Bitcoin Forum
May 27, 2018, 02:11:09 AM *
News: Latest stable version of Bitcoin Core: 0.16.0  [Torrent]. (New!)
 
   Home   Help Search Donate Login Register  
Pages: 1 2 [All]
  Print  
Author Topic: How should I cool my 5970 VRMs?  (Read 3178 times)
humanage
Member
**
Offline Offline

Activity: 64
Merit: 10



View Profile
February 15, 2012, 02:00:56 PM
 #1

I'm running my 5970 with an Accelero Extreme triple fan cooler, but the VRM temps are VERY high! I've read that this is a common 5970 bottleneck, and I'm trying to lower the temps. I'm now running the card at 1.111v, 910MHz Core/300MHz Memory, and getting 113c,117c,and 115c on the VRMs.
Here's what I've done so far:


I have an 80mm 60 CFM fan blowing air onto the first set of VRMs, and a Scythe fan(40 or so CFM) directing air in the direction. I have a 92mm blue gelid fan blowing onto the other set of VRMs. Finally, I have a Cooler Master fan blowing into the side of the GPU. Any other fan positions I should try? Should I maybe buy some 100+ CFM fans?
1527387069
Hero Member
*
Offline Offline

Posts: 1527387069

View Profile Personal Message (Offline)

Ignore
1527387069
Reply with quote  #2

1527387069
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
1527387069
Hero Member
*
Offline Offline

Posts: 1527387069

View Profile Personal Message (Offline)

Ignore
1527387069
Reply with quote  #2

1527387069
Report to moderator
1527387069
Hero Member
*
Offline Offline

Posts: 1527387069

View Profile Personal Message (Offline)

Ignore
1527387069
Reply with quote  #2

1527387069
Report to moderator
1527387069
Hero Member
*
Offline Offline

Posts: 1527387069

View Profile Personal Message (Offline)

Ignore
1527387069
Reply with quote  #2

1527387069
Report to moderator
malevolent
can into space
Staff
Legendary
*
Offline Offline

Activity: 2030
Merit: 1013



View Profile
February 15, 2012, 08:29:20 PM
 #2


What are vrm temps at stock voltage and core?
cablepair
Hero Member
*****
Offline Offline

Activity: 854
Merit: 1000


These days, i am Crazy.


View Profile WWW
February 15, 2012, 09:12:58 PM
 #3

i think 910 is a bit high and excessive

I can get 380 mhash per core at 825
humanage
Member
**
Offline Offline

Activity: 64
Merit: 10



View Profile
February 15, 2012, 11:47:49 PM
 #4

i think 910 is a bit high and excessive

I can get 380 mhash per core at 825

I'm getting 410MH/s on 910MHz.
MiningBuddy
Moderator
Hero Member
*
Offline Offline

Activity: 926
Merit: 1000


฿itcoin ฿itcoin ฿itcoin


View Profile
February 16, 2012, 12:02:12 AM
 #5

What kind of core temps are we looking at?

OgNasty
Donator
Legendary
*
Offline Offline

Activity: 2562
Merit: 1262


Quoth the Raven "Nevermore."


View Profile WWW
February 16, 2012, 12:06:22 AM
 #6

i think 910 is a bit high and excessive

I can get 380 mhash per core at 825

I'm getting 410MH/s on 910MHz.

I can assure you that those extra 30MH/s aren't worth the amount of additional electricity you're buying to run it at those clock speeds.  I can also assure you that at those clock speeds and temperatures, you'll be lucky if those cards are still working a month from now.  So basically, you're destroying your cards and overpaying for electricity for no reason other than turning your room into a sauna.  Turn down the clock speeds or get a water cooling setup.  What you're doing now is insane.

