I have 3 different computers running Rosetta. Different CPU's, Different OS. The two 2-core computers (Intel) is returning aprox. the same grantet as claimed credit to me (often a bit more) But the last computer with a 4-core (AMD) is mostley giving me less that half the claimed credit. Can any of you give me a hint about what is wrong, or what I am doing wrong.

---

I have 3 different computers running Rosetta. Different CPU's, Different OS. The two 2-core computers (Intel) is returning aprox. the same grantet as claimed credit to me (often a bit more) But the last computer with a 4-core (AMD) is mostley giving me less that half the claimed credit. Can any of you give me a hint about what is wrong, or what I am doing wrong.

It's possible your claimed credit is way out. Claimed credit is just based on the simple BOINC benchmark, which doesn't take memory speed or CPU cache size etc into account. In other words your claimed credit is artificially high and your CPU can't process rosetta (which is lots of data) as quickly as the benchmark (which is very small).

As transient says, it looks fine. You can compare the credit per CPU-hour against another similar CPU to check though. Just make sure you're comparing against one with the same size cache and adjust for any differences in clock speed.

HTH
Danny

---

AMD CPUs in general score very bad in Rosetta, one of the reason ive pretty much stoped crunching here (+the bugs ofc), cause I think the CPU time is somewhat wasted. Your results look very much like what I was getting with my old AMD K8 dual core, the new Phenom II scores a little bit better, but still is far away from what C2D CPUs get done.

You might want to try a different project on the AMD quad, SIMAP scales very good on AMD CPUs and has loads of work to send out. Its also in the same sciene area as rosetta.

---

http://www.MIAteam.eu

AMD CPUs in general score very bad in Rosetta, one of the reason ive pretty much stoped crunching here (+the bugs ofc), cause I think the CPU time is somewhat wasted. Your results look very much like what I was getting with my old AMD K8 dual core, the new Phenom II scores a little bit better, but still is far away from what C2D CPUs get done.

You might want to try a different project on the AMD quad, SIMAP scales very good on AMD CPUs and has loads of work to send out. Its also in the same sciene area as rosetta.

I don't see much difference between my Core2 Q6600 (3GHz) and my Phenom II 545 (3GHz) in credit granted per hour per core... The claimed credit is much lower on the Core2 machine so the difference between claimed and granted is much larger on that machine, but who cares about claimed!?!

And SIMAP is far from being the same science as Rosetta ;)

---

Where did I say Simap is the same science as Rosetta? Nowhere, please read carefully. What I said is that both projects are about proteins, so someone interessted in Rosetta might aswell be interessted in SIMAP; when looking for a replacement project.

Then I said, Phenom II are scoring better then older Phenom and K8. fact. he has an "old" Phenom I, that is wasted here. fact. Your Q6600 still scores better then a Phenom II Dual Core, a newer C2D like e8400 again scores better then your Q6600. fact. In SIMAP your P II 545 would process WUs faster then the Q6600 - prove that SIMAP is a better project then Rosetta for these type of CPUs:
PII 550 vs Intel q9450:
http://boinc.bio.wzw.tum.de/boincsimap/workunit.php?wuid=1628546

Where am I standing wrong?

---

http://www.MIAteam.eu

I don't see much difference between my Core2 Q6600 (3GHz) and my Phenom II 545 (3GHz) in credit granted per hour per core... The claimed credit is much lower on the Core2 machine so the difference between claimed and granted is much larger on that machine, but who cares about claimed!?!

I know how to settle this.

Whichever one made you "Predictor of the Day" is the winner (see Rosetta home page). Well done you! ;)

---

Where did I say Simap is the same science as Rosetta? Nowhere, please read carefully. What I said is that both projects are about proteins, so someone interessted in Rosetta might aswell be interessted in SIMAP; when looking for a replacement project.

My mistake - didn't read the 'area'! They are both protein related...

Then I said, Phenom II are scoring better then older Phenom and K8. fact. he has an "old" Phenom I, that is wasted here. fact.

I agree the PhII scores better than the PhI but it's certainly not wasted here- it's faster than a lot of my comps.

Your Q6600 still scores better then a Phenom II Dual Core, a newer C2D like e8400 again scores better then your Q6600. fact. In SIMAP your P II 545 would process WUs faster then the Q6600 - prove that SIMAP is a better project then Rosetta for these type of CPUs:
PII 550 vs Intel q9450:
http://boinc.bio.wzw.tum.de/boincsimap/workunit.php?wuid=1628546

Phenom systems are more suited to SIMAP than Core2 but I'll still put my CPUs on Rosetta because I believe the project is more deserving of the CPU time (not that I don't think SIMAP is a good/useful project - its just that Rosetta has such phenomenal potential), but I was referring to this:

the new Phenom II scores a little bit better, but still is far away from what C2D CPUs get done.

There isn't much difference at all: 22.6 credits per hour per core on the PhII and 22.9 on the C2Q. That 22.6 might also be higher on a normal Phenom II 545 but I've enabled the 3rd and 4th cores on that one so there were 4 cores sharing the 6MB L3 cache rather than just 2 as standard.

---

I don't see much difference between my Core2 Q6600 (3GHz) and my Phenom II 545 (3GHz) in credit granted per hour per core... The claimed credit is much lower on the Core2 machine so the difference between claimed and granted is much larger on that machine, but who cares about claimed!?!

I know how to settle this.

Whichever one made you "Predictor of the Day" is the winner (see Rosetta home page). Well done you! ;)

WOO! I wouldn't have seen that - ta Sid! I'm not sure how to find out which comp crunched that???

---

Go through all Workunits until you find the matching one, quite some work I guess ^^

About the Phenom II: My x2 550 wasnt getting those 22,6, I saw your numbers and they avarage out higher then mine did. I would agree that sharing the L3 to 4 instead of 2 should lower the credits granted, obvsiouly it doesnt? weird. What clock speed are you running? Mine is running at 3,1 Ghz (stock) and the avarage was around 20~21 credits per hour. My brothers C2D e8400 is getting a good ~30 credits here tho.

Maybe wasted is a bit too mean to say, but I´d like to use my CPUs where they are most usefull. I usualy look that up by comparing runtimes (like SIMAP) or credits (like Rosetta) to other type of CPUs. Then I see my x2 550 does better at SIMAP then a Q9450, which compared in Rosetta is up to 33% above me - easy choice for me, eventho I agree that Rosetta Sciene is very worth the crunching time.

btw my first reply to you might have been a bit harsh, rough day yesterday, sorry for that...

---

http://www.MIAteam.eu

Go through all Workunits until you find the matching one, quite some work I guess ^^

About the Phenom II: My x2 550 wasnt getting those 22,6, I saw your numbers and they avarage out higher then mine did. I would agree that sharing the L3 to 4 instead of 2 should lower the credits granted, obvsiouly it doesnt? weird. What clock speed are you running? Mine is running at 3,1 Ghz (stock) and the avarage was around 20~21 credits per hour. My brothers C2D e8400 is getting a good ~30 credits here tho.

Yeah - those things have a lot of cache so they're great for rosetta.

Maybe wasted is a bit too mean to say, but I´d like to use my CPUs where they are most usefull. I usualy look that up by comparing runtimes (like SIMAP) or credits (like Rosetta) to other type of CPUs. Then I see my x2 550 does better at SIMAP then a Q9450, which compared in Rosetta is up to 33% above me - easy choice for me, eventho I agree that Rosetta Sciene is very worth the crunching time.

I think ideally we'd be able to trade CPUs between BOINC projects - for every Intel CPU SIMAP transfer here we could send them an AMD and everyone wins ;)

btw my first reply to you might have been a bit harsh, rough day yesterday, sorry for that...

no problem - mine wasn't great either having re-read it ;)

---