50C? must be using some nice waterblocks.Nah cheapest ones I could find. They are full cover waterblocks though, does no good to cool the GPU and not the VRMs.
Topic: I think my 5970 should be working harder than it is. (Read 4075 times)
50C? must be using some nice waterblocks.Nah cheapest ones I could find. They are full cover waterblocks though, does no good to cool the GPU and not the VRMs.
50C? must be using some nice waterblocks.
410MH/s per core @ 920/300 (except one core which won't go above 900MHz).
It is all about the cooling. Running 1.125V @50C.
It is all about the cooling. Running 1.125V @50C.
sounds right. I am getting 377 mh 850/400 temps at 79
Running a 5970 at 810/150/1050 @ 380 MHash/S each core using Phoenix w/PhatK
Quote
DISPLAY=:0 aticonfig --od-setclocks=840,416 --adapter=2
DISPLAY=:0 aticonfig --od-setclocks=870,416 --adapter=3
DISPLAY=:0 aticonfig --od-setclocks=870,416 --adapter=3
Gives me 360-365MH/s respectively 379-386MH/s... could push the first one up to 845-850 something if I didn't have a 6990 over it. (the VRM's is getting bbq'd)
This is on the other hand very good... A shop gave me seven 6990 out of eight 5970 ordered and thought they were nice to me, they can't pull air from the sides, they blow hot air into my case which kills all my other equipment, pulls about 50W more energy a piece which ment I had to buy new PSU's and gives me the same 379-386MH/s! But hey, they cost more, so they must be better for everyone, right..?
Edit: Using DiabloMiner-Linux -v 2 -w 128, SDK 2.5 and driver 11.7. Just upgraded to 11.8 though which seems to give me slightly less hashes in average and alot more variance, allthough I just might need to balance the clocks a degree or two. The 6990 is now getting slightly higher speeds and seems more stable. Linux 11.8-drivers is not the same as the windows ones
Is it overheated at all?
No, temps are totally normal, in fact, low. I'm going to take the box apart today. I wish I knew someone locally (Seattle) with a 5970 so I could just swap it out, that would be the easiest way to isolate the problem.
The MB, PSU, CPU are all new. So I'd like to figure out if one of them is bad quickly so I can RMA if necessary.
Edit: Well, no dice. I took everything apart and put it back together and still can't seem to get this card working hard enough. I will try some other diagnostics on the card, but would definitely welcome any suggestions from the group.
Second edit: Solved it! So the problem is apparently with the 11.8 ATI drivers. I used the 11.5 ATI drivers instead and now things are working beautifully! I even added in an old 4800 card that a friend wasn't using anymore, so I'm getting (on stock clocks) about 680 Mhash out of the two cards. I'm sure I can tweak it upwards from there.
Is it overheated at all?
Code:
cgminer version 2.0.2 - Started: [2011-09-15 21:33:41]
--------------------------------------------------------------------------------
(5s):335.9 (avg):322.1 Mh/s | Q:29 A:0 R:18 HW:0 E:0% U:0.00/m
TQ: 2 ST: 2 SS: 0 DW: 0 NB: 1 LW: 0 GF: 0 RF: 0 I: 4
Connected to http://localhost:8332 without LP as user XXXXXXXX
Block: 000008152519c654c5385277c7576405... Started: [21:33:41]
--------------------------------------------------------------------------------
[P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit
GPU 0: 55.0C 2884RPM | 123.8/120.2Mh/s | A:0 R:5 HW:0 U:0.00/m
GPU 1: 210.0/203.6Mh/s | A:0 R:13 HW:0 U:0.00/m
--------------------------------------------------------------------------------
[2011-09-15 21:36:54] Rejected 7de7f450 GPU 0 thread 2
[2011-09-15 21:37:01] Rejected 83eb721b GPU 1 thread 3
[2011-09-15 21:37:09] Rejected 02a39c4f GPU 1 thread 3
[2011-09-15 21:37:13] Rejected 2b90d8b4 GPU 1 thread 1
--------------------------------------------------------------------------------
(5s):335.9 (avg):322.1 Mh/s | Q:29 A:0 R:18 HW:0 E:0% U:0.00/m
TQ: 2 ST: 2 SS: 0 DW: 0 NB: 1 LW: 0 GF: 0 RF: 0 I: 4
Connected to http://localhost:8332 without LP as user XXXXXXXX
Block: 000008152519c654c5385277c7576405... Started: [21:33:41]
--------------------------------------------------------------------------------
[P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit
GPU 0: 55.0C 2884RPM | 123.8/120.2Mh/s | A:0 R:5 HW:0 U:0.00/m
GPU 1: 210.0/203.6Mh/s | A:0 R:13 HW:0 U:0.00/m
--------------------------------------------------------------------------------
[2011-09-15 21:36:54] Rejected 7de7f450 GPU 0 thread 2
[2011-09-15 21:37:01] Rejected 83eb721b GPU 1 thread 3
[2011-09-15 21:37:09] Rejected 02a39c4f GPU 1 thread 3
[2011-09-15 21:37:13] Rejected 2b90d8b4 GPU 1 thread 1
Ugh, that doesn't look good. My bet is that your GPU is somehow
faulty. The reject rate also tends to point in that direction.
