HD5850 @ 340Mhash/s - page 3. | Bitcointalksearch.org

Basiley

newbie

Activity: 42

Merit: 0

Quote from: chungenhung on May 21, 2011, 11:44:48 PM

I thought you can't volt mode sapphire 5850 via BIOS, as it is not reference design.
I believe the risk of bricking the is very high.

sure. thats why RBE creators[one more or less-working modern Radeon firmware editors] note that in howtows.
someone, friendly enough with hardware/iron and have good hands/cold brain/less power-greedy/hungry, can try hardware voltmod instead, or ask someone[for BitCoins, obviously ;-]on hardware enthusiasts/overclockers forums/web-sites do do it for you.

chungenhung

legendary

Activity: 1134

Merit: 1005

I thought you can't volt mode sapphire 5850 via BIOS, as it is not reference design.
I believe the risk of bricking the is very high.

foggyb

legendary

Activity: 1708

Merit: 1006

Quote from: chungenhung on May 21, 2011, 09:14:08 PM

Quote from: Kluge on May 21, 2011, 07:22:03 PM

Trick to getting around Trixx/Smartdoctor not seeing second card -- start off in CFX, overvolt the cards, then switch CFX off. Assuming you have dummy plugs in, the overvolt should still hold on both cards. From there, use MSI AB to clock them up to where you want.

(dunno for sure if that works on Trixx. Maybe if you go CFX, overvolt, close program but keep settings, then switch CFX and OC using MSI AB)

That is a good trick.
However, I have three cards, and so can't CFX all three cards (5850 only has one bridge on top).

I suppose the workaround is to test each card for best settings, then flash the bios with the desired volts/clocks.

Basiley

newbie

Activity: 42

Merit: 0

Quote from: chungenhung on May 21, 2011, 02:55:02 PM

Quote from: Basiley on May 21, 2011, 02:42:50 PM

no offense, but F@H sucks in both GPU saturation and computing itself.
but hell, there offense, really. F@H sucks !!!
*beep*&blinking led*
/pop-corn

Quote from: ElectroGeek007 on May 21, 2011, 02:36:10 PM

My reference 5850 is currently getting a stable 395 MH/s(!) at 1.20v, 1000 core, 180 memory, and 71 degrees. Grin

I am using Phoenix 1.46 and the phatk kernel.

you "forget" to add you GPU power system temperature[not monitored by hardware/software], at that clocks, usually heating up to 95-110'C degrees, or more. so im strongly suggest you to add at least one 120/140mm fan against you GPU on you <-- case cover and/or instal heatsing on you GPU power IC/VRM.
1.2 is serious. esp for reference-alike 3+1, 4+1 power-subsystem.
IMO its safer stay beyond 900, even if its cost you about 50 MH/s.

Then that means you are probably just plain ignorant.
F@H and Furmark is considered to push GPUs to their highest limits.
You can run 3D Mark all day or any games, and your GPUs still won't be as taxing.
In fact, my system pulls 350W with mining at aggression=13, when I fire up F@H or Furmark, it sucks in 30W extra. The temperature also goes up by 10C higher than mining.

wrong! again !!
in not "just plain ignorant", im offensive plaing ignorant Tongue

*whoosh !!!* *f-@#%@/useless F&H application/network defeated, anyone happily dances*
note: GPU saturation!=GPU utilisation.
thats why so many "multi-threaded" products, actually wast CPU's cores cycles, instead of profiting from i in CPU-Wolrd and thats why you can't call [for example]FurMark "world best computation application !"
for thats, you need something, actually can/able seriously/very efficently use farmers resources to something useful. which isn't F@H-related project[and affilated forks]case, sloooooowly underdevelopen last years[w/o serious advance ~last 2.5 years]

chungenhung

legendary

Activity: 1134

Merit: 1005

Quote from: Kluge on May 21, 2011, 07:22:03 PM

ASUS cards, I just learned, can be overvolted easily with ASUS SmartDoctor. Bumped up from 1.168v to 1.224v and was able to get to 990MHz core still at temps under 80*c. Now @ >380MH/s per card using phatk on phoenix.