          ▄█████▄
        ▄█████████▄
      ▄████▀   ▀████▄
    ▄████▀   ▄ ▄█▀████▄
  ▄████▀   ▄███▀   ▀████▄
▄████▀   ▄███▀   ▄   ▀████▄
█████   ███▀   ▄███   █████
▀████▄   ▀██▄▄███▀   ▄████▀
  ▀████▄   ▀███▀   ▄████▀
    ▀████▄       ▄████▀
      ▀████▄   ▄████▀
        ▀███  ████▀
          ▀█▄███▀
.
|
.
|
          ▄█████▄
        ▄█████████▄
      ▄████▀   ▀████▄
    ▄████▀   ▄ ▄█▀████▄
  ▄████▀   ▄███▀   ▀████▄
▄████▀   ▄███▀   ▄   ▀████▄
█████   ███▀   ▄███   █████
▀████▄   ▀██▄▄███▀   ▄████▀
  ▀████▄   ▀███▀   ▄████▀
    ▀████▄       ▄████▀
      ▀████▄   ▄████▀
        ▀███  ████▀
          ▀█▄███▀
unthy
P4man
Hero Member
*****
Offline Offline

Activity: 518
Merit: 500



View Profile
February 16, 2012, 12:28:30 AM
 #7


I can assure you that those extra 30MH/s aren't worth the amount of additional electricity you're buying to run it at those clock speeds. 

Higher clocks at equal voltage == better energy efficiency.

@OP: how are you cooling the VRMs? With those small sticky alu fins?
I dont know what the 5970 ref cooler looks like exactly, I know its different, but with a 5870 I had great success using the original baseplate to cool VRMs and ram while using an accelero to keep the GPU cool. The supplied sinks were just not good enough for the VRMs.

Mind you, I did have to dremel the baseplate to clear the accelero heatpipes, but if that doesnt scare you, have a look at the stock cooler and see if it has a baseplate you can use for that purpose. This might be useless advice if the 5970 has a single plate cooling GPUs and VRMs..

If that doesnt work, its time to undervolt the cards and lower your clocks.

jjiimm_64
Legendary
*
Offline Offline

Activity: 1876
Merit: 1000


View Profile
February 16, 2012, 01:50:59 AM
 #8

i think 910 is a bit high and excessive

I can get 380 mhash per core at 825

I'm getting 410MH/s on 910MHz.

I never run my 5970's over 820.  most are at 800.  Mine will still be hashing in a year!

The only reason I restarted this rig on the 4th was to upgrade cgminer to the better rpc-api
Code:
cgminer version 2.2.1 - Started: [February 4, 2012, 4:33 pm]    Rig: miner9
(5s):2982.13  (avg): 2966.26 Mh/s  |    Q:690271   A:651964   R:6233   HW:0   E:?%   U:40.52/m
TQ:?   ST:722   SS:?   DW:17127   NB:1600   LW:25420   GF:601   RF:655
Connected to http://gpumax.com:8332 with LP as user ?
Value:
GPU 0: 59.0C 3285RPM 53% | 372.7/370.8Mh/s | 98% | 820Mhz 300Mhz 1.05V A:81531 R:760 HW:0 U:5.07/m I: 7
GPU 1: 61.5C 3285RPM 53% | 372.9/370.8Mh/s | 98% | 820Mhz 300Mhz 1.05V A:81809 R:733 HW:0 U:5.08/m I: 7
GPU 2: 66.5C 3045RPM 50% | 372.8/370.8Mh/s | 98% | 820Mhz 300Mhz 1.05V A:81030 R:751 HW:0 U:5.04/m I: 7
GPU 3: 63.5C 3045RPM 50% | 372.8/370.9Mh/s | 98% | 820Mhz 300Mhz 1.05V A:81674 R:801 HW:0 U:5.08/m I: 7
GPU 4: 63.5C 3125RPM 50% | 372.7/370.8Mh/s | 98% | 820Mhz 300Mhz 1.05V A:81576 R:769 HW:0 U:5.07/m I: 7
GPU 5: 60.5C 3125RPM 50% | 372.7/370.8Mh/s | 98% | 820Mhz 300Mhz 1.05V A:81537 R:802 HW:0 U:5.07/m I: 7
GPU 6: 57.0C 2987RPM 49% | 372.8/370.7Mh/s | 99% | 820Mhz 300Mhz 1.05V A:81292 R:836 HW:0 U:5.05/m I: 7
GPU 7: 59.0C 2987RPM 49% | 372.8/370.8Mh/s | 98% | 820Mhz 300Mhz 1.05V A:81515 R:781 HW:0 U:5.07/m I: 7




even at 810 they get 370, this one is in a warmer spot, hence the higher temps.

Code:
cgminer version 2.2.1 - Started: [February 4, 2012, 4:21 pm]    Rig: miner10
(5s):2967.27  (avg): 2934.61 Mh/s  |    Q:684834   A:646490   R:6218   HW:0   E:?%   U:40.15/m
TQ:?   ST:709   SS:?   DW:18067   NB:1602   LW:25981   GF:681   RF:680
Connected to http://gpumax.com:8332 with LP as user ?
Value:
GPU 0: 72.5C 3081RPM 50% | 371.0/357.9Mh/s | 99% | 810Mhz 300Mhz 1.05V A:79198 R:707 HW:0 U:4.92/m I: 7
GPU 1: 69.0C 3081RPM 50% | 370.9/368.1Mh/s | 99% | 810Mhz 300Mhz 1.05V A:80552 R:751 HW:0 U:5.00/m I: 7
GPU 2: 73.5C 2949RPM 48% | 370.9/368.1Mh/s | 99% | 810Mhz 300Mhz 1.05V A:81016 R:808 HW:0 U:5.03/m I: 7
GPU 3: 73.0C 2947RPM 48% | 370.9/368.1Mh/s | 99% | 810Mhz 300Mhz 1.05V A:81012 R:723 HW:0 U:5.03/m I: 7
GPU 4: 72.5C 2639RPM 44% | 370.9/368.1Mh/s | 99% | 810Mhz 300Mhz 1.05V A:81048 R:815 HW:0 U:5.03/m I: 7
GPU 5: 73.0C 2639RPM 44% | 370.9/368.1Mh/s | 99% | 810Mhz 300Mhz 1.05V A:81227 R:757 HW:0 U:5.04/m I: 7
GPU 6: 73.5C 2747RPM 49% | 370.8/368.0Mh/s | 99% | 810Mhz 300Mhz 1.05V A:80989 R:810 HW:0 U:5.03/m I: 7
GPU 7: 68.5C 2749RPM 49% | 371.0/368.1Mh/s | 99% | 810Mhz 300Mhz 1.05V A:81448 R:847 HW:0 U:5.06/m I: 7
.

1jimbitm6hAKTjKX4qurCNQubbnk2YsFw
humanage
Member
**
Offline Offline

Activity: 64
Merit: 10



View Profile
February 16, 2012, 03:55:14 AM
 #9

Alright, got down to 800Mhz Core/ 300MHz memory. I'm getting 49c and 47c on the GPU cores, so that's not a problem.
The VRMs are at 82c/81c//81c for the first core, and 74c/75c/73c on the 2nd core. Lets say I don't raise the voltage above stock, would a high clock speed damage/degrade the card?

ssateneth
Legendary
*
Offline Offline

Activity: 1330
Merit: 1004



View Profile
February 16, 2012, 06:37:04 AM
 #10

Alright, got down to 800Mhz Core/ 300MHz memory. I'm getting 49c and 47c on the GPU cores, so that's not a problem.
The VRMs are at 82c/81c//81c for the first core, and 74c/75c/73c on the 2nd core. Lets say I don't raise the voltage above stock, would a high clock speed damage/degrade the card?



Nope, high core speeds won't kill a card.

Temperatures are really the only thing that'll kill a card if you overclock smart. Most overvolting for mining is minor (maybe +0.025 to 0.075 in even the most extreme cases) and won't hurt the card either, anything above is generally overkill for 24/7 and generates too much heat to remain stable at the clock speed you may have aimed at. An example is one of my 5870's is stable at 1100 core and 1.200 volts with temperatures currently around 40c core and 68c VRM since it sits outside in the winter air. It requires more volts to hit a higher clock speed than 1100, but the higher temperature renders the higher speed unstable. I see a lot of gamers use even higher volts of 1.25+, probably because the cards aren't at 100% load with the games they're playing and temperatures are lower as a consequence.

ace24
Newbie
*
Offline Offline

Activity: 20
Merit: 0


View Profile
February 16, 2012, 07:02:12 AM
 #11

Higher clocks at equal voltage == better energy efficiency.

