Dabs
Legendary
Offline
Activity: 3416
Merit: 1912
The Concierge of Crypto
|
|
February 13, 2014, 09:11:49 AM |
|
sgminer 4.0.0 - Started: [2014-02-13 09:03:33] -------------------------------------------------------------------------------- (5s):1.753M (avg):1.795Mh/s | A:9216 R:128 HW:0 WU:1496.1/m ST: 2 SS: 0 NB: 7 LW: 84 GF: 0 RF: 0 Connected to Pool 0 (stratum) diff 512 as user Dabs.1 Block: 4a13c1dc... Diff:1.15M Started: [09:08:09] Best share: 328K -------------------------------------------------------------------------------- [P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit GPU 0: 67.0C 3774RPM | 452.9K/452.3Kh/s | R: 0.0% HW:0 WU: 354.0/m I:20 GPU 1: 60.0C 3855RPM | 452.8K/452.3Kh/s | R: 5.3% HW:0 WU: 411.4/m I:20 GPU 2: 71.0C 3852RPM | 452.0K/452.3Kh/s | R: 0.0% HW:0 WU: 387.4/m I:20 GPU 3: 75.0C 3894RPM | 453.2K/452.3Kh/s | R: 0.0% HW:0 WU: 361.8/m I:20 --------------------------------------------------------------------------------
[09:08:37] Accepted 1f0b3ddf Diff 2.11K/512 GPU 3 at Pool 0 [09:08:39] Accepted 0db1798a Diff 4.79K/512 GPU 0 at Pool 0 [09:08:41] Accepted 636ce3b0 Diff 659/512 GPU 2 at Pool 0 Finally got 4 GPUs working. Replaced the riser. Replaced the part that plugs into the slot. Replaced the USB cable. Replaced the SATA power cable. Plugged it in. Now it works. So, I'm not sure but maybe I might have a couple of defective risers. I'll do more testing at home. On the other hand, I upgraded to Litecoin BAMT 1.5 and the hash rates went lower with the same settings. I'll rewrite 1.3 and see if that improves, maybe it's the Catalyst driver versions.
|
|
|
|
24Kilo
|
|
February 13, 2014, 09:25:59 AM |
|
the latest veggie crate miner... 1.5Mh/s with ease...
|
|
|
|
Dabs
Legendary
Offline
Activity: 3416
Merit: 1912
The Concierge of Crypto
|
|
February 13, 2014, 09:41:18 AM |
|
That's really wide spacing, LOL! I think I'll have to space out mine a little bit more when it moves to its permanent home.
Watt Meter shows 650 ~ 670 watts for my 1.8 ~ 1.9 MH/s (R9 270s doing 455 kh/s each). I'll try to bump up this speed using the older BAMT, but this one actually seems more stable now.
|
|
|
|
24Kilo
|
|
February 13, 2014, 10:09:53 AM |
|
270's do seem to be the card to choose... I am getting 1.5Mh/s at 670W... hoping to trim that by 50W... but still not as efficient as 270's
|
|
|
|
vesperwillow
|
|
February 13, 2014, 01:54:49 PM |
|
Congrats Dabs. Heey Guys.
I have my Asrock Pro BTC. Now im looking for a goog GPU. What do you recommend as a good PSU?? I was thinking of a 1200 wat PSU sow must be enough for 5 R9270 of 280. What do you guys think??
280X's pull up to 275w each, not sure what 280's pull, I'd wager close to the same. No way you can run 5 of them on one PSU. You'd need at least 1500w for that. If I were running 5 gpu's, I'd be wanting to run 270/270X's. You could put those on a 1050w PSU.
|
|
|
|
ehalvers
|
|
February 13, 2014, 11:22:33 PM |
|
What would you guys suggest for a x6 280x rig?
I was planning on having 4 cards and the mobo powered by a 1200w and then the other 2 cards on a 650w. I feel like the 1200w might be overloaded though. Any suggestions?
