Message boards :
SETI@home Enhanced :
Retiring my old ATI HD 4850.
Message board moderation
Author | Message |
---|---|
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
Today I bought a new AMD/ATI Radeon HD R7870 2Gb Core Edition. When the weekend comes, I will retire my old HD 4850, and install this new toy in my Q8200 PC. I do believe that the HD 7870 will be at least a little bit faster than my old HD 4850 :-) |
![]() Send message Joined: 28 Jan 11 Posts: 619 Credit: 2,580,051 RAC: 0 ![]() |
Today I bought a new AMD/ATI Radeon HD R7870 2Gb Core Edition. When the weekend comes, I will retire my old HD 4850, and install this new toy in my Q8200 PC. Good choice!!! I look forward to see the run times for that card. |
![]() ![]() Send message Joined: 16 Jun 05 Posts: 2530 Credit: 1,074,556 RAC: 0 ![]() |
Very nice. I`m sure you will like it. With each crime and every kindness we birth our future. |
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
OK, first SETI@home v7 v6.99 (ati_opencl_sah) finished and validated with my new HD7870. resultid=12744254 I did have one problem though when trying to cram the card into my computer. I had to, (with some minor force) adjust the metal cover of the card (bend it out a little at one point) because there were some (unusually high) capacitors on the motherboard that was hitting the metal cover when I tried to put it in the PCI-E slot. I guess I can kiss goodbye to the warranty, but what the heck, no pain, no gain :-) I also had to reinstall the drivers (still on 12.4) though, because the card had some issues at first. After the reinstallation of the drivers, it behaves fine. Of course I had made sure that Boinc did not start up before I was certain the card behaved. And I had set Boinc to not allow any network connection, before I was sure everything did run fine. Also of course, I had made a full backup of the whole Boinc folder. I also noticed that the .bin files were recompiled automatically when the first v7 v6.99 (ati_opencl_sah) was started after card change. One example: now called: clFFTplan_Pitcairn_1024_r1643.bin, previously called clFFTplan_ATIRV770_1024_r1643.bin. Only strange stuff now, is that GPU-Z (the latest one yes) does not have any sensor field for GPU load any longer, as well as a few other sensor fields that showed up with the HD4850. Catalyst Control Center though, shows the GPU load. I guess it's some strange stuff that will be fixed in due time. Also Boinc at startup only shows the GPU as "ATI GPU 0: ATI unknown (CAL version 1.4.1720, 2048MB, 3200 GFLOPS peak)" Boinc's "Your computers" web page also shows the GPU as "AMD ATI unknown (2048MB) driver: 1.4.1720". Boinc doesn't seem to recognize my HD7870. Ah well, it's fast though, very fast compared to the old HD4850. I'm happy so far. |
Send message Joined: 29 May 06 Posts: 1037 Credit: 8,440,339 RAC: 0 ![]() |
Only strange stuff now, is that GPU-Z (the latest one yes) does not have any sensor field for GPU load any longer, as well as a few other sensor fields that showed up with the HD4850. Catalyst Control Center though, shows the GPU load. I guess it's some strange stuff that will be fixed in due time. Also Boinc at startup only shows the GPU as "ATI GPU 0: ATI unknown (CAL version 1.4.1720, 2048MB, 3200 GFLOPS peak)" Yep, you need Boinc 7.0.29 or later to detect pitcarn GPUs, AMD doesn't have an api to detect GPU versions like Nvidia has, so it needs coding in Boinc: client: Updates provided by Teamwork of Planet3Dnow.de to coproc_detect.cpp added CAL_TARGET_ID 21 as : AMD Radeon HD 78x0 series (Pitcairn) (from [P3D] Crashtest). Claggy |
![]() ![]() Send message Joined: 16 Jun 05 Posts: 2530 Credit: 1,074,556 RAC: 0 ![]() |
Looks nice Sten. With each crime and every kindness we birth our future. |
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
Looks nice Sten. And it runs like crazy, compared to the HD4850. It's been chewing through some MB 6.10 (ati13ati) on main now for some hours. In an hour or so, I will let it do some AP 6.04 (ati_opencl_100) here on Beta, and then some v6.99(ati_opencl_sah) Unfortunately, it does also get v6 v6.05 (cal_ati) CAL Brook+ AP's, but even though they will take forever, I think I will crunch them anyhow, in case Raistmer needs the data. |
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
This is one crazy card :-) First AstroPulse v6 v6.04 (ati_opencl_100) crunched. 0% blanked and finished in about 37 minutes. My old HD4850 would take about 3 hours and 45 minutes for a similar AP WU. resultid=12762017 |
![]() Send message Joined: 28 Jan 11 Posts: 619 Credit: 2,580,051 RAC: 0 ![]() |
OK, first SETI@home v7 v6.99 (ati_opencl_sah) finished and validated with my new HD7870. resultid=12744254 I use a program called SIV that is very good. I use Gpu-z as well. http://rh-software.com/ |
![]() Send message Joined: 28 Jan 11 Posts: 619 Credit: 2,580,051 RAC: 0 ![]() |
This is one crazy card :-) And you couldn't seew hat GPU load is had? I guess below 75% at 37minutes, maybe lower. |
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
Thanks for that idea. I downloaded SIV, and it is indeed very good. It shows the GPU load of the HD7870. So I guess they have to update GPU-Z to fully support the HD7870. |
![]() ![]() Send message Joined: 16 Jun 05 Posts: 2530 Credit: 1,074,556 RAC: 0 ![]() |
Sten, on MB set -sbs 256 in MB_cmdline.txt. It will speed up the app. With each crime and every kindness we birth our future. |
![]() Send message Joined: 28 Jan 11 Posts: 619 Credit: 2,580,051 RAC: 0 ![]() |
Sten, on MB set -sbs 256 in MB_cmdline.txt. One can wonder why app isn't self calibrating.... Or is that not possible? |
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
Sten, on MB set -sbs 256 in MB_cmdline.txt. Done, will see what difference it makes. |
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
-sbs 256 was certainly not faster on the AR's I run now (0.43). On the contrary much slower. The completion times went from around 22 minutes, and up to over 25 minutes. I'll revert to the default. -sbs 256 may be faster than default on other AR's, but for sure not on 0.43 AR WU's. At least not for me. |
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
Ok, some results. No upping of -sbs made any postitive difference whatsoever. Tried 128, 256, and 512. Adding -hp when also having added -sbs values made no positive difference. The thing though, that took the crunching times down was adding the following, and only having this in the MB_cmdline.txt file: -period_iterations_num 10 -hp Case closed for now. |
![]() ![]() Send message Joined: 16 Jun 05 Posts: 2530 Credit: 1,074,556 RAC: 0 ![]() |
Its curious, on a 5850 it gives 25% speedup and on a 7850 and 7970 its 30% speed up running with -sbs 256. With each crime and every kindness we birth our future. |
![]() Send message Joined: 10 Mar 12 Posts: 1700 Credit: 13,216,373 RAC: 0 ![]() |
Its curious, on a 5850 it gives 25% speedup and on a 7850 and 7970 its 30% speed up running with -sbs 256. Well, I could if I wanted probably fine tune the card to produce 30% more than now, but I see no reason to do that. Even at this level and almost silent (max 60 C temp), it's extremely faster than my old HD4850 when it was working like h***, and having a higher temp. I'm satisfied as it is, very satisified really. This card is hardly breaking a sweat, producing much more than the overworked hd4850. At this level, it will last forever :-) |
©2023 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.