Message boards :
Number crunching :
11.5 & 11.5a Hotfix Released
Message board moderation
Author | Message |
---|---|
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
11.5 & 11.5a Hotfix have been released. AMD went for a new record - 11.5a came out 15 mins after 11.5 (!). Dont yet know if it fixes 11.4 issues, so beware :) For the adventurous and/or 11.4 victims Download them from: 11.5 Catalyst 64 bit (85Mb) 11.5 Catalyst 32 bit (57Mb) 11.5a Hotfix Catalyst- all systems (135Mb) (Dont need both - 11.5a Hotfix is a complete CC & Driver set. Release Notes 11.5 Release Notes 11.5a (on driver download page) Regards Zy |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
11.5a works fine on a 5850 with a 0.62 WU Regards Zy |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
11.5a works on 2x5970s with 0.62, no bother, installed fine, validated fine ...... yey :) PG PPS Sieve also going through fine ..... double yey :) Not extensive tests, but 12 x MW WU put through and 8 x PPS Sieve put through, all ok ..... so 11.5a should be fine. Regards Zy |
Send message Joined: 1 Sep 08 Posts: 204 Credit: 219,354,537 RAC: 0 |
What about running 2 WUs concurrently? MrS Scanning for our furry friends since Jan 2002 |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
Not tried it yet - I'm crunching for our Team on the Pentathelon - PPS Sieve - dont want to take too much time away. I'll try it when the current PG phase is over. My guess is it should be fine (que murphy ....), it seemed solid enough on the few I did, albeit a limited run. Regards Zy |
Send message Joined: 1 Sep 08 Posts: 204 Credit: 219,354,537 RAC: 0 |
Oh well, taking one for the team ;) System: Win 7 64, HD6950@6970 1 WU: seems to work, as did Cat 11.4 for me. 2 WUs concurrently: immediate blue screen. This is new: with 11.4 the system would freeze immediately. MrS Scanning for our furry friends since Jan 2002 |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
Did a run of about 40 on the 2X5970s machine with 2 per GPU, total eight running in the box. Heat was less than with 11.3/11.4 - and worked fine. It was still a tad hot on one set of VRMs though, so had to bring the GPU down as well. Need to fiddle around a bit more, but bottom line is its ok, watch the VRM temps, expect to reduce GPU (I ended running at 775/300 with times around 250 secs for two WUs, 125 each of the 267 credit size. Thats about 10-15 sec saving on 11.3 singles - so worth doing. Carefull with VRMs though ..... The 5850 ran two per ok, temps not a worry at all, stable. At 775/500 was getting 260secs for 267 credit WUs, so 130 secs each on dual WU runs. Solid no problem. 11.5a is looking good for 5XXX at least, and dual WU per GPU definitely on the cards if individuals want to do it. 5970s watch those VRMs though .... Regards Zy |
Send message Joined: 10 Oct 07 Posts: 79 Credit: 69,337,972 RAC: 0 |
hi. what temps are your VRM'S sitting at with the app and 775 meg clock i'm interested because as you know i'm running the stock app but at 800 meg and i'm processing the 267 credit units in 130 seconds only 5 seconds slower than the alternitive app but at the end of each WU as you know theres a little gap which is giving my cards a little break long enough to take the edge of the temps on the VRM's i cleared some cables out the road of the twin 4 inch fans i have directed at the cards and something strange the temps of the VRM's stayed at exactly between 100 and 103 degrees depending how far they were through the work unit but the gpu temps dropped by 2 degrees accross the board just shows its alot harder to get the temps down on those VRM's than the GPU's which is where the trouble starts with these cards :( i'd be interested to see if the new drivers somehow do take a little bit more off the temps especially the VRM's which are now my only problem as the 4 GPU's sit betwee 84 to 89 degrees peak best regards Ian ....Please Join team Scotland HERE |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
I suspect the additional airflow provided benefited the GPUs more as the latter are covered by passive heat pipes/heat sinks, so any additional flow will be picked up more easily on those. The VRMs/other heatsinks, another nine I think, are small and not covered by the same kind of array of passive cooling pipes etc, just heat pads, will therefore benefit "less". 11.5a was definitely cooler than 11.3 when trying two per GPU. With 11.3 the VRMs went skyhigh in about 30 seconds to 126 degrees C ..... they got aborted pretty damn quick rofl, and I have two accelero aftermarket coolers on those beasts ... With 11.5a I found the temps of the VRMs were markedly lower. I started at 820/300 and VRMs were blipping at 105/106 on the infamous fourth GPU VRMs. Progressively lowered it down to 775/300, and not much above 85. Settled on 780/300 which were consistently at either side of 90. I need to fiddle some more as I only did a rushed quick "is the damn thing working" kind of check. I suspect I will get back up to 810/300 where the VRMs will be either side of 95 on the fourth GPU. The other three GPUs were at 48, and their VRMs at around 70-75 when at 820/300 (the effect of the coolers) So from a heat perspective, I found 11.5a to be cooler, so if only from that angle, its worth going for with 5970s. Regards Zy |
Send message Joined: 1 Sep 08 Posts: 204 Credit: 219,354,537 RAC: 0 |
I'm wondering, are you using any additional parameters in the app_info when you're running 2 WUs concurrently? I'm using "<cmdline>--gpu-target-frequency 60</cmdline>" to reduce GUI lag. This could explain why you're able to run 2 WUs whereas I am not (anymore). I'll give it a try when I'm feeling like I could use another blue screen ;) MrS Scanning for our furry friends since Jan 2002 |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
The one I used is below - all pretty straightforward and normal for an app_info <app_info> <app> <name>milkyway</name> </app> <file_info> <name>milkyway_0.62_windows_intelx86__ati14.exe</name> <executable/> </file_info> <app_version> <app_name>milkyway</app_name> <version_num>62</version_num> <plan_class>ati14</plan_class> <flops>1.0e11</flops> <avg_ncpus>0.05</avg_ncpus> <max_ncpus>1</max_ncpus> <coproc> <type>ATI</type> <count>0.5</count> </coproc> <file_ref> <file_name>milkyway_0.62_windows_intelx86__ati14.exe</file_name> <main_program/> </file_ref> </app_version> </app_info> Regards Zy |
Send message Joined: 15 Jul 08 Posts: 383 Credit: 729,293,740 RAC: 0 |
I'm wondering, are you using any additional parameters in the app_info when you're running 2 WUs concurrently? I'm using "<cmdline>--gpu-target-frequency 60</cmdline>" to reduce GUI lag. This could explain why you're able to run 2 WUs whereas I am not (anymore). I've been running 2 WUs/GPU on all 12 ATI GPUs ranging from the HD4770 to HD5870 ever since .62 hit the streets. No problems, GPU usage is much better and total WU throughput is significantly higher. All on them are running ATI 10.12 (W7-64) although 11.4 preview worked too (11.4 production release locked immediately upon completion of the first MW WU). A couple dedicated crunching machines have about a 30 second "pause" if I move the cursor over one of the Microsoft new type balloon messages. This seemed to coincide with one of the more recent service updates. Not sure what's causing it... |
Send message Joined: 1 Sep 08 Posts: 204 Credit: 219,354,537 RAC: 0 |
A couple dedicated crunching machines have about a 30 second "pause" if I move the cursor over one of the Microsoft new type balloon messages. This seemed to coincide with one of the more recent service updates. Not sure what's causing it... Yeah, that's what made me stop running 2 WUs with Cat 11.3, when it still worked. 11.4 production and 11.5 have broken support completely for me, though I still have to try without the 60 Hz adjustment. MrS Scanning for our furry friends since Jan 2002 |
©2024 Astroinformatics Group