I could purchase an 850w to power 2 cards and the mobo and have the 1200w just power the 4 cards, which would make me sleep a little better at night lol.
I could always use VBE7 and flash the gigabyte BIOS to lower the voltage, a little but I have never messed with that stuff before and don't want to wreck an cards.
|
|
|
|
AFK.MPEG
Member
Offline
Activity: 72
Merit: 10
|
|
February 13, 2014, 11:33:59 PM |
|
Congrats Dabs. Heey Guys.
I have my Asrock Pro BTC. Now im looking for a goog GPU. What do you recommend as a good PSU?? I was thinking of a 1200 wat PSU sow must be enough for 5 R9270 of 280. What do you guys think??
280X's pull up to 275w each, not sure what 280's pull, I'd wager close to the same. No way you can run 5 of them on one PSU. You'd need at least 1500w for that. If I were running 5 gpu's, I'd be wanting to run 270/270X's. You could put those on a 1050w PSU. Strange. Because i had 3 hd6990 running on 1200 watt with out a problem. And those cards are real energy users. But i gonna look further for a good psu
|
|
|
|
Dabs
Legendary
Offline
Activity: 3416
Merit: 1912
The Concierge of Crypto
|
|
February 14, 2014, 04:08:59 AM |
|
GPU Status Temp Load Rate Accept/Invalid GPU Family Fan% (rpm) Core Memory Power 0 63.0 c 99% 490 Kh/s 3 / 0 0.00% 200 Series 80% (3894) 1150 Mhz 1500 Mhz 1.163v 1 57.0 c 99% 490 Kh/s 1 / 0 0.00% 200 Series 80% (3830) 1150 Mhz 1500 Mhz 1.188v 2 63.0 c 99% 490 Kh/s 6 / 0 0.00% 200 Series 80% (3826) 1150 Mhz 1500 Mhz 1.163v 3 63.0 c 99% 490 Kh/s 5 / 0 0.00% 200 Series 80% (3685) 1150 Mhz 1500 Mhz 1.188v I copied the cgminer.conf settings. I took out the flash drive. I stuffed BAMT 1.3 back to it. Put it back. Turned it on. Edited cgminer.conf to be exactly the same. Now my hash rates are back to "normal". I'm just going to dial down either the intensity or the core speed as 1150 is just a bit tad too high I think.
|
|
|
|
Dabs
Legendary
Offline
Activity: 3416
Merit: 1912
The Concierge of Crypto
|
|
February 14, 2014, 06:20:49 AM Last edit: February 14, 2014, 07:07:52 AM by Dabs |
|
It seems to be crashing much too often in the past few hours, so I restart it and set the Core speed a bit lower, or set the intensity 1 down.
Right now, I = 18 and 1135, 1500 for 470 Kh/s. So from 1.9 Mh/s it went to 1.88 Mh/s. That's probably a good compromise for stability and here's hoping it won't crash anymore.
*edit* On second thought, this might be a temperature issue. I'll need to investigate on this a bit. In any case, if it runs too hot, I will have to slow it down so it runs cool enough.
Now down to 1125, 1500 at 480 Kh/s but back to Intensity 20.
|
|
|
|
vesperwillow
|
|
February 14, 2014, 02:41:06 PM |
|
What would you guys suggest for a x6 280x rig?
