Nehalem,One cpu to rule them all!

Thanks for the info Dave. Just what I was looking for.

Obviously now the choice is do I upgrade my gaming rig (I think so!!) or get a couple of Quads to add to the farm?

P35 gear might be getting cheaper soon!!!!:up::up:

Quote:

---

Originally Posted by jcool

I mean it can ofc (Dave said sth. about 31k today somewhere around here), but it shouldn't.
Or, the Harpers should get more. I already posted how they get put down by the quota here :shakes:

---

I was having the same problem with one of mine that was doing only HCC work. I switched it to HPF2 and am doing much better. Some machines just get better returns on different projects.

For example:
Attachment 88452
This machine (name blocked out as I was rather cross when I named it) was getting about HALF it's claimed credit on all HCC units. Since switching to HPF2 it's getting a LOT better return.

My Opty systems on the other hand:
Attachment 88456
do quite nicely on HCC work.

Only that I don't care about points, I care about how much HCC WUs per day are computed by my machines. If they decide to rate down the Harpers that badly it sucks, but it sure as hell doesn't stop me from crunching HCC.

That's fair enough, but it's not just the Harpers. Those stats I posted first are from an Athlon XP machine. I haven't noticed a real reason for this disparity across different hardware yet, either.

Just for clarification of my standpoint, I am constantly torn between running exclusively running HCC (personal reasons not unlike anyone else's here) and mixing the projects up with other valuable work that gets a better return for my team and which achieves the same goal albeit indirectly. I tend to vacillate between both strategies but always have mostly HCC work. As such I cannot and will not criticise anyone for running whatever project they feel better meets their needs/situation. I only offer the kind of comparisons I made before for those to whom the work selection is less personal.
It's all good.

smoken folding !!

the worlds gonna be in better shape quickly now :up:

Quote:

---

Originally Posted by D_A

Just for clarification of my standpoint, I am constantly torn between running exclusively running HCC (personal reasons not unlike anyone else's here) and mixing the projects up with other valuable work that gets a better return for my team and which achieves the same goal albeit indirectly. I tend to vacillate between both strategies but always have mostly HCC work. As such I cannot and will not criticise anyone for running whatever project they feel better meets their needs/situation. I only offer the kind of comparisons I made before for those to whom the work selection is less personal.
It's all good.

---

Sure, and your views and insights are very much appreciated.
I didn't mean to imply anything on your behalf with my last post, just stating my POV. Getting less points than deserved does bug me, after all - even if it's not the deciding factor here.

No problems ;)

No matter how many smilies we might use, there's no substitute for a friendly tone in conversation. :D (I get accused of things a lot when no issue is intended)

Quote:

---

Originally Posted by NaeKuh

i would think if ET was in space, he would also learn how distructive we are, and would avoid our race.

Also they would have good cloaking technology, because they need intergalatic flight, so until they want to say hi, i dont think theres a point in us trying to say hi first. :rofl:


J/K WHO!

Any DC project you do is a reward in its own special way! :up:

---

I ll get to Others, as soon as I am done checking how high Core i7 can get on SETI.

Francois

Quote:

---

Originally Posted by Drwho?

I ll get to Others, as soon as I am done checking how high Core i7 can get on SETI.

Francois

---

Don't forget f@h ;)

If you can beat my Dual Harpertown machine with one Bloomfield, I'll buy the 3200EE ;)

it's only ranked #5 in dailies out of 540000 rigs on WCG :hrhr:

jcool: wait for Dave to put it under water. It's already most of the way to Harpers. He said 4GHz should match his old Harpers, so let's see about 4.5 or 5ghz. :p:

You know, with the way these things are running, I might need to add in another loop just for the CPU and see how high I can get an i920 screaming ;) One MCR320 for CPU + NB ('sif WC the SB) + Mosfets and another MCR320 for 2x HD4870 1Gb's with those new Swiftech MCW60-4870 partial cover blocks... Damn, I need some money, and stat!

Eller

My new Core i7 System (i920 / P6T Deluxe / 3x 1GB Corsair) have to come in a few day's, then I'm back with crunching. :toast:

Quote:

---

Originally Posted by Bobsama

jcool: wait for Dave to put it under water. It's already most of the way to Harpers. He said 4GHz should match his old Harpers, so let's see about 4.5 or 5ghz. :p:

---

He's saying a lot these days.. but what I still see is 4:25h vs. 2:44h for 8 WUs...
No sub-5Ghz Bloomfield is gonna close that gap anytime soon, and even then I doubt it can be done (I'm gonna try myself ofc, phase + 3200EE FTW :wasntme: )

Where is he anyway? Waiting for him to verbally slaughter me :rofl: :rolleyes:

Quote:

---

My new Core i7 System (i920 / P6T Deluxe / 3x 1GB Corsair) have to come in a few day's, then I'm back with crunching.

