Just joined up for this topic alone, brilliant info so far.
I'm having issues with hashra, its showing my blizzard hashrate at 650 Kh/s, the pool reports 1,698.01 KH/s
How do I distinguish which of these values is correct? 650 appears far too low however 1.6 must but too high?
I'm a newbie in the mining world so sorry if i'm way off on something.
Pool hashrate will vary constantly unless your pool shows hashrate for specific time interval. The pool I'm on shows current and 24 hour avg. I watch the 24 hr avg because current hashrate jumps around too much. Also, sometimes it takes a while for the miner hashrate to stabilize. Up to an hour for me. |
|
|
|
This thread is getting confusing. Would someone mind putting together a small tutorial on what which resistor they replaced and the best clk with the least hardware errors there seems to be a mix of replacing one or the other or a mix of both? Then there's discussion about the crystal!?
Can I replace the resistor with the larger resistors or does it have to be a surface mounted resistor? what clock speeds are best with the blizzard with no mods and least hardware errors? Can I run a stock blizzard at clk 342 with no mods? Is BFG miner better or cgminer? I'm seeing mixed reviews on which produces less hw errors?
Thanks! Looking to order some more of these i like the style reminds me of a mini looking g blade.
We're replacing one resistor. It is in the R10 location on the board schematics. The photo in this post shows the resistor. https://bitcointalk.org/index.php?topic=644921.msg7297522#msg72975227.5k seems to be the best resistor choice. https://bitcointalk.org/index.php?topic=644921.msg7379971#msg7379971I ran 340 clk unmodified. You might be ok with 342. I had more errors. I run 381 clock now with extra cooling and 1.4 volt mod using approx 7.67k resistor value. I didn't have many resistors at home to choose from so I had to parallel a 47.5k axial resistor with the stock 9.1k resistor to get approx 7.67k. I used what I had. One of mine shows 3.2% HW errors after 22 hours. The other, after 8 hours, doesn't show a % yet. It'll be less than 5%. We're limited to 381 clock because of software. We can't go any higher. BFGMiner works great for me. Averaging 3,100 kH/s on the pool with two Fury's. |
|
|
|
|
Does hashra allow you to enter the number of chips?
|
|
|
|
There is now a branch available for folks who want to start testing ZeusMiner support and giving feedback:
I love to test in Windows if there is a build. I don't know anything about coding so bear with me but is this necessary in BFGMiner: int chip_clk = freq;
//max clock 381MHz, min clock 200MHz
if (chip_clk > 381)
chip_clk = 381;
else if (chip_clk < 200)
chip_clk = 200; earlier code from Zeus looked like this: uint32_t clk_reg;
uint64_t golden_speed_percore;
if(opt_chip_clk>(0xff*3/2)){
opt_chip_clk = 0xff*3/2;
}
else if(opt_chip_clk<2){
opt_chip_clk = 2;
}
clk_reg= (uint32_t)(opt_chip_clk*2/3); I'd like to try freq or clk 383 as explained by kramble here: https://bitcointalk.org/index.php?topic=644921.msg7380726#msg7380726 but I think the BFGMiner code will only allow 381. Can I set 382.5 clk? I know, tiny gains, but already running at 381 clk with low HW errors. |
|
|
|
|
First Fury 24hr avg just got to 1,572 kH/s
I have to try to remember what I did 24 hrs ago ..........
|
|
|
|
Improving the cooling should be helpful. Get some air flowing through the chamber would be the least intrusive. "Blade" style like the guys here did. Adding heatsinks help but now you're making visible changes. J4bberwock & nst6563 would know for sure if a "pencil mod" would be beneficial or too risky. Maybe it could help reduce HW errors at slight overclock speeds. The results would not be precise though. If you reduce the resistance too much the voltage will get too high and you could see smoke  There's no way to put the smoke back in once you let it out.  The volt mod so far is replacing one small resistor. I actually paralleled a resistor using what I had on hand. You need a steady hand and a very fine tip on a soldering iron. Cooling comes first. They're mini heaters. |
|
|
|
|
24hr averages for two Fury's: 1,565 kH/s + 1,535 kH/s = 3,100 kH/s 381 clk.
Same mods, 1.4v, not sure why the 2nd is a bit slower as it was the faster one before mod.
Now I can try diff software builds etc. and see how changes affect the 24 hr avg.
|
|
|
|
I don't know if it's over. Darkwinde and jstefanop were mentioned in Zeus' weekly update http://zeusminer.com/2014/06/18/1/And this post from Terry today on Litecointalk https://litecointalk.org/index.php?topic=16301.msg184567#msg184567 was interesting the chip docs were already up for like two weeks, no? and yes the schematics for the blades will be up as well. also we are going for more so you geniuses will be happy to be involved with. we are still working on the "old" cgminer and a new version just released. don't yet have time to put it online. |
|
|
|
right have reverted it, was a first shot without testing and understanding of kramble command protocoll, as it supports 8 Bit for setting the clock plus the complement as check sum, so if the HW implementation does not support any other clue we are limited here to 381 MHz. Sorry can not do anything against it...only Zeus...