Trick to getting around Trixx/Smartdoctor not seeing second card -- start off in CFX, overvolt the cards, then switch CFX off. Assuming you have dummy plugs in, the overvolt should still hold on both cards. From there, use MSI AB to clock them up to where you want.

(dunno for sure if that works on Trixx. Maybe if you go CFX, overvolt, close program but keep settings, then switch CFX and OC using MSI AB)

That is a good trick.
However, I have three cards, and so can't CFX all three cards (5850 only has one bridge on top).
If there is no workaround, I will just stick with the highest OC I can get from default voltage.

Saruto

newbie

Activity: 2

Merit: 0

I have a non-reference (not Sapphire though) 5850 and I was actually able to adjust the voltage in Trixx. Got it stable to 1.225 and I was trying to OC to 1000/1300 but I got a hard lock. Restarted and suddenly I lost voltage control in Trixx. Without voltage control the furthest I can OC my card is 860 core and around 1190 memory (underclocking the memory does nothing for my hashrate)...I wonder why I can't adjust it in Trixx anymore, the VRM's aren't blown out otherwise the card wouldn't run...I think it might be some kind of software issue.

Ah well, right now I have an acceptable hashrate of 325 MHash/s with the card running at around 66C on load. If it's a hot day it gets up to 70C. All in all, decent, but I could have definitely broken 400 MHash/s with a 1Ghz core.

frankiebits

full member

Activity: 155

Merit: 100

Cypress 1 5870
cypress 2 5850

Kluge

donator

Activity: 1218

Merit: 1015

ASUS cards, I just learned, can be overvolted easily with ASUS SmartDoctor. Bumped up from 1.168v to 1.224v and was able to get to 990MHz core still at temps under 80*c. Now @ >380MH/s per card using phatk on phoenix.

Trick to getting around Trixx/Smartdoctor not seeing second card -- start off in CFX, overvolt the cards, then switch CFX off. Assuming you have dummy plugs in, the overvolt should still hold on both cards. From there, use MSI AB to clock them up to where you want.

(dunno for sure if that works on Trixx. Maybe if you go CFX, overvolt, close program but keep settings, then switch CFX and OC using MSI AB)

chungenhung

legendary

Activity: 1134

Merit: 1005

Quote from: Jokah on May 21, 2011, 05:33:33 PM

This is excellent news. Looks like I might be purchasing two of these then.

Theres only one thing holding me back. I read that trixx will not allow you to adjust the voltage on each card individually. They have to be in cross fire or trixx will only recognize the top gpu.

Can anyone confirm this or has anyone been able to do them individually?

I ask because I have an asus z68 pro on the way and I will be running these cards in the bottom two slots. Problem is the second slot runs at either x16 or x8 speed and the bottom one runs at x4 speed so I will not be able to crossfire them. I dont think I will anyway?

I have 3 of them, and Trixx only adjust clock and voltage on the first card.
No, I don't have them in Crossfire.
While MSI AfterBurner can adjust all 3 cards individually, but can't adjust voltage.
So, really not that great.

Jokah

member

Activity: 69

Merit: 10

This is excellent news. Looks like I might be purchasing two of these then.

Theres only one thing holding me back. I read that trixx will not allow you to adjust the voltage on each card individually. They have to be in cross fire or trixx will only recognize the top gpu.

Can anyone confirm this or has anyone been able to do them individually?

I ask because I have an asus z68 pro on the way and I will be running these cards in the bottom two slots. Problem is the second slot runs at either x16 or x8 speed and the bottom one runs at x4 speed so I will not be able to crossfire them. I dont think I will anyway?

cyberfart

newbie

Activity: 7

Merit: 0

Quote from: Jokah on May 21, 2011, 04:50:51 PM

I'm looking to buy a couple of non reference sapphire 5850 xtremes.

I read in this thread, and it is my understanding, that you cannot adjust the voltage on non reference cards. Yet some people in here claim to have adjusted the voltage with the latest version of trixx.

I don't know what to believe because you read one post that says you can then you read another post that says you cant (this seems to repeat itself over and over as you go through this thread).

