Bitcoin Forum

Hello,

So it appears Sapphire made a change to their boards recently, all new stock of the Dual-X is arriving with BIOS 15.041 - Which has a really poor hash rate - I can't push it past 700kh/s.

Dropping it to 15.039 BIOS and you can forget about anything over 500kh/s.

15.040 has the temp throttling issue at 74c, this problem doesn't appear to be the case with the 15.041.

So I'm stuck on 15.041 and can't get high enough hash - remember these Dual-X's before the new batch arrived would push 750kh/s easily!

Has anyone else had this issue? is there a fix?

Thanks

15.041 BIOS does have throttle problem

When does it trigger then?

I took it to 84C and was still happy at 690kh/s

I have one with the new bios (41) and I can just about get 730 (max) with 1050 core 1500 mem 1130 undervolt

I'm having the same issue...getting about 680 khs on my 280x dual-x with 041 bios...is there any other bios we can use? maybe another brand?

It's the vrm temps. 

Anything we can do about it?

Same card, same bios, same problem. Can't get above 708 Kh/s with this card and a engine clock of anything above 1050 lowers the hashrate.
Card gets 64C, but can't read the VRM temps as there are no VRM sensors anymore. Thinking of returning this card if there isn't a fix available.

Some more people seem to think the problem is the VRMs. This is probably a related post on a different forum:
source: https://litecointalk.org/index.php?topic=12369.1215

And probably another related topic:
https://litecointalk.org/index.php?topic=13554.0

I am now getting about 730 kh/s on mine, using TC 8193, gpu clock 1065, memory clock 1500. Staying at 72-73C at the moment.
Also lowered my screen res to 1024x768, and made sure what is displaying on the screen is static. As soon as something is moving (cgminer window, or whatever) the hashes go down to about 690. Yes, I know, crazy.

With those exact settings I get 717 Kh/s, so it's a bit better than before. First setting where a clock above 1050 isn't counterproductive.

Has anyone tried flashing a different bios? Maybe from the ones on this thread: https://litecointalk.org/index.php?topic=12369.225
I am a complete noob with flashing..maybe if someone gets good results I'd man up and do it   ;D

Not tried it myself, but see my post above. Someone tried some different roms, but there were no changes for this new sapphire revision.
Some actually give worse results.

I tryed different bioses, different configs, i have 10 of this fuking cards  :-\
I dont know why they done this.
Now i allways go to think before buying this cards.
Vapor-x now comes also with this dumb 15.041 bios and trottles.

No guarantee that new cards wount come with this bug. Thinking about migration to other brand.

Any solution to this problem? I have the same card and i can't get past 680 khs :/

Same problem here. Bought Sapphire Dual-X this week and cannot get over 650kH/s.
Tried flash bios a no sucess.

What I've uncovered through reading many forum posts is that Sapphire "cheaped" out on their recent batches - Their cards are more inline with the likes of PowerColor, XFX, MSI etc...

The Memory Chips are no longer the better of the Hynix pairs, but are now from Elipda. And also the BIOS memory timings heat up the VRM's too much that the card throttles.

There are many optimised Dual-X BIOS's out there - but you'll need to extract your own one and allow TheStilt to modify from the litecoin forum [Google him]

Memory timings are confidential, and he works on the inside - donations is all he would like.

yeah 20 hours of optimizing cgminer.conf and now 730 kH/s stable per card. Orig bios 015.041.000.000,  no undervolting, GPU 1070, MEM 1500,  2 cards 660 W power consumption.  Seasonic bronze 850W.

What voltage you running at?

I am same but using the optimized SCRYPT kernel I get a steady 750 k/hs. I don't get throttled and havn't can't blame the BIOS for anything, it runs steady at 77 C it is set to overheat at 80 it hasn't gone higher than 81. Sometimes it runs at 75 C. The problem I am having is with the HW monitor and GPU-z doesn't show Vrm temps the sensors for it just don't seem to show anything.
 I am running the engine clock at 1070Mhz and the stock meclock and voltage and it has crashed while running cgminer due to errors but I have had running stable for about 48 hours wihtout a problem. However,  my PC has froze randomly while at very low load i.e. basically idle or browsing  possibly due to Ram or Vram.

Thats good, can we get a bit more detail?

Can you share the optimised scrypt file?

