Questions and Answers :
GPU applications :
ATI GPU loops at 21.611 % for almost all units
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 . . . 9 · Next
Author | Message |
---|---|
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
It shouldn't be a problem with adding a 3rd or even a 4th as long as your power supply can handle it and you have enough power connections. I run 4 GPUS in 2 of my computers, all but 1 GPU in each as an addition 6 pin connector to the power supply to supplement the Motherboard supply. Heck, you could even use the PCI if you had an adaptor and enough reserve on your PSU. At one time I ran 2 gpus off the motherboard and 2 more off the PCI via risers. I'm not saying you should but you could. So in short, as long as your PSU can supply enough power, you can add that 3rd GPU. Do you know what you PSU is rated to supply and if it's a bronze, gold, or platinum. The actual amount will vary depending on what it is. More on that later if need be. As far as PCIe, as I started above you could go all the way back the just a plain PCI and Seti will still be able to run. Almost all GPUs now are backward compatible to PCIe's. So no worries about Version 2. The amount of data going back and forth thru that connection isn't so big that it requires a X4, X8, or x16 so you are good with using any connection. Happy Crunching... Zalster |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
The power supply is 750watts. I don't know about gold, etc. But the manual says is has a min. efficiency of 80% at max load up to 85%. It a Rosewill RX750. I know that I have a shitload of power plugs that I haven't used at all 4, 6 & 8 pin. I originally bought the mobo and psu with the idea of having 2 or 3 video cards running boinc. Can I put in a Nividia card with my ati's? Have any suggestions? The mobo is AsRock x58 extreme. I only have 3GB of fast memory however. Thanks again. This is a lot of fun. |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
Ok, I'm not sure how much power each of your ati requires. But I'd say you are probably near the max for that PSU. 80% at 85% max load, sounds like a bronze unit. Means that while it may rated at a 750 Watts, it's not going to hit that consistently. It will hit 80% consistently, so that means 600 Watts consistently. This is what I use to figure out how much of a PSU I need. http://www.extreme.outervision.com/psucalculatorlite.jsp Can you put a Nvidia with ATIs....yes, but it's not easy. I've heard of people doing it but it's not easy at all. Better to just build a cheap computer and put the nvidia in it. Yes, this can quickly become addictive. Happy Crunching.. Zalster |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
http://www.amd.com/en-us/products/graphics/desktop/6000/6570# (Support tab) 400 Watt or greater power supply recommended (500 Watt for AMD CrossFireXâ„¢ technology in dual mode) Even if the two aren't in Crossfire mode, I don't see why the wattage needs to be more. |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
I used your link to the calculator and it told me I needed 388watts min. and 438 recommended. I have an older cpu LGA1366 socket 2.67 GHz, so I think I could add another 6570 if I can figure out about how these riser cards might fit into my mobo. What I told you about having only a x4 is not precisely correct. What it is is a x16 connection rated at a x4 speed, if I am saying that correctly. If that is correct I guess it means I could just buy another 6570 x16 and a throughput of x4 would be OK. Does it sound as though my thinking is clear? |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
Correction to last comments: The Manual says my "x4" is only for cross fire, not regular graphics cards but it is strange in that it is the same size as my x16 slots. At the same time it is calling it sometimes an x4 and sometimes an x1???? |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
If that other card can sit on your motherboard in that PCIe slot and you have the free pin connectors for it, then yes you can use it. The only time I used risers were was when I didn't have an available PCIe slot. But you have to be careful if you ever decide on one of those. I'm not going to go into detail unless you decide to go that route. For now, if you can sit another GPU in a slot and power it, then you can use it. The throughput for seti is very low so a x4 is more than adequate. Your thinking is correct. Normally you have 1 or 2 slots at x16, but there are other slots sometimes as well, when you connect to those other slots, it lowers the throughput to the original so then they become x8 or even x4 if you are connecting multiple slots. It gets pretty technical to explain. Let me find a link for you http://www.enthusiastpc.net/articles/00003/ Not the original article I wanted but close enough. Happy Crunching.. Zalster |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
