Test memory GF100 and many other NVIDIA videocards for DOS http://rapidshare.com/files/339752167/TEST.rar MD5: 034DD437A733463B855393E3E896AB94 mats.exe 110.460 bytes 01.01.2008 00:00 REV: 4.173.1 NV30 NV31 NV34 NV35 NV36 NV40 NV41 NV42 NV43 NV44 NV4C NV4E NV67 NV73 G70 G71 G72 G73 G80 G84 G86 G92 G94 G96 G98 GT200 GT206 MCP77 iGT206 MCP79 iGT209 GT212 GT214 GT216 GT218 GF100 Not work for GT215 Code: MATS version 4.173.1. MATS is short for Modified Algorithmic Test Sequence Based on an algorithm described in 'March tests for word-oriented memories' by A.J. van de Goor, Delft University, the Netherlands MATS Options: -e xx Forces endpoint of test to megabyte xx -c xx Only test xx percent of the framebuffer. -b xx Forces start of test to megabyte xx -t xx Use an alternate memory test algorithm (xx=1 to 7) 0 = Box Mats test (default) 4 = March test 1 = Linear Mats test 5 = 64-bit pattern test 2 = Address-On-Data 6 = MOD3 test 3 = March test 7 = Knaizuk-Hartmann -o Suppress file output -m xx Set the maximum number of errors to log -n xx Run the algorithm on the card with index xx -v Disable save & restore of VGA memory space -ver Display version information -x Don't use the box algorithm
Example Code: MATS version 4.173.1. Testing G92 with 384 MB of memory. Errors found. Use .Status() or .PrintErrors() for more info. This message will only appear once. LANE EB0 READ ERR EB0 WRT ERR UNKNOWN ERR ---------- ------------ ------------ ------------ INTERNAL MEMERROR PROBLEM PART=0 LANE=1 FBB[31: 0] 0 0 0 FBB[63:32] 0 0 0 FBC[31: 0] 0 0 0 FBC[63:32] 0 0 0 FBD[31: 0] 0 0 0 FBD[63:32] 0 0 0 External Bank 0 Failing bits: A41 Read Error Count: 0 Write Error Count: 2 Unknown Error Count: 0 BIT FRONT WRITE FRONT READ UNKNOWN ---- ----------- ----------- ----------- A41 2 0 0 ADDRESS EXPECTED ACTUAL REREAD1 REREAD2 FAILBITS TYPE P L E I ROW COL BO -------- -------- -------- -------- -------- -------- ----- - - - - ---- --- -- 10dd5d30 aaaaaaaa aaaaa8aa aaaaa8aa aaaaa8aa 00000200 WRITE A 1 0 5 0b3e 1e4 0 10dd5d30 0f0f0f0f 0f0f0d0f 0f0f0d0f 0f0f0d0f 00000200 WRITE A 1 0 5 0b3e 1e4 0
i tried memtestG80 and later i found out OCCT uses the same technique..Although memtestG80 is more sophisticated (can use random amount of vram and random number of repeats vs fixed 10 repeats in OCCT) its based on foding@home http://folding.stanford.edu/English/DownloadUtils i was concerned its my gpu mem that's screwing in dirt2,3 dx11 but its not, its something in stupid engine lol:nerd:
Mother of all necros? I seriously doubt the OP still has this file, but perhaps it is floating around on the internet somewhere.
http://vlab.su/viewtopic.php?f=48&t=96805 http://vlab.su/viewtopic.php?f=48&t=99855 http://vlab.su/viewtopic.php?f=48&t=124090 And here you have some slightly more legit guide/explanation.