Well, no. Voltage is only one part of power usage, higher clocked cards use more current than lower clocked cards at the same voltage. That's why overclocked cards get hotter even when not overvolted.
ssateneth
Legendary
*
Offline Offline

Activity: 1330
Merit: 1004



View Profile
February 16, 2012, 07:16:47 AM
 #12

Higher clocks at equal voltage == better energy efficiency.

Well, no. Voltage is only one part of power usage, higher clocked cards use more current than lower clocked cards at the same voltage. That's why overclocked cards get hotter even when not overvolted.

No, he's right, when taking the whole computer into the power efficiency picture. 10% more core frequency = 10% more power used (and generally mhash scales linearly with core speed), but the system has a baseline watts, so do some math with some assumed numbers...

100w baseline, 1000w used from video cards doing 2000 mhash. 1.8181 mhash/watt.
Increase core speed by 10% without increasing volts, system is now at...
100w baseline, 1100w used from video cards doing 2200mhash. 1.8333 mhash/watt.

P4man
Hero Member
*****
Offline Offline

Activity: 518
Merit: 500



View Profile
February 16, 2012, 07:18:55 AM
 #13

Well, no. Voltage is only one part of power usage, higher clocked cards use more current than lower clocked cards at the same voltage. That's why overclocked cards get hotter even when not overvolted.

Yes, power consumption scales linearly with clockspeed. But performance also scales linearly with clockspeed.
Higher performance at equal MH/W for the cards, means higher efficiency for the overall rig since CPU, MB, RAM, etc remain the same.

Power consumption scales quadratically with voltage. Clockspeed -and  thus performance- does not. Therefore undervolting almost always gives better MH/W, but overclocking is completely sane from an efficiency POV.

BOARBEAR
Member
**
Offline Offline

Activity: 77
Merit: 10


View Profile
February 16, 2012, 07:40:09 AM
 #14

Well, no. Voltage is only one part of power usage, higher clocked cards use more current than lower clocked cards at the same voltage. That's why overclocked cards get hotter even when not overvolted.

Yes, power consumption scales linearly with clockspeed. But performance also scales linearly with clockspeed.
Higher performance at equal MH/W for the cards, means higher efficiency for the overall rig since CPU, MB, RAM, etc remain the same.

Power consumption scales quadratically with voltage. Clockspeed -and  thus performance- does not. Therefore undervolting almost always gives better MH/W, but overclocking is completely sane from an efficiency POV.
Performance does not scale linearly with clock speed at all.  (try 500mhz and 800mhz and you will see)
P4man
Hero Member
*****
Offline Offline

Activity: 518
Merit: 500



View Profile
February 16, 2012, 08:14:58 AM
 #15

Performance does not scale linearly with clock speed at all.  (try 500mhz and 800mhz and you will see)

Of course it does. Unless perhaps if you are using SDK 2.6.

humanage
Member
**
Offline Offline

Activity: 64
Merit: 10



View Profile
February 16, 2012, 12:57:41 PM
 #16

Quote
I can assure you that those extra 30MH/s aren't worth the amount of additional electricity you're buying to run it at those clock speeds.  I can also assure you that at those clock speeds and temperatures, you'll be lucky if those cards are still working a month from now.  So basically, you're destroying your cards and overpaying for electricity for no reason other than turning your room into a sauna.  Turn down the clock speeds or get a water cooling setup.  What you're doing now is insane.

I pay a fixed power bill, so I'm just trying to squeeze as much power out as I can. Tongue Also, I'm hoping to get 890MHz core on 1.112v. With sub-100c VRM temps, would this be okay?
jjiimm_64
Legendary
*
Offline Offline

Activity: 1876
Merit: 1000


View Profile
February 16, 2012, 04:25:03 PM
 #17

Quote
I can assure you that those extra 30MH/s aren't worth the amount of additional electricity you're buying to run it at those clock speeds.  I can also assure you that at those clock speeds and temperatures, you'll be lucky if those cards are still working a month from now.  So basically, you're destroying your cards and overpaying for electricity for no reason other than turning your room into a sauna.  Turn down the clock speeds or get a water cooling setup.  What you're doing now is insane.

I pay a fixed power bill, so I'm just trying to squeeze as much power out as I can. Tongue Also, I'm hoping to get 890MHz core on 1.112v. With sub-100c VRM temps, would this be okay?


maybe you dont realize what the 30Mh extra will get you....

assuming 5% growth of btc, using this calculator http://striketeam.ath.cx/btccalc/btccalc.php
usd exchange: 4.5

Month BTC    USD    Difficulty
1   0.63   2.835   1521061.30173
2   0.57   2.565   1676970.08516
3   0.51   2.295   1848859.51889
4   0.45   2.025   2140286.00056
5   0.4   1.8   2359665.31561
6   0.36   1.62   2601531.01046
7   0.32   1.44   3011597.33599
8   0.29   1.305   3320286.06293
9   0.26   1.17   3660615.38438
10   0.23   1.035   4237619.88434
11   0.21   0.945   4671975.92248
12   0.18   0.81   5150853.45454

 
Total    4.41   19.845      


so after a year of burning up your 5970 for an extra 30Mh, you will make 4 btcs.....if the card lasts a year

1jimbitm6hAKTjKX4qurCNQubbnk2YsFw
humanage
Member
**
Offline Offline

Activity: 64
Merit: 10



View Profile
February 16, 2012, 08:51:19 PM
 #18

Quote
maybe you dont realize what the 30Mh extra will get you....

assuming 5% growth of btc, using this calculator http://striketeam.ath.cx/btccalc/btccalc.php
usd exchange: 4.5

Month BTC    USD    Difficulty
1   0.63   2.835   1521061.30173
2   0.57   2.565   1676970.08516
3   0.51   2.295   1848859.51889
4   0.45   2.025   2140286.00056
5   0.4   1.8   2359665.31561
6   0.36   1.62   2601531.01046
7   0.32   1.44   3011597.33599
8   0.29   1.305   3320286.06293
9   0.26   1.17   3660615.38438
10   0.23   1.035   4237619.88434
11   0.21   0.945   4671975.92248
12   0.18   0.81   5150853.45454

 
Total    4.41   19.845      


so after a year of burning up your 5970 for an extra 30Mh, you will make 4 btcs.....if the card lasts a year
I see you point. I'll definitely stay on stock voltage then. Smiley
The-Real-Link
Hero Member
*****
Offline Offline

Activity: 511
Merit: 500


View Profile
February 16, 2012, 11:56:51 PM
 #19


  I'd also highly advise against tons of overvolting on them.  Like others said, the efficiency may increase at higher core clocks but the power use and heat increase *dramatically* with overvolting.  Tried mine at 900mhz only once.  Power use for two of the cards increased by over 300w on load, VRMs were in the 120s nearly constantly - I could tell as in AB you'd see the core clocks keep throttling back and forth.  Sure you'd get some extra MH but the effort simply isn't worth it IMO.

  Kept mine all around 800-820 range on stock volts and they stay working for the most part exceptionally well and temps are good.  VRMs will still be in the 80s/90s but that's just how it works.  They throttle at 125C though so it's not abnormal for them to be hotter than the cores.

Oh Loaded, who art up in Mt. Gox, hallowed be thy name!  Thy dollars rain, thy will be done, on BTCUSD.  Give us this day our daily 10% 30%, and forgive the bears, as we have bought their bitcoins.  And lead us into quadruple digits
check_status
Full Member
***
Offline Offline

Activity: 196
Merit: 100


Web Dev, Db Admin, Computer Technician


View Profile
February 17, 2012, 01:04:52 AM
 #20

This should be stickied for n00bs to Bitcoin mining, like me, who might attemp to apply gamers tweaks to their setup.
I haven't pulled my 5970 out of its box to run yet and this thread has already helped me out. Wink Sweet.

