16.12.1-December5 is now the latest, but doubt it offers any improvement for mining over 16.11.5 |
|
|
|
i have experimented and 8 gig gpu's (or at least rx480's) will not hash with 1 gig system ram.. possibly 2 gig also i didnt take notes it was more of a lark really to try.
they dont crash though just hash at 0h/s
no reason to mine with less than 4 gig ram these days. and four is enough for the hobbyist miner.
On NEW builds sure - but have you priced DDR lately? It would cost me almost as much to completely replace the MB/RAM/CPU in a few of my existing rigs than to upgrade them to 4GB of RAM. |
|
|
|
Hmmm, yes, a 16 GB swap file, but only 2 GB of 800 MHz RAM for a 64-bit OS on the two boxes that are giving me trouble. The third and similar box which worked fine (with the exception having 3 PCIe slots but not taking a third GPU) has 4 GB of 1333 MHz RAM so you might be on to something. Should Claymore need more than 2 GB and do 8 GB cards require more than 4 GB cards? I'm running 8 GB ones now but have a couple of 4 GB ones I could try...
Lack of physical system RAM could be the issue, as all of the systems I've tried v9 on to date had 2 GB or less physical RAM. Seems odd if that is the case though that taskmanager doesn't SHOW Claymore v9 using a ton of RAM. |
|
|
|
Is it me the only one who is getting a lot of connection failed retry to connect in 20 sec?
All my miners started doing this, and my pool stats show 30% less hashrate like this.
Version 9/9.1 isn't worth the hassle if your rigs were working fine and are now playing up, even if you get a lesser fee then Version 8 stick to being stable and hashing stable. The newer versions seem nuked to me... You just need to reduce the voltage or the clock frequency so that it will stress the PSU less. That should solve the problem. NOT always, as I 've already documented in my testing on a single-card HD 7870 rig running at max UNDERCLOCK on a Seasonic X-1250. Might help in some cases where the rig was right at the margins for stability, but it does NOT help in all cases. |
|
|
|
some of my rigs reverted back to v8.
v9 has some serious stability issues that needs to be adressed.
Check clocks, volts, and PSUs. I haven't had a single crash on 24 cards (470/480). I have moderate undervolt and have PSUs at ~70% load (EVGA P2 1200w, 6 cards each). At one point my HD7870 rig was set to as far UNDERCLOCKED as I could get it, the card CAN'T be undervolted, and the PSU on it is a Seasonic X1250 Gold *LOAFING* at ballpark 15% of it's capasity. No, it's NOT the hardware when the same system runs at a serious OVERCLOCK with good stability on v8 running the same IDENTICAL hardware. |
|
|
|
|
May depend on the specific card model - one Gigabyte model I have a couple of increase the TDP from the 150ish "standard" to 180ish, and I wouldn't be supprised to see certain models from other manufacturers do the same thing.
|
|
|
|
|
Does GPU-Z show the card has having OpenCL enabled?
Might have to do a full uninstall of 16.x.x clean and reinstall 15.12 if it does not.
|
|
|
|
|
You're not going to find a replacement in the $10 range with comparable airflow - and even at wholesale pricing I doubt you'll find something comparable for $15 each.
You COULD try using AFB series Deltas instead, but they're not designed for the high back pressure that the S7 (and S9) design puts on a fan and might not flow as much even if they're RATED to do so at zero backpressure.
|
|
|
|
V9.0 has no stability issues! Your rigs do! Increase GPU voltage or lower clocks!
V9 has some MAJOR stability issues for at least some configurations. I tried lowering clocks on the 7870 test rig I DOCUMENTED my issues with during fairly extensive testing in the main CLaymore V9 thread - doing so made NO change even when I got the card down to about 200 Mhz UNDERclock on core and 300 on mem (afterburner limits for that card) the ONLY change was that the hashrates reported got lower than what I reported in that thread. I can't increase voltages as they're locked on that card (and they're at stock anyway). Same card was highly stable back in the days I used it on Scrypt mining, and for all the MooWrapper/Dnet work I did between it's Scrypt days and it's ETH/ZEC days - both of those drove that card a LOT harder than ETH or ZEC do (as measured by the MUCH HIGHER POWER DRAW and higher operating temp at higher fan settings in both cases). I've never played with the BIOS on that card in any way, so it's not a "bios issue". |
|
|
|
|
Do NOT impliment Claymore ZEC v9 in Nicehash by default - it has major crashyness and HIGH REJECTS issues for quite a few of us.