Can anyone tell me for sure if you can or cannot adjust the voltage on one of these 5850's http://www.yoyotech.co.uk/item-detail.php?products_id=4373117

I am running that particular card overvoltaged and overclocked.

TETOCONPOLERA

member

Activity: 118

Merit: 10

Quote from: Jokah on May 21, 2011, 04:50:51 PM

I'm looking to buy a couple of non reference sapphire 5850 xtremes.

I read in this thread, and it is my understanding, that you cannot adjust the voltage on non reference cards. Yet some people in here claim to have adjusted the voltage with the latest version of trixx.

I don't know what to believe because you read one post that says you can then you read another post that says you cant (this seems to repeat itself over and over as you go through this thread).

Can anyone tell me for sure if you can or cannot adjust the voltage on one of these 5850's http://www.yoyotech.co.uk/item-detail.php?products_id=4373117

of course you can, downloading latest Trixx version

but it's imposible with Afterburner

Jokah

member

Activity: 69

Merit: 10

I'm looking to buy a couple of non reference sapphire 5850 xtremes.

I read in this thread, and it is my understanding, that you cannot adjust the voltage on non reference cards. Yet some people in here claim to have adjusted the voltage with the latest version of trixx.

I don't know what to believe because you read one post that says you can then you read another post that says you cant (this seems to repeat itself over and over as you go through this thread).

Can anyone tell me for sure if you can or cannot adjust the voltage on one of these 5850's http://www.yoyotech.co.uk/item-detail.php?products_id=4373117

Basiley

newbie

Activity: 42

Merit: 0

at 86% stock GPU fan not only noisier, but live not so long. try help it as said above by improving airflow in case or investing into GPU watercooling.

cyberfart

newbie

Activity: 7

Merit: 0

Sapphire 5850:

920Mhz/320Mhz/1225mv
Phoenix 1.47 with phatk
Stable 365Mhash/s while using the computer. (82C, 86% Fan)

Basiley

newbie

Activity: 42

Merit: 0

Quote from: ElectroGeek007 on May 21, 2011, 02:53:06 PM

Quote from: Basiley on May 21, 2011, 02:42:50 PM

Quote from: ElectroGeek007 on May 21, 2011, 02:36:10 PM

My reference 5850 is currently getting a stable 395 MH/s(!) at 1.20v, 1000 core, 180 memory, and 71 degrees. Grin

I am using Phoenix 1.46 and the phatk kernel.

you "forget" to add you GPU power system temperature[not monitored by hardware/software], at that clocks, usually heating up to 95-110'C degrees, or more. so im strongly suggest you to add at least one 120/140mm fan against you GPU on you <-- case cover and/or instal heatsing on you GPU power IC/VRM.
1.2 is serious. esp for reference-alike 3+1, 4+1 power-subsystem.
IMO its safer stay beyond 900, even if its cost you about 50 MH/s.

The vram (I'm assuming that's what it is, it's the second sensor that shows up in HWMonitor) stays just under 100C, it gets too hot at any more than 1.2v.

VRM designed[more or less] for high-temp conditions[up to 120 or about], but PCB and nearby100'C is is seriously affected, by both PCB nature and airflow around.
try add/replace you case fans with something more powerful at least[usually rated in CFM(ie Cubic feets per minute)].

chungenhung

legendary

Activity: 1134

Merit: 1005

Quote from: Basiley on May 21, 2011, 02:42:50 PM

no offense, but F@H sucks in both GPU saturation and computing itself.
but hell, there offense, really. F@H sucks !!!
*beep*&blinking led*
/pop-corn

Quote from: ElectroGeek007 on May 21, 2011, 02:36:10 PM

My reference 5850 is currently getting a stable 395 MH/s(!) at 1.20v, 1000 core, 180 memory, and 71 degrees. Grin

I am using Phoenix 1.46 and the phatk kernel.

you "forget" to add you GPU power system temperature[not monitored by hardware/software], at that clocks, usually heating up to 95-110'C degrees, or more. so im strongly suggest you to add at least one 120/140mm fan against you GPU on you <-- case cover and/or instal heatsing on you GPU power IC/VRM.
1.2 is serious. esp for reference-alike 3+1, 4+1 power-subsystem.
IMO its safer stay beyond 900, even if its cost you about 50 MH/s.

