your device.You Need to go in BOOT Option before starting WINDOWS OS and enable VT-x from there>
The option of enabling Virtual technology is putted in different option depends on device manufacturer
Edit: Android Studio emulator won't run on Windows with an AMD processor. The error message is kind of misleading, as it suggests the problem is with your CPU. But it is within the troubleshoot message: "Windows/OSX computer with an Intel processor". Basicallly, that means it is not going to work properly in your current setup. Either try installing Linux and running Android Studio on that (which might come with its own issues), using a physical device for testing or use the slow ARM images.
You are using an AMD processor. SVM is AMD technology and VT-x is Intel technology. So you won't be able to get VT-x to run, but SVM might be possible.
As another poster had suggested, virtualization may have been disabled in the BIOS. There may be an option to enable virtualization. It does however seem to happen that virtualization is activated in the BIOS and Android-Studio does not recognize that. I have not figured out how to fix that either.
You could use the emulator with an ARM image, which will be very slow. Alternatively, you could use another emulator that is not integrated into Android-Studio.
I have seen this question many times but never found an answer for Windows.
I recently ported my CUDA code to OpenCL.
When testing with an ATI card, the Catalyst drivers contain a CPU OpenCL driver, hence I can run the OpenCL code on the CPU.
When testing with an NVIDIA card, there is no driver for the CPU.
Question is: how can I install (and deploy) a CPU driver when running with an Nvidia card?
Thanks a lot
To use OpenCL on CPU you don't need any driver, you only need OpenCL runtime that supports CPU, which (in case of AMD/ATI) is part of APP SDK. It could be installed no matter what GPU you have. Your end-users would also have to install the APP SDK: currently, there is no way to install OpenCL runtime only.
If you have Intel CPU, you better try Intel OpenCL SDK, which has separate installer. However, AMD APP SDK works on Intel CPUs quite well, but note vice versa.
In our application we have many versions of the same routine optimized for different kind of processor architectures. During install we run performance tests and select the best version of routine.
Latest processors can boost their frequencies if few cores are in use, so sometimes our tests peeking wrong version of routine. Is there some way to temporarily turn off Turbo Boost?
I don't know the opcode for enabling/disabling turbo boost directly, but apparently it's possible because TMonitor - which is a Windows application - can do it at runtime.
There are MSR registers (https://en.wikipedia.org/wiki/Model-specific_register) in Intel CPUs to disable/enable per-core turbo boost. In linux, there are msr-tools package which allow root to read/write to MSR registers. There are two posts about Turbo Boost disabling/enabling in Linux:
https://askubuntu.com/questions/619875/disabling-intel-turbo-boost-in-ubuntu "Disabling Intel Turbo Boost in ubuntu", answer by Maythux from 2015
http://luisjdominguezp.tumblr.com/post/19610447111/disabling-turbo-boost-in-linux
Disabling Turbo Boost in Linux, Luis J. Dominguez P, 2012-2013; Sandy Bridge - "38th bit of the 0x1a0 register."
http://notepad2.blogspot.com/2014/11/a-script-to-turn-off-intel-cpu-turbo.html
And there is kext for OSX to do the same: Disable Turbo Boost on Core i7 Mac?
In windows OS you can add the following reg key.
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Power\PowerSettings\54533251-82be-4824-96c1-47b60b740d00\be337238-0d82-4146-a960-4f3749d470c7]
"Attributes"=dword:00000002
Then in advanced power settings you will get new option that allows you turning off turbo boost.
i need to run a few visual studios on windows XP and it seems to take up a lot of memory. i am also running resharper which is a memory hog.
i am running 32 bit XP. How much memory can i put into my machine until i get to the point where the OS hits its limit.
Also, any other ways of running multiple visual studio without such slow performance.
32-bit Operating Systems are limited to 4 GB of RAM, which may or may not be enough for you. Also, I think Windows shows 3 GB of RAM if you install 4 GB.
I suggest you switch to 64-bit and upgrade to 8 GB if you can.
UPDATE: See Jeff's blog post on the subject: Dude, Where's My 4 Gigabytes of RAM?
The maximum amount of memory that can be seen by 32bit WinXP is somewhere between 3 and 4 gigabytes depending on your chipset.
I have also run into issues running multiple instances of VS when I had resharper installed. The only thing you can do is run 64bit XP with more memory, or not use resharper (which is a bummer).
32-bit Windows kernel divides the 4GB virtual addressing space in 2GB/2GB partitions. If you feed the /3GB switch to NTLDR it will offer 1GB kernel space / 3GB user mode space. Note that this NOT implies that you can't write software to take advantage of machines with 32-bit CPUs and address more than 4GB at once.
A workaround is the hardware-supported feature to access the remaining memory in banks or "windows" since the CPU still sees a maximum of 4GB addressable space at once. Some database and GIS software offer this possibility. This is called Physical Address Extensions and allows to use (not addressing at once) up to 64GB with 36-bit addresses. WinXP offers AWE, an API built on top of PAE.
That's the theory. For using Visual Studio you can get the full 4GB for your system or upgrade to a 64-bit OS with more RAM. This only if VS offers a 64-bit version.
"Also, any other ways of running multiple visual studio without such slow performance."
+1 trick: you should use a RAM disk (download) to accelerate I/O.
If you're using - and hopefully do - source-managament system (ie. Subversion), you must just checkout your projects there. VS.NET makes tons of I/O calls, and RAM disks are much faster than real disks.
CAUTION! If you turn off your computer, RAM Disk disappers.