|
cgminer + diablo kernel + v1 w256 = same performance as using diablominer (for me)
12.2pre + bundled sdk + win7 32-bit
core 1125 mhz -> 662 mh/s
|
|
|
|
|
Got my first hw error now after 5000 shares - and only running one instance of DM.
Also performance is acutally much worse, mh/s dropped to ~1435 after the initial 1600 mh/s. TDR did not trigger again, don't know what caused it the first time it happened.
Is there actually a reason I should be worried about hw errors if everything else runs smooth?
|
|
|
|
|
Ok, running one instance with -v1 -w256 now.
Interestingly performance isn't much worse (was 1626 mh/s with two instances, one with -v1 -w256 for gcn and one with -v2 -w128 for vliw), now with one instance I get 1604 mh/s...
CPU usage is still strange, though: 50% on a dualcore, affinity set to *both* cores. If I would run into the classic cpu bug, it should consume 100% of the cpu, right?!
And as I wrote this, TDR triggered on one of the 5850... ouch. Can this be caused by different vector/worksize settings? (Never used -v1 -w256 on the 5850 before)
|
|
|
|
|
Ok, sorry, did not know about that memclock thing...
|
|
|
|
|
DiabloMiner now gets the same speed from SDK 2.6 as it got from SDK 2.1.
No need to stick with an old SDK.
|
|
|
|
Thanks guys, but I don't have much motivation to lower the temps unless that's the reason for the hw errors... (I have to drive to the remote location and spend much time I need somewhere else...) One of my rigs has a 230mm fan blowing on the cards from the side, maybe I'll get another one and put it in my second rig, but as I said, all this stuff costs time. My mining operation already is consuming much of it in the last weeks, because fans began to fail and I had to issue some RMAs... different story  Card is running @ 74°C now and still getting hw errors (4/800)... I'm not really convinced that temperature is my problem, is it?! Its not. 85c is the maximum safe limit for 24/7 mining, the cards can survive up to 120c for short periods of time without damage. What arguments are you using? None and/or just -v 1 -w 256? I'm using -v1 -w256. Also note this is a "mixed" setup (5xxx, 6xxx and 7970) so I have to run two instances of DM. I don't know if this is relevant, but the 2nd instance is causing 50% load on a dualcore (affinity set to both cores, so it does not look like the "classic" cpu-bug) |
|
|
|
Temp along with voltage would be my guess. unless you ASIC quality is above 68% your volts are too low for that clock and that will give you plenty of HW errors. I'm running stock settings now and still getting hw errors What is your quality on your cards?
Sorry, can't say atm. Last time I started GPU-Z crashed the machine and I had to drive to the remote location and reset it :/ |
|
|
|
Thanks guys, but I don't have much motivation to lower the temps unless that's the reason for the hw errors... (I have to drive to the remote location and spend much time I need somewhere else...) One of my rigs has a 230mm fan blowing on the cards from the side, maybe I'll get another one and put it in my second rig, but as I said, all this stuff costs time. My mining operation already is consuming much of it in the last weeks, because fans began to fail and I had to issue some RMAs... different story Card is running @ 74°C now and still getting hw errors (4/800)... I'm not really convinced that temperature is my problem, is it?! |
|
|
|
I have 4 cards installed in an open case, so there is not much room to breathe  I'm on manual fan control because on auto the temp goes all the way up to 90°C... |
|
|
|
How many hw errors are ok for a 7970? I currently have 0,17% (52/29410) hw errors, card running overclocked (1125 Mhz) and undervolted (1100 mV).
Everything running stable for >2 days now.
Significant errors like that is insane. Set stock clocks and volts and see if it still does it, if it does its a driver bug, if it doesn't, you overclocked too far. You shouldn't see more than 1 or 2 HW errors per 10k shares give or take, if it starts skyrocketing you went too far. With stock clock/voltage I now have 10/1738 hw errors... so still way too much. What does this mean now? I'm using the latest driver (12.2 pre-cert), latest DM, Windows 7 32-bit. Can I ignore the hw errors and just live with a higher reject rate or is there something I should/can/must do? |
|
|
|
|
Depends. Latest DiabloMiner is optimized for SDK 2.6
|
|
|
|
|
How many hw errors are ok for a 7970? I currently have 0,17% (52/29410) hw errors, card running overclocked (1125 Mhz) and undervolted (1100 mV).
Everything running stable for >2 days now.
|
|
|
|
|
Paid out: 112 BTC.
Solved Blocks: 5.
-.-
|
|
|
|
|
Is there a way to use optimal settings for both VLIW and GCN without running two instances of DiabloMiner?
|
|
|
|
|
Now that 12.2 is available this thread is obsolete, thanks everyone.
|
|
|
|
Dude, you're not using the newest copy of DM. Feature Request: Print build number to the console window |
|
|
|
But you shouldn't be using -v 4 -w 64. That is not appropriate for the new kernel on vliw5. For me, -v 2 -w 256 is fastest, -v 2,1 -w 256 is slower, -v 2 -w 128 isn't faster (but -w is very sensitive to memory timings, so -w 128 being faster for others does not surprise me).
What kernel do you mean by "the new kernel"? I'm using the zip-package linked in the first post of this thread and this is still the "old kernel", isn't it?! I did not compile DiabloMiner from source so I guess I'm still using your "not optimized for SDK 2.6"-version. Maybe we can clear that confusion first Okay, I figured out how to do this on Windows. In the .bat file or in cmd.exe or whatever you use, before running DiabloMiner, do SET GPU_DUMP_DEVICE_KERNEL=3. It should cause it to dump the .isa files.
-v4 http://pastebin.de/23309-v1,2 http://pastebin.de/23310-v2 http://pastebin.de/23311 |
|
|
|
Lets say you're getting similar numbers to mine. I get 393 or so on my 5850 at 960/320, 393/960*850 is 348. You have two of those, so 696. Now, your 5870, would be 393/960*950/1440*1600 = 432. You should be getting around 1128. precisely... and that's about that what I got using SDK 2.5. Now with SDK 2.6 and using -v4 -w64 I get 1060 MH/s, which is a ~10% decrease as expected from the change SDK 2.5->2.6. Using layered vectors gives those ridiculous high number of 1370 MH/s, which just cannot be explained. So if there is anything I can do/check, tell me... until then I can only ask you to double-check your meter code even if you are sure everything is correct - but something isn't ;o |
|
|
|
|
I'm on Windows 7 32-Bit... As you can read in post #1201 I also averaged over 3 hours (2500 shares) and I'm still getting the same results. And beside these "hard numbers"... don't you think it's very strange I can achieve 1370 MH/s with 2x5850@850/1000 + 1x5870@950/1200, SDK 2.6 and your old "SDK 2.1 miner"? It does not make any sense at all.
And again: If I'm not using layered vectors, miner mh/s and pool mh/s are matching very good. Also if I calculate the mh/s by hand using the submitted shares in a specific timeframe (like I did in post #1201), the numbers match. Only when using layered vectors the miner mh/s are WAY higher than everything else suggests (pool mh/s, hand-calculated mh/s, 24h earnings)
|
|
|
|
|
Show Posts