I'm not convinced that it is the card, especially since each core works on its own @ 300ish Mhash. But I will pull everything apart this weekend and put it back together. Perhaps there is a loose connection.
Hi all,
I've got a little bit of a different problem, but thought I'd post here since it concerns a 5970 card.
I'm running ubuntu 11.04 with ati-sdk 2.4 and catalyst 11-8. mining is working on each core, but if i run workers on both cores at the same time, the hash rate drops by 50%. Do I need to get dummy plugs? Or is something else going on?
Once I get both cores working independently of each other, I'll work on upping the hash rate.
Any help appreciated!
I've got a little bit of a different problem, but thought I'd post here since it concerns a 5970 card.
I'm running ubuntu 11.04 with ati-sdk 2.4 and catalyst 11-8. mining is working on each core, but if i run workers on both cores at the same time, the hash rate drops by 50%. Do I need to get dummy plugs? Or is something else going on?
Once I get both cores working independently of each other, I'll work on upping the hash rate.
Any help appreciated!
There's no never been a need for dummy plugs on Linux.
Try to use cgminer. It should detect all cores at once and
mine on all of them. If a core doesn't get detected, you'll
know at once.
Hi,
Thanks for the response. Actually, the problem is not with detection:
Code:
[0] Cypress
[1] Cypress
[2] AMD Sempron(tm) 145 Processor
[1] Cypress
[2] AMD Sempron(tm) 145 Processor
When I run a phoenix miner on device0, i get ~ 316 Mhash/sec and loads on Device0=99%. But when I start a second phoenix miner on device1, it starts, works fine, but I get this result on the miner running on Device0:
Code:
Kernel error: Unusual behavior from OpenCL. Hardware problem?
And the loads look like this:
Code:
# aticonfig --odgc --adapter=ALL
Adapter 0 - ATI Radeon HD 5900 Series
Core (MHz) Memory (MHz)
Current Clocks : 750 300
Current Peak : 750 300
Configurable Peak Range : [550-1000] [1000-1500]
GPU load : 60%
Adapter 1 - ATI Radeon HD 5900 Series
Core (MHz) Memory (MHz)
Current Clocks : 750 300
Current Peak : 750 300
Configurable Peak Range : [550-1000] [1000-1500]
GPU load : 38%
Adapter 0 - ATI Radeon HD 5900 Series
Core (MHz) Memory (MHz)
Current Clocks : 750 300
Current Peak : 750 300
Configurable Peak Range : [550-1000] [1000-1500]
GPU load : 60%
Adapter 1 - ATI Radeon HD 5900 Series
Core (MHz) Memory (MHz)
Current Clocks : 750 300
Current Peak : 750 300
Configurable Peak Range : [550-1000] [1000-1500]
GPU load : 38%
I can run either miner independently of the other without getting the error. Also, FWIW, I have crossfire chains disabled:
Code:
# aticonfig --lsch
No CrossFire chains defined
No CrossFire chains defined
Also, I will go ahead and install cgminer... maybe all the problems will go away! I'll report back after I do so.
