![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Hmmm
- Thread starter Abuse
- Start date
C
CIWS
Guest
Tinker units can take longer because that core does not use the SSE optimizations that Gromacs does. P4's really saw the slowdown compared to Athlons without SSE.
When you first wonder - "Why the h3ll is it taking so long ?"
look at the
The due date and point value.
If you are running EMIII it's apparent
OR
If not look at the unitinfo.txt file.
Example:
OR
Find out what the protein is and look at the psummary page or the Weatherman's protein page.
For example the above protein was a p358 worth 51.4 points due in 18 days from psummary page, or 36 days from unitinfo.txt. (the difference in due dates is the unitinfo.txt date is when they will still award points, the psummary is when they are valuable to the project).
But some proteins like the p687 are due in 50/100 days and are worth 70 points, while others like the p547 are due in 5/10 days and only worth 18 points.
And if they are assigning genes instead of proteins (799) there is a whole different formula.
Or in summary - some proteins are chewier than others, and if you are chewing jeans it's another matter still.
look at the
The due date and point value.
If you are running EMIII it's apparent
OR
If not look at the unitinfo.txt file.
Example:
OR
Find out what the protein is and look at the psummary page or the Weatherman's protein page.
For example the above protein was a p358 worth 51.4 points due in 18 days from psummary page, or 36 days from unitinfo.txt. (the difference in due dates is the unitinfo.txt date is when they will still award points, the psummary is when they are valuable to the project).
But some proteins like the p687 are due in 50/100 days and are worth 70 points, while others like the p547 are due in 5/10 days and only worth 18 points.
And if they are assigning genes instead of proteins (799) there is a whole different formula.
Or in summary - some proteins are chewier than others, and if you are chewing jeans it's another matter still.
The -advmethods flag tells the client that you want to use the advanced methods (duh!
) which currently translates to 'gimme proteins that use the Gromacs core.' The sweet thing about Gromacs is the built-in support of SSE and 3DNow! to speed up the calculations considerably. Without SSE, the P4 is a pretty poor folding CPU, so that 3 days per WU is probably spot on. If you were processing Gromacs proteins with SSE, a comparably scored protein would probably complete in a day and a half. Like I said check psummary or Larry's page for value. Both show the p692 as being 71 pointer Tinkers.Originally posted by Abuse
One has just completed, it was folding p692
I tried to look that project up and it doesnt appear to be there?
Oh and it was for 51 points
Bullseye_blam
Limp Gawd
- Joined
- Apr 6, 2001
- Messages
- 215
well, to be honest, it's good to use those flags (-advmethods and -forceSSE) on P4 CPU's simply because P4's will be greatly hindered by not using them. It's not a question of points, it's a question of how efficiently is your machine folding.Originally posted by Abuse
Im not bothered overly about the points
Im still folding, regardless of how long it takes
Originally posted by Bullseye_blam
well, to be honest, it's good to use those flags (-advmethods and -forceSSE) on P4 CPU's simply because P4's will be greatly hindered by not using them. It's not a question of points, it's a question of how efficiently is your machine folding.
Not true - from the Console User Guide
"-advmethods" is for those who are willing to be the first to get new Cores or work units. Does not mean they will be faster. They may be slower because they crash your system.
Bullseye_blam
Limp Gawd
- Joined
- Apr 6, 2001
- Messages
- 215
Yeah, I agree... but no matter how slow they are, they're certainly faster than the tinkers.