Taking 840-846.

[QUOTE=Thomas11;242138]I already tried GPU sieving for this drive on a relatively cheap ATI Radeon 5670. It reaches only about 85% of the speed compared to a single cpu core (Q6600) running sr2sieve, and the GPU sieve thereby utilizes a full cpu core too! Not to mention the memory requirement (about 2 GB vs. a few MB using sr2sieve)...[/QUOTE]

From Ken's post on Anandtech forum:

[URL]http://forums.anandtech.com/showpost.php?p=30909706&postcount=10[/URL]

[quote]OK, here's how hits work. On most sieving projects, you have a sieve file, and every hit is a factor that wasn't in that sieve file before. So each one usually eliminates a primality test.

But this project doesn't use a sieve file, because it would be too big and could take too much memory. The numbers that have been factored are checked for small factors (which would already have been found), but each case where no such factors are found counts as a hit, and the server sorts out the results later. [/quote]So why don't run:

tpsieve[COLOR=Red](version)[/COLOR] -p45000G -P1000000G -k10000 -K35000 -n[B]0.8[/B]M -N[B]2[/B]M -ffactors.txt -t# [B][U]-M2 -T[/U][/B] -q -R

Use the -t# switch to spawn process across # cores. i.e. quad core would be -t4

EDIT: Thomas, please try the above command and check sieve speed. The command says that you will be sieving 10000 < k < 35000 (all k's) from 0.8M to 2M on Riesel side starting at 45T to 100T

[QUOTE=em99010pepe;242447]
So why don't run:

tpsieve[COLOR=Red](version)[/COLOR] -p45000G -P1000000G -k10000 -K35000 -n[B]0.8[/B]M -N[B]2[/B]M -ffactors.txt -t# [B][U]-M2 -T[/U][/B] -q -R
[/QUOTE]

This would be about 10 times slower than using sr2sieve on just the 128 K's in this drive! Actually I'm getting about 118 kp/sec vs. 1220 kp/sec when using sr2sieve (single core each).
And the cpu version of ppsieve is only slightly faster by yielding about 150 kp/sec.

BTW.: The above command line will not work, since there is no "-R" switch for tpsieve, and the "-n0.8M" doesn't work either...

[QUOTE=Thomas11;242454]This would be about 10 times slower than using sr2sieve on just the 128 K's in this drive! Actually I'm getting about 118 kp/sec vs. 1220 kp/sec when using sr2sieve (single core each).
And the cpu version of ppsieve is only slightly faster by yielding about 150 kp/sec.

BTW.: The above command line will not work, since there is no "-R" switch for tpsieve, and the "-n0.8M" doesn't work either...[/QUOTE]

Now I am confused, how to you sieve riesel side with tpsieve? What about -n800000?

[QUOTE=em99010pepe;242458]Now I am confused, how to you sieve riesel side with tpsieve? What about -n800000?[/QUOTE]

As far as I know, you cannot sieve just the Riesel side with tpsieve.
My test (and the 118 kp/sec) was for a combined +1/-1 sieve.

EDIT: And, of course, I used -n800000.

I believe this is true- tpsieve is both, ppsieve is either one and set by flags. The point remains that while we can sieve this drive on a GPU, it doesn't really make sense to do so unless one has a truly fast video card- fast enough that the CPU overhead running the GPU sieve is less per factor than simply running sr2sieve. If anyone has such a card, it would be nice to boost this drive's files a bit.
-Curtis

[QUOTE=VBCurtis;242475]If anyone has such a card, it would be nice to boost this drive's files a bit.
-Curtis[/QUOTE]

Actually there is no need for any speedup, since the sieve is almost at it's optimum (for the whole range up to n=2M). The elimination rate is currently (p=50T) around one factor in about 2200 secs (64 bits mode), while an LLR test takes only 775 secs (at n=825k) on the same machine.

So you should rather pick some files for LLR testing... :smile:

What's the speed of your machine?

[QUOTE=em99010pepe;242564]What's the speed of your machine?[/QUOTE]

You know this has no relevance to optimal sieve depth, right?

Thomas- looks like 75T will be optimal.
-Curtis

[QUOTE=VBCurtis;242641]You know this has no relevance to optimal sieve depth, right?

Thomas- looks like 75T will be optimal.
-Curtis[/QUOTE]

Yes I know, just want to compare the LLR speed with my core i5.

776-778 complete, no primes.