If you play it safe, figure 350watts per card. That's 2100 watts, not counting about 100 for the system itself. You'd be looking at two 1,200w PSU's. It's overkill and inefficient IMO, unless there were other reasons for the 280's (such as availability or a great price). If you get 6 Sapphire Toxic 280X's, the finest 280X I've personally seen yet, and you push it to its limits, you could expect 4800kh/sec, and you'll be at the limit of 2400 watts of PSU. If you get 10 MSI 270 (or 270X, they hash the same), you would hash at 4900kh/sec, at around 1900w total. This would still require 2 PSU's and would be split up into 2 systems of course. But you can see the benefit of this setup vice a 280 setup. If I could, I would switch out my 280 collection for MSI 270's in a heartbeat and save wattage. Strange. Because i had 3 hd6990 running on 1200 watt with out a problem. And those cards are real energy users. But i gonna look further for a good psu
Yeah, I must have been thinking about the 79xx series or maybe underclocked 280X's. Looking back on my notes, my 280X's draw 340-380 watts each, depending on which one it is. Power-wise here's what I've personally noticed, real-world mining power usage per GPU/model: 79xx: 270-275 280X: 340-380 270X: 175 270: 167-175 It seems to be crashing much too often in the past few hours, so I restart it and set the Core speed a bit lower, or set the intensity 1 down.
Right now, I = 18 and 1135, 1500 for 470 Kh/s. So from 1.9 Mh/s it went to 1.88 Mh/s. That's probably a good compromise for stability and here's hoping it won't crash anymore.
*edit* On second thought, this might be a temperature issue. I'll need to investigate on this a bit. In any case, if it runs too hot, I will have to slow it down so it runs cool enough.
Now down to 1125, 1500 at 480 Kh/s but back to Intensity 20.
Usually it's heat-related, keep an eye out on the temps. Also make sure your power connections are solid.
|
|
|
|
ehalvers
|
|
February 14, 2014, 08:42:35 PM |
|
Thanks for the tips Vesper! I think I may end up doing a 4x 280x (3 on the 1200w and on the 650w 1 card/system) With that set up I may be able to squeeze one or two more 270x's or undervolted 7950's on there but I will most likely just play it safe. EDIT: Was able to pick up 2 Gigabyte 270x for $210 at microcenter. 270's were $199.99 so I said what the hell haha. Can only get them up to around 470 kh/s though. Was hoping for 10-15 more then that but I am satisfied
|
|
|
|
Dabs
Legendary
Offline
Activity: 3416
Merit: 1912
The Concierge of Crypto
|
|
February 15, 2014, 02:49:02 AM |
|
Usually it's heat-related, keep an eye out on the temps. Also make sure your power connections are solid.
GPU 0: 65.0C 3847RPM | 485.6K/484.7Kh/s | R: 0.0% HW:0 WU:445.1/m I:20 GPU 1: 60.0C 3799RPM | 485.4K/484.9Kh/s | R: 0.8% HW:0 WU:415.1/m I:20 GPU 2: 69.0C 3803RPM | 485.7K/484.9Kh/s | R: 0.2% HW:0 WU:470.5/m I:20 GPU 3: 72.0C 3748RPM | 482.8K/484.9Kh/s | R: 0.0% HW:0 WU:443.5/m I:20 I've been keeping temps below 75 C. But it should run fine even if it went to 85 C right? Fans are at 80%. Running it at 1125 Engine, 1500 Memory. I haven't had the chance to undervolt it, I might need to install windows to do that, or plug the cards to another system, but I could also just leave it alone for now. It's not broke, so ... ...
|
|
|
|
vesperwillow
|
|
February 17, 2014, 11:10:05 PM |
|
Ok I still need help here please I followed this thread as it got him decent results with my cards https://bitcointalk.org/index.php?topic=453067.20I've got a complied version of sgminer, downloaded his .cl file and .bin file and written the .bat file like his (as below) and all I get is "unexpected extra commandline arguments" from sgminer.. Why? color 02 timeout /30 setx GPU_MAX_ALLOC_PERCENT 100 setx GPU_USE_SYNC_OBJECTS 1 sgminer kernel : scrypt130302 --thread concurrency 8192,8192,8192 -g 2 -I 13 -w 256 --auto-fan --temp-cutoff 90,90,90 --temp-overheat 85,85,85 --temp-target 68,68,68 --gpu-memclock 1500,1500,1500 --gpu-engine 1065,1065,1065 --expiry 1 --scan-time 1 --queue 0 that thread has --lookup-gap 2 in his .bat file as well, though that has already been pointed out it doesn't need to be there, but I get the same problem whether its included or not. The kernel file is simply the name of the kernel file, but again i've tried the zuikkis.cl and that doesn't change anything. So where am I going wrong.. still... please help! I presume you're running windows. What version? Have you tried removing the kernel info from the commandline and then run sgminer?