IMO let folks copy it WHEN it works for them.
|
|
|
|
What area are you in?
i'm in south east asia. I'm supprised it's dry enough where you're at for evaporative effects to be noticeable at all. I'm mostly all-in on ETH now - my dual 280x rig is still ZEC 'cause it's a tossup either way, and the older HD7750/HD78x0 cards with 2GB or less are ZEC 'cause they can't mine ETH. XMR is WAY far down on profitability on every comparison I'v ever done. THe only thing I have running XMR is my older XP-based gaming machines 'cause it's the only thing I CAN run on the GTX 750 Ti in that machine - but the pools I've tried seem to only like giving me about half of my hashrate no matter what I try, and the payout thresholds are crazy high. Claymore v9 seems to be VERY iffy - everything I've tried it on is NOTHING BUT REJECTS when it runs at all, and quite a few other folks reporting it don't work well if at all. |
|
|
|
|
I'm not impressed with that 16.12.1 driver even when it does work - did some extensive testing and I find it loses hashrate compared to the 15.12 drivers on my older hardware.
I'm not exactly shocked it doesn't work well with new gear too.
|
|
|
|
|
Try the 15.12 drivers on your R9 380x - though if it's a 2GB card you may be out of luck as the DAG file is hitting the range that "DAG file + misc other needed stuff" won't fit on 2GB cards any more.
Someone suggested pairing 2GB cards up with a 3GB+ card as primary still lets the 2GB cards mine for now.
2GB cards work find for mining ZEC, though the only decent reputable software for ZEC with competitive hashrates is Windows based (Claymore for AMD, NiceHash eqm for NVidia).
|
|
|
|
They not accepting any orders for R4 ATM. and no wire for current limited batch of S9.
Who knows what will be at next sale of any miner.
There are still S7-LN and S5 units available - from NewEgg. |
|
|
|
There are new drivers, called Radeon ReLive. Anyone tested it with Claymore?
I lose 1-2% on hashrate vs 15.12 with the win7-64bit-radeon-software-crimson-relive-16.12.1-dec7 drivers under Win7 on my HD7870 I tested them on. They might work better on newer cards - AMD doesn't seem to test their new drivers on anything older than the R9 3xx series since 15.12 or so. |
|
|
|
Which is more profitable now ZEC or ETH.?
ETH has been more profitable for a while now Depends on the rig and your electric cost. You have to crunch YOUR OWN numbers - in my case, some of my cards are more profitable on ZEC most are more profitable on ETH the last week. The lower end cards have to do ZEC 'cause they can't do ETH any more (my HD 7750s never could do ETH with 1 GB ram). |
|
|
|
|
Did a bit more testing on my HIS Ice-Q 2GB HD 7870 single-card rig.
X240 dual core CPU, 2 Gig RAM, Win 7, Catalyst 15.12, mining on flypool.
Machine has been stable running V8 at about 110 H/s for a few days straight.
Switch to V9 (no other changes):
-i 6 miner reports "too high intensity" and reverts to -i 5, see below.
-i 5 miner stable, reports 96 give or take H/s, TONS OF REJECTS (bloody near 100% rejected).
-i 4 miner starts up, then fairly quickly crashes after never getting far enough to report hashrate or any shares. Watchdog then reboots miner, and repeat in a seemingly infinite cycle.
-i 3 miner starts up, reports 336 H/s (NO TYPO) but every single share is rejected.
-i 2 miner starts up, reports 96 H/s give or take, but tons of rejects (seems to be acting identically to -i 5)
didn't bother testing -i 1
[edit] Remembered seeing a comment about "16 GB virtual memory needed for multi-gpu machines" and decided to try bumping that up from 2GB.