For Bitcoin to be a true global currency the value of BTC needs always to rise.
If BTC became the global currency & money supply = 100 Trillion then ⊅1.00 BTC = $4,761,904.76.
P2Pool Server List | How To's and Guides Mega List |  1EndfedSryGUZK9sPrdvxHntYzv2EBexGA
BOARBEAR
Member
**
Offline Offline

Activity: 77
Merit: 10


View Profile
February 17, 2012, 01:06:31 AM
 #21

Performance does not scale linearly with clock speed at all.  (try 500mhz and 800mhz and you will see)

Of course it does. Unless perhaps if you are using SDK 2.6.
A simply test shows it does not.

870Mhz  397Mhash/s 870/397=2.191
800Mhz 359Mhash/s 800/359=2.222
500Mhz 202Mhash/s 500/202=2.475

In theory it can't be linear, same reason why cpu does not scale linearly with core speed.
Don't spread false information
cuz0882
Sr. Member
****
Offline Offline

Activity: 392
Merit: 250


View Profile
February 17, 2012, 01:52:04 AM
 #22

870 works great for me, 5970's stay under 70c. I've only had one pc that could run 910 and stay cool but it was only a 4 gpu rig. If they don't go over 80c I would not worry. Nice to shoot for 70c though. I think some people forget, not everyone is concerned with power consumption ratios.

Did you lose your 5970's stock fan cooler? Those Accelero Extreme coolers are terrible for mining. There not made for steady high temps.
DeathAndTaxes
Donator
Legendary
*
Offline Offline

Activity: 1218
Merit: 1003


Gerald Davis


View Profile
February 17, 2012, 02:27:15 AM
 #23

Well, no. Voltage is only one part of power usage, higher clocked cards use more current than lower clocked cards at the same voltage. That's why overclocked cards get hotter even when not overvolted.

Yes, power consumption scales linearly with clockspeed. But performance also scales linearly with clockspeed.
Higher performance at equal MH/W for the cards, means higher efficiency for the overall rig since CPU, MB, RAM, etc remain the same.

Power consumption scales quadratically with voltage. Clockspeed -and  thus performance- does not. Therefore undervolting almost always gives better MH/W, but overclocking is completely sane from an efficiency POV.
Performance does not scale linearly with clock speed at all.  (try 500mhz and 800mhz and you will see)

I did and performance scaled linearly.  Not sure why you think it wouldn't.

A 5870/5970 gets roughly 0.46 MH (+/- .04) per Mhz.  That hold true from 500 Mhz to 1000 Mhz+.
Now for max efficiency you are going to need to find the sweet spot on memclock which varies based on coreclock.   If you see significantly less than that at any clock your settings are non-optimal or you sample is too small (should be at least 2000 shares, 5000 would be better). 
DeathAndTaxes
Donator
Legendary
*
Offline Offline

Activity: 1218
Merit: 1003


Gerald Davis


View Profile
February 17, 2012, 02:38:44 AM
 #24

A simply test shows it does not.

870Mhz  397Mhash/s 870/397=2.191
800Mhz 359Mhash/s 800/359=2.222
500Mhz 202Mhash/s 500/202=2.475

In theory it can't be linear, same reason why cpu does not scale linearly with core speed.
Don't spread false information

I agree YOU should not spread false information.

Your numbers are garbage so likely you have suboptimal settings and the most likely culprit is memclock.

Do those results seem plausible to you. 


870Mhz  397Mhash/s 397/870= 0.41 MH per Mhz
800Mhz 359Mhash/s 359/800 = 0.44 MH per Mhz
500Mhz 202Mhash/s 202/500 = 0.46 MH per Mhz

So the card is getting more efficient at higher clock (and likely higher temp)?  Does that seem plausible to you?

Likely you normally run @ 870Mhz and have found a more optimizes memclock.  You sloppily moved core clock without modifying memclock introducing timing delays making the card less effective the further you move it from 870 Mhz.

For the record @ 500 MHz I get 225 MH (0.450 MH/MHz) and @ 820 I get 375 MH/s (0.457 MH/MHz).
BOARBEAR
Member
**
Offline Offline

Activity: 77
Merit: 10


View Profile
February 17, 2012, 05:36:57 AM
 #25

A simply test shows it does not.

870Mhz  397Mhash/s 870/397=2.191
800Mhz 359Mhash/s 800/359=2.222
500Mhz 202Mhash/s 500/202=2.475

In theory it can't be linear, same reason why cpu does not scale linearly with core speed.
Don't spread false information

I agree YOU should not spread false information.