|
|
|
|
FortuneSeeker
Member
Offline
Activity: 107
Merit: 10
|
|
February 17, 2014, 11:36:09 PM |
|
Yes I have and it got it working. I was just removing things and trying it and removing the kernel bit did the trick, So now i'm wondering if I delete all the other .cl files and just leave the optomised .cl file that i downloaded for my card in the kernel directory I also have to flash the bios on the cards yet. but I was getting 2 cards at about 707kh/s and 1 at 680 kh/s, this is an improvement on the cgminer results. temps are about 70/81/78 but as you can see from the picture cards 2 + 3 are very close together. I found a fan I can attach to the wall (box fan) which seems to get the temps i mentioned rather than a pc smouldering 90C so I'm hoping with the undervolting the temps will be more stable and I can start fine tuning the settings http://prntscr.com/2tha6g
|
|
|
|
vesperwillow
|
|
February 18, 2014, 12:17:20 AM |
|
Yeah they're pretty close. If you didn't mind not being able to close the case, you could get some risers and move them up and away from each other and lower the temps naturally. I'm not familiar with how sgminer handles kernels, but here's a link of the official modified kernels, which includes announcement sublinks to the various forums for each which describes how to properly use them (as each one is for different driver versions): https://github.com/veox/sgminer/blob/master/doc/KERNEL.mdIf this was a dedicated system solely for mining, I would personally suggest a linux-based approach for stability and higher output, but if this is your personal system I can understand wanting Windows.
|
|
|
|
Dabs
Legendary
Offline
Activity: 3416
Merit: 1912
The Concierge of Crypto
|
|
February 18, 2014, 12:47:17 AM |
|
@FortuneSeeker, I've had good experience with Litecoin BAMT. Grab 1.3 (or 1.5, or both), write it to your USB flash drive, and boot.
Then just get it's ip address on your network, and SSH to it using PuTTy (windows). Edit /etc/bamt/cgminer.conf and put in your own pool details, delete everything else, you can fine tune it later.
Reboot. Watch it mine for a few minutes.
|
|
|
|
FortuneSeeker
Member
Offline
Activity: 107
Merit: 10
|
|
February 18, 2014, 01:07:32 AM |
|
Well I am no planning to attach a fan to the wall above pointing into the case, and leaving all the panels off. testing it i.e. holding it above the rig, seems to keep temps to the 70s C. some progress there and might enable me to leave the cards where they are. I haven't decided if I can actually find a way of taking the middle card out and suspending it some how with risers. I reduced the VDDC to 1100 on all 3 cards using VBE7, the 1131 cryptobadger suggested for his 280x cards wasn't an option on the drop down menu. Now does anyone have some understanding off undervolting? I've never tried it before. My orginal VDDCs for each card were 1175/1225/1175. I set them all at 1100 and at one point I was getting 710/724/724.. though it fluctuates. Then card 1 started getting 800+ like this http://prntscr.com/2ti12hwhich as much as I wish I could get that stably I doubt thats right and the screen crashes and windows recovers from a driver failure with sgminer saying something about card 1 having a problem and can't be restarted. Does it stand to reason that as it already had a VDDC that was higher than the other 2 cards (no idea why they are all the same card bought at the same time) I've set the VDDC too low on that particular card? I'm going to try chaning the bios again tomorrow i think because the 1100 VDDC on that card clearly doesn't work. Also has anyone set a VDDC lower than 110 on a 280x card. Or is it a case or changing it around untill the driver doesn't crash?