-i 6 miner stable, reports 89 H/s, 100% rejects.
-i 5 miner stable, reports 80 H/s, 100% rejects.
-i 4 miner starts up, reports 111 H/s, then goes into constant driver crash / watchdog recycle mode and never starts hashing again.
-i 3 miner stable, reports 330 H/s, 100% rejects.
-i 2 miner stable, reports 84-87 H/s, 100% rejects.
-i 1 miner stable, reports 0 H/s and never finds a share.
Switch back to v8
-i 6 miner reports too high intensity going to 5, stable, reports 110 more or less H/s and lots of accepted shares 0 rejects.
GPU-Z also reported some odd behavior on the clocks under the -i 4 settings on both runs with v9 - a short spike early, then a "ramp-up" 'till the driver crashes, then just stays at idle after that.
[edit] I tried updating that machine to the current win7-64bit-radeon-software-crimson-relive-16.12.1-dec7 drivers.
Loses 1-2% hashrate and no other changes, on both V8 AND v9.
I'm starting to wonder if this v9 version requires a ton more system ram for some reason than V8, though it doesn't SEEM to use very much when I look at task manager memory usage.
It's definitely VERY PICKY about something, as I've seen quite a few other comments about it not working well from a few other folks.
|
|
|
|
|
I think I'm running the "3'd step down" TheStilt bios varient on them - haven't thought about it in months as they've just been working rock-solid.
I don't remember what the voltages were on each varient, but it was a fairly small incriment per step.
Did I remember to mention that they run 15 degrees C or so cooler despite being able to overclock to 1100 instead of thermal limiting at the stock 947 before the bios change?
|
|
|
|
running mining rigs makes the air in my mining room drier than the rest of the house, the water i spilled on the floor evaporated quickly, it is quicker to dry clothes in my mining room compared in drying them outside, and a portion of the room becomes cooler when clothes are being dried. I defrost stake in more or less 5 mins  ...those observations for me makes me think that swamp cooling here is feasible..i may enjoy the cooler climate now but i know what summer will bring to my mining room i have no way of measuring humidity here but based from my observations above, a constant pump of water to the cooling pads or columns of towels(custom made  ) and a fan would make a difference here. i haven't researched more about humidity but i think humidity or water in the air helps GPUs to transfer more heat to the air just like how water in the radiator helps transfer heat from the car engine to the radiator.. Humidity level of the air as such has insignificant effect on heat transfer from a heatsink to air. Your mining room "gets drier" because air can hold more moisture as it gets warmer, so the "relative humidity" drops even if the actual amount of moisture in the air per CC stays the same. drying clothes acts as an impromptu evaporative cooler. What area are you in? |
|
|
|
i thought it was useless that was a 7870 i wish i had kept to see how it would do ZEC mininng.
Ballpark 110 sol/s with Claymore v8 (v9 FOR ME in my testing IS BROKEN) - one HIS IceQ blower-type card, one Sapphire reference blower-type card. I also had issues with non-blower type HIS cards - only lost 1 card out of the 5 I bought but the FANS on all of them died in 3-6 months, cheap sleeve-type garbage. They used a good ballbearing blower on the Ice-Q HD 7870 I have from them. The living HIS fan-type cards now have leftover 60mm CPU ball-bearing fans twist-tied to them and run TONS cooler. I have never seen "compatability issues" with their cards, they just go way cheap on fans. I don't remember if the PowerColor cards I USED to have were ball bearing or sleeve - the half-broken HD 7850 survivor is ball-bearing but I think it was the last Power Color card I ever bought. My primary go-to brand is Sapphire, though I've had good luck overall with Gigabyte and EVGA as well. All 3 of those brands always use ball-bearing fans IME with them. Oddly enough, some of the EVGA 1070 cards I have are the "overheat and die" issue ones they had the recall on - but if you put a SANE fan profile in them they have no issue. The only REAL issue with those cards is they aimed TOO FAR towards "quiet" at the cost of "cool" in the default in-BIOS fan profile. |
|
|
|
|
Show Posts