---

Nice to see someone else from Germany chiming in :up:

Any word on what this pulls out of the wall for wattage yet?

While I'm at it....maybe we could start a thread/database of hardware, clocks, and power draw. I know 45nm is more efficient than 65nm....but by how much? Is it worth it for me to replace a Q6600 with a Q9650? Will it pay for itself in a year or two or three at my current electric rate? All rhetorical questions - with such a database, we could do the math and answer these questions for ourselves.

Anyone know of such a database already in existence? I'm sure this isn't an original idea of mine....especially among crunchers/folders.

Maybe I'll post this separately:shrug:....sorry for the off topic.

Wait....didn't riptide do something with excel somewhere on here?

.....off to find that thread

Quote:

---

Originally Posted by Bobsama

jcool: wait for Dave to put it under water. It's already most of the way to Harpers. He said 4GHz should match his old Harpers, so let's see about 4.5 or 5ghz. :p:

---

I've got a block, barbs and tubing sitting here waiting for Dave to return! :up:

Quote:

---

Originally Posted by jcool

He's saying a lot these days.. but what I still see is 4:25h vs. 2:44h for 8 WUs...
No sub-5Ghz Bloomfield is gonna close that gap anytime soon, and even then I doubt it can be done (I'm gonna try myself ofc, phase + 3200EE FTW :wasntme: )

Where is he anyway? Waiting for him to verbally slaughter me :rofl: :rolleyes:



Nice to see someone else from Germany chiming in :up:

---

No, you are right. There is no common sense to the claims involved for the timeframe taken EXCEPT I think it's claiming based on using an entire core or two for the benchmark.
That is the issue and it's leveled out claiming at a rate of app 33,477 per day and being awarded at a rate of app 25,339 per day based on the last 10 days.
This is at 3733mhz on air and almost 24/7 crunching.
A little time taken during these 10 days for the occasional bench but not much time lost.
This is all on the HCC project.
OOPS, forgot..Factor in another TWO FULL pages of WU pending validation...:wasntme:

So it looks like the benchmarks with HT on are ... dubious. How do the claims average out based on actual working time as opposed to wall clock time?

If you don't know what I mean just grab me a full page of valid work with crunch time, claimed and awarded points and I'll work it out myself.

Quote:

---

Originally Posted by D_A

So it looks like the benchmarks with HT on are ... dubious.

---

What I can guess is that the BOINC Benchmark only stresses four cores for the test, or perhaps the workload assigned to the benchmark doesn't keep the low level caches very busy, whereas the workunits themselves are more complicated and cache/ram stressful, thus making the difference between four and eight threads sucking bandwidth from the L's much more tangible in the "real world" scenario as opposed to the benchmark scenario.

Just a congecture.

http://img.3d-nexus.de/main.php?g2_v...serialNumber=1

http://img.3d-nexus.de/main.php?g2_v...serialNumber=1

http://img.3d-nexus.de/main.php?g2_v...serialNumber=1

Now I'm still waiting of my P6T. :rolleyes:

Ahhh teasers.

Stop posting pictures!! It's so hard to resist buying one :(

Quote:

---

Originally Posted by Movieman

No, you are right. There is no common sense to the claims involved for the timeframe taken EXCEPT I think it's claiming based on using an entire core or two for the benchmark.
That is the issue and it's leveled out claiming at a rate of app 33,477 per day and being awarded at a rate of app 25,339 per day based on the last 10 days.
This is at 3733mhz on air and almost 24/7 crunching.
A little time taken during these 10 days for the occasional bench but not much time lost.
This is all on the HCC project.
OOPS, forgot..Factor in another TWO FULL pages of WU pending validation...:wasntme:

---

Yes, that looks to be about right. Still impressive. Means a single Bloomfield could do 30k+ under water if pushed far enough.

Which would make a Dual Gainsy a 60K/d monster.. which would make it 25-30% faster than my fastest Harper. Yummie :hrhr:

PS: Good to have you back Dave :up:

Quote:

---

Originally Posted by jcool

Yes, that looks to be about right. Still impressive. Means a single Bloomfield could do 30k+ under water if pushed far enough.

Which would make a Dual Gainsy a 60K/d monster.. which would make it 25-30% faster than my fastest Harper. Yummie :hrhr:

PS: Good to have you back Dave :up:

---

That will be my next and last system for a while.
I figure a dual Gainstown, a Bloomfield, 2 harpers, the Q6600 and sammy split up between my and Mesh's accounts wil be it.
The clovers are going.
The bunch should do close to 170K a day.