Okay, here's my cgminer install, which is working, but no better than the other miners:
Code:
cgminer version 2.0.2 - Started: [2011-09-15 21:33:41]
--------------------------------------------------------------------------------
(5s):335.9 (avg):322.1 Mh/s | Q:29 A:0 R:18 HW:0 E:0% U:0.00/m
TQ: 2 ST: 2 SS: 0 DW: 0 NB: 1 LW: 0 GF: 0 RF: 0 I: 4
Connected to http://localhost:8332 without LP as user XXXXXXXX
Block: 000008152519c654c5385277c7576405... Started: [21:33:41]
--------------------------------------------------------------------------------
[P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit
GPU 0: 55.0C 2884RPM | 123.8/120.2Mh/s | A:0 R:5 HW:0 U:0.00/m
GPU 1: 210.0/203.6Mh/s | A:0 R:13 HW:0 U:0.00/m
--------------------------------------------------------------------------------
[2011-09-15 21:36:54] Rejected 7de7f450 GPU 0 thread 2
[2011-09-15 21:37:01] Rejected 83eb721b GPU 1 thread 3
[2011-09-15 21:37:09] Rejected 02a39c4f GPU 1 thread 3
[2011-09-15 21:37:13] Rejected 2b90d8b4 GPU 1 thread 1
--------------------------------------------------------------------------------
(5s):335.9 (avg):322.1 Mh/s | Q:29 A:0 R:18 HW:0 E:0% U:0.00/m
TQ: 2 ST: 2 SS: 0 DW: 0 NB: 1 LW: 0 GF: 0 RF: 0 I: 4
Connected to http://localhost:8332 without LP as user XXXXXXXX
Block: 000008152519c654c5385277c7576405... Started: [21:33:41]
--------------------------------------------------------------------------------
[P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit
GPU 0: 55.0C 2884RPM | 123.8/120.2Mh/s | A:0 R:5 HW:0 U:0.00/m
GPU 1: 210.0/203.6Mh/s | A:0 R:13 HW:0 U:0.00/m
--------------------------------------------------------------------------------
[2011-09-15 21:36:54] Rejected 7de7f450 GPU 0 thread 2
[2011-09-15 21:37:01] Rejected 83eb721b GPU 1 thread 3
[2011-09-15 21:37:09] Rejected 02a39c4f GPU 1 thread 3
[2011-09-15 21:37:13] Rejected 2b90d8b4 GPU 1 thread 1
Hi all,
I've got a little bit of a different problem, but thought I'd post here since it concerns a 5970 card.
I'm running ubuntu 11.04 with ati-sdk 2.4 and catalyst 11-8. mining is working on each core, but if i run workers on both cores at the same time, the hash rate drops by 50%. Do I need to get dummy plugs? Or is something else going on?
Once I get both cores working independently of each other, I'll work on upping the hash rate.
Any help appreciated!
I've got a little bit of a different problem, but thought I'd post here since it concerns a 5970 card.
I'm running ubuntu 11.04 with ati-sdk 2.4 and catalyst 11-8. mining is working on each core, but if i run workers on both cores at the same time, the hash rate drops by 50%. Do I need to get dummy plugs? Or is something else going on?
Once I get both cores working independently of each other, I'll work on upping the hash rate.
Any help appreciated!
Also don't forget some tweaks can push you to the 370+ range. Here is a modified phatk kernel I use http://forum.bitcoin.org/index.php?topic=7964.0 and here is a tweak +3-5% hashrate http://forum.bitcoin.org/index.php?topic=23067.0
my 6950 has lost around 60Mh/s when i installed the most recent catalyst drivers, so maybe thats also an issue with your card? after rverting to the old driver version I gotthe 60mhs back!
Yea, I just noticed that this card can go over 820, but the clock rate is at 950. Not even going to think about pushing it that hard.
Your hash-rate sounds about right.
Anybody posting the "extreme" hash rates is most definitely using water cooling. You cannot achieve those rates without water cooling.
Anybody posting the "extreme" hash rates is most definitely using water cooling. You cannot achieve those rates without water cooling.
Quote
I can't get under 500 regardless of how much I play with afterburner 
You have to set it to 500, hit apply, then close it. Then open it back up and you can go much lower than 500.
I will try the new kernel and let you know how it went.
Edit: I'm hitting 370 a card now. I hate myself for not doing this earlier.
360 per core is actually pretty good.
I can't get my 6990 cores much past that without significant overclocking.
Even then it will hit about 380 max, maybe 800mhash/s total if I really tweaked around with the settings.
Consider that a 5970 is at least 100-150 bucks cheaper so it's a bargain if you find one.
I can't get my 6990 cores much past that without significant overclocking.
Even then it will hit about 380 max, maybe 800mhash/s total if I really tweaked around with the settings.
Consider that a 5970 is at least 100-150 bucks cheaper so it's a bargain if you find one.
i always thought this page is undervalued/used
i try to aim for the top 3 mostly, if i can get around those numbers i think ive pretty much maxed the card
i try to aim for the top 3 mostly, if i can get around those numbers i think ive pretty much maxed the card
I'm using Phoenix Miner, and my settings are -k poclbm DEVICE=0 AGGRESSION=12 WORKSIZE=128 VECTORS BFI_INT
I've overclocked to 850mhz and my ram is at 300mhz.
Temps are currently 75 and 79 degrees C
I'm actually getting around 350 mhashes a GPU, not the 360 previously reported.
I've overclocked to 850mhz and my ram is at 300mhz.
Temps are currently 75 and 79 degrees C
I'm actually getting around 350 mhashes a GPU, not the 360 previously reported.
What settings are you using and what miner?
fastest I can get mine is 376 mh/s per core at 850 mhz. If I push it any more the vrms overheat.
How any Mhashes are you guys pulling per GPU on this card? I get around 360 when I know I should be getting more like 400. I don't water cool though, instead I have a fan keeping it around mid 70s.
Jump to: