If my CPU and GPU have to interact a lot more often I might be in trouble since the only machine I have access to right now has a pentium D CPU. I'll just hope for the best until I can get my hands on a newer machine.

---

Cruncher/Learner in progress.

Hm okay I can move some stuff from long to short queue. I was stretching the definition a bit but Gianni said to go with it. But if it's an issue I will send only 4+ hours to long and under that to short.

Hm okay I can move some stuff from long to short queue. I was stretching the definition a bit but Gianni said to go with it. But if it's an issue I will send only 4+ hours to long and under that to short.

Stefan, I don't see it as an issue at all. Leave them in the long queue.

Yeah I switched to the 4 hour thing but Gianni also told me to revert it back. I guess for this project we can "suffer" some shorter sims in the long queue.

I am sending now some thousands more simulations including a new protein called alpha3D (a3D).

Awesome job by the way on the crunching. I guess since I never did many simulations before I hadn't realized just how awesome the throughput of GPUGRID is :P Totally enjoying it now, hehe.

Yeah I switched to the 4 hour thing but Gianni also told me to revert it back. I guess for this project we can "suffer" some shorter sims in the long queue.

I am sending now some thousands more simulations including a new protein called alpha3D (a3D).

Awesome job by the way on the crunching. I guess since I never did many simulations before I hadn't realized just how awesome the throughput of GPUGRID is :P Totally enjoying it now, hehe.

Thumbs up on the long queue decision and a double thumbs up on being a DOCTOR!
BTW, I've a krick in my knee...

Yeah I switched to the 4 hour thing but Gianni also told me to revert it back. I guess for this project we can "suffer" some shorter sims in the long queue.

It's ok to have the 50ns and the 20ns long simulations in the long queue, until there are any in the long queue; as the 5ns and the 1ns long simulations take only 14 minutes and 3 minutes (respectively) to process, the data transfer of a long task takes more time than the calculation of a short task, which makes my hosts to download tasks from my backup project (because there's the limit of two GPUGrid workunits).

I am sending now some thousands more simulations including a new protein called alpha3D (a3D).

This should be no problem. :)

Awesome job by the way on the crunching. I guess since I never did many simulations before I hadn't realized just how awesome the throughput of GPUGRID is :P Totally enjoying it now, hehe.

Besides enjoying it, it is your job and responsibility to use and nourish this huge computing power as wisely as possible. As most of us (your volunteer crunchers) are not into biochemistry this job partly consists of making us motivated to support a research whose results we can't comprehend as much as you do (to be polite :) ).

BTW, I've a krick in my knee...

Zoltan, once we have some results I will make a post about it to explain it. I know we are not really top at the communication aspect. Right now it's still a bit in the making and a bit under the covers to avoid competition etc. But if the results are nice it will probably make a quite important publication.

CASP WU on (2) GTX 970 at 1.5GHz are 2.5~2.7x slower with PCIe 2.0 x1 compared to PCIe 3.0 x4.
WU's require PCIe 3.0 x8 for proper OC scaling.

1ns: 900/sec vs. 350/sec
5ns: 4770/sec vs. 1761/sec

PCIe 2.0 x1: 46% GPU / 7% MCU / 80% BUS usage / 75W GPU power
PCIe 3.0 x4: 57% GPU / 12% MCU / 40% BUS / 107W

CASP runtimes (atom and step amount) vary so this just a general reference.

GTX 1070 (PCIe 3.0 x8) CASP runtimes:

-- 1ns (ntl9) 600 credits = 240/sec @ 2.1GHz / 59% GPU / 15% MCU / 37% BUS / 78W power
-- 1ns (a3d)= 1,350 credits = 330/sec @ 2.1GHz / 70% GPU / 24% MCU / 39% BUS / 96W power

-- 5ns (ntl9) 3,150 credits = 1,200/sec @ same usage and power numbers as 1ns
-- 5ns (a3d) 6,900 credits = 1,600/sec @ same usage and power numbers as 1ns

GTX 1060 (3GB) PCIe 3.0 x4 CASP runtimes:

-- 1ns (ntl9) 600 credits = 300/sec @ 2.1GHz / 63% GPU / 17% MCU / 51% BUS / 74W power
-- 1ns (a3d) 1,350 credits = 450/sec @ 2.1GHz / 74% GPU / 24% MCU / 59% BUS / 88W power

-- 5ns (ntl9) 3,150 credits = 1,500/sec @ same GPU usage and power as 1ns
-- 5ns (a3d) = 6,900 credits = 2.275/sec @ same GPU usage and power as 1ns

IMO: a (1152CUDA GTX 1060) is on par with (2048CUDA GTX 980) and ~20% faster than a (1664CUDA GTX 970).
The (1920CUDA GTX 1070) is as (if not) ~5% faster than a (2816CUDA GTX 980ti).

IMO: a (1152CUDA GTX 1060) is on par with (2048CUDA GTX 980) and ~20% faster than a (1664CUDA GTX 970).
The (1920CUDA GTX 1070) is as (if not) ~5% faster than a (2816CUDA GTX 980ti).

This is true if you compare these cards under WDDM os.
I guesstimate that my GTX980Ti@1390MHz/3500MHz will be ~6% faster under Windows XP x64 than my GTX1080@2000MHz/4750MHz under Windows 10x64 (while processing CASP11_crystal_contacts_**ns_a3D workunits). I also guesstimate that the TITAN X (Pascal) cards won't scale well under WDDM os especially with low atom-count workunits, the TITAN X (Pascal) will be just slightly faster than a GTX 1080. In general, under a WDDM os the TITAN X (Pascal) GPUs won't be as faster as they should be taking the ratio of the CUDA cores of the GTX 1080 and the TITAN X (Pascal) GPUs in consideration.

BTW, I've a krick in my knee...

Dr. Who?

hah!