How many cards you running on the rig? and what Driver version?

750kh/s on this card seems too good to be true.

Can we check that you are in fact using the same card as this one?

What is your memory manufacturer? Hynix AFR MFR or Elpida?

And lastly, your Card version are you using, should be written in the ( )'s on the BIOS version field on GPU-z

Thanks

are you sure you had 15.041 bios ?

I would also be very interested to hear about your setup Vyker

I flash an optimised BIOS from TheStilt thread [google it]

Then I undervolted to 1100mv using VBE7.

And finally I run this config for cgminer  "-g 2 --thread-concurrency 8192 -g 2 -I 13 -w 256 --gpu-fan 90 --temp-cutoff 90 --temp-overheat 85 --temp-target 85"

Pushing 726kh/s per card now - 6 card rig, 13.1TP driver - Win8.1Pro

HTH

r9 280x Dual-x  760kh/s but 2 of my cards burned. I don't know the exact reason. The don't run hot as you see. I'll try to RMA them.

http://s28.postimg.org/wko9813i5/screenshot.png (http://postimage.org/)

burned??? XD not so funny :P how much consuming undervolt dual x? (W)

I'm having the same card 280x Dual-x and i cant get more then 720 but the temo is going around 77 78c.

can you share more details about your conf.

can you share your Sensor screen on GPU-Z?

Regards,

i have a problem, just flashed bios at 039 but one have 700khs with elpida, the second have only 550Khs with hynx, WHY??

here are my settings:

"intensity" : "13",
"thread-concurrency" : "8192",
"temp-cutoff" : "85",
"temp-overheat" : "80",
"temp-target" : "75",
"vectors" : "1",
"worksize" : "256",
"kernel" : "scrypt",
"lookup-gap" : "2",
"shaders" : "2048",
"api-port" : "4028",
"expiry" : "120",
"gpu-dyninterval" : "7",
"gpu-threads" : "2",
"gpu-engine" : "1080",
"gpu-memclock" : "1500",
"gpu-powertune" : "0",
"log" : "5",
"no-pool-disable" : true,
"queue" : "1",
"scan-time" : "60",
"scrypt" : true,
"shares" : "0",
"auto-fan": true

But since the 3 of my cards burned due to VRM overheat I turned back the clock to stock and pointed a fan to the VRM to prevent more damage. I don't know may be the cards were defective but 3 of 8 were damaged.
I don't see VRM temp in GPU-Z may be there isn't sensor, also tried to undervolt with MSI afterburner but couldn't unlock it. I hope to RMA the burned ones.

So i'll try to help the ppl whitch havent conf or reach the 700+khs.



What I've found (for the ver41 cards)
TC: 8194 is the figure you don't touch
GPU Fan: I've lowered it to 70 to save its life, you can run it at 85 if your temp gets too hot
Auto GPU: don't use it or don't set GPU range
Powertune: just try various +20 or -20 until you see your card running optimal ( mine is on -20)

What you do:
Start your GPU clocks at 1050, and slowly push them up to find the optimal, no crashing speed.
My cards optimize at 1067 with best khs around 735~

My final setting looks like this:
-I 13 --lookup-gap 2 -g 2 -w 256 --thread-concurrency 8195 --gpu-powertune -20 --queue 0 --expiry 1 --scan-time 1 --no-submit-stale


http://s1.postimg.org/uvqe11zej/Afterburner.jpg (http://postimg.org/image/uvqe11zej/)

http://s1.postimg.org/qnvlsaxyz/cgminer.jpg (http://postimg.org/image/qnvlsaxyz/)

http://s1.postimg.org/isl2dhocb/GPU_z.gif (http://postimg.org/image/isl2dhocb/)

http://s1.postimg.org/dsj0rstbv/gpu_z1.gif (http://postimg.org/image/dsj0rstbv/)

If any one can answer me this question:
Flashed the version 41 original legacy bios to VDDC 1.093 (get the original from this thread)

The main problem is actually i can pass more then 735khs and i cant reach 740 or at least to get low temp.
I think 77c is too hight for now...

If anyone have any suggestions. ??? ??? ;D

I burned 5 of them out of 20 cards on stock everything. They were running at 77C. They were < 1 month old so I was reluctant to flash them. Now that the initial reseller warranty is over I'll flash them and undervolt them
Piss poor card...