I am facing a strange problem. I wrote a piece of code in Matlab which uses parfor. For testing, I ran the code on a 4 core Linux-machine. Once I was satisfied that it was working fine, I tried to run the same code on a Windows machine which has higher number of cores.
To my dismay, the parfor loop just gets stuck and runs even slower than the serial code. Its so slow, that the Windows-Task-Manager shows 0% usage on all 28 cores while the code is still running. The Matlab session doesn't generate any error but runs at a snails pace.
Similar (but not same) problems were recently reported on the Mathworks website here and here. Unlike the issue mentioned on Mathworks website, my code uses only Matlab commands and libraries.
Both machines (linux as well as Windows) have Matlab-2018a 64 bit installed.
Related
I'm using f3, a Linux version of the Windows tool H2testw, to test on a Mac the actual capacity of some flash memory I bought. Trouble is that the quicker test, done via f3probe, is only available for Linux, so I'm using the standard test, via the GUI F3X, which does a full integrity test with f3write/f3write rather than just a total capacity test. Trouble is that the flash I bought is claimed to be 512GB large, so it's taking forever. What are alternative best bets? Running f3probe in a VirtualBox? Running H2testw through Wine?
I am writing a huffman coding and decoding in python and when I run it, windows task manager goes to 100% and jumps up to 5GB of RAM usage under my python project. I installed Memory_profiler for python, to see what is going on and there was not much going on. Windows jumps to those high values in places where profiler showed me 7MB or 180MB increments. What should I trust? My laptop does get laggy a lot when I get to those values.
No, this isn't another rant question about NVidia vs AMD; I'm genuinely interested in having my demo running well with both vendors. I've tested my code with four configurations:
MacBook Pro (NVidia GT650M) - fine
Desktop with CentOS 6.5 (Nvidia Quadro FX) - fine
Desktop with Windows 7 64 bit (AMD HD7950 with Catalyst 14.4) - slow
Desktop with Fedora 19 (AMD HD7950 with catalyst 14.4) - slow
3 and 4 are actually the same machine. The code is not highly optimized but it's not doing anything too complex either: I have a grid (which I render using GL_POINTS), a line that represents the path found by A* and a moving agent. The grid has about 10k elements, if I remove that the demo runs better, but still not perfectly.
I guess it's a driver issue, as on 3 and 4 it seems it's running with software rendering; I profiled the code on Windows with CodeXL and a frame take ~400ms and seems to be using mostly the CPU rather than the GPU.
As final information, I'm using GLEW and GLFW for cross-platform development. The full code is available here: https://bitbucket.org/theWatchmen/behaviour-trees
Let me know if you need any further information.
Original answer posted here: http://www.gamedev.net/topic/659012-issue-with-opengl-demo-using-catalyst-drivers-linux-and-windows/#entry5168395
It seems that for this particular card GL_POINTS are emulated in software and that's causing the demo to slow down. I will change the grid to triangles to make sure it runs smoothly on all cards.
Well, there is a strange problem occured in my working project. It is written on Delphi. When I try to compile it, it takes 8 hours to compile about 770 000 lines (and it is not the end), while my colleague needs only 15-20 seconds.
I've tried everything suggested in Why does Delphi's compilation speed degrade the longer it's open, and what can I do about it?
Shorten the path to project
Defragment disc with MyDefrag
Use Clear Unit Cache (do not sure, if it worked at all)
I also turned off the optimization and I use debug mode. My PC is pretty fast (i5-2310 3.1 GHz, 16 Gb RAM, usual SATA HDDs), the bottle neck could be the HDD, but my collegue has usual one too. So, it is very mysterious, what is the reason of so slow compilation.
Edit: I apologize for lack of information. Here is additional info:
I use debug mode, release one works same.
We use Delphi XE version.
I've copied my collegue's folder with project initially.
I do not use network drive, and I tried to move project to another HDD.
Additional info about system: I use Windows 7 Enterprise N 64 bit, while my collegue uses Windows 7 32 bit, Also, Delphi XE is 32-bit (dunno, if it can be 64-bit). May be it is the reason in some way?
Edit 2: I found solution! The problem was that I installed Delphi on my Windows 64 bit system. Installing it on virtual Windows 7 x86 made it work: compiling in seconds. Dunno, why is there so big gap in perfomance.
Are you sure this is not some hardware problem, e.g. your hard disk having a bad sector? Try to put the source code on a different disk and see if the problem goes away. Or maybe the search path points to a network drive that is very slow or not even available?
Background: On machines >2.1 Ghz, Win 3.11 cannot load protected-mode drivers for TCP/IP ("Wolverine") because of a timing problem in "IOS.386". There is a similar problem with Win95 and 98 1st release. There was a patch release from MS for 95/8, but never for WFW.
I found an old post here by Rob Cowell about this problem. He said that he used a different stack. I was curious about which one. I have made it work with the 16-bit DOS mode stack from MS, but it is a real pig for memory <640K. (Also, no native DHCP, but NBT does work, so you can map network drives. You also need to get rid of IPX & NETBEUI support for similar reasons.)
And just to prevent questions of the type "why would anyone want to ....", I'm trying to put together a WFW Live CD for modern hardware, just for fun. I did cobble together DHCP support using WATTCP modules and some complicated scripting before windows is started, but I need to be very circumspect about running any DOS programs or it runs out of low memory. (WFW actually sees 256 Mb of high memory; I was quite surprised.)
I voted to close, because it's not programming related. But it is an interesting project to see if you can pull it off. I'd be keen to see a link or two to see how you go.