Brother John
Limp Gawd
- Joined
- Oct 30, 2012
- Messages
- 283
WOW, 38,666,408 points for Friday! WTF [H] !
Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
looks like we're holding them off .... for now
No... we are about neck and neck at current.
When GPU QRB goes full swing we shall see what happens then. We are also at the beginning of the month... so they have a bit of a boost going on.
It's time to reload.
Looks like EVGA is nipping at our heels again. Time to step things up.
Looks like EVGA is nipping at our heels again. Time to step things up.
It is that time of the month again.
My production is down a bit due to lack of 8102's. I haven't seen one in almost a week now.
My production is down a bit due to lack of 8102's. I haven't seen one in almost a week now.
Heads up everyone, they are well and truly back on the Radar, current overtake is 5.1 Months
EVGA bucks time?
*looks at calendar* yup. First half (or third) of each month is bucks crunch time. Nothing to worry about, drops down like a rock once everyone finishes their monthly dues.
GPUs don't have qrb yet. It was in public beta for a while, but now its gone again.
I'm not sure whats causing this spike. Somebody should look into it.
uh oh. Sounds like I'm leaving points on the tableBucks spike is normally over by now however they have upped the points requirements.
They have about 30 folders more than they do at the start of a normal month.
GPU QRB is available if you run with the beta flag
We are also missning some heavy hitters
uh oh. Sounds like I'm leaving points on the table
I've only got 1 machine aimed at Team 33. We have a small team, and most are pointed there.
I going to drive over and aim more. You folk have been a lot of help to me.
It won't do a whole bunch, but probably about 400k ppd starting in a couple days.
Maybe, maybe not.
The current beta is very different...
It is a new OpenCL core built from the ground up using OpenMM 5...
Both AMD and nvidia Use OpenCL as it requires JIT compiling and cuda does not support that.
This is the first viable AMD core in a very long time.
Warning... roles have been reversed. Nvidia uses excessive cpu for gpu folding on the new core due to their poor OpenCL implementation. AMD uses < 1% cpu.
QRB is enabled currently but the base system is higher up and the Work units are chewy.
22k atoms explicit. This causes lower end gfx cards to have lower ppd but the higher end ones do better. My 7970 gets 45k ppd and the GTX 580 is about the same... kepler is still trailing due to its castrated gpgpu arch.
I suggest starting a new thread if you want to discuss this further...
Well despite my kepler (3xGTX680 4GB)'s castrated GPGPU, I'll install client and point all the 3 GPUs at the team.
680 is still tops for non-beta GPU WUs.
Technically Kepler is not bad at compute. The architecture itself is faster than Fermi. But, on consumer cards, NVidia decided to nerf compute capability significantly (primarily by giving them low memory bandwidth, but also in drivers (see CUDA performance of K5000 for example, using quadro drivers)).
With QRB, Titan has the potential to be insane.
Yeah, I wasn't happy about that which I figured out after making my move from 570's to a 680. LiveNLearn, I guess.
What is QRB, by the way?