Thanks Zalster & Ageless. I think now I see how this all fits. That last link and explanation did it. I think I'm good to go now with another 6570. What would we do without help from those who know more and take the time and have the patience? |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
One of my ati work units failed. I forgot and aborted it rather than sending in the stderr.txt file. Sorry, a bad habit that I have to break. I am going to have to keep my eye on my work units to see if this is a recurring thing or not before I get that 3rd 6570. What is the range of completion times for the v7 work units? Let say the cpus times should range between 1.0 units and 1.8 units and the gpus times should range between 1.0 and 2.5 units. Do you understand the template of mine? |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
I just had another wu fail: below is the stderr.txt file: Do you need my first 30 lines of my log too? Running on device number: 0 Priority of worker thread raised successfully Priority of process adjusted successfully, below normal priority class used OpenCL platform detected: Advanced Micro Devices, Inc. BOINC assigns device 0 Info: BOINC provided device ID used Build features: SETI7 Non-graphics OpenCL OCL_CHIRP3 FFTW AMD specific USE_SSE x86 CPUID: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz Cache: L1=64K L2=256K CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3 SSSE3 SSE4.1 SSE4.2 OpenCL-kernels filename : MultiBeam_Kernels_r1843.cl ar=0.447621 NumCfft=191331 NumGauss= 1055964852 NumPulse= 113599462911 NumTriplet= 15681210875904 Currently allocated 145 MB for GPU buffers In v_BaseLineSmooth: NumDataPoints=1048576, BoxCarLength=8192, NumPointsInChunk=32768 Windows optimized S@H Enhanced application by Alex Kan Version info: SSEx (Intel, Core 2-optimized v8-nographics) V5.13 by Alex Kan SSEx Win32 Build 1843 , Ported by : Raistmer, JDWhale SETI7 update by Raistmer OpenCL version by Raistmer, r1843 Number of OpenCL platforms: 1 OpenCL Platform Name: AMD Accelerated Parallel Processing Number of devices: 2 Max compute units: 6 Max work group size: 256 Max clock frequency: 650Mhz Max memory allocation: 536870912 Cache type: None Cache line size: 0 Cache size: 0 Global memory size: 1073741824 Constant buffer size: 65536 Max number of constant args: 8 Local memory type: Scratchpad Local memory size: 32768 Queue properties: Out-of-Order: No Name: Turks Vendor: Advanced Micro Devices, Inc. Driver version: 1445.5 (VM) Version: OpenCL 1.2 AMD-APP (1445.5) Extensions: cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_atomic_counters_32 cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_d3d10_sharing cl_khr_d3d11_sharing cl_khr_dx9_media_sharing cl_amd_image2d_from_buffer_read_only cl_khr_spir cl_khr_gl_event Max compute units: 6 Max work group size: 256 Max clock frequency: 650Mhz Max memory allocation: 536870912 Cache type: None Cache line size: 0 Cache size: 0 Global memory size: 1073741824 Constant buffer size: 65536 Max number of constant args: 8 Local memory type: Scratchpad Local memory size: 32768 Queue properties: Out-of-Order: No Name: Turks Vendor: Advanced Micro Devices, Inc. Driver version: 1445.5 (VM) Version: OpenCL 1.2 AMD-APP (1445.5) Extensions: cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_atomic_counters_32 cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_d3d10_sharing cl_khr_d3d11_sharing cl_khr_dx9_media_sharing cl_amd_image2d_from_buffer_read_only cl_khr_spir cl_khr_gl_event Work Unit Info: ............... Credit multiplier is : 2.85 WU true angle range is : 0.447621 Used GPU device parameters are: Number of compute units: 6 Single buffer allocation size: 64MB max WG size: 256 period_iterations_num=20 Spike: peak=24.35341, time=87.24, d_freq=1420995159.07, chirp=10.269, fft_len=128k Spike: peak=24.09945, time=6.711, d_freq=1421000101.98, chirp=-12.65, fft_len=128k Spike: peak=24.65407, time=6.711, d_freq=1421000101.97, chirp=-12.651, fft_len=128k Spike: peak=24.0889, time=6.711, d_freq=1421000101.97, chirp=-12.652, fft_len=128k Running on device number: 1 Priority of worker thread raised successfully Priority of process adjusted successfully, below normal priority class used OpenCL platform detected: Advanced Micro Devices, Inc. BOINC assigns device 1 Info: BOINC provided device ID used Build features: SETI7 Non-graphics OpenCL OCL_CHIRP3 FFTW AMD specific USE_SSE x86 CPUID: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz Cache: L1=64K L2=256K CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3 SSSE3 SSE4.1 SSE4.2 