|
|
|
|
vesperwillow
|
|
February 18, 2014, 01:42:54 PM |
|
I reduced the VDDC to 1100 on all 3 cards using VBE7, the 1131 cryptobadger suggested for his 280x cards wasn't an option on the drop down menu. Now does anyone have some understanding off undervolting? I've never tried it before. My orginal VDDCs for each card were 1175/1225/1175. I set them all at 1100 and at one point I was getting 710/724/724.. though it fluctuates. Then card 1 started It's basically finding an equillibrium between efficiency and stability, reducing the voltage the card needs to lower heat and sometimes increase or decreash hash in a stable fashion. You typically do it in slight increments per card to see what works best. restarted. Does it stand to reason that as it already had a VDDC that was higher than the other 2 cards (no idea why they are all the same card bought at the same time) I've set the VDDC too low on that particular card? I'm going to try chaning the bios again tomorrow i think because the 1100 VDDC on that card clearly doesn't work. Also has anyone set a VDDC lower than 110 on a 280x card. Or is it a case or changing it around untill the driver doesn't crash?
I don't undervolt but I understand it. Every card is different just because of component variance, even if you buy 10 of the same model/brand. Generally speaking same-model cards should have the same default operating voltages, so if you have 2 cards which have different "stock" voltages, make sure they're actually the same model. Yes, you can undervolt too much and cause an issue. You typically want to do it in increments. Lots of people go in steps of .05, sometimes .1
|
|
|
|
FortuneSeeker
Member
Offline
Activity: 107
Merit: 10
|
|
February 18, 2014, 03:54:36 PM |
|
I reduced the VDDC to 1100 on all 3 cards using VBE7, the 1131 cryptobadger suggested for his 280x cards wasn't an option on the drop down menu. Now does anyone have some understanding off undervolting? I've never tried it before. My orginal VDDCs for each card were 1175/1225/1175. I set them all at 1100 and at one point I was getting 710/724/724.. though it fluctuates. Then card 1 started It's basically finding an equillibrium between efficiency and stability, reducing the voltage the card needs to lower heat and sometimes increase or decreash hash in a stable fashion. You typically do it in slight increments per card to see what works best. restarted. Does it stand to reason that as it already had a VDDC that was higher than the other 2 cards (no idea why they are all the same card bought at the same time) I've set the VDDC too low on that particular card? I'm going to try chaning the bios again tomorrow i think because the 1100 VDDC on that card clearly doesn't work. Also has anyone set a VDDC lower than 110 on a 280x card. Or is it a case or changing it around untill the driver doesn't crash?
I don't undervolt but I understand it. Every card is different just because of component variance, even if you buy 10 of the same model/brand. Generally speaking same-model cards should have the same default operating voltages, so if you have 2 cards which have different "stock" voltages, make sure they're actually the same model. Yes, you can undervolt too much and cause an issue. You typically want to do it in increments. Lots of people go in steps of .05, sometimes .1 Using VBE7 I only get a drop down menu to choose new figures from 1275, 1256, 1250, 1225, 1200, 1188, 1175, 1163, 1150, 1131, 1120, 1100 an so forth. With TRIXX I can adjust it down by 1 unit.... not sure how 0.05 and 0.1 relates to these figures. Later tonight I was just going to play around with VBE7 and the figures above increaseing to the next option higher on card 1 till it is stable My cards are all the same model all Sapphire R9 280X DUAL-X OC version with the same sku and PN numbers on the label
|
|
|
|
FortuneSeeker
Member
Offline
Activity: 107
Merit: 10
|
|
February 18, 2014, 09:58:39 PM |
|
this may be a dumb question but my figures i see on screen fluctuate alot while i'm watching
when looking at for the hash rate my cards produce and I see 705.2K/710.8Kh/s I am assuming its the rights side figure thats important the 710.8kh/s
Because sometimes the left side is higher and sometimes the right side figure is higher but it never stays around 1 figure for long enough for me to decide what i'm getting
|
|
|
|
|