Your numbers are garbage so likely you have suboptimal settings and the most likely culprit is memclock.

Do those results seem plausible to you. 


870Mhz  397Mhash/s 397/870= 0.41 MH per Mhz
800Mhz 359Mhash/s 359/800 = 0.44 MH per Mhz
500Mhz 202Mhash/s 202/500 = 0.46 MH per Mhz

So the card is getting more efficient at higher clock (and likely higher temp)?  Does that seem plausible to you?

Likely you normally run @ 870Mhz and have found a more optimizes memclock.  You sloppily moved core clock without modifying memclock introducing timing delays making the card less effective the further you move it from 870 Mhz.

For the record @ 500 MHz I get 225 MH (0.450 MH/MHz) and @ 820 I get 375 MH/s (0.457 MH/MHz).


Your calculation is garbage.
202/500=0.404 not 0.46
397/870=0.45 not 0.41
These are just the multiplicative inverse of the number I gave above.
And I tested all above in identical situation except of the core frequency.
DeathAndTaxes
Donator
Legendary
*
Offline Offline

Activity: 1218
Merit: 1003


Gerald Davis


View Profile
February 17, 2012, 05:48:13 AM
 #26

And I tested all above in identical situation except of the core frequency.

Oops on me reversing the order of the values.  Now that WAS foolish on my part. Smiley

I guess it still hasn't sunk in that the optimal memory clock at 500Mhz isn't the same as the optimal memory clock at 870 Mhz.  All you have proven if you are really bad at getting optimal performance from a 5870/5970 as your efficiency (in terms of MH/s per MHz) is about 10% lower than most miners get.

In case you ignored it from the prior post
Quote
"For the record @ 500 MHz I get 225 MH (0.450 MH/MHz) and @ 820 I get 375 MH/s (0.457 MH/MHz)."

and lastly (from prior post) this this stands:

Quote
A 5870/5970 with optimal settings gets roughly 0.46 MH (+/- ~.04) per Mhz.  That hold true from 500 Mhz to 1000 Mhz+.
Now for max efficiency you are going to need to find the sweet spot on memclock which varies based on coreclock.   If you see significantly less than that at any clock your settings are non-optimal or you sample is too small (should be at least 2000 shares, 5000 would be better).
ssateneth
Legendary
*
Offline Offline

Activity: 1330
Merit: 1004



View Profile
February 17, 2012, 07:14:38 AM
 #27

And I tested all above in identical situation except of the core frequency.

Oops on me reversing the order of the values.  Now that was foolish. Smiley

Still I guess it still hasn't sunk in that the optimal memory clock at 500Mhz isn't the same as the optimal memory clock at 870 Mhz.  All you have proven if you are really bad and getting optimal performance from a 5870/5970 as you efficiency (MH per MHz) is about 10% lower than most miners get across the board.

Once again in case you ignored it from the prior post
"For the record @ 500 MHz I get 225 MH (0.450 MH/MHz) and @ 820 I get 375 MH/s (0.457 MH/MHz)."

and lastly

Quote
A 5870/5970 with optimal settings gets roughly 0.46 MH (+/- ~.04) per Mhz.  That hold true from 500 Mhz to 1000 Mhz+.
Now for max efficiency you are going to need to find the sweet spot on memclock which varies based on coreclock.   If you see significantly less than that at any clock your settings are non-optimal or you sample is too small (should be at least 2000 shares, 5000 would be better).

Thank you for laying the beatdown on this poor misguided soul. It was deserved.

P4man
Hero Member
*****
Offline Offline

Activity: 518
Merit: 500



View Profile
February 17, 2012, 07:16:21 AM
 #28

Performance does not scale linearly with clock speed at all.  (try 500mhz and 800mhz and you will see)

Of course it does. Unless perhaps if you are using SDK 2.6.
A simply test shows it does not.

870Mhz  397Mhash/s 870/397=2.191
800Mhz 359Mhash/s 800/359=2.222
500Mhz 202Mhash/s 500/202=2.475

In theory it can't be linear, same reason why cpu does not scale linearly with core speed.