OpenCL-kernels filename : MultiBeam_Kernels_r1843.cl ar=0.447621 NumCfft=191331 NumGauss= 1055964852 NumPulse= 113599462911 NumTriplet= 15681210875904 Currently allocated 145 MB for GPU buffers In v_BaseLineSmooth: NumDataPoints=1048576, BoxCarLength=8192, NumPointsInChunk=32768 Restarted at 19.30 percent. Used GPU device parameters are: Number of compute units: 6 Single buffer allocation size: 64MB max WG size: 256 period_iterations_num=20 Running on device number: 0 Priority of worker thread raised successfully Priority of process adjusted successfully, below normal priority class used OpenCL platform detected: Advanced Micro Devices, Inc. BOINC assigns device 0 Info: BOINC provided device ID used Build features: SETI7 Non-graphics OpenCL OCL_CHIRP3 FFTW AMD specific USE_SSE x86 CPUID: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz Cache: L1=64K L2=256K CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3 SSSE3 SSE4.1 SSE4.2 OpenCL-kernels filename : MultiBeam_Kernels_r1843.cl ar=0.447621 NumCfft=191331 NumGauss= 1055964852 NumPulse= 113599462911 NumTriplet= 15681210875904 Currently allocated 145 MB for GPU buffers In v_BaseLineSmooth: NumDataPoints=1048576, BoxCarLength=8192, NumPointsInChunk=32768 Restarted at 34.29 percent. Used GPU device parameters are: Number of compute units: 6 Single buffer allocation size: 64MB max WG size: 256 period_iterations_num=20 |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
Do you need my first 30 lines of my log too? No, I'm asking for the partial stderr.txt in the hopes that someone who can read those things can glance at them and see where things go wrong. But that's probably something that isn't in the log. Have you ever tried to continue to run such a task? Have you tried to fully exit BOINC & restart it? |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
The answers are No and No. The next time it happens I will try and restart boinc. I reduced my cpu threads from 7 to 6 in the hopes that might help. It was hung bad around 43% at 2.24 hours which is almost 2x normal and the gpu temp was too low for a running wu so I don't think waiting with it hung would do any good. Do you? Thanks |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
Next time check the CPU temperature. Some of these tasks fall back to running on the CPU when the GPU can't handle them. |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
I restarted boinc by restarting my computer and it worked.The wu for the gpu restarted. Is there a better way to restart boinc that is safe. What about the completion times on the gpu units. Do they vary greatly? Thanks. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
Is there a better way to restart boinc that is safe. See the BOINC FAQs: How to exit BOINC completely? Then after that you just restart BOINC by going Start->(All) Programs->BOINC->BOINC Manager. What about the completion times on the gpu units. Do they vary greatly? Thanks. The length of run time depends on the angle at which the tasks were recorded against the night sky. The lower the angle range, the slower the task will run. The higher the angle range, the faster the task will process. |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
I had ignored the message to use boinc 7.0.64, duh! I now am running with it installed and reinstalled lunatics. So far I look like that may have been my trouble. I hope. |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
I am again having problems running a second gpu (ati 6570), sometimes it works and sometimes not. Since no one has commented on my last stderr.txt, I suppose there isn't any sense to keep sending those in. |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
 Try to set TdrDelay (to more than 2 seconds which is the default on Windows 8) http://setiathome.berkeley.edu/forum_thread.php?id=72035&postid=1383819#1383819 http://setiathome.berkeley.edu/forum_thread.php?id=72035&postid=1384758#1384758 http://setiathome.berkeley.edu/forum_thread.php?id=74512&postid=1505950#1505950 http://setiathome.berkeley.edu/forum_thread.php?id=74843&postid=1523444#1523444   - ALF - "Find out what you don't do well ..... then don't do it!" :)  |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
Error - ignore post |
merle van osdol Send message Joined: 23 Oct 02 Posts: 809 Credit: 1,980,117 RAC: 0 |
This is what I posted using TrdDelay.reg: Windows Registry Editor Version 5.00 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers] "TdrDelay"=dword:00000005 When I look back to check the registry it said: TrdDelay =dword 00000005 [5] {with the second five in square brac's} I assume that would be correct? Any comment? Thanks for your help!!!! |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.