Then that means you are probably just plain ignorant.
F@H and Furmark is considered to push GPUs to their highest limits.
You can run 3D Mark all day or any games, and your GPUs still won't be as taxing.
In fact, my system pulls 350W with mining at aggression=13, when I fire up F@H or Furmark, it sucks in 30W extra. The temperature also goes up by 10C higher than mining.

ElectroGeek007

member

Activity: 79

Merit: 10

Quote from: Basiley on May 21, 2011, 02:42:50 PM

Quote from: ElectroGeek007 on May 21, 2011, 02:36:10 PM

My reference 5850 is currently getting a stable 395 MH/s(!) at 1.20v, 1000 core, 180 memory, and 71 degrees. Grin

I am using Phoenix 1.46 and the phatk kernel.

you "forget" to add you GPU power system temperature[not monitored by hardware/software], at that clocks, usually heating up to 95-110'C degrees, or more. so im strongly suggest you to add at least one 120/140mm fan against you GPU on you <-- case cover and/or instal heatsing on you GPU power IC/VRM.
1.2 is serious. esp for reference-alike 3+1, 4+1 power-subsystem.
IMO its safer stay beyond 900, even if its cost you about 50 MH/s.

The vram (I'm assuming that's what it is, it's the second sensor that shows up in HWMonitor) stays just under 100C, it gets too hot at any more than 1.2v.

Basiley

newbie

Activity: 42

Merit: 0

no offense, but F@H sucks in both GPU saturation and computing itself.
but hell, there offense, really. F@H sucks !!!
*beep*&blinking led*
/pop-corn

Quote from: ElectroGeek007 on May 21, 2011, 02:36:10 PM

My reference 5850 is currently getting a stable 395 MH/s(!) at 1.20v, 1000 core, 180 memory, and 71 degrees. Grin

I am using Phoenix 1.46 and the phatk kernel.

you "forget" to add you GPU power system temperature[not monitored by hardware/software], at that clocks, usually heating up to 95-110'C degrees, or more. so im strongly suggest you to add at least one 120/140mm fan against you GPU on you <-- case cover and/or instal heatsing on you GPU power IC/VRM.
1.2 is serious. esp for reference-alike 3+1, 4+1 power-subsystem.
IMO its safer stay beyond 900, even if its cost you about 50 MH/s.

drgr33n

sr. member

Activity: 308

Merit: 251

Quote from: chungenhung on May 21, 2011, 01:42:48 PM

Quote from: Basiley on May 21, 2011, 01:12:12 PM

Quote from: chungenhung on May 21, 2011, 01:09:40 PM

Quote from: Basiley on May 21, 2011, 06:48:18 AM

75'C is too much, IMO for 24/7 load.
try keep it under 70'C or preferably under 65'C :[
PC cases with 3-4-5 120mm/140mm mounts inexpensive and start cost from $30. so put some investment there !! 30+10x4~$70 for 10-15-20 temp drop isn't that much ? :-|

Wrong.
Modern GPUs can handle minimum of 105C.
I ran 9800GX2 before, and it runs 105-110C. Been running 24/7 for over a year, and no problem at all.

"can't handle"!="work efficient/long/reliable" and thats true even for Nvidia junk.
even for gaming workload, let alone 24/7 95% GPU saturation.

Unless you have something to back that up, otherwise you are just spreading false rumors.
I own and run all my nVidia GPUs at 99-100% usage (as measured by GPU-Z). It runs folding @ home projects.
I have used 8800GS, 8800GT, 9800GT, 9800GTX, 9800GX2, 9600GSO, GTX295, GT240, GTX 260, GTX 460.
All high end cards have reached 90C and above, and no problems at all. And all running 24/7 for over a year.
Just go to folding@home forums and you will see tons of people running at such high temperatures.
How hot do you think tesla clusters reach when they are packed together in a server rack? I would tell you well over 90C if not 100C+.

if your hitting 100 / 105c then your card will be probably going into throttled mode.

Topic: HD5850 @ 340Mhash/s - page 3. (Read 43272 times)