Windows S5R3 SSE power App 4.36 available

Jord
Joined: 26 Jan 05
Posts: 2,952
Credit: 5,878,802
RAC: 6,460

I see a small niggle...

I see a small niggle...

See what I mean.
Einstein ran to 100% and is paused at Waiting to run.

KSMarksPsych
KSMarksPsych
Moderator
Joined: 15 Oct 05
Posts: 2,702
Credit: 4,090,227
RAC: 0

RE: I see a small niggle...

Message 79656 in response to message 79655

Quote:

I see a small niggle...

See what I mean.
Einstein ran to 100% and is paused at Waiting to run.

Isn't that just a quirk of the scheduler where it'll preempt at a checkpoint and if that checkpoint just happens to be at 100%? There are similar reports at Ralph.

Kathryn :o)

Einstein@Home Moderator

Jord
Joined: 26 Jan 05
Posts: 2,952
Credit: 5,878,802
RAC: 6,460

Probably so. It ran for

Message 79657 in response to message 79656

Probably so. It ran for another 25 seconds before it uploaded.
There you go. 44390.68 seconds for yet another 838.85

ErichZann
ErichZann
Joined: 11 Feb 05
Posts: 120
Credit: 81,582
RAC: 0

I just finished my first WU

I just finished my first WU with 4.36 and it was really fast, about 20% faster compared to old versions.

This finally is the Speed increase i waited for.

Greetings

hoarfrost
hoarfrost
Joined: 9 Feb 05
Posts: 207
Credit: 103,053,349
RAC: 7

Hello! My results with

Hello!

My results with near sequence number:

_____________WU___________ __Time (sec)__
4.32:
h1_0902.95_S5R3__310_S5R3b 17737.56

4.32 && 4.36
h1_0902.95_S5R3__309_S5R3b 17088.93

4.36:
h1_0902.95_S5R3__307_S5R3b 14211.28
h1_0902.95_S5R3__306_S5R3b 14136.14
h1_0902.95_S5R3__305_S5R3b 14135.56

14211.28 / 17737.56 ~ 0.8!

Jord
Joined: 26 Jan 05
Posts: 2,952
Credit: 5,878,802
RAC: 6,460

One thing I wonder about.

One thing I wonder about. What will this new application eventually do to my result duration correction factor (RDCF)? I just checked in BOINC Manager and the new 838.85 I have has now run for an hour, it's still estimating to run another 25h56m03s and that on an RDCF of 0.604165

Taking the other 838.85s as a lead, the run time will be around 12 hours, especially with this new application. So to have the correct estimated time to completion, I should have an RDCF of 0.302000, which in my opinion is way too low for a project like Einstein.

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3,522
Credit: 694,429,372
RAC: 127,448

RE: Taking the other

Message 79661 in response to message 79660

Quote:

Taking the other 838.85s as a lead, the run time will be around 12 hours, especially with this new application. So to have the correct estimated time to completion, I should have an RDCF of 0.302000, which in my opinion is way too low for a project like Einstein.

Isn't that quite common for "optimized" apps? RCDF is as low as 0.2 on one of my hosts (Core 2 Duo)

CU
Bikeman

Ed1934158
Ed1934158
Joined: 10 Nov 04
Posts: 62
Credit: 14,481,483
RAC: 0

While with my laptop centrino

While with my laptop centrino 1.7GHz, 2Mb chache 4.32 needed 46733s for h1_0918.80_S5R3__360_S5R3b_1, 4.36 needed 34131 for h1_0918.85_S5R3__360_S5R3b_0!
Now, that is great! If you compare these results with my +50000s average per unit by standard E@H application, this gain is even more impressive.
When will these ultramegacool applications be available for general public? I want to see +100TFlops of cumulative performance!

These application uses just SSE? What are potential gains for SSE2, SSE3, SSSE3, SSE4, 3dnow etc.? What are the chances that you will include some of them, lets say, in a year?

RandyC
RandyC
Joined: 18 Jan 05
Posts: 6,476
Credit: 111,139,797
RAC: 0

RE: While with my laptop

Message 79663 in response to message 79662

Quote:

While with my laptop centrino 1.7GHz, 2Mb chache 4.32 needed 46733s for h1_0918.80_S5R3__360_S5R3b_1, 4.36 needed 34131 for h1_0918.85_S5R3__360_S5R3b_0!
Now, that is great! If you compare these results with my +50000s average per unit by standard E@H application, this gain is even more impressive.
When will these ultramegacool applications be available for general public? I want to see +100TFlops of cumulative performance!

These application uses just SSE? What are potential gains for SSE2, SSE3, SSSE3, SSE4, 3dnow etc.? What are the chances that you will include some of them, lets say, in a year?

Before this app can be released for general use it needs to be modified to detect whether SSE is available on the system. If it doesn't, we'll have perhaps thousands of hosts bombing as soon as it starts.

Seti Classic Final Total: 11446 WU.

Winterknight
Winterknight
Joined: 4 Jun 05
Posts: 1,417
Credit: 370,154,899
RAC: 130,285

RE: One thing I wonder

Message 79664 in response to message 79660

Quote:

One thing I wonder about. What will this new application eventually do to my result duration correction factor (RDCF)? I just checked in BOINC Manager and the new 838.85 I have has now run for an hour, it's still estimating to run another 25h56m03s and that on an RDCF of 0.604165

Taking the other 838.85s as a lead, the run time will be around 12 hours, especially with this new application. So to have the correct estimated time to completion, I should have an RDCF of 0.302000, which in my opinion is way too low for a project like Einstein.


The Task duration correction factor = 0.227386 on my E6600 over on Seti Beta running standard apps, running 4.26 here it is at 0.424755.
I've always regarded the RDCF as a measure of how efficient the computer is compared to the benchmarks, running that project. And as a way of comparing project apps.
On Seti running optimised apps the RDCF's are:
E6600 = 0.140119
Q6600 = 0.127601
AMD2000 = 0.738527

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.