Thanks for trying. I'm hoping something will turn up. Can you allow 382 in a future build without affecting all the other improvements you did? I know it's only like 4kh/s but it might just give us a steady 1,600 kH/s. |
|
|
|
Aloa, as promised I implemented an override for the 381MHz cap and pushed it to git: https://github.com/Darkwinde/bfgminer.gitAdd to your command line: --zeus-clk-override and you can set any clock you like with "--zeus-clk". This is done on your own risk  Example Linux: ./bfgminer --scrypt --zeus-cc 6 --zeus-clk 400 --zeus-clk-override -S zeus:/dev/ttyUSB1 -S zeus:/dev/ttyUSB0 -c /opt/minepeon/etc/miner-zeus.conf Have fun and looking for feedback if everything works fine. Darkwinde I tried --zeus-clk-override on Windows using this build: http://cryptomining-blog.com/2791-new-bfgminer-4-2-1-release-with-support-for-zeusminer-scrypt-asics/but it doesn't start. The BFGMiner window flashes up and closes before I can read anything. Without --zeus-clk-override the build runs fine. |
|
|
|
I like the blade idea you and J4bberwock are doing. If I get two more I'll do that also. On my pool, the 24 hour average is 1,566 kH/s for the Fury I modded yesterday. Not bad but I'd like to go higher. We are so ready on the hardware side. 380 clk  That 3rd hashrate in BFGMiner really gives a good idea of poolside hashrate. |
|
|
|
|
One thing I noticed was the almost instant heat on the board when I had it out of the enclosure to test voltage. It was not even hashing. I didn't measure the heat or anything but you could notice it.
Another thing I notice is how porous the the pcb is. Heatsink paste just oozes up through every hole. Your earlier advice to use non conductive paste is the only way to go.
|
|
|
|
|
I might go ahead with the cooling mods on my other Fury today if I have time.
The overvolted & double fan Fury handles the heat much better. A temperature probe touched to one heat sink measures 35C. Air coming out of the chamber 22.5C. The unmodded Fury chip temp can reach 50C at 342 clk.
Do we know for sure what the minimum voltage is required for 382 clk 1.5+Mh/s?
If these chips ever hit 300kh/s then Zeus used a different driver. We now know that Zeus could not hit 300kh/s with the driver they released. I'm keeping my fingers crossed hoping we get a breakthrough.
|
|
|
|
|
That's a Week 1. The GAWMiners logo was upside down on Week 1's. They must have known I'd stand it on it's side.
I have a Week 3 also. Logo is right side up when the Fury is laying flat after Week 1.
I going to run jstefanop's BFGMiner 4.0 overnight. I'm not 100% sure but I think that one allows 382 clk.
Can't wait to get 24hr poolside number.
|
|
|
|
I think you'll like the cooling    The second fan is pumping air into the chamber. There's a really good flow of air out of both open ends. The 3rd hashrate number has climbed to 1.55Mh/s at 2 hours running. |
|
|
|
I measured Zeus R9 R10 R11 on my board
R11 is in the correct spot on my Week 1 GAW Fury
R9 measures 4.93K on the board
R10 measures 4.92K on the board
I checked like 5 times and cross checked my multimeter with several spare resistors.
R10 = ( R9 * Vfb ) / ( Vout - Vfb )
4.92 = ( 4.93 * 0.591 ) / ( 1.183v - 0.591 )
Looks like mine is slightly under volted at 1.183v (I did not measure the voltage)
I'll have to check my other Fury (Week 3).
edit:
Almost the same on the other Fury
R11 .906k
R10 4.91k
R9 4.91k
so 1.182v
You will have to make measurements on components removed from the board. Read the value written on them. I'm almost sure you have 9.3k and 9.1k too. You're right. I didn't remove the resistors but ran the markings through an online decoder. 912 = 9.1k 93A = .909k (909 Ohm) 03C = 10.5k Looks like the reference resistor is 10.5k I figured I could get 1.4v if I paralleled a 47.5k with the 9.1k R10. It was the only resistor I had on hand that could work for the volt mod. R10 = ( R9 * Vfb ) / ( Vout - Vfb ) 7.67 = ( 10.5 * 0.591 ) / ( 1.4v - 0.591 ) 47.5k paralleled with 9.1k = 7.64 Measured voltage on the board at 1.4v I'm running it now with an extra cooling fan and cut up heatsinks glued to the chips. Clock is 380. 5% HW errors, 67 watts with a second fan. 1 hour running so far. Local hashrate is steady 1.7mh/s but poolside looks like it would only be 1.46mh/s  Edit: Looks like it made it to 1.5mh/s after 1 1/2 hours  |
|
|
|
In the early days of these chips it was thought that they were capable of 300kh/s each. Remember this:  The current chips might be a completely new design OR performance was limited as the number of chips in each miner grew. Zeus would have had to limit performance to balance power use, hash rate, heat removal etc. 300 kh/s would need a clock rate of approx 430. 6 chips could potentially hit 1.8mh/s. |
|
|
|
|
The schematic was dated June 14 - printed on the pdf. I wonder if new miners will match it?
|
|
|
|
|
Show Posts