Of course CPU bound cache resident benchmarks do scale linearly with clockspeed. Try benchmarking Coremark, its linear from 0.1 Hz to 1 GHz.

As for your results, like I said, unless perhaps you are using SDK 2.6. In that case, you become memory bandwidth bound. Hashrate will still scale linearly with gpu clock but only if you can scale memory clock as well.

If you use SDK 2.1 or 2.4,  hashrate is not bandwidth bound, in fact you typically get a small increase as you lower memory clocks down to ~150-200 MHz. Yes negative scaling. Check the link and try yourself. But the result is perfectly linear performance scaling with GPU clock.


johnyj
Legendary
*
Offline Offline

Activity: 1848
Merit: 1000


Beyond Imagination


View Profile
February 17, 2012, 03:45:10 PM
 #29

Stay stock voltage without too much overclock are important IMO

With stock voltage and decent overclock, machine can run several weeks without interrupt, but if it crashes from time to time because of the stability issues caused by high temp or high frequency, then the time spent trouble shooting and fixing things will easily cost you many BTCs

DeathAndTaxes
Donator
Legendary
*
Offline Offline

Activity: 1218
Merit: 1003


Gerald Davis


View Profile
February 17, 2012, 03:51:35 PM
 #30

Stay stock voltage without too much overclock are important IMO

With stock voltage and decent overclock, machine can run several weeks without interrupt, but if it crashes from time to time because of the stability issues caused by high temp or high frequency, then the time spent trouble shooting and fixing things will easily cost you many BTCs

Good point and as you get more cards and more rigs it makes more sense to become even more conservative.  That was something I learned the hard way.  You can't run 38 GPUs (19x 5970) at the same clocks you can run a single card that you can baby.  Well not without losing your sanity.  Not only are there stability issues but there are power and heat issues to which only get harder as the farm scales.  Dissipating 1KW of heat.  Meh.  Get a good fan.  Dissipating 5KW?  A little tougher. Smiley
The-Real-Link
Hero Member
*****
Offline Offline

Activity: 511
Merit: 500


View Profile
February 17, 2012, 09:13:52 PM
 #31

Stay stock voltage without too much overclock are important IMO

With stock voltage and decent overclock, machine can run several weeks without interrupt, but if it crashes from time to time because of the stability issues caused by high temp or high frequency, then the time spent trouble shooting and fixing things will easily cost you many BTCs

Good point and as you get more cards and more rigs it makes more sense to become even more conservative.  That was something I learned the hard way.  You can't run 38 GPUs (19x 5970) at the same clocks you can run a single card that you can baby.  Well not without losing your sanity.  Not only are there stability issues but there are power and heat issues to which only get harder as the farm scales.  Dissipating 1KW of heat.  Meh.  Get a good fan.  Dissipating 5KW?  A little tougher. Smiley

*drools*.  Yeah same.  I've kept my clocks reasonable and I'm still having to troubleshoot a good amount here and there.  Cards get buggy, network crashes, etc.  I'd much rather be slightly conservative but have the systems run for a week or more without a hitch than reset things every day.  When you have that many machines you'd think it'd be a given.

Oh Loaded, who art up in Mt. Gox, hallowed be thy name!  Thy dollars rain, thy will be done, on BTCUSD.  Give us this day our daily 10% 30%, and forgive the bears, as we have bought their bitcoins.  And lead us into quadruple digits
johnyj
Legendary
*
Offline Offline

Activity: 1848
Merit: 1000


Beyond Imagination


View Profile
February 21, 2012, 03:14:47 PM
 #32

Unfortunately even at stock clock and voltage, I still get some crash recently when running cgminer (never had issues when running phoenix miner), so a stable miner is also critical

It's also time to look at VRM cooling. I have some very efficient heat pad came with water blocks for GTX295 years ago, they could reduce the VRM temp by at least 10 degrees. As I understand, the VRM location on 5970 card are very weak, maybe a heatsink on the back of the plate will help

Pages: 1 2 [All]
  Print  
 
Jump to:  

Sponsored by , a Bitcoin-accepting VPN.
Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!